Path:
Periodical volume

Full text: Annual Report Issue 2005

BESSY - Annual Report

2005

Berliner Elektronenspeicherring-Gesellschaft für Synchrotronstrahlung m.b.H.

Published by: Berliner Elektronenspeicherring-Gesellschaft für Synchrotronstrahlung m.b.H. (BESSY) Albert-Einstein-Str.15 12489 Berlin Tel. +49 30 6392 2999 Fax +49 30 6392 2990 info@bessy.de www.bessy.de Edited by: Dr. Kai Godehusen

BESSY Inside ‘Physics is not one of the favorite subjects of German school children!’ titled a newspaper recently. This is not a surprise. It probably never was during the last decades – but it became more apparent that physics education in schools and pre-schools is lacking time, interest and enthusiasm. This is a dramatic issue also for a scientific institution such as BESSY, since science needs well educated students. For the realization of our future plans, like our soft Xray Free Electron Laser, we also need clever new scientists ready to eagerly venture into new fields and address entirely new scientific questions. In order to involve the public especially young people more in science and what scientists do, several initiatives have been started in recent years reaching back to the Year of Physics in 2000 and finding another climax celebrating the centennial of Einstein’s ‘magic five papers’ in 1905. During this ‘Einstein Year’, BESSY was involved in manifold activities. We demonstrated the photoelectric effect with hands-on experiments on the exhibition ship MS Einstein. The exhibition ‘Highlights der Physik’ and likewise the events ‘Lange Nacht der Wissenschaften’ and ‘Physik zum Frühstück’ had a focus on Einstein’s work. The television science programme – ‘nano’- visited Adlershof with their ‘nanoCamper’, a group of twelve selected high school students who were introduced to the daily work of scientists. At their day at BESSY they took control of the machine, injecting electrons in the storage ring and kicking them out afterwards. They crystallized proteins and built a 3d protein model from scratch, and they learned about non-destructive material analysis using X-rays. Every event has been very successful and we feel encouraged to keep on introducing children and students of all age into the fascinating world of physics, chemistry and biology. That even ‘good old’ school and textbook science can be exciting, especially when this science is revamped by the use of new experimental evidence has been shown by Uwe Becker’s group from Fritz-Haber-Institute. They downsized the classical double slit experiment onto atomic scale leaving electrons ‘in doubt’ of their state of origin (see Highlights). This is only one of the exciting research results obtained by our steadily increasing user community. There is good news for our current and future users. The construction of the Willy-WienLabor is proceeding with the new Metrology Light Source of the PTB which will start operations in 2007. BESSY itself passed the stringent evaluation of the Leibniz Association with flying colors in the year 2005 and we are very pleased that following the positive evaluation report we received funding to hire more desperately needed beamline scientists to improve our user support. Finally, we would like to thank all our users and staff members who through their efforts and dedication made 2005 yet another successful year. Enjoy browsing through the Annual Report and the Highlights 2005.

3

4

Content

Introduction Scientific Reports
Radiometry Laboratory Reports Report Application Center for Microtechniques Basic Research Reports
Atoms and Molecules Clusters and Nanoparticles Electronic Structure of Solids and Interfaces Chemical Reactions, Catalysis Materials Research Life Sciences Instrumentation, Experimental Techniques Others

5 7
7 36 58
58 86 124 256 291 411 443 489

List of Projects List of Publications Keyword Index Authors Index

520 545 566 575

First measurements at the BAM 7 T wavelength shifter for its use as a primary source standard
R. Klein, G. Brandt, L. Cibik, M. Gerlach, M. Hoffmann, M. Krumrey, G. Ulm Physikalisch-Technische Bundesanstalt, Abbestraße 2 – 12, 10587 Berlin, Germany M. Scheer BESSY GmbH, Albert-Einsteinstraße 15, 12489 Berlin, Germany

For more than 20 years PTB has been using the calculable radiation of bending magnets from the BESSY I and BESSY II electron storage rings in the visible, UV, VUV and X-ray spectral range for radiometry [1], especially for the calibration of radiation sources [2] and energy-dispersive detectors [3]. The calculation of the spectral photon flux φ for a given photon energy E requires the measurement of all parameters that enter the Schwinger equation [4] with high accuracy [5]. These parameters are the electron beam current I, the electron energy W, the magnetic induction at the radiation source point B, the vertical electron beam size and divergence and the geometric parameters defining the angular acceptance. Nevertheless, the usable spectral range is limited on the high energy side to about the 10 to 20 fold of the characteristic energy. This is on the one hand due to the low photon flux because of the exponentially decreasing flux at these energies and on the other hand due to the increasing relative uncertainty in the calculation of the spectral photon flux. For photon energies E that are large compared to the characteristic energy Ec the uncertainty in the calculation of φ is dominated by the relative uncertainty in the measurement of B and W, which are in the order of 10-4. For a small angular acceptance in the forward direction of the synchrotron radiation emission the relative uncertainty ∆φ/φ scales roughly as ∆φ/φ ≈ (E/Ec -1) · ∆B/B and ∆φ/φ ≈ 2 · E/Ec · ∆W/W, (1) respectively. To extend the radiometry based on calculable synchrotron radiation sources to higher photon energies, a source with a higher characteristic photon energy as compared to a bending magnet is needed. Already at BESSY I, PTB had used a 6 T superconducting wavelength shifter (WLS) [6] for that purpose. At BESSY II, PTB has access to the BAM 7 T WLS, which has a more than five-fold higher characteristic energy as the BESSY II bending magnets. The WLS was procured according to the PTB and BAM specifications and is operated by the BESSY GmbH. Unfortunately, the magnetic field map of a WLS is more complicated than in the case of a bending magnet. The bending magnet has a large area of homogeneous magnetic field, whereas a wavelength shifter shows a magnetic field map as illustrated in fig. 1 with large field gradients in the direction of the orbit. So the determination of the magnetic induction at the radiation source point is difficult, not only because the field map must be determined, but also because it strongly depends on the exact location of the source point in the device. The magnetic field in the center of the WLS along the orbit s can be approximated by B(s) = B0 · cos(2πs/λs). (2)

5

For small variations of the observation angle in the horizontal plane by ∆αh, the shift ∆s of the source point along the orbit is ∆s ≈ R · ∆αh, (3) with R being the bending radius, which is approximately 1 m for the 7 T WLS. At the specification of the WLS emphasis was put on a rather flat top of the main pole field (λs = 0.45 m), so that for a source point located at s0 = 0, the relative change of the magnetic induction ∆B/B will stay below 1 ·10-4 for |∆s| < 1 mm. The field of the wavelength shifter was characterized before installation and with NMR probes which are mounted in the joke, the exact value of the maximum of the main pole field B0 can be monitored. Unfortunately, these NMR probes are not working right now for what ever reason, so a precise determination of B0 is not possible at the moment. Moreover, as can be seen in fig. 1, the upstream steering magnet was moved closer to the WLS by BESSY as was originally designed in order to gain space for the installation of a Landau cavity. This had the side effect, that the trajectory in the WLS is not symmetric with respect to the middle of the WLS anymore. Calculations of the trajectory for this setting predict, that the source point is shifted to s0 ≈ - 4 mm [7], i.e. upstream away from the flat top into a region with larger field gradients. This has little influence on the performance of the WLS monochromator beamline, but has a major disadvantages for the radiometric use of the undispersed radiation of the device: Even if the NMR probes would work properly,

Fig. 1: Top: magnetic induction of the WLS along the electron orbit. In black color the standard setting of the 7T WLS with its asymmetrically placed steering magnets is shown. In red color, a special setting during a PTB shift is shown. The steering magnets were switched off, resulting in a symmetric field map. At this setting, the maximum field could only be set to 6 T in order to avoid to large deviations of the electron trajectory in the horizontal plane; Bottom: the electron trajectory in the WLS for the different field settings.

6

the magnetic induction could not be precisely determined anymore because it is more sensitive to the exact location of s0, which can not be determined precise enough.To experimentally detect a possible shift of the source point from the center position, the undispersed radiation of the WLS was measured by a HPGe-detector at different horizontal positions within the beamline acceptance during a special PTB shift with reduced electron beam current. The slightly different observation angles ∆αh mean slight shifts of the radiation source point along the orbit and therefore slightly different magnetic inductions at the source points (see eq. 3). The result will be a variation in the photon flux according to (1), which is largest for high values of E/EC. In fig. 2 the relative change of the photon flux in the photon energy interval from 90 keV to 140 keV is shown. A clear indication for a position depended variation of the flux is seen. The observed relative change in the photon flux is in accordance with the calculated shift of the radiation source point of -4 mm.

Fig. 2: Relative count rate for photon energies summed from 90 keV to 140 keV, measured with a HPGe-detector with a 4 mm Cu-filter. The detectors aperture of 1 mm diameter was placed approx. 36 m from the radiation source point. The error bars are given by the counting statistics and by the possible variation of the angular acceptance of the aperture during the horizontal movement. The electron beam current was 108 nA.

In another PTB shift, the asymmetrically placed WLS steering magnets were switched off in order to obtain a symmetric field map. In this setting the WLS main pole field could only be set to 6 T maximum magnetic induction due to the large horizontal deviation of the electron beam trajectory. With this setting the trajectory in the WLS (fig. 1 bottom, red curve) is symmetric again and the source point is expected to be at the maximum of the main pole field, thus showing a much smaller field gradient for a variation of the horizontal observation angle. Fig. 3 shows a similar measurement as described above. There no significant inhomogenity is observed. With this symmetric setting, the WLS behaves as expected and - given, the NMR probes are repaired - the use of the WLS as a calculable source of radiation in the hard X-ray spectral range can be further pursued.

7

Fig. 3: Relative count rate for photon energies summed from 90 keV to 140 keV, measured with a HPGe-detector with a 4 mm Cu-filter for the special operation mode of the WLS. The detectors aperture of 3 mm diameter was placed approx. 36 m from the radiation source point. The error bars are given by the counting statistics and by the possible variation of the angular acceptance of the aperture during the horizontal movement. The electron beam current was 5.1 nA.

[1] [2] [3] [4] [5] [6] [7]

G. Ulm, Metrologia 40 (2003) S101 – S106 M. Richter et al., Nucl. Instrum. Methods A467/468 (2001) 605 - 608 F. Scholze et al., Metrologia 38 (2001) 391 - 395 J. Schwinger, Phys. Rev. 75 (1949) 1912 R. Klein et al., Proc. of EPAC 2004 (2004) 273 – 275 R. Thornagel et al., Rev. Sci. Instrum. 67 (1996) 653 – 657 M. Scheer, PhD thesis, BESSY 2006, in preparation

8

Comparison of spectral responsivity scale realizations at 157 nm
A. Gottwald, S. Kück1, F. Brandt1, M. Richter
1

Physikalisch-Technische Bundesanstalt, Abbestr. 2-12, 10587 Berlin Physikalisch-Technische Bundesanstalt, Bundesallee 100, 38116 Braunschweig

Excimer lasers are nowadays widely used for industrial applications such as photolithography of semiconductor devices. For production control, standardization and safety regulations during testing and production, calibrated radiometers with uncertainties of a few percent in the measurement of pulse energies are needed. At the Physikalisch-Technische Bundesanstalt (PTB), high precision measurement and calibration services are provided for the main excimer laser wavelengths in the UV and DUV spectral regions. Recently, at PTB’s laser radiometry facility in the Optics Division a vacuum compatible setup and a calibration chain for the measurement of the average power for F2-laser radiation at 157 nm was established (Fig. 1) [1]. On the other hand, to maintain a scale of spectral responsivity based on semiconductor photodiodes at wavelengths below 400 nm, PTB uses monochromatized synchrotron radiation of the electron storage ring BESSY II in combination with cryogenic electrical substitution radiometers as primary detector standards [2, 3]. This allows determination of radiant power with relative uncertainties well below 1 % also at 157 nm, however, limited to a few µW cw power as available at the Normal-IncidenceMonochromator (NIM) beamline for detector calibration [4]. To extend the calibration facilities, a F2-laser system was installed, with the possibility of being alternatively coupled into the beamline behind the NIM. This allows the direct measurement of the average laser power emitted by the pulsed excimer laser using the primary detector standard SYRES II, and for the calibration of suitable transfer detector devices, since the SYRES II radiometer had been shown to be insensitive to the radiation pulse structure [5].
LM7
10 10 10 10 10 10 10
1

LM8

LM4

Figure 1.

0

-1

-2

-3

PE-10
-4

CR-II Trap
400 600 800 1000

CR-I
-5

100

200

Wavelength (nm)

Schematic calibration chains: CR-I: SYRES II cryogenic radiometer at the PTB laboratory at BESSY II. CR-II: cryogenic radiometer at the PTB laser radiometry facility; LM4, LM7, LM8: Standard detectors at the PTB laser radiometry facility. Trap: Si-trap detector at the PTB laser radiometry facility. PE-10: pyroelectric detector as transfer standard in this comparison

The two realizations of the spectral responsivity scale at 157 nm have been compared – worldwide for the first time – using an OPHIR PE-10 pyroelectric detector, which is able to span over the gap in intensities between laser-based and synchrotron radiation-based instrumentation (for details, see [6]). At the two facilities, this detector was calibrated regarding its pulse energy responsivity, i. e. the detector output voltage per laser pulse energy.

Average Power (W)

9

7100

6900 6700 6500 0.010

Figure 2. Results of the calibration run for the PE-10 detector in vacuum at PTB laser radiometry facility: Responsivity (200 µJrange, not corrected for the switching factor) as a function of the applied pulse energy. The diamonds correspond to single measurements, the large circle corresponds to the mean value.
0.015 0.020 0.025 0.030 pulse energy / mJ

sLM8,meas / VJ

-1

sSYRES,meas. / V mJ

70 65 60 55 50 0 2 4 6 pulse energy / µJ 8 10

Figure 3. Results of the PE-10 powermeter calibration with the SYRES II primary standard in the PTB laboratory at BESSY II. The responsivity (20 µJ-range, without correction for plasma contribution) is shown for different pulse energies (diamonds); the large circle indicates the mean value.

At the PTB laser radiometry facility, a total set of 16 measurements at 157 nm in vacuum atmosphere (about 1 × 10-6 mbar) were performed (Fig. 2). No significant dependence of the responsivity on the applied average power, repetition rate, or pulse energy was observed. Thus, we determined the responsivity of the OPHIR PE-10 at 157 nm in the 200-µJ measurement range. The responsivity in the 20-µJ-range is obtained by taking into account the switching factor between the two ranges. Finally, we obtain for the responsivity of PE-10 at 157 nm in vacuum atmosphere in the 20-µJ range sLM8 = (66700 ± 1090) V/J (standard uncertainty, k = 1). With the SYRES II radiometer at BESSY II, a total of eight measurements at different pulse energies were performed in the lowest pulse energy range of the PE-10, i.e. 20 µJ (Fig. 2). For all measurements, a contribution of (29 + 2) % non-VUV plasma glow in the radiant power was determined by using an UV-blocking glass filter, which has to be taken into account for a correction of the measured values. This relatively high contribution is specific for the laser system used at BESSY II. The average responsivity (corrected for the non-157 nm contributions) was determined from all measurements to be sSYRES = (69890 ± 5510) V/J (standard uncertainty, k = 1). In conclusion, the agreement between the responsivities determined is about 4.8 % and, therefore, within the combined standard uncertainties. Hence, this result validates both physical scale realizations. The uncertainty at the PTB laser radiometry group arises mainly from the measurement itself, caused by the requirement to measure at low average powers. The uncertainty at the PTB laboratory at BESSY II is mainly caused by the determination of the fraction of non-VUV radiation from the laser source.
[1] [2] [3] [4] [5] [6] S. Kück, F. Brandt, M. Taddeo, Appl. Opt. 44, 2258 – 2265 (2005). A. Gottwald, U. Kroth, M. Krumrey, M. Richter, F. Scholze, G. Ulm, accepted for Metrologia (2006), H. Rabus, V. Persch, G. Ulm, Appl. Opt. 36, 5421 – 5440 (1997). M. Richter, J. Hollandt, U. Kroth, W. Paustian, R. Thornagel, G. Ulm, Nucl. Instr. and Meth. A 467-468, 605 – 608 (2001). M. Richter, U. Kroth, A. Gottwald. Ch. Gerth, K. Tiedtke, T. Saito, I. Tassy, K. Vogler, Appl. Opt. 41, 7167 – 7172 (2002). S. Kück, F. Brandt, H.-A. Kremling, A. Gottwald, A. Hoehl, M. Richter, accepted for Appl. Opt. (2006)

-1

10

Polarization dependence of multilayer reflectance in the EUV spectral range
Frank Scholze, Christian Laubis, Christian Buchholz, Andreas Fischer, Sven Plöger, Frank Scholz, and Gerhard Ulm Physikalisch-Technische Bundesanstalt, Abbestraße 2-12, 10587 Berlin, Germany The Physikalisch-Technische Bundesanstalt (PTB) with its laboratory at the electron storage ring BESSY II1 is the European centre of EUV radiometry and supports the national and European industry by carrying out high-accuracy at-wavelength measurements in the EUV spectral region, particularly to support the development of Extreme Ultraviolet Lithography (EUVL), which holds the key to the next generation of computer technology. EUVL imaging requires multilayer optics which have been manufactured with extreme accuracy, in order to ensure a uniform illumination of the wafer plane and an optimized throughput. PTB’s EUV reflectometer allows the measurement of mirrors having diameters up to 550 mm, a height of 230 mm, and a weight of up to 50 kg2. In response to the increasing demands of the industry,, substantial improvements in the total measurement uncertainty and repeatability at PTB have been demonstrated3,4. The measurements at PTB are carried out using highly polarized synchrotron radiation5; whereas EUVL uses EUV pulsed plasma sources emitting unpolarized radiation. For a full understanding and specification of the optics it is therefore essential to know the polarization properties of the optical components. We will demonstrate here that taking advantage of all possible movements of detector and sample in the EUV reflectometer enables sound measurements of polarization properties up to angles of incidence of 20° to the normal. The polarization vector of the incident radiation is parallel to the X-axis. This results in pure S-polarization in our normal mode of operation, i.e. using the Θ and 2Θ axis for the setting of the incidence and exit angle of the radiation. There is, however, also the possibility of using the X-movement of the detector and the tilt-motion of the sample to obtain a P-polarization geometry. This is shown schematically in Fig. 1. As the tilt-axis was only designed to compensate for the curvature of mirrors, its range is restricted to ±10°. If a wedge is used to offset the angle of the sample, it allows measurements of angles of incidence between 0° and 20°. The detector manipulator allows the distance of the detector to the Θ-axis to be set to 150 mm and the detector to be moved 125 mm horizontally out of the plane. This corresponds to an angle of 39.8° with respect to the incoming beam and thus matches very well the possible range of the tilt-movement. A further option is to use both axis of rotation and set, e.g. equal angles for Θ and tilt, resulting in a plane of deflection at an orientation of 45° to the electrical field vector. This would be equivalent to a measurement with unpolarized radiation. The radiation is measured with normal incidence at the detector for the incoming direct beam and also for the reflected beam if we use Θ and 2Θ to position sample and detector in the Spolarization geometry2. For the P-polarization geometry shown in Fig. 1, however, the reflected radiation is incident at non-normal direction on the detector diode. Therefore, the signal for the reflected beam must be corrected for the influence of different angles of incidence. This correction is determined experimentally. The diodes used in our measurements are Hamamatsu G1127 GaAsP/Au Schottky diodes. Here, the absorbing front layers are the top Au layer and a thin layer of GaAsP to account for recombination losses at

11

the interface6. To interpolate the measurements for different angles of incidence, we assumed that the change in responsivity is only due to the change in the top-layer transmittance: 2 −1 s (Θ) (1) = ∏ exp µ i d i 1 − cos sin −1 [sin (Θ / n i )] , s(0°) i =1

( [

{

}

])

where s is the responsivity, Θ the angle of incidence (in vacuum), µ the absorption coefficient, d the thickness, and n the refractive index of the respective layer; i=1,2 stands for Au and GaAsP with d(gold)=5.7 nm and d(GaAsP)=3 nm. With this equation an excellent fit of the measured dependence is achieved for the range of incidence angles used.
Fig. 1 Scheme of the P-polarization geometry in the EUV-reflectometer. Using a wedge of 10°, the tilt range of ±10° is transferred to 0° to 20° for the angle of incidence. The direction of the E-vector for the incoming radiation is indicated. The detector movement is linear in this direction.

Using the method as described above, we measured the reflectance as a function of the wavelength for a Mo/Si multilayer mirror with 60 bi-layers with diffusion barriers7. We performed measurements for angles of incidence ranging from 2° to the normal up to 20° with S-, P- and a 45° polarization orientation.
Fig. 2 Reflectance measured at an angle of incidence of 17.5° for S-polarized radiation (▲), P-polarized radiation (●) and with a 45° polarization angle (♦), i.e. equivalent to unpolarized radiation.

Fig. 2 shows results of measurements at an angle of incidence of 17.5° for all 3 polarization orientations. The measurement uncertainties are 0.1 % for absolute reflectance and 2 pm in wavelength. The results clearly show the behaviour which had been expected: the reflectance is lower for P- than for S-polarization and also the bandwidth of the reflectance curve decreases. The measurement at 45° polarization as a representation for unpolarized radiation falls in-between both curves. A comprehensive presentation of the trends of peak wavelength and reflectance is given in Fig. 3. Here, the peak reflectance is plotted as a function of the peak wavelength. By doing so, the cos-dependence of the peak wavelength on the angle of incidence disappears and the curves are linear within our experimental uncertainty of 0.1 %

12

for the reflectance which is approximated by the height of the plot symbols. The linear dependence allows a reliable interpolation of the measurements at 45° polarization and thus a more detailed comparison of the average values for S- and P-polarization. It turns out that the measured values are slightly but significantly lower than the average value. It must, however, be noted, that the peak reflectance values of each scan were averaged. These appear at slightly different wavelengths for S- and P-polarization, as can be seen in Fig. 2. Averaging the values for each wavelength in the scan therefore results in a slightly lower peak reflectance. This is basically what was measured for the 45° orientation.
Fig. 3 Peak reflectance versus peak wavelength for S-polarized radiation (▲), P-polarized radiation (●) and with a 45° polarization angle (♦). The average of the S- und Ppolarization (○) is shown for comparison. The lines are linear fits of the respective data.

All in all, taking advantage of all possible movements of detector and sample in the EUV reflectometer, we demonstrated the ability to measure detailed polarization properties of EUV optics up to angles of incidence of 20° to the normal. By moving the detector out of plane and tilting the sample we can model any degree of linear polarization by properly balancing the tilt and Θ rotation angles. The results shown here reveal significant differences between reflections in S- and P-polarization geometry already for rather small angles of incidence, such as 5°. References
1 G. Ulm, B. Beckhoff, R. Klein, M. Krumrey, H. Rabus, and R. Thornagel, "The PTB radiometry laboratory at the BESSY II electron storage ring," Proc. SPIE 3444, 610-621 (1998) 2 J. Tümmler, G. Brandt, J. Eden, H. Scherr, F. Scholze, G. Ulm, "Characterization of the PTB EUV reflectometry facility for large EUVL optical components," Proc. SPIE 5037, 265-273 (2003) 3 F. Scholze, C. Laubis, C. Buchholz, A. Fischer, S. Plöger, F. Scholz, H. Wagner, and G. Ulm, "Status of EUV Reflectometry at PTB," Proc. SPIE 5751, 749-758 (2005) 4 F. Scholze, J. Tümmler, G. Ulm, "High-accuracy radiometry in the EUV range at the PTB soft X-ray radiometry beamline," Metrologia 40, S224-S228 (2003) 5 F. Scholze, B. Beckhoff, G. Brandt, R. Fliegauf, R. Klein, B. Meyer, D. Rost, D. Schmitz, M. Veldkamp, J. Weser, G. Ulm, E. Louis, A. Yakshin, P. Goerts, S. Oestreich, F. Bijkerk, "The new PTB-beamlines for high-accuracy EUV reflectometry at BESSY II," Proc. SPIE 4146, 72–82 (2000) 6 F. Scholze, H. Henneken, P. Kuschnerus, H. Rabus, M. Richter, and G. Ulm, "Determination of the electron-hole pair creation energy for semiconductors from the spectral responsivity of photodiodes," Nucl. Instrum. Meth A439, 208 - 215 (2000) 7 S. Braun, H. Mai, M. Moss, R. Scholz, and A. Leson, "Mo/Si Multilayers with Different Barrier Layers for Applications as Extreme Ultraviolet Mirrors," Jpn. J. Appl. Phys. 41, 4074–4081 (2002)

13

Planned infrared beamlines at the Metrology Light Source of PTB
Ralph Müller1, Arne Hoehl1, Roman Klein1, Gerhard Ulm1, Ulrich Schade2, Karsten Holldack2, Godehard Wüstefeld2
2

Physikalisch-Technische Bundesanstalt, Abbestraße 2-12, 10587 Berlin, Germany Berliner Elektronenspeicherring-Gesellschaft für Synchrotronstrahlung m.b.H., AlbertEinstein-Str. 15, 12489 Berlin, Germany

1

Electron storage rings have proven to be brilliant radiation sources in the VUV, the soft and the hard X-ray regions of the spectrum. In recent years, their strength as a unique source of IR radiation has also become apparent and is increasingly being exploited at synchrotron radiation facilities around the world. A rapidly growing number of IR beamlines at several synchrotron storage rings have been realized taking advantage of these unique IR light sources [1-3]. In particular, IR radiation in the mid infrared wavelength region is increasingly used in research by means of Fourier transform spectroscopy on biological tissues down to single cells, high-pressure and micro-sample measurements and in investigations on surfaces and thin films applying IR ellipsometry with a high lateral resolution. Applications of IR synchrotron radiation are multidisciplinary, and the number continues to increase as IR beamlines become accessible to an increasing number of users. Here we present concepts for obtaining infrared synchrotron radiation at the Metrology Light Source (MLS), the new low-energy electron storage ring of the Physikalisch-Technische Bundesanstalt (PTB). The MLS is designed in close cooperation with BESSY and will be located adjacent to the BESSY II facility in Berlin [4]. Its electron energy can be adjusted between 200 MeV and 600 MeV, maximum stored current will be 200 mA. The building which is going to accommodate the MLS is presently almost completed. First electrons will be stored in 2007, user operation is scheduled to begin in 2008. PTB will use the MLS as a dedicated facility for radiometry and photon metrology in the UV and VUV range, especially also as a calculable radiation source, a primary source standard in the VIS, UV and VUV. Furthermore, synchrotron radiation in the IR range will be used. At the MLS three beamlines dedicated to the use of IR synchrotron radiation are planned [5] (see Fig. 1): (1) The long period undulator U180 provides radiation with high flux in the MIR spectral range (up to 20 µm), (2) a special THz beamline optimized for the FIR/THz spectral range, and (3) an IR beamline optimized for the MIR to FIR.

IR

UV VUV

EUV

D6 D5

MLS
U180
Undulator beamlines

Figure 1: Planned beamlines at the MLS dedicated to the use in the infrared spectral range: (1) Undulator IR, (2) THz beamline, and (3) IR beamline.

14

The front ends of all IR beamlines are on the roof of the tunnel of the MLS, so there is more space for other beamlines dedicated to the use in the UV, extreme UV and VUV spectral range left in the experimental hall. In the following we will describe the optical design concept of the IR beamline. The scheme of the optical design concept of the IR beamline is shown in Fig. 2. The aim is to develop an IR experimental station at the MLS for the near, mid and far IR wavelength region and to make it accessible to the PTB and to an interested scientific community for a broad field of applications.

Figure 2: Optical design of the IR beamline (number 3 in Fig. 1) at the MLS.

The IR beamline will be located at the bending magnet D6 at the MLS (Fig. 1). The plane extraction mirror allows - in combination with a special port of the dipole chamber - a horizontal and vertical collecting angle of 64 mrad (h) × 43 mrad (v). At the maximal ring operation conditions of 600 MeV and 200 mA a mirror heat load of about 16 W is expected. Due to this relatively low heat load and with an additional cooling system we expect to have no deformation or destruction effects at the first mirror. So there is no need to split the mirror as, for example, at the IRIS beamline at BESSY II [3]. The first optical component M1 will be placed at a distance of 1550 mm from the source, the first position possible outside the vacuum chamber of the dipole magnet. M1 deflects the photon beam upwards by 90° to a combination of mirrors which focus the beam outside the radiation shielding wall near the plane of a CVD diamond window. The second mirror M2 and the third mirror M3 focus the IR radiation vertically and horizontally, respectively. Both mirrors are cylindrical and deflect the beam by 90° towards the storage ring (M2) and upwards (M3). The beam passes the tunnel roof at a distance of 700 mm after M3. The fourth optical element M4 is a planar mirror and transports the beam to the parabolical mirror M5. M5 collimates the beam and sends it to the remaining optical system. The IR photon beam has an intermediate focus between M4 and M5 at a distance of 4900 mm from the center of M1. After all these reflections the polarization is horizontally oriented. Ray tracing calculations for an energy of 500 cm-1 indicate a focal spot size of 2.5 mm (h) × 1.0 mm (v) which should easily pass through the diamond window (30 mm clear aperture). This focus serves as a new source point for the remaining optical system. The diamond window separates the UHV of the storage ring from the remainder of the beamline. The subsequent optical elements should direct the light to the different experiments. By mounting the optics and experiments on the massive storage ring tunnel itself the mechanical stability required for vibration sensitive IR experiments can be achieved. Fig. 3 shows the expected brilliance at the MLS compared with other IR beamlines. In an optimal beamline design this brilliance is transferred to the experiment with minimal losses. The brilliance is calculated here using the far-field approach taking into account the intrinsic beam size in the center of the bending magnet as well as the source size due to projection and diffraction. The kink in the brilliance curves together with a steeper slope at lower wavelength results when the vertical collecting angle falls short of the natural vertical opening angle of the radiation. The 2 cm-1 bandwidth provides suitable resolution in typical spectra of condensed matter.

15

brilliance / W/(mm sr 2 cm BW)

10 10 10 1x10 1x10 10 10 10 10

-1 -2 -3 -4 -5 -6 -7 -8 -9

2

-1

ALS BESSY II MLS NSLS
1 10 100 1000
-1

10000

wavenumber / cm

Figure 3: Calculated brilliance of the MLS IR beamline compared to the IR beamlines IRIS at BESSY II [3], U4IR at the NSLS [1], and the IR beamline 1.4 at the ALS [2]. The brilliance is quoted per 1 A ring current.

Intense coherent synchrotron radiation (CSR) can be generated if electron bunches are compressed to 1 mm rms length at the BESSY II storage ring [6]. This option is also envisaged for the MLS by tuning the quadrupoles in an appropriate way. The emitted spectral range of the bunches compressed to 1 mm will be the same as for BESSY II. Recent calculations have shown that the bunch length at the MLS operated in the low-alpha mode is smallest in the straight sections and largest in the section between the bending magnets. Therefore, a higher intensity and broader spectrum of the CSR is expected at the bending magnet ports that are located most closely to a straight section, as is the case for the dedicated THz beamline at bending magnet D5 (Fig. 1) [5]. The optical design is comparable with the new THz beamline at BESSY II [7]. The plane extraction mirror will be placed at about 1550 mm from the source, and deflects the beam upwards to a combination of mirrors which focus the beam outside the radiation shielding wall to the experiment located at the roof of the tunnel. A quartz window after the first mirror separates the UHV of the storage ring from the rest of the beamline. The production of stable, high power CSR opens a new region in the electromagnetic spectrum which can be applied for imaging, spectroscopic and microscopic methods in solid state physics, biology, chemistry and medicine. Additionally, we plan the use of CSR and incoherent radiation for the characterization and calibration of THz detectors and optical components. Construction of the IR, the THz, and the undulator IR beamline should be finished by the end of 2006. Commissioning of the MLS and of the beamlines will start in 2007. Special emphasis has to be placed on the commissioning of the low-alpha mode necessary for the use of the THz beamline. The MLS complements the measurement potential available for the PTB at BESSY II [8] in the lower photon energy range and thus enables PTB to use synchrotron radiation from the THz up to the hard X-ray region for high accuracy photon metrology.

[1] [2] [3] [4] [5] [6] [7] [8]

W.R. McKinney et al., LBNL, LSBL-414 (1997) G.L. Carr et al., Il Nuevo Cimento D 20, 375 (1998) U. Schade et al., Rev. Sci. Instrum. 73, 1568 (2001) R. Klein, G. Ulm, M. Abo-Bakr, P. Budz, K. Bürkmann-Gehrlein, D. Krämer, J. Rahn, G. Wüstefeld, Proceedings of EPAC 2004, Lucerne, Switzerland, 2290 (2004) R. Müller, A. Hoehl, R. Klein, U. Schade, G. Ulm, U. Schade, K. Holldack, G. Wüstefeld, Infrared Physics & Technology (2006), in press M. Abo-Bakr, J. Feikes, K. Holldack, G. Wüstefeld, H.-W. Hübers, Phys. Rev. Lett. 88, 254801 (2002) K. Holldack, S. Khan, R. Mitzner, T. Quast, G. Wüstefeld, Proc. of the EPAC 2004, Lucerne, Switzerland, 2284 (2004) G. Ulm, Metrologia 40, S101 (2003)

16

CD characterization of EUV masks by EUV scatterometry
Frank Scholze, Bernd Bodermann, Christian Laubis, Gerhard Ulm, Matthias Wurm Physikalisch-Technische Bundesanstalt, Braunschweig and Berlin, Germany Uwe Dersch and Christian Holfeld AMTC GmbH&CoKG, Dresden, Germany Extreme Ultraviolet Lithography (EUVL) holds the key to the next generation of computer technology. When exposing an EUV mask with EUV radiation of 13.5 nm, the radiation is reflected by a Mo/Si multilayer stack which is about 300 nm thick. In comparison to visible and UV light, where the radiation is reflected at the top interface of the mask, for EUV all layers in the stack contribute to the reflection. Therefore, only EUV radiation provides direct information on the mask performance relevant for an EUVL tool. Scatterometry, the analysis of light diffracted from a periodic structure, is a versatile metrology for characterizing periodic structures, regarding critical dimension (CD) and other profile properties. With respect to the small feature dimensions on EUV masks, the short wavelength of EUV is also advantageous since it minimizes diffraction phenomena. The Physikalisch-Technische Bundesanstalt (PTB) with its laboratory at the electron storage ring BESSY II1 is the European centre of EUV metrology and supports the national and European industry by carrying out high-accuracy at-wavelength measurements in the EUV spectral region. PTB’s EUV reflectometer allows the measurement of mirrors having diameters up to 550 mm, a height of 230 mm, and a weight of up to 50 kg2. Iz allows mask surface scanning in Cartesian coordinates at 10 µm positioning reproducibility3,4. The probed area (photon beam size) is about 1 mm². The detector can be moved in 3 dimensions (rotation around the sample, radial movement to adjust the distance to the sample and a linear movement out of the reflection plane). In this contribution we present measurements on prototype EUV masks.
Fig. 1

Sketch of the experimental setup. The illumination angle of the incident synchrotron beam is 6° with respect to the surface normal. The diffraction angle is varied by moving the detector. The investigated line pattern is oriented perpendicular to the scattering plane.

The scheme of a scatterometry measurement is shown in Fig. 1. For homogeneous regions and with equal angles of incidence and exit, it is a reflectometry measurement. Already such

17

measurements at open multilayer (ML) or absorber areas can be used to obtain detailed information on layer thicknesses (see Fig. 2 and 3). By comparison of the reflectance in ML and absorber areas, information can be obtained on the absorber thickness and the thinning of the capping layer due to oxidation in the open ML regions after buffer etch.
Fig. 2

Spectral reflectance measured in a bright field region (○, red, left scale) and at the absorber layer (○, blue, right scale). The respective lines are model calculations using IMD5.The parameters are given in Fig. 3.

Fig. 3

Scheme of the EUV mask coating layers. The substrate is coated with a Mo/Si multilayer stack with a Si-capping layer. The absorber stack consists of a SiO2 buffer layer and a Ta-based absorber.

Results of a scatterometry measurement are shown in Fig. 4 with the corresponding scheme of the geometry in Fig. 5. A structure of lines&spaces was illuminated at an angle of 6° to the normal and perpendicular to the orientation of the lines, see Fig. 1. The pitch of the structure was 840 nm and the designed duty cycle 5:1 (dark:bright). It was already known from the process development that the bright areas are smaller than designed. Fig. 4 shows the measured diffraction orders at the angular positions corresponding to the pitch of the structure and the wavelength. It should be noted that this correlation is valid for any geometry. The position of the diffraction orders is only defined by the lateral period (pitch) of the structure. For planar structures, the Fraunhofer-approximation can be used to calculate the so-called form factor. It depends for a given pitch only on the width of the lines. The structures at EUVmasks, however, are 3-dimensional. Due to the oblique angle of incidence, the bright MLregions are partially shadowed by the absorber lines. In a first approximation, the dark lines are broadened by this shadowing and the dark:bright ratio changes. Although this is a rather crude approximation, it works quite well as illustrated in Fig. 4. Here we measured the diffracted intensity for positive as well as negative diffraction orders. The form factor envelope as given by the Fraunhofer approximation can only be fitted to match either positive or negative orders. The difference in effective line width of 13 nm obtained using the Fraunhofer approximation is indeed very close to the direct geometrical shadowing effect of 14 nm, as shown in Fig. 5. It is therefore obvious that information on the absorber line height

18

and – using more sophisticated electromagnetic field based calculations – also on the line profile can be obtained from scatterometric measurements.
Fig. 4

Diffraction measured at lines&spaces with 840 nm pitch. The 0th order is suppressed by top-to-bottom interference effects. The envelopes shown are calculated using the Fraunhofer approximation for an open width of 92 nm (red) and 105 nm (green). For geometry see Fig. 5.

Fig. 5

Scheme of the EUV mask lines&spaces. Due to the finite height of the absorber stack and the oblique angle of incidence, the bright area is partly shadowed. The geometrical difference in apparent width for scattering anlges of ±6° is 14 nm.

The investigations shown here are part of the “ABBILD” project, supported by the Bundesministerum für Bildung und Forschung. References
1 G. Ulm, B. Beckhoff, R. Klein, M. Krumrey, H. Rabus, and R. Thornagel, "The PTB radiometry laboratory at the BESSY II electron storage ring," Proc. SPIE 3444, 610-621 (1998) 2 J. Tümmler, G. Brandt, J. Eden, H. Scherr, F. Scholze, G. Ulm, "Characterization of the PTB EUV reflectometry facility for large EUVL optical components," Proc. SPIE 5037, 265-273 (2003) 3 F. Scholze, C. Laubis, C. Buchholz, A. Fischer, S. Plöger, F. Scholz, H. Wagner, and G. Ulm, "Status of EUV Reflectometry at PTB," Proc. SPIE 5751, 749-758 (2005) 4 F. Scholze, J. Tümmler, G. Ulm, "High-accuracy radiometry in the EUV range at the PTB soft X-ray radiometry beamline," Metrologia 40, S224-S228 (2003) 5 D. Windt, "IMD—Software for modeling the optical properties of multilayer films," Computers in Physics 12, 360-370 (1998)

19

Alignment of large off-axis EUV mirrors in the EUV reflectometer of PTB
Christian Laubis, Christian Buchholz, Andreas Fischer, Sven Plöger, Frank Scholz, Heike Wagner, Frank Scholze, and Gerhard Ulm Physikalisch-Technische Bundesanstalt, Abbestraße 2-12, 10587 Berlin, Germany Extreme Ultraviolet Lithography (EUVL) is the key to the next generation of computer technology. EUVL imaging requires multilayer mirror optics being manufactured with extreme accuracy to ensure uniform illumination at the wafer plane and optimized throughput1. The wavelength for the peak mirror reflectivity should be matched to better than 2.5 pm and the uniformity of reflectance should be better than 0.08 %2. In response to this demand, substantial improvements in the total measurement uncertainty and repeatability at the Physikalisch-Technische Bundesanstalt (PTB) have been achieved and reported3,4. The EUV reflectometer allows the measurement of large mirrors with a diameter as large as 550 mm, a height of 230 mm, and a weight of up to 50 kg5. We present here work on the alignment of aspheric off-axis mirrors which is crucial to achieve accurate measurement results for wavelength and reflectivity needed for the validation of the coating development for EUV optics6,7. The SX700 soft X-ray bending magnet beamline was designed first of all for a parallel beam with a reasonably low spot size. To use this high quality beam for accurate measurements3,8, the mirror under test has to be properly aligned. The centre wavelength λ is related to the double-layer thickness d of the coating by the Bragg equation: (1) λ = 2 n λ d cosΘ ML = λ 0 cosΘ ML , with nλ the (average) index of refraction at wavelength λ, and ΘML the propagation angle in the multilayer coating. Assuming constant nλ and ΘML = Θ we get a cos-dependence for approximation. To keep the shift in measured wavelength below 2 pm, the angular scales have to be adjusted to better than 0.3° for measurements close to normal incidence (Θ=1.5°) and better than 0.02° for Θ=20°. Thus, for higher angles of incidence (AOI), the alignment requires accurate scales in Θ and 2Θ (see Fig. 3) and the surface of the sample to coincide with the main axis of the goniometer within a few tenth of a millimeter.
Fig. 1 Determination of zero reflection angle: A 180° rotation around a pivot point (indicated by the black dot) enables the measurement of direct beam (orientation with bright colors) and reflected beam (transparent). The pivot is designed to be in the center of the surface of the detector, and the offset is measured and corrected.

Our method of measurement involves turning the detector (Ψ-axis) between direct-beam and reflected-beam measurements5. Particularly for the alignment, this approach has strong advantages. We rotate the detector (Ψ-axis) to look into the direct beam at 2Θ=0°, see Fig. 1, and define the zero angle to be at the position of the incoming beam. A possible offset of the detector switching axis is accounted for.

20

Fig. 2 Scheme of the alignment for the reflection angle. The detector is aligned in the direct beam to define “zero” angle of incidence, left. Due to the symmetry of the reflection geometry, any offset of the beam with respect to the axis of rotation (solid circle in the centre) causes only a fixed offset in detector angle, right.

After defining zero for the detector axis, we define the angle of incidence for the mirror, typically at 1.5° off normal. The detector is moved to 3° and the angle of the sample is scanned to position the beam in the centre of the detector. The main advantage of this approach is illustrated in Fig. 2. Even if the beam is not incident at the correct height, i.e. it does not go through the axis of rotation, the alignment between detector and sample rotation is not disturbed due to the symmetry of the reflection. The angular offset for the reflected beam at the detector is the same for any angle of incidence as well as for the direct beam. In contrast to the height of the incoming beam which does not affect the determination of the angle of incidence if the axes are properly aligned, as just shown, the surface height of the mirror does. It must coincide with the central axis of the goniometer to assure proper definition of angle of incidence (AOI). The influence of an offset in height (∆z) on the AOI is shown in Fig. 3. Especially AOI scans are susceptible to this misalignment. Equation (2) gives the correlation between the angular offset of the detector axis and the offset in the height of the mirror. The designations are taken from Fig. 3. ∆z sin(2 ∗ AOI) 2∆z sin(∆ 2Θ) = sin(AOI) = (2) R Det cos(AOI) R Det It is seen that the shift in detector angle, respectively misalignment in AOI, is proportional to sin(AOI). In order to minimize this misalignment, we align the AOI at a small angle (1.5°) and use the Θ-axis (which has been qualified with an auto collimation set-up9) to go to the desired AOI. Here, the detector-angle is measured to verify the sample height. Any mismatch in detector angle can be converted to a misalignment in sample height. It should be mentioned that any possible offset in incoming photon beam height with respect to the axis of rotation does not compromise this procedure as shown before, see Fig. 2.
Fig. 3 Visualization of the height adjustment. The mirror surface is not in the center of rotation of the goniometer leading to an inaccurate AOI. Red arrows symbolize the beam path, the black circle is the main axis (Θ) of the goniometer. The dashed black line from the main axis to the detector symbolizes the detector-arm, the dotted red line shows the direction of the reflected beam. The difference between the direction of the reflected beam path and the angle of the detector-arm is indicated as ∆2Θ. ∆z is the distance from the mirror surface to the center of rotation.

21

To align a large off-axis EUV mirror, we use alignment mirrors, see Fig. 4. First, Φ (rotation around the optical axis) is adjusted using a pair of alignment mirrors mounted against one edge of the sample at the same position in y (and different positions in x). Using the x-drive of the goniometer to place these alignment mirrors in the beam, their positions are measured with respect to the R–coordinate of the reflectometer. From this measurement, we obtain the angle between R and the y-axis of the mirror. Φ is then set to orientate R parallel to y. In this orientation, Φ0, the known position of the alignment mirrors is used to correlate the ycoordinates of the sample to the goniometer R-values. Now, at Φ = Φ0+90° where R is parallel to the mirror x-coordinate, we use the third alignment mirror to correlate the xcoordinates of the mirror under test to the goniometer R-values for this value of Φ. Then, we align the angle of incidence in the centre of the mirror under test. Last, we check the height of the mirror. Now, given the shape of the mirror surface, every point on the mirror is known in lateral and angular position of the goniometer coordinates.
Fig. 4 The different coordinate systems: Mirror sample: x and y (in red) Goniometer: R, x, and Φ (black) Three alignment mirrors are shown touching the mirror side.

References
1 H. Meiling, V. Banine, N. Harned, B. Blumb, P. Kürz, and H. Meijer, "Development of the ASML EUV alpha demo tool," Proc. SPIE 5751, 90-101 (2005) 2 E. Gullikson, S. Mrowka, and B. Kaufmann, "Recent developments in EUV reflectometry at the Advanced Light Source," Proc. SPIE 4343, 363 – 373 (2001) 3 F. Scholze, C. Laubis, C. Buchholz, A. Fischer, S. Plöger, F. Scholz, H. Wagner and G. Ulm, "Status of EUV Reflectometry at PTB", Proc. SPIE 5751, 749-758 (2005) 4 F. Scholze, B. Beckhoff, G. Brandt, R. Fliegauf, A. Gottwald, R. Klein, B. Meyer, U. Schwarz, R. Thornagel, J. Tümmler, K. Vogel, J. Weser, and G. Ulm, "High-accuracy EUV metrology of PTB using synchrotron radiation," Proc. SPIE 4344, 402-413 (2001) 5 J. Tümmler, G. Brandt, J. Eden, H. Scherr, F. Scholze, G. Ulm, "Characterization of the PTB EUV reflectometry facility for large EUVL optical components," Proc. SPIE 5037, 265-273 (2003) 6 E. Louis, A. Yakshin, E. Zoethout, R. van de Kruijs, I. Nedelcu, S. van der Westen, T. Tsarfati, F. Bijkerk, H. Enkisch, S. Müllender, B. Wolschrijn, B. Mertens, "Enhanced performance of EUV multilayer coatings," Proc SPIE 5900, 1-4 (2005) 7 E. Louis, E. Zoethout, R. van de Kruijs, I. Nedelcu, A. Yakshin, S. van der Westen, T. Tsarfati, F. Bijkerk, H. Enkisch, S. Müllender, "Multilayer coatings for the EUVL process development tool," Proc SPIE 5751, 1170-1177 (2005) 8 F. Scholze, J. Tümmler, G. Ulm, "High-accuracy radiometry in the EUV range at the PTB soft X-ray radiometry beamline," Metrologia 40, S224-S228 (2003) 9 C. Buchholz, A. Fischer, S. Plöger, F. Scholz, F. Scholze, H. Wagner, "High-accuracy measurement of the angular sample positioning in the PTB EUV-reflectometer," BESSY Annual Report, 11-13 (2003)

22

Characterization of large off-axis EUV mirrors with high accuracy reflectometry at PTB
Christian Laubis, Christian Buchholz, Andreas Fischer, Sven Plöger, Frank Scholz, Heike Wagner, Frank Scholze, and Gerhard Ulm Physikalisch-Technische Bundesanstalt, Abbestraße 2-12, 10587 Berlin, Germany Hartmut Enkisch, Stephan Müllender, Marco Wedowski Carl Zeiss SMT AG, Carl Zeiss Straße 22, 73446 Oberkochen, Germany Eric Louis, Erwin Zoethout FOM-Institute for Plasma Physics, Edisonbaan 14, NL-3439 MN Nieuwegein, The Netherlands Extreme Ultraviolet Lithography (EUVL) is the key to the next generation of computer technology. EUVL imaging requires multilayer mirror optics being manufactured with extreme accuracy to ensure uniform illumination at the wafer plane and optimized throughput. As EUVL matures1,2, the requirements for the accuracy of reflectivity and wavelength measurements become tighter. Especially the wavelength for the peak mirror reflectivity should be matched to better than 2.5 pm and the uniformity of reflectance should be better than 0.08 %3. In response to this demand, substantial improvements in the total measurement uncertainty and repeatability at the Physikalisch-Technische Bundesanstalt (PTB), Germany's national metrology institute, have been achieved and reported4,5,6. PTB’s EUV reflectometer allows the measurement of large mirrors7 with a diameter as large as 550 mm, a height of 230 mm, and a weight of up to 50 kg8. We present here work on the alignment of aspheric off-axis mirrors which is crucial to achieve accurate measurement results for wavelength and reflectivity needed for the validation of the coating development for EUV optics9,10. Carl Zeiss SMT AG produced large off-axis EUV mirrors as they are used e.g. in ASML's alpha demo tools, the predecessor for EUVL production tools by ASML. The coating development and a large part of the actual coatings were done in collaboration with the FOMInstitute. Critical issues for EUVL mirrors are the high reflectivity close to the theoretical limit, the matching of the period to the operating wavelength of the stepper (13.5 nm) and the imaging properties of the EUV optics. The full multilayer stack needs to be controlled laterally to such extend that the initial sub-nanometre surface figure of the substrate is preserved. The so-called added figure error should not exceed 100 pm in order to ensure faultless imaging at 13.5 nm wavelength. For large off-axis EUV mirrors, measurements have to be done at angles significantly off normal, which dramatically increases the influence of angular alignment errors of the sample on the measured peak wavelength. Furthermore, according to the optical design, these optics have gradients of the coating thickness which require exact knowledge of the measurement position in the mirror coordinates. Extensive studies were done to develop alignment procedures of the sample with respect to the polar coordinates of PTB’s EUV reflectometer. The range of the reflectometer sample Raxis from -10 mm to 290 mm allows the centre area of the mirror to be positioned redundantly, i.e. using the setting Φ=Φ1 and R=x as well as Φ=Φ1 + 180° and R=-x, for |x| < 10. If everything is well aligned, this is physically the same spot on the sample. Thus for mirrors with a graded coating, a test of the alignment can be done by measuring several points

23

on the mirror surface (sufficiently close the centre) twice. If the mirror was misaligned, the curves measured would not coincide. Particularly for mirrors with a strongly graded coating, this is a sensitive test of the alignment.
0.7

0.6

0.5

reflectivity /%

0.4

Figure 1 Alignment test of a mirror with graded coating. The figure shows the measured reflectance on the center spot of the mirror, measured at Φ0 (□), and with the mirror turned by 180° around its optical axis (x).

0.3

0.2

0.1 13.7 13.8 13.9 14.0 14.1 14.2 14.3 14.4 14.5 14.6 14.7

wavelength /nm

Figure 2 Relative thickness profile of the coating on a projection optics element, measured in two perpendicular directions.

Figure 3 Non correctable added figure error determined in two perpendicular directions. The rms value amounts to 15 pm.

24

To validate the lateral alignment uncertainty, we used a large off-axis EUV test mirror with a large gradient in its multilayer coating thickness. Figure 1 gives the data for the test. After alignment (Φ0, R=x) and (Φ0+180°, R=-x) were measured. The value for center-wavelength of these measurements differs by only 0.6 pm which is, however, well beyond our short term repeatability of 0.06 pm4. Given the local gradient of the multilayer of about 3 pm/mm, the positional offset between the two measurements shown is 0.2 mm. This is twice the offset in y-alignment due to the rotation by 180°. This result verifies the uncertainty of 0.1 mm for the lateral alignment of mirrors. The most critical issue for the multilayer system on the projection optics is the noncorrectable added figure error. The wavelength of the mirrors can also be used to determine the multilayer period Λ(r) and thus the total film thickness. As an example of an optical element, which is representative for the projection optics, we present results for a concave mirror. The active optical surface has a diameter of 120 mm and the optical design requires a slightly parabolic profile. Figure 2 shows the relative thickness profile obtained, determined in two perpendicular directions where both curves nicely coincide, demonstrating the required rotational symmetry of the coating profile. The design profile can be corrected through lower order Zernike polynomial corrections to determine the alignment-adjusted design period Λa. Taking the difference between the measured period Λ(r) and the alignment-adjusted design period Λa(r), the non-correctable added figure error is calculated. The distribution of this figure error is shown in Figure 3 and amounts to an rms value of 15 pm only, which is seven times better than specified. Our results clearly show that it is possible to meet and verify the tight specifications for the lateral coating profiles of EUV multilayer mirrors. The non-correctable added figure error is significantly better than required and the overall reflectance of the coatings with a special protective capping layer is 65%. References
1 H. Meiling, V. Banine, P. Kürz, N. Harned, "Progress in the ASML EUV program," Proc. SPIE 5374, 31 - 42 (2004) 2 H. Meiling, V. Banine, N. Harned, B. Blumb, P. Kürz, and H. Meijer, "Development of the ASML EUV alpha demo tool," Proc. SPIE 5751, 90-101 (2005) 3 E. Gullikson, S. Mrowka, and B. Kaufmann, "Recent developments in EUV reflectometry at the Advanced Light Source," Proc. SPIE 4343, 363 – 373 (2001) 4 F. Scholze, C. Laubis, C. Buchholz, A. Fischer, S. Plöger, F. Scholz, H. Wagner and G. Ulm, "Status of EUV Reflectometry at PTB", Proc. SPIE 5751, 749-758 (2005) 5 R. Klein, A. Gottwald, F. Scholze, R. Thornagel, J. Tümmler, G. Ulm, M. Wedowski, F. Stietz, B. Mertens, N. Koster, and J. Elp, "Lifetime testing of EUV optics using intense sychrotron radiation at the PTB radiometry laboratory," Proc. SPIE 4506, 105-112 (2001) 6 F. Scholze, B. Beckhoff, G. Brandt, R. Fliegauf, A. Gottwald, R. Klein, B. Meyer, U. Schwarz, R. Thornagel, J. Tümmler, K. Vogel, J. Weser, and G. Ulm, "High-accuracy EUV metrology of PTB using synchrotron radiation," Proc. SPIE 4344, 402-413 (2001) 7 U. Dinger, G. Seitz, S. Schulte, F. Eisert, C. Münster, S. Burkart, S. Stacklies, C. Bustaus, H. Höfer, M. Mayer, B. Fellner, O. Hocky, M. Rupp, K. Riedelsheimer, P. Kürz, "Fabrication and metrology of diffraction limited soft x-ray optics for the EUV microlithography," Proc SPIE 5193 18-28 (2004) 8 J. Tümmler, G. Brandt, J. Eden, H. Scherr, F. Scholze, G. Ulm, "Characterization of the PTB EUV reflectometry facility for large EUVL optical components," Proc. SPIE 5037, 265-273 (2003) 9 E. Louis, A. Yakshin, E. Zoethout, R. van de Kruijs, I. Nedelcu, S. van der Westen, T. Tsarfati, F. Bijkerk, H. Enkisch, S. Müllender, B. Wolschrijn, B. Mertens, "Enhanced performance of EUV multilayer coatings," Proc SPIE 5900, 1-4 (2005) 10 E. Louis, E. Zoethout, R. van de Kruijs, I. Nedelcu, A. Yakshin, S. van der Westen, T. Tsarfati, F. Bijkerk, H. Enkisch, S. Müllender, "Multilayer coatings for the EUVL process development tool," Proc SPIE 5751, 1170-1177 (2005)

25

Re-calibration of the SUMER calibration source
M. Richter, W. Paustian, R. Thornagel Physikalisch-Technische Bundesanstalt, Berlin, Germany Characterization of space instrumentation is one of the major activities in the PTB laboratory at BESSY [1,2]. In many cases, calibration of devices is based on operating the storage ring as a primary source standard of calculable synchrotron radiation [3]. In the spectral range of vacuum-UV (VUV) radiation, e. g., the use of transfer source standards calibrated by comparison with BESSY has been established, for many years, as a powerful tool for absolute characterization of solar telescopes at the respective home laboratories under cleanroom conditions or within large vacuum tanks. Outstanding examples are the calibration of the SUMER and CDS telescopes of the Solar and Heliospheric Observatory (SOHO), more than ten years ago, within the framework of scientific cooperations with the Max Planck Institute for Solar System Research (MPS) and the Rutherford Appleton Laboratory (RAL). For this purpose, two hollow cathode rare gas discharge plasma sources with different collimating optics were developed as transfer standards [4,5]. Both standards have widely been used also for the characterization of further instruments (SOL-ACES, SERTS, EIS, MOSES) at different laboratories during the last years.

Fig. 1. Scheme of the SUMER calibration source consisting of a hollow cathode plasma source (A) with collimating normal-incidence optics (B).

Fig. 1 shows a scheme of the SUMER calibration setup together with the SUMER calibration source which has recently been re-calibrated in the PTB laboratory at BESSY after more than ten years of operation. The top part of Fig. 2 displays the re-calibration results, i.e.

26

the radiant power for various atomic and ionic emission lines in the wavelength range from 50 nm to 130 nm when operating the source with different rare gases (He, Ne, Ar, Kr). The data are

compared to those obtained in 1994. At the bottom of Fig. 2, both datasets have been normalized to the mean value for each emission line, respectively. In summary, the source seems to be slightly degraded by about 10 Fig. 2. Top: Radiant power of the SUMER calibration source for various atomic and ionic emission lines of He, Ne, Ar, and Kr as obtained in 1994 and 2005. Bottom: Ratio between the 2005 and 1994 data. The dashed line indicates the mean of the ratio values. The grey areas display the combined relative standard measurement uncertainty (k=1) varying from 14 % to 17 %. %, but within the

combined relative standard measurement uncertainty

only, which varies from 14 % to 17 % depending on the strength of the respective emission line.

As a consequence, the SUMER calibration source has been proven to be a stable VUV radiation source standard over a long period of time. Similar investigations are scheduled for 2006 on the CDS calibration source which currently is part of the VUV calibration facility for space instrumentation at RAL. Its grazing-incidence optics is expected to be even more stable than the normal-incidence optics of the SUMER calibration source. Next instrument, scheduled for characterization with the help of the CDS calibration source at RAL, is the Extreme Ultraviolet Normal Incidence Spectrometer (EUNIS) of NASA. [1] M. Richter et al., Advances in Space Research, in press [2] M. Richter et al., PTB Mitteilungen 115 (3), 218-221 (2005) [3] R. Thornagel, R. Klein, G. Ulm, Metrologia 38, 385-389 (2001) [4] J. Hollandt, M.C.E. Huber, M. Kühne, Metrologia 30, 381-388 (1993) [5] J. Hollandt, M. Kühne, B. Wende, Appl. Opt. 33, 68 - 74 (1994)

27

Environmental Analyses by TXRF - NEXAFS and IR-Spectroscopy: Speciation of Br in Organics and Characterization of the Organic Matrix
B. Beckhoff1, O. Hahn2, J. Weser1, M. Wilke3, G. Ulm1, O. Jann2
Physikalisch-Technische Bundesanstalt (PTB), Abbestraße 2-12, 10587 Berlin, Germany Bundesanstalt für Materialforschung und –prüfung, (BAM) Unter den Eichen 87, 12205 Berlin
3 1

2

Universität Potsdam, Institut für Geowissenschaften, Karl-Liebknecht-Str. 24, 14476 Golm

Introduction: Many polymers in building materials and consumer products contain flame retardants
(FR). The most common organic FR are brominated organic compounds. Due to their worldwide output several FR are now ubiquitous and can be found in sediments, biota, and – moreover - in fine dusts. The major task of FR, reduction of risk of fire, is contrary to possible risks from the toxicity and eco-toxicity of FR [1]. The aim of former studies was the characterization of the emission of certain polybrominated FRs from selected products to determine their contribution to the contamination of the indoor environment. On the basis of these results it was not possible to answer convincingly the question if the distribution of polybrominated FR in the environment is really caused by emission into air. So the aim of this present study is the analysis of trace constituents (e.g. polybrominated FR in polymer matrices) in fine dusts.

Experimental: Due to detection limits of conventional methods the analysis of trace constituents in
fine dusts is complicated. In order to simulate prospective experiments based on cascade impactors for sample collecting, some polymers containing brominated flame retardants were abraded mechanically on an ultra-clean silicon wafer surface. The chemical speciation of bromine is obtained by use of Total-Reflection X-ray Fluorescence Analysis (TXRF) in combination with Near-Edge X-ray Absorption Fine Structure (NEXAFS) analysis [2-4]. Synchrotron radiation based infrared spectroscopy (SRFTIR) in reflection mode is used for the characterization of the polymer matrix. Analysis of the X-ray Absorption Near Edge Structure at K and L absorption edges in the hard x-ray range are employed frequently in the characterisation of element speciation. Due to the small penetration and information depth of only a few nanometers in total reflection beam geometry, self absorption effects are largely negligible even for the analysis of L edges in the soft x-ray range. In addition, the TXRF method offers drastically reduced scattering contributions and thus lowest detection limits. In the present investigation monochromatized undulator radiation available at the PTB PGM-U49 beamline was employed to probe the Br-L3 and Br-L2 absorption edges of some samples deposited on a silicon wafer surface. For the excitation of the present specimens the PGM-U49 beamline provides an energy resolution of about 300 meV and permits the photon energy to be varied in steps of 250 meV. The PTB employes a reference-free TXRF arrangement providing knowledge of the incident radiant power, of the solid angle of detection and of both the detection efficiency and response. The latter allows for a very reliable spectra deconvolution.

28

The SR-FTIR measurements were performed at the synchrotron infrared beamline IRIS at BESSY II. A FTIR spectrometer (Bruker 66/v) and an IR microscope (Thermo Nicolet Continuum and Nexus) were used. Due to the fact that synchrotron infrared light is much brighter than a conventional infrared source (e.g. globar) it is possible to carry out FTIR measurements in reflectance with high lateral resolution. A gold sputtered silicon wafer was measured for background correction.

Results and discussion: SR-FTIR spectroscopy in reflection mode provides the classification of
different polymers within reasonable measurement times. Figure 1 shows the corrected FTIR spectrum of an artificial “dust” sample in comparison with a polystyrene reference spectrum. It is obvious that the identification of the polymer is possible but there is no reliable indication of the existence of a polybrominated FR.

1,0 0,8 0,6 0,4 0,2 0,0
3500 3000

Sample Polystyrene

Absorption (norm.)

2500

2000
-1

1500

1000

Wavenumber, ν/cm

Fig.1: Corrected FTIR spectrum of an artificial “dust” sample in comparison with a polystyrene reference spectrum. The identification of the polymer polystyrene is possible. Due to the fact that characteristic bands for brominated compounds (e.g. ν = 1050-1040 cm-1 for aromatic compounds) are not clearly visible there is no reliable indication for the existence of PBFR. The initial results demonstrate the potential of the TXRF-NEXAFS method to successfully contribute to the elemental speciation, even of trace elements. The shape of the various TXRF-NEXAFS structures reflects the chemical environment of the element bromine in the different organic matrices (Fig. 2, 3). It is possible to distinguish between an aliphatic polybrominated compound and an aliphatic polybrominated compound: the chemical shift of the L-edge (as well as the K-edge) is caused by different hybridisation of the neighbour atom carbon (sp2 and sp3). Due to the fact that sp2 hybridisation is more electrophilic the electron density at an „aromatic“ bromine is a little bit diminished as a consequence the binding energy of the electron which has to be excited is increased. The combination of TXRF-NEXAFS with SR-FTIR offers the option for reference measurements without need for any chemical sample preparation, thus reducing undesired modifications (e.g. such as in the case of GC/MS).

29

1,0
HBCD (aliphatic) TBBPA (aromatic)

norm. Br-La CR / (nWs)

-1

0,8

0,6

reference substances
0,4

0,2
ΔE = 0.5 eV

0,0 1545 1550 1555 1560

phton energy / eV

Fig. 2: TXRF-NEXAFS spectra of two different polybrominated flame retardants

1,0 HBCD PS with HBCD

norm. Br-La CR / (nWs)

-1

0,8

0,6

0,4

0,2

0,0

1547

1554

1561

photon energy / eV

Fig.3: TXRF-NEXAFS spectra of pure HBCD in comparison with PS containing < 2 % HBCD
References
[1] S.Kemmlein, O.Hahn, and O.Jann, Emissions of organophosphate and brominated flame retardants from consumer products and building materials, Atmos. Environ. 39-40 (2003) 5485-5493. [2] G.Pepponi, B.Beckhoff, T.Ehmann, G.Ulm, C.Streli, L.Fabry, S.Pahlke and P.Wobrauschek, Analysis of organic contaminants on Si wafers with TXRF-NEXAFS, Spectrochim. Acta B 58 (2003) 2245-2253. [3] B.Beckhoff, R.Fliegauf, G.Ulm, J.Weser, G.Pepponi, C.Streli, P.Wobrauschek, T.Ehmann, L.Fabry, C.Mantler, S.Pahlke, B.Kanngießer, W.Malzer, Ultra-trace analysis of light elements and speciation of minute organic contaminants on silicon wafer surfaces by means of TXRF in combination with NEXAFS, Electrochemical Society Proceedings ‘Analytical and Diagnostic Techniques for Semiconductor Materials, Devices, and Processes’, 2003-03, (2003) 120-128. [4] S.Török, J.Osan, B.Beckhoff, G.Ulm, Ultra-trace speciation of nitrogen compounds in aerosols collected on silicon wafer surfaces by means of TXRF-NEXAFS, Powder Diffraction Journal 19 (2004) 81-86.

30

X-Ray Resonant Raman Scattering (RRS) on Ni
M. Müllera , Ch. Zarkadasb , B. Beckhoffa, A. G. Karydasb
b

Physikalisch-Technische Bundesanstalt, Abbestraße 2-12, 10587, Berlin, Germany Institute of Nuclear Physics, N.C.S.R Demokritos, Aghia Paraskevi 15310, Athens, Greece

a

The X-ray resonant Raman scattering (RRS) effect on nickel was studied by means of monochromatic polarized exciting radiation. Experiments were carried out at the four crystal monochromator beamline [1] of the Physikalisch-Technische Bundesanstalt (PTB) for synchrotron radiation from 4 to 10 keV at the electron storage ring BESSY II. Resonant Raman spectra of thin Ni foils were recorded at the Cu-Kα (8041 eV) exciting beam energy. For a pure monoatomic target, and assuming an isotropic emission, the theoretical intensity of the KL-RRS photons in the energy interval ES, ES+dES, can be expressed in terms of experimental and fundamental parameters by the following formula [2]:
RRS dN the ( E o , E s ) N d 2σ RRS 1 = N 0 ( E 0 ) ⋅ ΔΩ d ⋅ AV ⋅ ⋅ M ξ ( E0 , Ε s ) ⋅ ε d ( E s ) ⋅ sin ϕ1 dE s AW dΩ ⋅ dE s

(1)

where N0(E0) is the number of the incident monoenergetic photons on the sample surface, having energy E0, ΔΩd is the detection solid angle defined by an aperture placed in front of the detector entrance window, NAV is the Avogadro number, AW is the atomic weight of the target, εd(ES) the detector efficiency [3,4] at the scattered energy ES, φ1 is the angle between the target surface and the incoming exciting beam, and Mξ(E0,ES) is the self absorption factor expressed as follows:

M ξ ( E0 , Ε s ) =

1 − exp[− μ tot ( E o , E s ) ⋅ ξ ] μ tot ( E o , E s )

(2)

In Equation (2), ξ denotes the sample thickness and µtot(E0,ES) the total mass absorption coefficient of the target. The study of the RRS effect under different combinations of the incident X-ray energy – target elements can contribute significantly to several areas of X-ray spectrometry. From a fundamental aspect, the knowledge of the RRS cross sections can improve the accuracy of the tabulated theoretical attenuation coefficients for energies below the absorption edge of an element and provide possible explanations regarding the discrepancies (up to 5-6 %) observed between experimental and theoretical mass absorption coefficients [5]. The results obtained are compared to existing literature values, whereas the uncertainties associated with the different experimental and analytical approaches are critically evaluated. For the RRS studies on Ni, a thin Ni foil (purity of 99.99% for the initial bulk material) of 2 µm nominal thickness was measured at exciting beam energy of 8041 eV. During these measurements, transmission scans were also performed for the experimental determination of the absorption correction required for the calculation of the RRS cross sections [2]. The Ni KL-RRS spectrum obtained at the energy of Cu-Kα (8041 eV) is shown in Fig. 4. The spectrum is composed clearly from the pronounced Ni KL-RRS structure, by the bremstrahlung radiation induced by Ni L-shell photo-electrons and also from small interfering fluorescent peaks arising from trace impurities in the thin Ni target. The contribution of the scattered Cu-Kα exciting radiation is minimized, due to the positioning of the detector in the polarization plane and at an angle of 90° with respect to the incident beam axis. The tailing of the Cu-Kα scattered peak is produced mostly by the coexisting Ni KM-RRS structure.

1
31

Fig. 1: RRS spectrum and associated components obtained as a result of the fitting procedure

For the extraction of the corresponding Ni KL-RRS cross-sections Eq. (1) was convoluted with the known detector’s energy response function and then fitted to the actual experimental data to obtain the net RRS distributions, as follows:
RRS dN exp ( E o , E s )

dE s

=

max ES

∫
0

RRS dN the ( E o , E s ) ⋅ Rd ( E s , E ′) ⋅ dE ′ dE s

(3)

The height of the RRS distribution was treated as one fitting parameter, whereas in order to generalize the analytical procedure, the onset energy of Eq. (3) was introduced as a second fitting parameter. The onset energy is expected to be equal to, if the Ni target atoms do not participate into any chemical bonding [6]. The small contribution of the Ni L-shell photoelectrons in the spectrum was also taken into account by introducing a simple functional dependence of the electron bremstrahlung cross section on the electron’s kinetic energy and the emitted photon’s energy [7]. The absorption correction factors used were experimentally determined from transmissions measurements covering an energy region of 4 keV to the highest possible incident photon energy just below the Ni-K absorption edge. The total number of the incident exciting photons was determined from the photodiode measurements, whereas the solid angle subtended by the detector was calculated using the distance and diameter of the detector’s calibrated diaphragm. The KL-RRS cross section (dσ/dΩ) was next deduced by integrating the double differential KL-RRS cross section to the maximum energy of the scattered RRS photons. Our results are presented in Table 1 in comparison with the Ni KL-RRS cross sections already reported in the literature. Evidently, there is an excellent agreement between the cross sections determined for polarized and unpolarized radiation, verifying the theoretical predictions regarding the polarization state of the exciting beam. In principle, the results deduced in this work are also in good agreement with previously published values [8-9, 10], within the overall uncertainties achieved. Referring to our work, the total relative uncertainties of the cross sections, presented in Table 1, are defined by the uncertainty of the first fitting parameter. For the polarized radiation, the constant term, after the end of the fitting procedure with the experimental detector response function, was found to have a relative uncertainty of about 5%. The number N(E0) of the inci-

2
32

Table 1. Ni KL-RRS cross sections obtained in this work compared to the literature values.

Reference

Methodology adopted for the extraction of the KL-RRS cross sections

(dσ/dΩ)KL-RRS

ro2 /sr
This work, polarized radiation complementary study with unpolarized radiation Ref. [10] Ref. [8] Ref. [9] Integration of the double differential cross sections from 5.0 keV upwards Integration of the double differential cross sections from 5 keV upwards Integration of the net RRS spectral data from 4.8 keV upwards Integration of the net RRS spectral data from 4.8 keV upwards Not reported 7.6 ± 0.5 7.7 ± 0.9 7.0 ± 0.2 7.3 ± 0.2 6.8 ± 0.3

dent exciting photons was measured with a relative uncertainty of 1%, the efficiency εd(E) of the detector used has been determined with a relative uncertainty of 1.5%, and the solid angle of detection ΔΩd was calculated to have a relative uncertainty of 0.7%. Furthermore, the relative uncertainty in the determination of the sample’s thickness ξ, is estimated to about 4%, attributed mostly to the uncertainties of the theoretical mass absorption coefficients employed. Therefore, a total relative uncertainty of 7% was calculated for the Ni KL-RRS cross section dσ(E0)/dΩ obtained for polarized radiation. Acknowledgments This work is partially supported by the Greek State Scholarships Foundation IKY and the German Academic Exchange Service DAAD within the IKYDA program (project 314-ikydadr), which is gratefully acknowledged. Furthermore the authors would like to thank their colleagues R. Fliegauf, M. Kolbe, G. Ulm, J. Weser and N. Divis for supporting the present work.

[1] [2]

[3] [4]

[5] [6 ]

[7] [8] [9] [10]

M. Krumrey, G. Ulm, High-accuracy detector calibration at the PTB four-crystal monochromator beamline, Nucl. Instr. Meth. A 467-468 (2001) 1175-1178 B. Beckhoff and G.Ulm, Determination of fluorescence yields using monochromatized undulator radiation of high spectral purity and well-known flux, Adv. X-Ray Anal. 44 (2001) 349-354 F. Scholze, M. Procop, Measurement of detection efficiency and response functions for an Si(Li) x-ray spectrometer in the range 0.1-5 keV, X-Ray Spectrom. 30 (2001) 69-76 B. Beckhoff, R. Klein, M. Krumrey, F. Scholze, R. Thornagel, G. Ulm, X-ray detector calibration in the PTB radiometry laboratory at the electron storage ring BESSY II, Nucl. Instr. Meth. A 444 (2000) 480-483 L. Unonius and P. Suortti, Mass Attenuation Coefficients of the Elements Ti, V, Fe, Co, Ni, Cu and Zn for the K Emission Lines between 4.51 and 10.98 keV, J. Appl. Cryst. 22 (1989) 46-52 A.G Karydas, S. Galanopoulos, Ch. Zarkadas, T. Paradellis and N. Kallithrakas-Kontos, Chemical State speciation by resonant Raman scattering, J. Phys.: Condens. Matter 14 (2002) 12367-12381 H. Ebel, X-ray Tube Spectra, X-Ray Spectrom. 28 (1999) 255 - 266 Y. B. Bannett, D.C. Rapaport and I. Freund, Resonant X-Ray Raman Scattering and the infrared divergence of the Compton effect, Phys. Rev. A 16 (1977) 2011-2021 P. Suortti, Scattering of X-Rays near the K Absorption Edge, Phys. Status. Solidi (b). 91 (1979) 657-666 A.F. Kodre and S.M. Shafroth, Resonant Raman scattering of X-rays: Evidence for K-M scattering, Phys. Rev. A 19 (1979) 675-677

3
33

Determination of the mechanical properties of electroplated metals
NICOLAS BEISSER, DANIEL SCHONDELMAIER, IVO RUDOLPH, JOSEF KOUBA, BERND LOECHEL BESSY GmbH, Anwenderzentrum Mikrotechnik, Albert-Einstein-Strasse 15, 12489 Berlin, Germany
E-mail: Nicolas.Beisser@bessy.de, Tel.: +49-30-6392-4681, Fax: +49-30-6392-4682

The characteristics of materials and alloys used in LIGA technology differ considerably from conventionally produced metals with regard to Young’s modulus and tensile strength. Furthermore, alloy compositions are used which are not commonly applied in macro mechanics. For constructing and determining micro components, it is essential to be familiar with their properties. When using FEM analyses for the construction of a micro-module it is crucial to know about the mechanical characteristics of a material like Young’s modulus, tensile strength, yield stress and hardness. Due to the low dimensions of the micro components, the defects resulting from the production process are of particular importance. Likewise, the structure created during an electrodeposition process, is different with respect to material of same composition but fabricated with conventional processes. To determine the mechanical characteristics of materials used in MEMS it is necessary to employ specially adapted measuring tools with a resolution appropriate for determining the measured values. For this purpose, at BESSY two measurement tools were constructed and realized. The first tool is a tensile-stress testing machine which allows a resolution of 0.1 N for the tensile force and 70 nm for the tensile path [1]. Special tensile samples have to be developed for the new measurement tool with a cross-section of 0.015 mm2 in the tensile area. The second measurement tool is a bending-stress machine with a resolution of 0.01N for the bending-stress and 70 nm for the bending path. Likewise special bending samples had been developed with a cross-section of 0.09 mm² in their bending area. The samples were produced on substrate surfaces by means of a LIGA process. In order to compare the mechanical properties of both kinds of samples, they were placed on the same area on the substrate. Presently the determination of the mechanical data for all electrodeposited materials used at BESSY (Ni, NiFe, Cu and Au) is under work. During the deposition process of the tensile samples (Fig. 1), crucial parameters like current density and temperature were varied in order to determine their influence on tensile strength, Young’s modulus, yield strength and hardness. The crystalline structure of the electroplated metal varies based on deposition conditions and can be observed by means of a micrograph. These experiments were done to understand the correlation between the crystalline structure and measured material properties. Likewise, the process related non homogenous current density distribution on the substrate surface is taken into account. By this means, the influence of current density distribution on the material properties within a single fabrication batch can be determined. Resulting from non homogeneous growth, mechanical properties of electrodeposited materials vary in the growth and substrate plane direction. In order to characterize this effect for different
34

materials, using the bending test, strain is applied on micro-beams both in and perpendicular to the growth direction [2, 3]. Force-flexion diagrams are created using employed force and measured flexion. Based on the accomplished analyses, stress-strain diagrams (Fig. 2) for electrodeposited Ni, NiFe (90/10), Cu and Au are extracted in order to determine tensile strength, yield strength, Young’s modulus and elongation limit. Furthermore, the Vickers hardness of the materials was measured and the influence of the deposition parameters (current density and temperature) on the mechanical properties was determined. As in the case of steel, comparison tables for electrodeposited Ni, NiFe, Cu and Au are created illustrating the connection between hardness and tensile strength. The mechanical properties in different directions of electrodeposited Ni, NiFe (90/10), Cu and Au were determined by means of bending test. For applications like mold inserts or other mechanical parts, determined date will be used for the design development.
electrodeposited Copper
400
Rm = 338MPa

350 tensile strength [MPa]

cross-section

300 250 200 150 100 50 0 0 2

Rp02 = 277MPa

E-Modul = 164GPa

A = 14,%

4

6

8

10

12

14 [%]

FIGURE 1: sample setup for tensile strength measurement

FIGURE 2: force flexion diagram for electrodeposited copper

electrodeposited Ni
1400

1200

bending strenght [MPa]

1000

800

FIGURE 3: sample setup for bending strength measurement

600

400

200

0 0 20 40 60 bending path [µm] 80 100

120

FIGURE 4: force bending diagram for electrodeposited Nickel

35

References
1. 2. 3. M. Hoffmann, “Ermittlung materialspezifischer Eigenschaften von galvanisch abgeschiedenen Schichten”, Diploma thesis 2004, TFH Berlin S. Timoshenk, Theory of Plates and Shells, McGraw-Hill Education, ISBN 0070858209 N. Beisser, “Entwicklung, Konstruktion und Fertigung einer Justiervorrichtung für eine Gradientenlinse in der Mikrotechnik”, Diploma thesis 2003, TFH-Berlin

36

Photonische Kristalle für das sichtbare Spektrum
KOUBA, J.
BESSY GmbH, Anwenderzentrum für Mikrotechnik; Albert-Einstein-Str. 15, 12489 Berlin E-Mail: josef.kouba@bessy.de Tel.: +49 30 6392 3125 Fax: +49 30 6392 4709 Abstrakt: Photonische Kristalle bilden Grundelemente vieler zukünftiger photonischen Komponenten. Bei BESSY werden im Rahmen einer Doktorarbeit photonische Kristalle für das sichtbare Spektrum untersucht. In diesem Beitrag werden erste Ergebnisse der theoretischen Untersuchungen, der Herstellung und der experimentellen Untersuchungen vorgestellt.

Einleitung
Photonische Kristalle sind komplexe Strukturen mit örtlich periodischer Dielektrizitätsfunktion. Diese ermöglichen in Bezug auf Licht die gleichen Funktionalitäten, wie sie Halbleiter für elektrische Ströme zur Verfügung stellen. Man bezeichnet photonische Kristalle daher auch als photonische Bandlückenmaterialien. Sie bestehen aus strukturierten Halbleitern, Gläsern oder Polymeren und zwingen das Licht mittels ihrer spezifischen Brechungsindexstruktur dazu, sich in der für die Bauteilfunktion notwendigen Art und Weise im Medium auszubreiten. Ein wesentliches Merkmal dieser Technik ist die Möglichkeit, Licht auf kleinste Abmessungen unterhalb der Lichtwellenlänge zu führen um lokal hohe Feldintensitäten zu erzeugen und somit eine starke Wechselwirkung zwischen der Materie und Licht zu gewährleisten. Von den dabei auftretenden, z.T. quantenelektrodynamischen Effekten werden neuartige Anwendungen erwartet. Damit rückt ein entscheidender Durchbruch auf dem Gebiet der optischen Miniaturisierung, die mit konventionellen optischen Techniken eine prinzipbedingte Grenze nicht unterschreiten kann, in greifbare Nähe. Die erfolgreiche Nutzung photonischer Bandlückenmaterialien eröffnet die Möglichkeit einer weiteren Steigerung der Datenübertragungen bei gleichzeitiger Verkleinerung der Geräte und Reduzierung der Kosten. Generell wird die Zielrichtung verfolgt, eine vollständig optische Kommunikationstechnologie zu verwirklichen, da sich gegenwärtig die Elektronik als hauptsächlicher Engpass hinsichtlich der Geschwindigkeit und Übertragungskapazität der Nachrichtenübertragung darstellt. Es existieren bereits Konzepte für rein optische Schaltelemente, die eines Tages als optische Hochgeschwindigkeits-Transistoren genutzt werden könnten. In diesen Schaltelementen werden photonische Kristalle eine zentrale Rolle spielen. Eine Vielzahl von Arbeiten befasste sich bereits mit photonischen Kristallen und Resonatoren im nahen Infrarot. Diesen Arbeiten stehen bisher jedoch nur sehr wenige Untersuchungen im sichtbaren Spektralbereich gegenüber. Gründe

37

hierfür liegen vor allem in der Technologie der Herstellung der Resonatoren. Zum einen muss das verwendete Material im sichtbaren Spektralbereich transparent sein und gleichzeitig einen möglichst hohen Brechungsindex aufweisen, zum anderen liegen die Strukturgrößen der Resonatoren teilweise unter 100 nm. Dabei wären optische Resonatoren im Sichtbaren insbesondere in Verbindung mit gängigen Fluoreszenzmarkern und Laserfarbstoffen mit Emissionswellenlängen im sichtbaren Spektralbereich äußerst interessante Kandidaten, um an ihnen Effekte der Wechselwirkung zwischen Licht und Materie zu studieren. Im Rahmen einer Doktorarbeit werden bei BESSY zweidimensionale photonische Kristalle für sichtbares Licht entwickelt, hergestellt und untersucht. Zentraler Punkt der Untersuchung sind die mit Hilfe photonischer Kristalle erzeugten optischen Resonatoren. Diese zeichnen sich durch eine charakteristische optische Modenfrequenz aus, nämlich die des im Resonator erlaubten Modes. Solche Resonatoren können dann als Grundelemente in verschiedenen Anwendungen eingesetzt werden. Die möglichen Anwendungen reichen dabei von selektiven Wellenlängenfiltern über hocheffiziente Leuchtdioden oder Mikrolaser bis hin zu Einzelphotonenquellen für die Quanteninformationsverarbeitung in den zukünftigen Quantenrechnern.

Ergebnisse der Untersuchungen
Die Arbeit befasst sich mit theoretischen Untersuchungen, der Herstellung der Resonatoren und deren Charakterisierung. Abb. 1 zeigt die Bandstruktur eines photonischen Kristalls für sichtbare Wellenlängen. Die der Berechnung zugrunde liegenden Materialdaten entsprechen den von in der Mikroelektronik standardmäßig verwendetem Siliziumnitrid. Bei der Struktur handelt es sich um eine periodisch perforierte Membrane wie in Abb. 1 dargestellt. Die Periode der Struktur beträgt 280 nm und der Lochradius 112 nm. Die Bandstruktur für das in der Kristallebene linear polarisierte Licht zeigt eine Bandlücke mit etwa 10 % Breite. Die Bandlücke wird auch in dem Transmissionsspektrum im Bereich von 560 bis etwa 700 nm deutlich sichtbar, siehe Abb. 2. Am einfachsten kann ein optischer Resonator durch Verzicht auf eine oder mehrere Löcher in dem Kristallgitter realisiert werden. Solch eine Verletzung der Periodizität wird in Analogie zu Kristallgittern als Punktdefekt bezeichnet. Bei geeignet konzipierten Punkdefekten ergibt sich eine Eigenfrequenz die innerhalb der Bandlücke des umgebenden Kristalls liegt. Die Bandstruktur einer solchen Struktur ist in der Abb. 3 zu sehen und der zugehörige Eigenmod in der Abb. 4. Die Kavität zeichnet sich durch ein sehr flaches Band über die ganze BrilloneZone aus, was auf die geringe Frequenzbreite hinweist. Die Resonanzfrequenz der Kavität liegt bei 630 nm. Wie in der Abb. 4 deutlich, wird das Licht sehr stark in der Kavität lokalisiert.

38

Abb. 1 – Bandstruktur von einem hexagonal angeordneten Lochmuster im SiN; x-Abszisse – Ränder der Brillone-Zone, y-Abszisse – normalisierte Wellenlänge; Bandlücke mit 10 % Breite sichtbar

Abb. 2 – Berechnetes Spektrum des photonischen Kristalls mit 5 Perioden; gleiche Abmessungen wie in Abb.1; Bandlücke aus Abb. 1 ist vom 700 nm bis 560 nm deutlich zu erkennen

Abb. 3 – Bandstruktur einer Kavität; Strukturabmessungen siehe Abb. 1; 1 Loch ausgelassen; flaches Band korrespondierend zu dem Kavitätsmod bei 630 nm

Abb. 4 – Kavitätsmod des Punktdefekts aus Abb. 3 Resonanzfrequenz bei 630 nm; starke Lokalisierung vom Licht sichtbar

Eine der einfachsten Anwendungen solcher optischen Resonatoren ist ein Wellenlängenfilter. Der Resonator wirkt dabei als eine Art Schaltelement und lässt nur Licht mit der Frequenz durch die der Eigenfrequenz des Resonators entspricht wie in Abb. 7 und Abb. 8 dargestellt.

39

Abb. 5 – Photonische Kavität kombiniert mit Wellenleiterstruktur angeregt mit der Resonanzfrequenz;

Abb. 6 – Analoge Struktur wie aus Abb. 5 angeregt mit 0.95xResonanzfrequenz

Die Herstellung der photonischen Kristalle für das sichtbare Spektrum stellt mit Perioden kleiner als 200 nm und einzelnen Strukturdetails unterhalb von 30 nm eine technologische Herausforderung dar. Hochauflösende Elektronenstrahllithographie und besondere Trockenätzprozesse mussten entwickelt und optimiert werden, um die Strukturen in geeigneten Materialien herstellen zu können. Beispiele der bereits realisierten Strukturen sind in den Abb. 7 und 8 zu sehen.

Abb. 7 – Photonischer Kristall aus perforiertem SiN für sichtbare Wellenlängen; Freigeätzte Membranstruktur; Membrandicke 300 nm, Periode 400 nm, Radius der Löcher 150 nm

Abb. 8 – Kavitätsstruktur aus perforiertem Silizium für IR optische Wellenlängen; Teststruktur; Membrandicke 500 nm, Periode 500 nm, Radius der Löcher 420 nm

Die Überprüfung der vorhergesagten optischen Eigenschaften erfolgt mir Hilfe von Transmissionsmessungen. Dabei wird in die Kristallstruktur Licht mit passender Wellenlänge eingekoppelt und das Transmissionsspektrum aufgenommen, aus in dem die optischen Eigenschaften nachweisbar sind. Das Einkoppeln des Lichts in die Struktur stellt erneut eine Herausforderung dar,

40

denn die Schichtdicken der photonischen Kristalle betragen nur wenige Hundert Nanometern. Um die Einkopplung zu ermöglichen, ist ein präziser mechanischer Aufbau notwendig. Die ersten Ergebnisse der Untersuchungen werden in kürze erwartet. Zusätzlich zu den Transmissionsmessungen sind in Zusammenarbeit mit der Humboldt Universität zu Berlin und dem Max Born Institut Nahfeldmessungen geplant, um die lokalen Feldintensitäten abbilden zu können. Zusammenfassung Im Rahmen einer Doktorarbeit werden bei BESSY photonische Kristalle für den sichtbaren Spektralbereich untersucht. Dabei stehen vor allem optische Resonatoren im Vordergrund, die es ermöglichen starke Lichtintensitäten auf lokal sehr engem Raum zu realisieren und dadurch die Wechselwirkung des Lichts mit der Materie zu studieren. Eine weitere Anwendung der Resonatoren liegt in deren Verwendung als Wellenlängenfilter. Insgesamt zielt die Untersuchung auf die Erprobung der optischen Eigenschaften neuartiger photonischen Komponenten, die mit Hinblick auf die Entwicklung in der Optoelektronik in naher Zukunft breite Anwendung finden werden. Die Arbeiten beinhalten theoretische Untersuchungen, Herstellung der Proben und deren optische Charakterisierung. Durch die Berechnungen wurden mehrere Resonatortypen vorcharakterisiert. Dabei wurden Daten von Siliziumnitrid verwendet, welches ein branchenübliches Material ist. Durch Optimierung der Herstellungsprozesse wurden erfolgreich Prototypen der optischen Resonatoren im Siliziumnitrid realisiert. Die Herstellung erwies sich als extrem herausfordernd, denn die Strukturgrößen liegen im unteren sub-100 nm Bereich. Durch einen experimentellen Aufbau basierend auf Messung der Transmission sollen in der kommenden Phase die optischen Eigenschaften bestätigt werden. Durch Zusammenarbeit mit weiteren Institutionen sind auch Nahfeldmessungen an den Resonatorstrukturen geplant.

41

SU-8 – Promising Resist for Advanced Direct LIGA Applications
KOUBA, J. a, ENGELKE R. b, BEDNARZIK M. a, AHRENS G. b, HEINZ-ULRICH SCHEUNEMANN a, GRUETZNER G. b, LOECHEL B. a, MILLER, H. c, HAASE D. d BESSY GmbH, Albert-Einstein-Str. 15, 12489 Berlin, Germany micro resist technology Koepenicker Str. 32,D-12555 Berlin, Germany c MicroChem Corp., 1254 Chestnut Street, Newton, MA 02464, USA d Jenoptik Mikrotechnik GmbH, Goeschwitzerstr. 40, 07745 Jena, Germany
b a

Email: josef.kouba@bessy.de Phone: +49 30 6392 3125 Fax: +49 30 6392 4709
This work widely uses the contents of the presentation “Comparative study of the Sidewall Profile of PMMA and SU-8 Moulds Made by UDXRL and of Electroformed Metallic Counterpart” to the High Aspect Ratio Micro Structure Technology workshop HARMST 2005 held in Gyeongyu (Republic of Korea), June 10-13, 2005.

Abstract: A case study of use of negative type SU-8 X-ray sensitive resist for fabrication of advanced, highly precise, ultra tall Direct LIGA mechanical micro parts is presented in this paper. Using Direct LIGA technique, ~1 mm tall highly precise metallic gear wheels are being fabricated, previously using PMMA based process. Starting from a non-optimized non satisfying SU-8 process, significant process parameters for process optimization were identified using statistical design of experiment. By varying the significant process parameters, SU-8 process was further optimized with respect to critical aspect of sidewall bow and tilt of metallic structures. After the optimization, metallic parts fabricated using SU-8 process showed comparable quality as those fabricated using PMMA based process.

Introduction Currently, conventional precision engineering technologies are reaching their limits concerning the feature size and precision demands, so that alternative technologies such as LIGA gain on importance [1]. An excellent example is the fabrication of metallic micro gear wheels, which are used in precision, zero backlash micro harmonic drives [2]. In order to reach their performance, metallic components of the these complex and small gear systems require the highest level of geometrical and dimensional precision, which can only be reached by means of LIGA technology. An assembled micro harmonic drive is depicted in

1
42

Fig. 11.

Fig. 1 - Assembled micro harmonic drive with the driven wheel in the middle, two planet wheels and the flexible element called ‘flex-spline’

The customer-given specifications including precision requirements are given in
Substrate Resist/metall height Smallest feature size Largest aspect ratio Linewidth error Vertical tolerances per sidewall Allowed roughness (Ra/Rz) 4 “ Si wafer, 90 % patterned 1100/1000 µm 25 µm (flex-spline) 40 (in metall) ±1 µm ±1 µm 100 nm

Tab. 1 and clearly illustrate the need of use of LIGA.
Substrate Resist/metall height Smallest feature size Largest aspect ratio Linewidth error Vertical tolerances per sidewall Allowed roughness (Ra/Rz) 4 “ Si wafer, 90 % patterned 1100/1000 µm 25 µm (flex-spline) 40 (in metall) ±1 µm ±1 µm 100 nm

Tab. 1 – Dimensional parameters for critical parts of harmonic drive units

As shown in
Substrate Resist/metall height Smallest feature size Largest aspect ratio Linewidth error Vertical tolerances per sidewall Allowed roughness (Ra/Rz) 4 “ Si wafer, 90 % patterned 1100/1000 µm 25 µm (flex-spline) 40 (in metall) ±1 µm ±1 µm 100 nm

1

With courtesy of Micromotion GmbH, Mainz, Germany; www.micromotion.de

2
43

Tab. 1, fabrication of the most critical parts of the gear assembly – the flex spline – poses the largest challenge, especially regarding the sidewall tolerances. The fabrication of all the components occurs by the means of Direct LIGA [3]. This reduces the LIGA process on X-ray Lithography and first electroforming step only thus posing high requirements on process stability, repeatability and cost effectiveness.

Fig. 2 – Scheme of Direct LIGA process; (a) exposure of the resist via X-Ray mask; (b) development of the exposed resist; (c) electroplating of the metallic micro parts; (d) removal of the resist

Up to recently, the gear wheels such as the ~1 mm tall flex-splines were fabricated using PMMA-based process with overall sufficient quality and yield. But due to the requirements to cut the processing costs, switching to SU-8 process would be preferable since the process costs are smaller as illustrated in In Fehler! Ungültiger Eigenverweis auf Textmarke. the costs for X-ray exposure are the bottle neck of the whole process such that cutting the costs of X-ray exposure together with parallelizing of the pre- and post processing steps would reduce the single-part costs significantly. But using the initial SU-8 resist process failed and the precision requirements couldn’t be fulfilled, giving so an impulse for this study. The aim of the work was to optimize the SU-8 process in order to reach the quality of final micro gear wheels similar to those of made using PMMA process.

3
44

Fig. 3. In Fehler! Ungültiger Eigenverweis auf Textmarke. the costs for X-ray exposure are the bottle neck of the whole process such that cutting the costs of Xray exposure together with parallelizing of the pre- and post processing steps would reduce the single-part costs significantly. But using the initial SU-8 resist process failed and the precision requirements couldn’t be fulfilled, giving so an impulse for this study. The aim of the work was to optimize the SU-8 process in order to reach the quality of final micro gear wheels similar to those of made using PMMA process.

Fig. 3 – Comparison of process costs for PMMA and SU-8 process; (a) – preprocessing costs, (b) – X-Ray exposure related costs, (c) – post processing costs; 4 “ Si substrate, 1 mm thick resist; electroforming and inspection not included; all costs related to the preprocessing costs of one SU-8 substrate

SU-8 versus PMMA SU-8 is nowadays a well known X-ray resist and compared to PMMA, it shows some advantages as well as disadvantages as illustrated in
SU-8 Liquid system, spinning or casting easy Chemically amlipfied, highly sensitive Narrow process window Excellent thermal stability Difficult to remove Lithographical performance unclear PMMA Solid system, casting difficult, gluing necessary Non amplified, low sensitivity Robust process Limited thermal stability Easy to remove Excellent lithographical performance

Tab. 2.
SU-8 Liquid system, spinning or casting easy Chemically amlipfied, highly sensitive Narrow process window Excellent thermal stability Difficult to remove Lithographical performance unclear PMMA Solid system, casting difficult, gluing necessary Non amplified, low sensitivity Robust process Limited thermal stability Easy to remove Excellent lithographical performance

4
45

Tab. 2 – Comparison of SU-8 and PMMA with respect to the Direct LIGA process.

One of the main problems of SU-8 resist is its high sensitivity, which at the same time is its main advantage over PMMA. As ilustrated in A summarizing overview gives the
Threshold dose Contrast Exposure dose - 1100 µm Exposure time – 4 “ wafer SU-8 < 0.75 J/cm3 1.4 80 mAmin/cm 10 min. PMMA < 250 J/cm3 1.2 1600 mAmin/cm 3,5 hrs

Tab. 3. As can be seen, the exposure time is about 20 times shorter in case of SU8.
SU-8 < 0.75 J/cm3 1.4 80 mAmin/cm 10 min. PMMA < 250 J/cm3 1.2 1600 mAmin/cm 3,5 hrs

Threshold dose Contrast Exposure dose - 1100 µm Exposure time – 4 “ wafer

Tab. 3 – Comparison of SU-8 and PMMA with regards to sensitivity and exposure time (calculated for BESSY synchrotron and with respect to typical process conditions)

Fig. 4, the threshold dose of SU-8 is in the order of 300 times smaller than that of PMMA, making the resist much more sensitive but at the same time making the process window in case of SU-8 much narrower than that of PMMA. Further, by comparing the slope of the curves, one can see that both resists show about the same contrast. A summarizing overview gives the
Threshold dose Contrast Exposure dose - 1100 µm Exposure time – 4 “ wafer SU-8 < 0.75 J/cm3 1.4 80 mAmin/cm 10 min. PMMA < 250 J/cm3 1.2 1600 mAmin/cm 3,5 hrs

Tab. 3. As can be seen, the exposure time is about 20 times shorter in case of SU8. 5
46

Threshold dose Contrast Exposure dose - 1100 µm Exposure time – 4 “ wafer

SU-8 < 0.75 J/cm3 1.4 80 mAmin/cm 10 min.

PMMA < 250 J/cm3 1.2 1600 mAmin/cm 3,5 hrs

Tab. 3 – Comparison of SU-8 and PMMA with regards to sensitivity and exposure time (calculated for BESSY synchrotron and with respect to typical process conditions)

Fig. 4 – Gradation curves for SU-8 (left) and PMMA (right); 1 mm resist thickness, standard processing conditions

Experimental The study consisted of several steps. Starting with the insufficient structure quality produced using ‘standard’ SU-8 process, identifying the most significant SU-8 process parameters was the first task in order to be able to start the optimization of the SU8 process. Inspection of fabricated gear wheels and comparison to their PMMA-based counterparts followed. For SU-8 process, 4 “ Si wafers with Ti/Au plating base were used as a substrate. Standard, commercially available SU-8 resist as well as modified SU-8 resist were used in this study. Commercially available resist was SU-8 100 obtained from micro resist technology, chemically modified resist consisted of standard SU-8 epoxy with varying amount of photo active component (PAC). The content of PAC varied from 2 to 6%. The SU-8 resist layers was prepared by casting of well defined amount of liquid resist on the substrate and baking it on leveled hot plate. By precise control of the process, resist layers of 1100 µm thickness with variation of less than 0.25 % in thickness could be repeatable prepared [4]. By varying the softbake temperature and time between 105 ºC and 130 ºC and 8 to 40 hrs respectively, residual solvent content as an amount of remaining solvent in the resist film after softbake was varied between 2% and 10%. By adjusting the softbake regimes, some variation in the solvent content distribution was also achieved [5]. X-ray exposures were realized at BESSY dipole lithographic beamline DIP 06.1a, featuring an Jenoptik DEX02 scanner. BESSY ring is operated at 1.7 GeV, the 6
47

radius of the dipole magnet of 4.359 m and magnetic field of 1.35 T result in critical energy of 2.5 keV or critical wavelength of 0.5 nm respectively. 200 µm thick Be window is used to separate the high vacuum region from the scanner. Typically, the ring is operated with 150-250 mA. During the experimental study, exposures of SU-8 were performed using bottom dose ranging from 10 to 40 J/cm3 using various kinds of filtration of spectra. A special testing mask for the process optimization was fabricated at BESSY. The mask consisted of 180 µm thick graphite substrate carrying 60 µm gold absorber structures. The mask carried 10 identical fields allowing multiple exposures with different conditions on single wafer. For the fabrication of metallic gears, X-ray masks fabricated at CAMD [6] consisting of 550 µm thick Be substrate and 50 µm thick gold absorber were used. Post exposure bake was realized on the hot plate using temperatures from 60 ºC – 95 ºC and times from 20 min to 60 min. In case of PMMA process, 4 “ Si substrates with oxidized Ti as a plating base were used. High molecular, 2 mm thick PMMA sheets obtained from Goodfellow were glued to the substrate, using a method as described in [7] and fly-cut to the final thickness of 1050 µm. X-ray exposures were realized at BESSY wavelength shifter lithographic beamline ID06, featuring an Jenoptik DEX01 scanner. Critical energy of the spectra is 7.69 keV corresponding to 0.16 nm critical wavelength. A 200 µm thick Be window is used to separate the high vacuum region from the scanner. X-ray masks fabricated at CAMD consisting of 550 µm thick Be substrate and 50 µm thick gold absorber were used. Bottom dose of 4.0 kJ/cm3 was used. In order to adjust the spectra, 160 µm thick graphite filter was used. Electroforming was in both cases performed by Micromotiong GmbH using commercially available Kissler plating bench and sulphite based Ni-Fe plating solution. Since the line width control and surface roughness could be well controlled as a result of a long lasting and intensive work concentrating on the fabrication of highly precise, high contrast Be based X-ray masks, the inspection concentrated on the remaining aspects which have significant impact on the quality of fabricated gear wheels. These are as our experience show the sidewall bow and the sidewall tilt. In order to evaluate the sidewall bow, white light interferometery using Wyko NT 1000 was used at group of professor Neyer, University of Dortmund [8]. In order to evaluate the tilt of the gear wheel sidewalls, video based contact angle measurement method as proposed in [9] was used. In order to evaluate the occurrence of top scum, resist residue or structure collapse during the DoE study (see results), LEO 1560 scanning electron microscope was used. Results The unacceptably large deviation as shown in Fehler! Ungültiger Eigenverweis auf Textmarke. was the starting point of the SU-8 process optimization. Because of large number of processing parameters in case of SU-8, statistical design of experiment was used at first in order to limit the optimization to the most significant parameters only. As described above, amount of PAC, remaining solvent content, post exposure bake temperature, post exposure time and bottom dose were varied. As responses, 7
48

the top scum, also called skin effect, amount of observable residue after development of resist, collapse of the structures and sidewall bow were considered as responses of the variation of process conditions. Since it would take too long time to perform electroplating on each of substrates in order to evaluate shows the typical sidewall bow of gear wheel fabricated using initial SU-8 and PMMA fabrication process respectively. As can be seen, the amplitude of the bow in case of SU-8 approaches 5 µm, which is an unacceptably large dimensional deviation and at the end would even disable the assembly of the gears. The unacceptably large deviation as shown in Fehler! Ungültiger Eigenverweis auf Textmarke. was the starting point of the SU-8 process optimization. Because of large number of processing parameters in case of SU-8, statistical design of experiment was used at first in order to limit the optimization to the most significant parameters only. As described above, amount of PAC, remaining solvent content, post exposure bake temperature, post exposure time and bottom dose were varied. As responses, the top scum, also called skin effect, amount of observable residue after development of resist, collapse of the structures and sidewall bow were considered as responses of the variation of process conditions. Since it would take too long time to perform electroplating on each of substrates in order to evaluate the metallic structures, the evaluation was limited to resist forms only. As a result of the study, most significant parameters were identified.

Fig. 5 – Sidewall bow on flex-splines fabricated using SU-8 and PMMA fabrication process.
Top Scum Residue 0.85 0.95 ** *** *** *** *** *** NS *** * *** ** ** *** *** ** * Structure Collapse 0.5 * *** *** NS NS * ** NS Sidewall profile 0.69 NS *** *** NS ** NS ** **

Correlation Coefficient PAC content Solvent content PEB Temperature PEB Time Bottom dose Solvent content x Bottom dose Solvent content x PEB Temp. PAC content x Solvent content Tab. 4

shows a brief summary of each independent factor and several interactions.

8
49

Fig. 5 – Sidewall bow on flex-splines fabricated using SU-8 and PMMA fabrication process. Top Scum Residue 0.85 0.95 ** *** *** *** *** *** NS *** * *** ** ** *** *** ** * Structure Collapse 0.5 * *** *** NS NS * ** NS Sidewall profile 0.69 NS *** *** NS ** NS ** **

Correlation Coefficient PAC content Solvent content PEB Temperature PEB Time Bottom dose Solvent content x Bottom dose Solvent content x PEB Temp. PAC content x Solvent content

Tab. 4 – Significance factors for various processing parameters; * ~ 90 % confidence interval for significance, ** ~ 95 % confidence interval for significance, *** ~ 99 % confidence interval for significance, NS – not significant

Although the predictive model correlation for structural collapse and sidewall bow were low, two independent variables were found to be highly significant, namely the solvent content, PEB temperature and bottom dose. As indicated from the results, solvent content of 6 % and PEB temperature of 60 ºC should produce the least structure collapse and lowest sidewall bow. Using the results of the previous study, by further varying the significant parameters, SU-8 process was optimized with respect to the sidewall bow. Each point of the graph represents the average value of the amplitude of all the measurements at metallic flex-splines produced with the same conditions. The error bars represent the maximum and minimum of all measured values. The left part of the chart in Fehler! Ungültiger Eigenverweis auf Textmarke. shows the best and the worst corresponding values found on metallic flex-splines fabricated using PMMA process as described above. In both of these two cases, the assembled gear wheels were fully functional and exceeded their required life time, so that for this study, these two cases represented the working range considering the sidewall bow.

9
50

Fig. 6 shows the evolution of the sidewall profile along the time scale of SU-8 process optimization. Each point of the graph represents the average value of the amplitude of all the measurements at metallic flex-splines produced with the same conditions. The error bars represent the maximum and minimum of all measured values. The left part of the chart in Fehler! Ungültiger Eigenverweis auf Textmarke. shows the best and the worst corresponding values found on metallic flex-splines fabricated using PMMA process as described above. In both of these two cases, the assembled gear wheels were fully functional and exceeded their required life time, so that for this study, these two cases represented the working range considering the sidewall bow.

Fig. 6 – Progress of sidewall bow along the process of optimization (right); PMMA reference measurements (left); working range highlighted

As can be seen, the initial results in case of SU-8 process were fully nonsatisfying with amplitudes of sidewall bow approaching 12 µm! As the process of SU-8 optimization was carried forward, both the average values and the variation decreased, so that at the end, the amplitudes of bow of SU-8 based fabricated flexsplines reached those of flex-splines fabricated using PMMA process. 10
51

As predicted before, the most significant parameter was the solvent content together with the post exposure bake temperature and the bottom dose. It was also found, that the distribution of solvent content in the resist film plays an important role and some specific softbake procedures yielded better results.

Fig. 7 – Sidewall bow on flex-splines fabricated using optimized SU-8 and PMMA fabrication process; significant improvement in comparison with initial non-optimized SU-8 process; tolerance range of ±1 µm highlighted;

Fig. 7 depicts the typical sidewall bow of a metallic flex-spline fabricated using an optimized SU-8 process in comparison to the corresponding sidewall bow in case of non-optimized SU-8 process and the PMMA process respectively. The significant improvement in case of optimized SU-8 process can be clearly observed. The overall sidewall profile consists both of the sidewall bow and the sidewall tilt. Using white light interferometer, only the bow can be measured. Using the contact angle measurement method, sidewall tilt was estimated by several flex-splines fabricated using the optimized SU-8 process and compared to sidewall tilt of flexsplines fabricated using PMMA process. The results of this measurement are shown in

11
52

Fig. 8. Square mean deviation of the sidewall tilt was estimated to 3.5 Deg in case of the optimized SU-8 process and 3.9 Deg in case of PMMA process.

Fig. 8 – Sidewall tilt measured on metallic flex-splines fabricated using optimized SU-8 process (left) and using standard PMMA process (right)

Comparing the square mean values, one can see that the sidewall tilt is comparable. The overall large value of the tilt of approximately 4 Deg is believed to result from the electroforming process and is currently under investigation. Is to be said though, that same as in the case of sidewall bow, the investigated flexsplines fabricated using PMMA process were all well working and acceptable parts, so that these high values of tilt obviously do not disturb the application. Summary SU-8 process for fabrication of precise metallic gear wheels with heights of 1 mm, smallest feature size of 25 µm, aspect ratios exceeding 40 and overall very tight tolerances was established. Starting from a non-optimized SU-8 process, by careful design of the experiment, significant process parameters for process optimization were identified. Remaining solvent concentration, post exposure bake temperature and bottom dose were identified among the most significant parameters. By varying significant parameters, SU-8 process was optimized with respect to the sidewall bow. Among optimal conditions, 6 % of remaining solvent content and post bake temperature of 60 ºC were found. Using the optimized SU-8 process, the sidewall bow of the electroplated gear wheels was found comparable to the one of electroplated gear wheels fabricated using PMMA process. Further, the sidewall tilt of metallic gear wheels fabricated using optimized SU-8 process 12
53

was found comparable to the one of gear wheels fabricated using PMMA process. The results are summarized in
Critical Dimensions Sidewall bow on metallic parts Sidewall tilt of metallic parts PMMA process (Reference) Optimized SU-8 process 1100 µm high, 25 µm wide, aspect ratio 44 < 4 µm < 4 µm < 4 Deg < 3.5 Deg

Tab. 5.
PMMA process (Reference) Optimized SU-8 process 1100 µm high, 25 µm wide, aspect ratio 44 < 4 µm < 4 µm < 4 Deg < 3.5 Deg

Critical Dimensions Sidewall bow on metallic parts Sidewall tilt of metallic parts

Tab. 5 – Table summarizing the main results of the study

Concluding, the main task of the study was achieved and SU-8 was proven to be well suitable for presented application. The authors of this study see further potential in the optimization of the whole fabrication process, both on the resist processing side as well as on the electroplating side, so that further studies and optimization processes are ongoing. Acknowledgements Authors would like to thank to all participants and contributors to this work, especially to co-workers from BESSY GmbH, micro resist technology GmbH, Micromotion GmbH, CAMD and University of Dortmund. This research was partially supported by the initiative of the Federal Ministry of Education (BMBF), Berlin. References
1. W. Menz, J.Mohr (1997) Mikrosystemtechnik fuer Ingenieure, VCH Verlaggesselschaft mbH, Weinheim 2. R. Degen, R. Slatter (2004) Zero Backlash Micro-Gears and Actuators for Microassembly Applications, www.harmonicdrive.de/de/pdf/fachauf_18.pdf 3. H.-U. Scheunemann, B. Loechel, L. Jian, D. Schondelmaier, J. Goettert, Y. M. Desta, Z.-G. Ling, T. Morris, J. Joonyoung, V. Singh, R. Degen, U. Kirsch, G. Gruetzner, R. Ruhmann, G. Ahrens, Providing a Direct-LIGA Service – A Status Report, Book of Abstracts HARMST 2005, Monterey, CA, pp. 281-282 4. G. Grützner, R. Ruhmann, G. Ahrens, M. Bednarzik, B. Loechel, P. Limbecker, J. Goettert, Y. Desta, V. Singh (2003), Improved process parameters and impact on structure quality of high aspect ratio SU-8 micro-strustructures, Book of Abstracts HARMST 2005, Monterey, CA, pp. 11-12 5. R. Ruhmann, G. Ahrens, G. Grützner, N. A.-Staufenbiel, H. Schoeder (2005), Residual solvent content distribution in ultra-thick SU-8 films and its influence on the imagin quality, HARMST 2005, in press 6. http//www.camd.lsu.edu 7. S. Achenbach (2000), Optimierung zur Herstellung von Mikrostrukturen durch Ultratiefe Röntgenlithographie (UDXRL), PhD Thesis, University Karlsruhe 8. http://www-mst.e-technik.uni-dortmund.de/

13
54

9. R. Engelke, G. Ahrens, R. Ruhmann, S. Kopetz, J. Kastner, K. Wiesauer, D. Stifter, B. Loechel, A. Neyer, G. Gruetzner (2005), Possibilities of Inline Process Inspection of High Aspect ratio LIGA Micro Structures, HARMST 2005, in press

14
55

1

THE PHOTOABSORPTION AND PHOTOIONIZATION OF HALOGENATED ETHYLENES: BROMINATED DERIVATIVES. R. Locht1, B. Leyh1, H.W. Jochims2, H. Baumgärtel2. 1 Molecular Dynamics Laboratory, Department of Chemistry, Institute of Chemistry, University of Liège, Sart-Tilman par B-4000 Liège 1, Belgium. 2Institut für Physikalische und Theoretische Chemie, Freie Universität Berlin, Takustrasse 3, D-14195 Berlin, Germany.

In our program on the dynamics of highly excited states of the halogenated derivatives of ethylene, particular attention has been paid to the brominated derivatives [1]. The main motivation is not only their importance in chemical and technological applications but also the scarcity, or even the lack of spectroscopic data related to these compounds. In 2000, a thorough investigation of C2H3Br has been presented. The vacuum UV photoabsorption spectrum [2] as well as the threshold and He(I)photoelectron spectrum [3] have been published. The CIS-spectra of the first six electronic states were measured [3]. The unimolecular photodissociation dynamics of + in the C H ++Br channel has C2H3Br 2 3 been studied by the maximum entropy (MEM) method [4]. Finally, the ion-pair formation and dissociative photoionization of this molecule have been investigated by mass spectrometry and TPEPICO techniques [5]: the C2H3++Br¯, C2H3++Br, C2H2++[H,Br] and Br++[C2, H2] channels were considered. This research program has been continued by recording the photoabsorption spectrum and the dissociative photoionization mass spectrometry of brominated deriv-

Figure 1.

56

2 atives of ethylene, e.g. C2HBrF2 and C2BrF3 in the 5-20 eV photon energy range. The present work has been performed on the 3m-NIM-2 beamline equipped with a 600 ℓ/mm Al grating and entrance and exit slits of 30 µm and 50 µm widths respectively. In addition, the photoelectron spectrum and the kinetic energy distributions of the fragment ions observed at the energy of the HeI (21.2 eV), NeI (16.67-16.85 eV) and ArII (13.47 eV) resonance lines have been recorded in the laboratory. Fig. 1 shows an example of photoabsorption spectra obtained respectively for (1) C2HBrF2, (2) C2BrF3 and (3) C2ClF3 in the 6-12 eV photon energy range. The spectra related to the latter two compounds clearly differ from the former by the π→π* transition. This transition, located at about 6.8 eV for C2HBrF2, is blue-shifted by about 0.8 eV when the H atom is substituted by F. By contrast, the substitution of the Br atom by Cl has no significant influence on either the shape or the energy position of the spectral features. Furthermore, the H-substitution Figure 2. leads to a severe quenching of the Br (or Cl) lone pair excitation (9-11 eV photon energy range). A deeper analysis of these spectra, combined with all our previous results on about more than ten halogenated ethylene derivatives, is in progress. The dissociative photoionization of C2HBrF2 and C2BrF3 has been recorded by mass spectrometry at variable wavelength. Almost all fragment ions present in the 20 eV photon energy mass spectrum have been considered. Fig. 2 shows a typical example of results obtained by HeI photoelectron spectroscopy and mass spectrometric photoionization. The photoionization efficiency curve (PIC) of the molecular ion C2HBrF2+ and its first derivative (dI+/dE) as observed

57

3 in the threshold region are displayed in fig. 2 (bottom). This figure clearly shows the complex pattern of autoionization and direct ionization. The best way to disentangle these data is to record both the HeI photoelectron spectrum (PES) and the photoabsorption spectrum (PAS) shown in fig. 1. The latter clearly shows that the ionization threshold region is dominated by Br-lone pair Rydberg transitions. Fig. 2 (upper part) shows the structure for the first HeI PES-band. Several features present in this spectrum correlate with those observed in the PIC by resonant photoionization and are ascribed to direct ionizing transitions. A systematic study is in progress.

Acknowledgments. The authors gratefully acknowledge the Freie Universität Berlin. R.L. and B.L. are indebted to the Belgian "Fonds National de la Recherche Scientifique" (FNRS) for financial support. The authors wish to thank the BESSY staff, and particularly Dr. G. Reichardt and Dr. W. Braun, for their essential collaboration. References. [1]. R. Locht, B. Leyh, H.W. Jochims, H. Baumgärtel, Jahresbericht (2003), 48. [2]. A. Hoxha, R. Locht, B. Leyh, D. Dehareng, K. Hottmann, H.W. Jochims, H. Baumgärtel, Chem.Phys. 260 (2000) 237. [3]. A. Hoxha, R. Locht, B. Leyh, D. Dehareng, K. Hottmann, H. Baumgärtel, Chem.Phys. 256 (2000) 239. [4]. A. Hoxha, R.Locht, A.J. Lorquet, J.C. Lorquet, B. Leyh, J.Chem.Phys. 111 (1999) 9256. [5]. A. Hoxha, B. Leyh, R. Locht, H.W. Jochims, M. Malow, K.-M. Weitzel, H. Baumgärtel, to be published.

58

Photophysical characteristics of a series of donoracceptor-polyenes of different length
Hani El-Gezawy, Wolfgang Rettig
Institut für Chemie, Humboldt-Universtät zu Berlin, Brook-Taylor-Str. 2, D-12489 Berlin, Germany

Abstract:
The photophysical behaviour of a series of donor-acceptor-polyenes of different length has been investigated: 4-dimethylamino-4`-cyanostilbene (DCS), 4-dimethylamino-4`-cyano1,4-diphenylbutadiene (DCB), 4-dimethylamino-4`-cyano-1,4-diphenylhexatriene (DCH). These compounds show a maximum of the fluorescence yield in the most polar solvents. The three dyes show a pronounced solvatochromic behaviour only for the emission spectra whereas the absorption spectra show a very weak solvatochromic shift which correlates better with the polarizability of the solvents. The nonradiative decay decreases sizeably in the order DCS>DCB>DCH.

Introduction:
For several decades, the photophysical, photochemical and also solvatochromic properties of donor-acceptor-stilbenes have attracted interest. The excited state behavior of trans-stilbene is one of the best-known examples of an adiabatic photoreaction: it reacts to a twisted conformation, which is often referred to as the “phantom-singlet state”. This conformation corresponds to a maximum on the ground state surface to which it is strongly coupled, and this pathway thus provides an effective deactivation funnel [1]. The main focus of this project is to study the photophysical properties of a structurally related series of molecules in which the donor (dimethylamino) and acceptor (cyano) substituents are separated by trans-stilbene (DCS), all-trans-diphenyl-butadiene (DCB) and all-trans-diphenylhexatriene (DCH). The experimental of UV absorption, steady state emission, time-resolved fluorescence and low temperature techniques are employed here in conjunction with quantum chemical calculation.

The Investigated Compounds:
NC N CH3 DCS DCB N CH3 DCH N CH3 CH3 NC CH3 CH3 NC

Temperature dependence of the fluorescence spectra and nonradiative losses:
Fig. 1 shows the temperature dependent emission spectra and the room temperature absorption spectrum of DCS, DCB and DCH in diethyl ether. All compounds show a similar behaviour. As can be seen in Fig. 1, with decreasing temperature the fluorescence intensity increases, and a weak red shift of the fluorescence maximum is observed, explainable by the increasing dielectric constant ε of the solvent at low temperatures. Table 1 contains the fluorescence decay times and the derived temperature dependent nonradiative decay rates knr as extracted from the fluorescence intensities. It can be seen that knr decreases with decreasing temperature but also with a lengthening of the π-electronic
59

system. The former is indicative of an activation barrier leading to the phantom singlet state, the latter indicates that this activation barrier increases for longer diphenylpolyenes.
4 2

DCS

abs.

flu o .
at at at at 25 °C 0°C -2 0 ° C -4 0 ° C at at at at -6 0 ° C -8 0 ° C -1 0 0 ° C -1 2 0 ° C

abs./fluo. (a. u.)

0 6 4 2 0 6 4 2 0 200

DCB

DCH

300

400

500

600

700

800

W a v e le n g th ( n m )

Fig. 1 Absorption spectrum at room temperature and fluorescence emission spectra of DCS, DCB and DCH at different temperatures in diethyl ether. Table 1. Photophysical parameters of DCS, DCB and DCH in diethyl ether at low temperatures kf (s-1) Compound T (K) knr (s-1) Φf τf (ns) λf (nm) 298 475 0.032 0.10 DCS 9.68x109 3.20x108 253 482 0.098 2.95x109 213 487 0.42 4.42x108 173 490 0.68 1.51x108 153 492 0.84 6.09x107 77 1.7 298 509 0.085 0.32 DCB 2.62x108 2.82x109 253 525 0.26 7.38x108 213 538 0.51 2.56x108 173 548 0.79 6.88x107 153 550 0.92 2.19x107 77 1.65 DCH 298 253 213 173 153 77 567 589 595 613 617 0.091 0.28 0.54 0.81 0.95 0.47 1.94x108 1.94x109 4.99x108 1.65x108 4.55x107 1.02x107

1.7

References:
[1] D.H. Waldeck, Chem. Rev. 91 (1991) 415; J. Saltiel and Y.-P. Sun, in: Photochromism - molecules and systems, Eds. H. Dürr and H. Bouas-Laurent, Elsevier, Amsterdam, 1990, p. 64.

60

Investigation of photoionization processes in the 3d-transition metal (TM) compounds FeCl2 , FeBr2 and CoCl2
Mohamed Al-Hada, Tobias Richter and Peter Zimmermann Institut f¨r Atomare Physik und Fachdidaktik, Technische Universit¨t Berlin u a Michael Martins Institut f¨r Experimentalphysik, Universit¨t Hamburg u a ¨ Omer G¨rg¨l¨er and Ali Tutay o uu Science Faculty, Istanbul University

TM compounds FeCl2 , FeBr2 and CoCl2 are the systems that have been selected for this investigation. They are important in many different areas of technology, for examples, the magnetic properties of the thin films that have lead to significant improvement in the storage capability of the magnetic devices during the last decade. To understand the physics of such complex systems, deep analysis of the electronic and magnetic structure is needed. The key of the magnetic and electronic properties of these compound system are the 3d electrons. Analysis of these systems can be accomplished by exciting 3p electrons into the 3d shell or by a direct excitation of the 3p electrons. Photoion and photoelectron spectra of the 3d- TM compounds FeCl2 , FeBr2 and CoCl2 were studied by photoionization spectroscopy using molecular beam technique by thermal evaporation of the metal compounds at a temperature of approx. 700 K and monochromatized synchrotron radiation at the BESSY II beamline U125/2-SGM. A time-of-flight spectrometer in pulsed voltage mode has been used to detect partial ion yield spectra of the produced ions in the excitation energy range (50-80) eV. The photoelectrons are created by photons of 100 eV and analyzed using a hemispherical electron spectrometer (Scienta SES-2002).
storage ring

undulator

monochromator

eelectron analyzer ions atomic beam

polarization axis

Figure 1: Experimental Setup

time-of-flight spectrometer

Experimental results of in singly charged ions of Fe, FeCl2 , FeBr2 in the 3p (Fe) excitation region and Co, CoCl2 in the 3p (Co) excitation region are shown in the Fig 2. 3p photoion spectra of FeCl2 , FeBr2 and CoCl2 show a close similarity when compared to the corresponding atomic spectra of Fe and Co. The atomic spectra are dominated by strong resonances. These are due to discrete transitions 3p6 3dN − 3p5 3dN +1 with a subsequent emission → of a 3d electron 3p5 3dN +1 − 3p6 3dN −1 ι. This is caused by the large overlap of the 3p and 3d → wavefunction. For the photoelectron spectra of FeCl2 , FeBr2 and CoCl2 the validity of the central field at least for the 3d valence electrons is not existing as a reason of the molecular binding. A charge transfer (CT) model has been applied successfully to the 2p photoelectron spectrum of the FeCl2 [2] Fig

61

50

55

60

65

70

Fe

FeCl2

Intensity (arb. units)

FeBr2

Co

Figure 2: Partial ion yield spectra of atomic Fe [1] with corresponding molecular FeCl2 , FeBr2 and the atomic Co [1] with the corresponding molecular CoCl2 in the region of the 3p (Fe) and (Co) respectively

CoCl2

55

60

65

70

75

80

Photon energy (eV)
3 (left panel). This also results in good agreement with the corresponding solid spectrum. The 3p photoelectron spectrum shows a similarity with the solid 3p FeCl2 , thus the interpretation with the charge transfer model by introducing an additional 3dN +1 configuration will be the key of analyzing the 3p photoelectron spectrum.
75

Binding energy (eV) 70 I 65

Binding energy (eV)
60 750 740 730 720 710

3p
Intensity (arb. units)
FeCl2

2p
FeCl2

Calculated data using (CT) model

Solid FeCl2

Figure 3: Experimental 3p FeCl2 taken at a photon energy of 100 eV with the corresponding solid spectrum [3](right panel) and 2p photoelectron spectrum taken at photon energy of 800 eV with calculated data using a CT model.

[1] H. Feist, M. Feldt, Ch. Gerth, M. Martins, P. Zimmermann Phys. Rev. A 53 760 (1995) [2] T. Richter, K. Godehusen, T. Wolf, M. Martins, P. Zimmermann Phys. Rev. Lett. 93 023002 (2004) [3] M. Okusawa. Phys. stat. sol. (b) 124 673 (1984)

62

Predissociation of the N+ C 2 Σ+ state observed via 2 u 2 + 2 + C Σu → X Σg fluorescence after resonant 1s−1 π ∗ excitation of N2
L Werner1 , S Klumpp1 , A Ehresmann1 , Ph V Demekhin2,4 , M P Lemeshko2 , V L Sukhorukov2,4 , K-H Schartner3 and H Schmoranzer4
1 2 3 4

Institut für Physik, Universität Kassel D-34132 Kassel, Germany Rostov State University of Transport Communications, 344038 Rostov-on-Don, Russia I Physikalisches Institut, Justus-Liebig-Universität, D-35392 Giessen, Germany Fachbereich Physik, Technische Universität Kaiserslautern, D-67653 Kaiserslautern, Germany

E-mail: ehresmann@physik.uni-kassel.de

In our recent studies we investigated the vibrationally selective inner-shell photoexcitation of the ∗ N2 (1s−1 πg )(v )-resonances and its subsequent autoionization into different levels of the N+ C 2 Σ+ u 2 electronic state via the photon-induced dispersed fluorescence spectroscopy (PIFS) (Ehresmann et al 2006). One peculiarity of the C(v ′ ) → X(v ′′ ) emission bands with v ′ ≥ 3 is that they are very weak compared to bands with v ′ = 0, 1, 2. This behaviour was first observed by (Carroll 1959). A first explanation was given by (Carroll 1959) with the assumption of (Douglas 1952) that a weak predissociation occurs for the vibrational levels of the C(v ′ ) state with v ′ ≥ 3. The pathway of this indirect predissociation was clarified by (Lorquet and Lorquet 1974) and supported by computing the respective potential curves (Langhoff et al 1988). Extended calculations performed in (Ehresmann et al 2006) predicted the maximum of the fluorescence intensity for the vibrational sequence with ∆v = −5 if this indirect predissociation is taken into account and for ∆v = −6 if the indirect predissociation is neglected delivering, therefore, an experimental means to unequivocally decide whether the indirect predissocation works here or not. Unfortunately, the vibrational sequence with ∆v = −6 was outside the sensitivity range of the detector used in (Ehresmann et al 2006). The main goal of the present work was therefore to cover the fluorescence wavelength range of vibrational sequences between at least ∆v ≤ −2, . . . , −8. The experimental setup in the present work is essentially the same as in (Ehresmann et al 2006). The gold coated 600 lines/mm grating of the undulator beamline U49/2 PGM 1 at BESSY II was used to monochromatize the synchrotron radiation which was then focused into a differentially pumped target cell filled with molecular nitrogen at room temperature and at a pressure of 27.6 µbar. The exciting-photon energy was varied from 400 eV to 403 eV in steps of 20 meV with a bandwidth of about 70 meV FWHM at 400 eV, determined mainly by a reasonable signal-to-noise ratio for the observed fluorescence intensities. The fluorescence was dispersed by a 1 m-normal-incidence monochromator (McPherson 225) equipped with a 1200 lines/mm grating (blaze wavelength 150nm; coating Al) and recorded in the fluorescence range between 165 nm and 208 nm by a position-sensitive CsTe microchannel-plate detector (manufactored by Quantar Technology Inc.). The exciting photon energy was calibrated to known (Chen et al 1989) energy ∗ positions of the 1s−1 πg (vr ) levels. The resolution of this ’monochromator-detector’ combination was about
up to

1.5

0,8 0,6 0,4 0,2 0,0

-3

-4

-5

-6

-7

-8

12

3 4 5

6

7 8

9

10

11

12

Exciting-photon energy [eV]

402,0 401,6 401,2

a)
v=4 v=3 v=2 v=1

[arb. un.]

-2

Sum of intensities

N

+

2

C(v')

X(v''),

v = v'

- v''

b)

1,0

c)

400,8 400,4 165 170 175 180 185 190 195 200 205

v=0

1s

-1

*(v )
r

Fluorescence wavelength [nm]

Figure 1. a) Dispersed fluorescence yield as a function of the exciting-photon energy. b) Fluorescence intensities integrated over the exciting-photon energies between 400 and 403 eV. Computed positions of the N+ (C − X) vibronic bands are shown by vertical bars. Numbering of lines and their assignments 2 as listed in table 1. c) Fluorescence intensities integrated over all recorded fluorescence wavelengths.

63

Table 1. Assignment of the observed fluorescence lines after resonant 1s−1 π ∗ excitation of N2 . The line numbering is the same as in figure 1b. No 1 2 3 4 5 6 7 8 9 10 11 12 ∆λexp [nm] 165.98−116.86 167.57−167.59 171.59−172.97 174.03−174.32 174.27−174.53 177.82−179.49 184.23−186.69 185.79−186.26 188.50−188.52 188.74 190.51−192.98 197.78−200.31
a)

∆λcalc [nm] 165.9−166.2 171.8−171.9

b)

Parts N+ 2 NII N+ 2 NII NI N+ 2 N+ 2 NII NIII NII N+ 2 N+ 2 N+ 2

Transition C 2 Σ+ , v ′ u 2s2 2p1 (2 P)3p 3 SJ ′ C 2 Σ+ , v ′ u 2s2 2p1 (2 P)3p 3 DJ ′ 2s2 2p2 (3 P)3s 2 PJ ′ C 2 Σ+ , v ′ u C 2 Σ+ , v ′ u 2s2 2p1 (2 P)4p 3 DJ ′ 2s2 (1 S)4f 2 FJ ′ 2s2 2p1 (2 P)4p 1 PJ ′ C 2 Σ+ , v ′ u C 2 Σ+ , v ′ u C 2 Σ+ , v ′ u → → → → → → → → → → → → → X 2 Σ+ , v ′′ g 2s1 2p3 3 PJ ′′ X 2 Σ+ , v ′′ g 2s1 2p3 3 PJ ′′ 2s2 2p3 2 PJ ′′ X 2 Σ+ , v ′′ g X 2 Σ+ , v ′′ g 2s2 2p1 (2 P)3s 3 PJ ′′ 2s2 (1 S)3d 2 DJ ′′ 2s2 2p1 (2 P)3s 1 PJ ′′ X 2 Σ+ , v ′′ g X 2 Σ+ , v ′′ g X 2 Σ+ , v ′′ g

v ′ − v ′′ −2 −3

177.7−178.1 184.0−184.8

−4 −5

190.5−191.8 197.5−199.2 204.8−207.0

−6 −7 −8

a) Wavelengths for the transitions in N+ - present measurement (see figure 1); the NI, NII and NIII 2 fluorescence wavelength intervals represent the fine structure components (Kelly 1987) not resolved in present measurement; b) Present calculations (only v ′ = 0 . . . 3 vibrational levels are taken into account due to the indirect predissociation);

∆λf l = 0.1 nm which is 3 times better than in (Ehresmann et al 2006). A two-dimensional fluorescence yield spectrum in the exciting-photon energy range around the 1s−1 π ∗ resonance and the fluorescence wavelength range between 165 and 208 nm is displayed in figure 1. Figure 1a shows the 2D- plot recorded by the CsTe detector normalized for the exciting-photon flux. Assignments of the observed fluorescence lines are listed in table 1 where the numbering of the lines corresponds to figure 1b. Wavelengths for the NI, NII and NIII transitions have been assigned by comparing the experimentally determined wavelengths to known spectroscopic data (Kelly 1987). The C (v ′ ) → X (v ′′ ) fluorescence bands observed in the present fluorescence wavelength range are strongly overlapping with a weak fluorescence emission which starts at approximately 165 nm and can be seen as a background in figure 1. Since this emission was not resolved it was tentatively connected with the N+ D 2 Πg → A 2 Πu Janin-d’Incan bands when 2 first observed by (Holland and Maier 1971). An exploratory calculation performed in the present work showed that it stems from strongly overlapped ro-vibronic lines of weak D 2 Πg → A 2 Πu transitions and more intense (2) 2 Πg → A 2 Πu transitions in N+ . In the present work we corrected the observed 2 fluorescence intensities by a constant background and by the quantum efficiency of the CsTe detector when extracting information on the C → X emission from figure 1. Relative doubly integrated emission cross sections for the C (v ′ ) → X (v ′′ ) fluorescence sequences (∆v = v ′ − v ′′ = const), σC (∆v), have ¯X been determined similar to (Ehresmann et al 2006) by integrating the fluorescence intensities over the fluorescence wavelength interval ∆λexp (listed in table 1) and over the present exciting-photon energy range.
Table 2. Doubly integrated cross sections, σC (∆v), for band sequences (relative to ∆v = −5 in ¯X percent) observed in the C (v ′ ) → X (v ′′ ) fluorescence of N+ after N2 (1s−1 π*) resonant Auger decay 2 and calculated in different approximations.

Approximation Exper. a Exper. Present Theory c Theory d Theory Present
b

3

2 6

1

N+ (C 2 Σ+ , v ′ → X 2 Σ+ , v ′′ ), ∆v = v ′ − v ′′ u g 2 0 –1 –2 –3 –4 –5 –6 17 24 22 20 25 48 16(9) 33 35 38 59 52(7) 47 47 53 69 69(6) 65 70 83 100 100 100 100 100 64(5) 159 74 66

–7 19(2) 85 22 19

–8

e

5 3 1

7 5 5

10 9 10

15 15 14

4 1 1

a)

Data corrected for the quantum efficiency of the CsI detector, Paper I; Data corrected for the quantum efficiency of the CsTe detector; c) The fluorescence from all v ′ vibrational levels is taken into account neglecting the indirect predissociation of the v ′ ≥ 3 ones; data from table 5 of Paper I; d) The fluorescence from v ′ = 0, 1, 2, 3 vibrational levels only is taken into account due to the indirect predissociation of the v ′ ≥ 3, data from table 5 of Paper I; e) The indirect predissociation of the v ′ ≥ 3 levels is taken into account.
b)

64

5

4

3

2

1 0

400,5 1,2

401,0

401,5

402,0

191

192

193
N
+

0

1
2

2
N 1s

3
-1

4
r

5

65

74 8 3

2

1 0

2

C(v')

1,2
[arb. un.] Cross section

*(v )

[arb. un.]

0,8

v
0,4
app

=-6
app

v
=0.1 nm

=-6

0,8

0,4
=100 meV

v

0,0

0,0

0
0,3 0,2 0,1

1
2

2
N 1s

3
-1

4
r

5

64

7

38

2

1

0

C X

N

+

2

C(v')

Cross section

*(v )

0,3

v

=-7
app

v
=0.1 nm

=-7

0,2 0,1 0,0

app

=100 meV

R

P
200

0,0 400,5 401,0 401,5 402,0 198 199
Exciting-Photon Energy [eV]

Fluorescence wavelength [nm]

X Figure 2. Measured (open circles) and calculated (solid lines) cross sections σC (ω, ∆v) (left panels) X and σC (λ, ∆v) (right panels) for band sequences with ∆v = −6 and −7. Partial cross sections for individual C(v ′ ) → X(v ′′ ) bands (dashed lines) calculated with accounting for the APD and total X σC (λ, ∆v) calculated neglecting indirect predissociation (dash-dot lines) are also shown in the right panels. Typical rotational structure (represented by R and P branches) calculated in the present work ∗ for the C → X band is also shown. Experimental positions of 1s−1 πg (vr ) vibrational levels (Chen X et al 1989) are marked in the left panels. Computed σC (ω, ∆v) are convolved with a Gaussian of 100 X meV FWHM, and σC (λ, ∆v) - of 0.1 nm FWHM.

The σC (∆v) values normalized to the σC (∆v = −5) value are listed in the upper part of table 2 together ¯X ¯X with our previous experimental data from (Ehresmann et al 2006). The cross section σC (∆v = −8) ¯X can not be determined since the weak fluorescence from ∆v = −8 sequence is blended by unresolved emission stemming from (D, (2)) 2 Πg → A 2 Πu band systems of N+ . One can see that cross sections 2 σC (∆v = −3, −4, −5) obtained in the present work agree with the previous measurement within the ¯X error bars. The difference between ‘old’ and ‘new’ values of σC (∆v = −2) is, possibly, connected with a ¯X sharp drop of the quantum efficiency of the CsTe detector at the edge of its sensitivity range. Measured cross sections σC (∆v) exhibit a maximum at ∆v = −5 that agrees with the prediction of (Ehresmann ¯X et al 2006) based on the calculation accounting for the predissocation (see line ‘Theoryd ’). As one can see from figure 1b the improved resolution allows us to observe the individual C(v ′ ) → X(v ′′ ) bands X for ∆v = −6 and −7 sequences. In large scale the observed cross sections for fluorescence, σC (λ, ∆v), are depicted in figure 2 for sequences ∆v = −6 and −7 (right panels, open circles). In the same figure X the cross sections σC (ω, ∆v) are also shown (left panels, open circles). Experimental positions of the −1 ∗ 1s πg (vr ) vibrational levels (Chen et al 1989) are marked in the left panels. In order to illustrate X the influence of the indirect predissociation on the cross sections σC (λ, ∆v) we performed calculations ′ neglecting the predissociation of v ≥ 3. The resulting ‘model’ cross sections are shown in figure 2 by dash-dot lines, in contrast to the solid lines accounting for predissociation. This work was supported by the Deutsche Forschungsgemeinschaft (DFG) and by the Bundesministerium für Bildung und Forschung (BMBF) (Förderkennzeichen 05 ES3XBA/5 and IB/DLR RUS 02/037) References
Baltzer P, Larsson M, Karlsson L, Wannberg B and Carlsson Göthe M 1992 Phys. Rev. A 46, 5545-5553. Carroll P K 1959 Can. J. Phys. 37, 880-889. Chen C T, Ma Y and Sette F 1989 Phys. Rev. A 40, 6737-6740. Douglas A E 1952 Can. J. Phys. 30, 302-313. Ehresmann A, Werner L, Klumpp S, Lucht S, Schmoranzer H, Mickat S, Schill R, Schartner K-H, Demekhin P V, Lemeshko M P and Sukhorukov V L 2006 J. Phys. B: At. Mol. Phys. 39, 283-304. Holland R F and Maier II W B 1971 J. Chem. Phys. 55(3), 1299-1314. Huber K P and Herzberg G 1979 Molecular Spectra and Molecular Structure. IV Constants of Diatomic Molecules Van Nostrand Reinhold Comp. New York. Kelly R L 1987 J. Phys. Chem. Ref. Data 16 (Suppl.1), 1-649. Langhoff S R, Bauschlicher C W and Jr. 1988 J. Chem. Phys. 88(1), 329-336. Lorquet A J and Lorquet J C 1974 Chem. Phys. Let. 26(1), 138-143. Schmoranzer H, Liebel H, Vollweiler F, Müller-Albrecht R, Ehresmann A, Schartner K H and Zimmermann B 2001 Nucl. Instrum. Methods in Phys. Res. A 467-468, 1526-1528.

65

C X

v

Photoemission in the Molecular Frame induced by soft X ray Elliptically Polarized Light
W.B. Li1, A. Haouas2, J.C. Houver1, L. Journel2, M. Simon2 and D. Dowek1 Laboratoire des Collisions Atomiques et Moléculaires (UMR 8625 UPS-CNRS), Bat 351, Université Paris-Sud, 91405 Orsay, France 2 Laboratoire de Chimie Physique-Matière et Rayonnement (UMR 7614 UPMC-CNRS), 11 rue Pierre et Marie Curie, 75231 Paris Cedex 05 France Work supported by the European Community-Research Institute Action (E.U. RII 3 CT-2004506008).
1

Angular distributions from fixed in space molecules like CO and N2 have been reported in recent years for photoelectrons [1] as well as for fast Auger electrons [2] emitted after K shell photoionization. In this project we have used the vector correlation method [3] and the linearly or elliptically polarized light delivered by the UE56-1 PGM beamline at BESSY in the soft X ray range. The energy and angular properties of photoelectron emission and multiple Auger decay following non-resonant (e.g. eq.1 or 2) and resonant (e.g. eq.3) dissociative photoionization (DPI) of the NO, CO, N2 and N2O molecules have been investigated in the neighborhood of the N 1s and O 1s ionization thresholds: NO + hν(ê) → NO+(1sN or O)-1 + eph → N+ + O+ + eph + eAu (eq.1) → N++ + O+ + eph + esAu + efAu (eq.2) NO + hν(ê) → NO*(1sN or O→π∗) → N+ + O+ + esAu + efAu (eq.3) esAu and efAu refer to the situation where a slow and a fast Auger electrons are produced in double Auger decay, as observed e.g. for Ar 2p and Ne 1s inner-shell photoionization [4]. The experimental approach consisted of measuring the velocity vectors of the two ionic fragments and one of the slow electrons eph or esAu collected in a 4π solid angle [5], leading to e.g. the (VN, VO, Ve, ê) quadruplet, where ê represents the k propagation axis of elliptically polarized light or the P polarization axis of linearly polarized light. The spatial analysis of the (VN, VO, Ve, ê) vector correlation provides in particular the molecular frame photoelectron (or slow Auger electron) angular distributions (MFPADs) for each selected process, an observable which is the most sensitive probe of the molecular photoionization dynamics. The multidimensional data analysis is in progress and several papers are in preparation. Here we report as an example the results of the vector correlation analysis for the DPI reaction according to (eq.4) [6] at the hν = 418 eV photon excitation energy. NO + hν(ê) → NO+(1Π, 3Π,1sN-1) + eph → NO++ + eph + eAu → N+ + O+ + eph + eAu (eq.4) Figure 1 displays the electron-ion kinetic-energy correlation diagram (KECD) (a) which represents a bidimensional (2D) histogram of the (eph, N+, O+) coincident events as a function of the electron energy and the kinetic energy release (KER) of the reaction, as well as the onedimensional (1D) histograms namely the photoelectron (b) and KER (c) energy spectra, corresponding to the projections of the 2D histogram on the Ee and KER axes, respectively. The two NO+(1Π) and NO+(3Π) are partially resolved here, and the KER distribution indicates the contribution of several NO++ electronic states [7].

66

1600

(a)
I(arb.u)

1400 1200 1000 800 600 400 200 0

NO+(1Π)

NO+(3Π)

(b)

0

2

4

6

8 10 E (eV)
e

12

14

16

1400 1200 1000

B2Σ+

C2Σ+

I(arb.u)

800 600 400 200 0

(c)

4Π

6

8

10

12 14 KER (eV)

16

18

20

Figure 1 : (a) Electron-ion kinetic-energy correlation diagram (KECD) for the (eph,N+,O+) coincident events
produced at a photon energy hν ≈ 418 eV, and the 1D projections of the KECD corresponding to the photoelectron and KER energy spectra (see text).

When the light is elliptically polarized, the theory of measurement leading to the molecular frame photoelectron angular distributions (MFPADs) previously developed for linear [3] or circular [8] polarization involves additional terms. If the polarization is described by the three (S1, S2, S3) Stokes parameters, the general analytical form of the complete angular distribution describing a DPI reaction I(χ,γ,θe,φe) [8] is a function of four angles: (χ,γ) characterize the fragment ion ejection direction in the laboratory frame and (θe,φe) the photoelectron ejection direction in the molecular frame, defined by the molecular axis and the polarization axis. The reference axis in the laboratory frame is the propagation axis of the light k. Integrating the I(χ,γ,θe,φe) distribution over the ion azimuthal angle γ leads to an expression identical to the one used for circularly polarized light, except for the F11(θe) term which is multiplied by the Stokes parameter S3. IS3 (χ, θe, φe) = F00(θe) – 1/2 F20(θe) P2 (cosχ) – 1/2 F21(θe) P2 (cosχ) cos (φe)
– S3 x F11(θe) P1 (cosχ) sin (φe) – 1/2 F22(θe) P2 (cosχ) cos (2φe)
1 2 0 1

(eq.5)

Here the Stokes parameter S3 is defined as S3 = –1 for circularly polarized of helicity +1. This demonstrates that the four FLN(θe) functions which provide the MFPADs for any orientation of the molecular axis with respect to the polarization axis of linearly polarized light [3], as well as the product of the Stokes parameter S3 by the fifth F11(θe) function characterizing the circular dichroism for electron emission in the molecular frame [8], can be derived from a single experiment performed with undefined elliptically polarized light [9]. The S1 and S2 Stokes parameters of the light are also determined in such an analysis, by fitting the ion polar and azimuthal angle dependence of the I(χ,γ,θe,φe) distribution, after integration over the electron angles [9]. S3 is subsequently determined if the degree of polarization is close to 1. Figure 2 displays three specific MFPADs derived from the measured I(χ,γ,θe,φe) distribution for eq.4 at hν = 418 eV: they correspond to a molecular axis orientation parallel and perpendicular to the polarization axis P of linearly polarized light as shown, and to a molecular axis perpendicular to the propagation axis k of elliptically polarized light.

67

Figure 2 : Measured I(θe,φe) MFPADs for a molecular axis parallel and perpendicular to linearly polarized light, and for a molecular axis perpendicular to the propagation axis of elliptically polarized light, derived from a single experiment performed with elliptically polarized light.. The MFPADs display strong electron emission anisotropies, with a clear contribution of up to l=3 (fσ) partial waves. A significant circular dichroism is found, revealed by the leftright asymmetry maximum in the plane perpendicular to the propagation axis k. For a molecule parallel to the linear polarization of the light, the results are in good agreement with those obtained previously [10] and multi-channel Schwinger configuration interaction calculations (MCSCI) [10]. The MFPADs measured for a comparable photoelectron energy in the region of the O 1s ionization threshold display a strikingly mirror like shape with respect to those measured at the N 1s edge. These results compare well with recent MCSCI calculations performed in the group of R.R. Lucchese (Texas A&M University, USA) [6]. An example of Stokes parameter set found in this photon energy range is S1 ≈ -0.4, and S2 ≈ 0, leading to S3 ≈ 0.92 in the assumption of a negligible component of unpolarized light.
References

1. e.g. E. Shigemasa et al, Phys. Rev. Lett. 80 1622 (1998); S. Motoki et al, J. Phys.B 33 4193 (2000); T. Weber et al, J. Phys.B 34 3669 (2001). 2. R. Guillemin et al, Phys. Rev. Lett. 87 203001 (2001); T. Weber et al, Phys. Rev. Lett. 90 153003 (2003). 3. A. Lafosse et al, Phys. Rev. Lett. 84 5987 (2000); R.R. Lucchese et al, Phys. Rev. A 65, 020702 (2002). 4. J. Viefhaus et al, J. Electron Spectrosc. Relat. Phenom. 141 121 (2004); Phys. Rev. Lett. 92 083001 (2004). 5. M. Lebech et al, Rev. Sci. Instrum. 73 1866 (2002). 6. W.B. Li et al, in preparation 7. D. Edvardsson et al, Chem. Phys. Lett. 256 341 (1996). 8. M. Lebech et al, J. Chem. Phys. 118 9653 (2003). 9. W.B. Li et al, in preparation 10. K. Hosaka et al, J. Phys.B 37 L49 (2004).

68

AUGER RELAXATION PROCESSES IN KRYPTON STUDIED WITH A MAGNETIC BOTTLE J.Palaudoux1, L.Andric1, P.Lablanquie1, F.Penent1, U. Becker2, M. Braune2, J.Viefhaus2, J.H.D.Eland3 LCPMR, Université Pierre et Marie Curie, 75252 Paris 5, France, 2 Fritz-Haber-Institut, Faradayweg 4-6, 14195 Berlin, Germany 3 Physical and Theoretical Chemistry Laboratory, Oxford, OX1 3DW, U.K. Inner shell ionization leads to the creation of a short lived hole that can often decay non radiatively by emission of Auger electron(s). Photoelectron spectroscopy (PES) and Auger Spectroscopy are powerful analysis tools that reveal many details of such processes. However open questions remain. Are more than one Auger electron released? If this occurs, are they released simultaneously or sequentially? What multiply charged final states are populated by this process? What happens when the two-step approximation, according to which photoelectron and Auger electron emissions are subsequent and independent events, fails? In particular, a detailed description of the Post Collision Interaction (PCI) process is needed, when Auger electrons and photoelectron interact and exchange energy with mediation of the ionic core. The proper way to study such processes is to perform a coincidence detection of all
1

released electrons. However, conventional coincidence experiments are often limited to the detection of only two electrons with low efficiencies due to a low angular acceptance and consequently a complete picture of the process is difficult to obtain. Our approach is to use a high efficiency time of flight apparatus having full solid angle (4π) collection efficiency which enables us to detect in coincidence (and resolving in energy) all electrons created in a multiple ionization event.

Figure: Decay of krypton 3d holes by emission of two Auger electrons. All three electrons of the process are detected in coincidence. The two-dimensional plot of the Auger double continua are filtered according to the associated photoelectron. Intensity colour scale was selected to enhance structures. Maximum count is 321 (3d3/2 case) or 534 (3d5/2). Random coincidences are not subtracted.

69

Our experiment uses a 2.4 m long time of flight magnetic bottle electron spectrometer that was initially developed for pulsed VUV laboratory sources by J.H.D.Eland in Oxford [1]. We adapted this technique to synchrotron radiation and made the first experiments at the end of 2003 during the last days of operation of Super ACO [2]. The setup allows the coincident detection of the photoelectrons with all (1, 2, 3…) subsequently released Auger electrons. An essential point is the time structure of the light pulses, which has to have sufficiently long time window in order to avoid overlapping time-of-flights of photoelectrons emitted by the different primary processes. Single bunch at BESSY having a pulse period of 800 ns is ideally suited for this purpose. Therefore 2 weeks in the single bunch mode were used on beamline U56/2-PGM-1 for this project, in June and December 2005 The figure reveals the energy map of the Double Auger decay of the Kr 3d hole. It displays events in which 3 electrons (the 3d photoelectron and the two Auger electrons) have been detected in coincidence. Thus Auger spectra are automatically filtered according to which hole (3d5/2 or 3d3/2) is involved. The diagonal lines correspond to a constant sum of the energies of the two Auger electrons and are associated with well defined Kr3+ final states. The spectroscopy of the triply charged ion is then readily obtained by this method. The analysis reveals that the Kr triple ionization potential is by 1 eV lower than the value retained in the literature, in agreement with a recent investigation [3]. Intensity along the lines reveals weak continua due to direct Double Auger process and peaks associated with a dominant process of cascade decays. Analysis is underway to characterize the intermediate Kr2+ states. The previous result was obtained at 140eV photon energy, when PCI effects can be neglected. In addition a series of measurements were performed in the immediate vicinity of the 3d thresholds, to study the influence of the energy excess on the PCI process. Coincidence detection of the photoelectron and the Auger electron allow us to study the final state dependence of this phenomenon. Using the same method it has already been shown that in Xenon the PCI profile depends on the Auger kinetic energy [4]. Preliminary analysis gives clear indication that we will be able to reveal the details of the recapture process of the photoelectron in autoionizing states. Finally this method was extended to study the 3p hole decay in Kr 3p, the 2p hole decay in Argon, and first tests on molecular targets were employed: the Carbon 1s hole decay in CO and the Iodine 3d hole decay in CH3I were chosen as examples.

References [1] Eland, J. H. D et al, Phys. Rev. Lett. 90, 053003 (2003). [2] Penent, F et al, Phys. Rev. Lett. 95, 083002 (2005) [3] J.Viefhaus et al, J. Phys. B, 38, 3885 (2005) [4] S.Sheinerman et al, J. Phys. B, 39, 1017 (2006)

70

High-resolution photoelectron spectroscopy on laser-excited potassium atoms M. Meyer1, D. Cubaynes1, F. Wuilleumier1, E. Heinecke2, T. Richter2, P. Zimmermann2, S.I. Strakhova3, and A. N. Grum-Grzhimailo3
1 2

LIXAM, Université Paris-Sud, Bâtiment 350, F-91405 Orsay cedex, France

Institut für Atomare Physik und Fachdidaktik, TU Berlin, 10623 Berlin, Germany Institute of Nuclear Physics, Moscow State University, Moscow 119992, Russia

3

The combination of laser and synchrotron radiation (SR) has been used to study electron correlations in atomic potassium. In particular, the photoionization in the 3p-shell from the K 3s23p64s ground state as well as from the K* 3s23p64p 2P1/2 and 2P3/2 laser-excited states has been investigated [1]. Photoelectron spectroscopy on laser-excited alkaline atoms has already a long history and first experiments on Na date back to the year 1982 [2], but the advent of third generation SR sources and high-resolution electron analyzers has given new impetus to these studies allowing now to explore in detail fine-structures in the photoelectron spectrum. In two recent experiments on laser-excited Na [3] and Rb [4], it was shown that it is possible to completely resolve the individual components belonging to different total angular momenta of the np5(n+1)p electronic configuration of the photoion. Both experimental results, especially the strong intensity variation observed after respective excitation of the 2P1/2 and 2P1/2 state by the laser, were well described by theory, but by using different coupling schemes. The aim of the present study was therefore to understand these differences and to fill the gap between Na and Rb for a systematic analysis of the electronic interactions. The experiments have been performed at the UE125/2-SGM beamline of BESSY II using monochromatized synchrotron radiation in the photon energy region between 35 and 70 eV. The potassium vapor produced by a radiatively heated oven was crossed in the source volume of a high-resolution electron analyzer (Scienta SES2002) by the counter-propagating laser and synchrotron radiation. The pumping of the K* 3s23p64p 2P1/2 and 2P3/2 excited states was obtained with a continuous wave single-mode Ti:Sa ring laser, which was delivering typically about 800 mW at the wavelengths of interest (765.70 and 770.11 nm). Only electrons emitted under the magic angle (54°44’) with respect to the polarization vector of the SR have been analyzed. Since SR and laser are linearly polarized, this geometry enables us to measure directly the photoionization cross sections for the 2P1/2 state, but for the 2P3/2 state the alignment of the target has to be taken into account in the final analysis. A complementary series of measurements has therefore been undertaken measuring the photoelectron spectra for different values of the relative angle between both polarization vectors.

71

One of our main results is displayed in figure 1 showing the photoelectron spectra of laser-excited potassium in the region of the 3p5 4p states after initial excitation to the 4p 2P1/2 and 4p 2P3/2 state, respectively. Strong variations of the relative intensities of the different multiplet components are observed. The most pronounced changes are found for the 4p[5/2]3 line, which is completely missing in the 2P1/2 spectrum, and for the 4p[3/2]2 line, which shows opposite behavior and is only present in the 2P3/2 spectrum. As for the case of laserexcited Na [2], the former can be attributed to a quasi-forbidden, the later to a dynamically forbidden transition. This interpretation is corroborated by theoretical results using the general geometrical model (GGM) [1], which are reproducing perfectly the experimental data.

Figure 1: Photoelectron spectrum of laser-excited K in the region of the 3p54p states recorded at a photon energy of 42 eV after laser excitation to the 4p 2P3/2 (solid line) and 4p 2P1/2 (dotted line). Strong variations of the relative line intensities have also been observed for the 3p55p shake-up satellites, in close similarity to the situation in Na. In general, our analysis demonstrates that the 3p54p electron spectrum of potassium is closer to the case of sodium than to rubidium, since it cannot be well described by any pure coupling scheme. For the 4p55p spectrum of Rubidium the jK coupling model has successfully been applied [4]. References [1] M. Meyer, D. Cubaynes, F. J. Wuilleumier, E. Heinecke, T. Richter, P. Zimmmermann, S. I. Strakhova, and A. N. Grum-Grzhimailo, J. Phys. B (2006), submitted. [2] J.M. Bizau, F.J. Wuilleumier, P. Dhez, D. Ederer, J.L. Picqué, J.L. Le Gouët, and P. Koch, in “Laser Techniques for Extreme Ultraviolet Spectroscopy”, AIP 90, 331982 (1982). [3] D. Cubaynes, M. Meyer, A. N. Grum-Grzhimailo, J.-M. Bizau, E. T. Kennedy, J. Bozek, M. Martins, S. Canton, B. Rude, N. Berrah, and F. J. Wuilleumier, PRL 92, 233002 (2004). [4] J. Schulz, M. Tchaplyguine, T. Rander, H. Bergersen, A. Lindblad, G. Öhrwall, S. Svensson, S. Heinäsmäki, R. Sankari, S. Osmekhin, S. Aksela and H. Aksela, Phys. Rev. A 72, 032718 (2005).

72

Angular-resolved partial cross sections of doubly excited helium
Y. H. Jiang,1 R. P¨ttner,1 M. Braune,2 J. Viefhaus,2 u M. Poiguine,1 R. Hentges,2 U. Becker,2 and G. Kaindl1
1

Institut f¨r Experimentalphysik, Freie Universit¨t Berlin, u a Arnimallee 14, D-14195 Berlin-Dahlem, Germany

2

Fritz-Haber-Institut Berlin, Faradayweg 4-6, D-14195 Berlin-Dahlem, Germany

Double-excitation states in helium have been considered to be prototypical for two-electron systems with strong electron correlation since their first observation in the 60’s of last century. So far, total photoionization cross sections (TCS) and partial cross sections (PCS) have been studied intensively [1, 2]. The angular distribution parameters (ADP), β, provide additional information on the coupling of the outgoing channels in comparison to the TCS and the PCS. Therefore, the resonances show a different behavior in the ADPs as compared to the PCSs and additional resonances are expected to be observable in the ADP. These information can provide additional tests for the quality of theoretical methods that are employed to study quantum chaos in the region close to double ionization threshold of helium [3]. Menzel at al. were the first group who reported on the measurements of the ADPs up to the SIT I5 [4]. Czasch et al. [5] studied the ADPs in the photon energy region from the single ionization thresholds (SIT) I9 to I16 . The SIT IN describes singly ionized helium with the remaining electron possessing the principal quantum number N . In this report, we completed experimental studies for the ADPs from the SITs I5 to I7 , which fill the gap between Menzel’s measurements and the experimental results by Czasch et al.. The experiments were performed at the undulator beamline U125/2-SGM (BUS-beamline) of the Berliner Elektronenspeicherring f¨r Synchrotronstrahlung (BESSY) using a photon enu ergy resolution of Ω ∼ 6 meV (FWHM). A number of time-of-flight (TOF) spectrometers = were mounted to detect photoelectrons at various angles in the dipole plane, which is oriented perpendicular to the propergation direction of the light. A needle (10 cm long, less than 500 µm inner diameter) directs an effusive jet of gas to the interaction region; the background pressure in the chamber was ∼ 10−4 mbar. TOF spectra were taken for different photon ener= gies using a step-width of 3 meV and converted into photoemission spectra by the well-known time-to-energy conversion procedure. The double-excitation 1 P 0 resonances in helium can be assigned by a simplified classification scheme N, Kn , with N and n being ionization threshold of the channel and the running

73

5,3n’ 6
0.8 0.6 0.4

7

8 7

5,1n’
β2

6

(a)

Angular parameters

0.2 0.2 0.0 -0.2 -0.4 -0.2 -0.4 -0.6
4 4

β3

(b)
FIG. 1: Angular distribution parameters
5 βn with (a) n = 2, (b) n = 3, and (c)

β4

(c)

n = 4, respectively, together with (d) the
5 PCS σ2 below the SIT I5 ; the measure-

ments of Menzel et al. were displayed by dash lines. The two vertical-bar diagrams

8×10

Counts

σ2

(d)

in the upper part of the figure give the assignments of the double-excitation resonances specified by n’; with the widths of

7×10 6×10 5×10

4 4

76

76.2

76.4

76.6

bars being proportional to the linewidths of the corresponding resonances.

Photon Energy (eV)

index of Rydberg series, respectively; K represents the angular-correlation quantum number. Preliminary measurements of the β-parameters below the SITs I5 to I6 are given in Figs. 1 and
N 2, respectively. For a better identification of the resonances, the PCSs σ2 are also plotted in N N these figures. Note that the PCSs and ADPs are labeled by σn and βn , respectively, with n

being the principal quantum number for the remaining electron in He+ (n). Our measurements below the SIT I5 agree well with experimental results of Menzel et al. [4]. The first two members 5, 36 and 5, 37 of the principal series below the SIT I5 were observed for the first
5 5 5 time in the β2 and β3 . These resonances could not be observed in β4 due to the low cross

section in combination with a low transmission rate for slow electrons, which are related to this decay channel. For the results presented in Figs. 1 and 2, up to four TOFs are employed and the error bars on absolute value of β are estimated to be less than 15%. Considering these error bars, the present absolute values of β 5 are in agreement with previous measurements and calculations published in Ref. [4]. Interestingly, the resonance 6, 28 was clearly observed for
6 6 the first time by variations in β2 and β3 . In contrast to the variations in β, this resonance

is strongly suppressed in σ2 given in Fig. 2(d) as well as in the experimental and theoretical σn presented in Ref. [2]. This can be understood by the different dependences of β and σ on the matrix elements for the excitation and decay of the autoionization states and on the phase

74

6,4n (a)
0.5

8

9 8 9 10

6,2n

β2

Angular parameters

0.4 0

(b)

β3

-0.1

-0.2 -0.4 -0.5 -0.6 80

(c)

β4

Counts (×10 )

(d)
75 70 65 77

σ2

3

FIG. 2: Angular distribution parameters
6 βn with (a) n = 2, (b) n = 3, and (c)

n = 4, respectively, together with (d) the
5 PCS σ2 below the SIT I6 . For details see

77.1

77.2

77.3

77.4

Photon Energy (eV)

Fig. 1.

shifts of the various outgoing channels [6]. In oder to improve the signal-to-noise ratio, in the last beamtime the measurement in regions below the SITs I6 to I7 were repeated with 12 TOFs mounted at various angles in the dipole plane. These data are, however, not yet fully analyzed.

This work was supported by the Bundesministerium f¨r Bildung und Forschung, project u no. 05 KS1EB1/2, and the Deutsche Forschnungsgemeinschaft, project no. PU 180/1-2.

[1] M. Domke et al., Phys. Rev. A 53, 1424 (1996). [2] Y. H. Jiang et al., Phys. Rev. A 69, 042706 (2004). [3] R. P¨ ttner et al., Phys. Rev. Lett. 86, 3747 (2001). u [4] A. Menzel et al., Phys. Rev. Lett. 75, 1479 (1995). [5] A Czasch et al., Phys. Rev. Lett. 95 243003 (2005). [6] Y. H. Jiang et al., J. Phys. B 39, L9 (2006).

75

New Vibrational Satellites in Highly Resolved Spectra of Xe/CF4 Mixtures
V.A. Alekseev
Institute of Physics, St.Petersburg State University, Peterhof, 198504 Russia

N. Schwentner
Institute für Experimental Physik, Freie Universität Berlin, Arnimallee 14, D-14195, Germany

Motivation.
Perfluoromethane is one of the very few molecular gases which are transparent in the range of Xe resonance transitions in the Vac UV region. Studies of pressure broadened contours of these lines provide information on Xe* - CF4 interaction potentials. In the present work we explored opportunities offered by a recently commissioned 10m normal incidence monochromator1 at the synchrotron radiation facility BESSY II to record high resolution mixtures in the Vac UV transmission and fluorescence excitation spectra of Xe/CF4 region. Here we report on observation and assignment of satellites of dipole forbidden transitions in Xe atom.

Experimental.
Intensity of Vac UV radiation in transmission was recorded by a GaAs photodiode and picoampermeter (Keithley Instruments). Fluorescence was observed at 90 degree via Mg2F side-on window with use of a solar blind photomultiplier. The optical pathway of SR radiation in the gas cell prior to the fluorescence observation zone was reduced with a special adapter. Perfluoromethane does not deactivate excited Xe atom. The major collisioninduced energy transfer process is relaxation to the lowest resonance Xe 6s[3/2]1 and metastable Xe 6s[3/2]2 states. Typically mixture consisted of ~1-10 mbar of Xe and ~ 0.5-1 bar of CF4. Spectra of Xe/He mixtures were recorded for comparison. Because of fast relaxation in these high pressure mixtures, Vac UV fluorescence excitation spectra closely followed in shape the absorption spectra, with nearly constant quantum yield over the spectral range 115-150 nm. Most likely emitters are the Xe 6s[3/2]1 atoms and Xe2* dimers formed in the three body recombination process. CF4 absorbs radiation in the λ< 115 nm region and the cell passes no light at a CF4 pressure ~ 1 bar. However, all broadened Xe bands up to LiF cutoff at 104 nm are easily observed in the excitation spectra of Vac UV emission.

Results and discussion. Figure 1a shows fluorescence excitation spectrum of Xe/He mixture on the red wing of the Xe 6d[3/2]1 resonance transition. This region covers dipole-forbidden transitions to numerous states from Xe 6p’-, 7p- and 6d-manifolds. Their energies are indicated by the lower set of sticks in Figure 1a. Most of these transitions are not visible in the spectrum of pure Xe at low pressure. Addition of ~ 1 bar of He weakens the optical selection rules and a sequence of bands appears in the spectrum. These bands in mixture with CF4 (Fig.1b) are much less intense, most of them are hardly seen on the background of Xe 6d[3/2]1 pressure broadened band. More striking, however, is the appearance of several new bands in spectral regions which are not overlapped with any dipole forbidden transitions. In particular, three bands appear on the blue wing of Xe 6d[3/2]1 resonance band (Fig.1c and Fig. 1d). The band closest to the resonance transition is so intense that at PXe=5 mbar (Fig.1d) it transforms to a spectral dip due to complete absorption of SR radiation prior fluorescence observation zone.

76

The upper sticks in Figures 1 indicate positions of Xe atom levels displaced by the energy of CF4 ν3 vibrational quanta. Each extra band in experimental spectra has its counterpart in the group of these displaced energy levels. We conclude that these bands are Xe (G)..CF4 (ν3=0) + hν à Xe(F)..CF4 (ν3=1) satellite transitions where Xe (G) states for the ground state and Xe(F) for a dipole forbidden state. In particular, the strongest band on the blue wing of the 6d[3/2]1 resonance corresponds to the 7p[1/2]0 satellite. This and other intense satellites are located on the wing of pressure broadened resonance bands, implying an intensity borrowing mechanism. Some dipole forbidden transitions are also close in energy to resonance transitions, but they are much less intense than their satellites or absent at all and it seems that the major intensity enhancing factor is excitation of CF4 vibrations. The CF4 ν3=0 à ν3=1 is a very strong infra red transition, an order of magnitude stronger than in CH4 molecule.3 Considering that the satellites are spectrally narrow bands in close proximity to the Xe(F) + CF4 (ν3=1) asymptotic energies, the observed effect may be viewed as coupling between the resonance and dipole forbidden atomic states in the electric field of a transient molecular dipole. The selection rules are the same as for the optical coupling. A qualitative explanation of the relative intensities of some satellites may be given in terms of perturbation theory. In particular, coupling of the 6d[3/2]1 resonance state with the 7p[1/2]0 forbidden state is allowed in the first order and yields the most intense satellite (Fig.1c,d ). Coupling with the 6d[7/2]3 state is allowed in the second order, while coupling with the 6d[7/2]4 state is a higher order effect. It presents a tentative explanation for the high intensity of the remote 6d[7/2]3 satellite (Fig.1c,d ). The spectrum in Figure 1d also displays a satellite of the 6d[3/2]1 resonance transition. Although this transition is optically allowed, the band is relatively weak. Satellites on the far blue wings of pressure broadened Xe 6s[3/2]1, Xe 6s’[1/2]1 and Xe 7s[3/2]1 resonance bands have been reported earlier.2 No satellites due to excitation of anyone of the three other vibrational modes of CF4 molecule, v1= 909 cm-1, v2 = 435 cm-1 , v4=632 cm-1, were found in the present study. The appearance of very strong satellites of dipole forbidden transitions is a striking manifestation of the vibronic coupling phenomena. A large variety of new experiments should be explored to characterize it in detail.

Acknowledgement.
The authors are grateful to the BESSY staff, especially to Dr. G. Reichardt for help with running 10m NIM References. 1. G. Reichardt et al, NIM A 467-468 (2001) 462-465 2. V. A. Alekseev , Opt. Spectrosc. 96 (2004) 492 and references therein 3. D.A. Dixon, J.Phys.Chem. 92 (1988 )86.

77

78

Photoionization of Free Chiral Molecules
ID.05.1.212: Weeks 4/5 2005 UE56/2 PGM2 EU R II 3-CT-2004-506008

Chris Harding, Elisabeth Mikajlo, Ivan Powis, University of Nottingham, UK Silko Barth, Uwe Hergenhahn, IPP Garching

Introduction
It was predicted in 1976 that an un-oriented sample of a chiral molecule could have an angular distribution I p (θ ) = 1 + b1p P1 (cos θ ) + b2p P2 (cos θ ) with an ‘additional’ cosθ term when ionized with circularly polarized light (helicity, p = ±1). Only recently has this been experimentally verified. A new chiral asymmetry results, seen as a difference in the forward-backward photoemission yield for a given enantiomer and polarization and expressed by the chiral parameter b1±1. This should reverse on changing either enantiomer or circular polarization. Alternatively, for a given enantiomer and a fixed detection direction, θ, one can see a new form of circular dichroism i.e. the difference between photoelectron yield for the two polarizations. These asymmetries are expected to be of the order 0.1, exceeding by several orders of magnitude previous natural dichroism effects since they arise in the pure electric dipole approximation. Our programme at BESSY II aims to examine these predictions and the general phenomenon in the core ionization region. C 1s core ionization is particularly appropriate for initial investigations since the initial orbital is localized and achiral (spherical), so any observed dichroism and chiral asymmetery can be interpreted as a purely final state effect. Using the rapid switching, two-beam mode of the UE56/2 twin undulators, and methodology developed in 2004, two molecular studies of carvone and fenchone, were undertaken, examining in each case both the molecular enantiomers. Polarimetry scans were made at the photon energies used for these measurements to confirm the degree of circular polarization.

Carvone

O

Fig. 1

Fig 1 shows a typical carvone C 1s XPS with, beneath, the difference spectrum, σ(+)- σ(-) . Results for both enantiomers are included, and show the anticipated opposite dichroism under the small carbonyl C=O peak.1

79

Fig. 2

The variation of the C=O dichroism, expressed as a % asymmetry, with photon energy is shown in Fig.2. The solid curves in this Figure are CMS-Xα theoretical calculations of the expected dichroism, assuming only the lowest energy conformers of the flexible carvone molecule are populated. However, better agreement with experiments is obtained when contributions from higher energy conformations are included, although in previous studies these have been discounted.2 A surprising feature is the sensitivity of the measured C=O dichroism to the orientation of the removed isopropenyl tail in the carvone molecule; not only is the dichroism a final state scattering effect, but evidently it operates over a long range.

Fenchone

Fig. 3

A similarly extensive set of data were obtained for fenchone. Figure 3, shows (top row) the C=O XPS band recorded with different helicity radiation, for each of three example photon energies.

80

Beneath are the difference spectra for R- (x) and S- (+) enantiomers, with the normalized % asymmetry in the bottom row. Figure 4 shows the photon energy dependence of the dichroism. It is interesting to compare these observations with earlier measurements made for the closely related camphor molecule.3 These species differ only in the removal of two methyl; group s from the bridging C atom (7) in camphor to the C5 position in fenchone (above). Fig. 4 Yet the energy dependence of the dichroism is quite different. Again this shows that the final state electron scattering sensitively probes centers that are not immediately adjacent to the initial site.

References
1

2 3

C. J. Harding, E. A. Mikajlo, I. Powis, S. Barth, and U. Hergenhahn, J. Chem. Phys. 123, 234310 (2005). C. J. Harding and I. Powis, J. Chem. Phys. (submitted to J. Chem. Phys.). U. Hergenhahn, E. E. Rennie, O. Kugeler, S. Marburger, T. Lischke, I. Powis, and G. Garcia, J. Chem. Phys. 120, 4553 (2004).

81

Experimental Observation of Spin-orbit Activated Interchannel Coupling Effects in Cesium
T. Richter, P. Zimmermann Institut f¨r Atomare Physik und Fachdidaktik, Technische Universit¨t Berlin u a K. Godehusen Berliner Elektronenspeicherring-Gesellschaft f¨r Synchrotronstrahlung mbH u M. Yalcinkaya Department of Physics, Istanbul University Stimulated by an experimental investigation on the Xenon 3d photoionization ([1]) a new type of configuration interaction of continuum states was predicted [2, 3]. This interchannel coupling is a supposedly general effect fed by the spin orbit interaction. The calculations use a modified version of the spin-polarized random phase approximation with exchange (SPRPAE) methodology. While it was possible to quantitatively describe the Xe photoionization cross section with that new theory, its major impact on other atomic photoionization processes was not noticed before. We now report on experimental data for the Cesium 3d photoionization, that strongly supports predictions based on the spin orbit activated intra channel coupling. Both the partial cross section of the Cs 3d5/2 photoelectrons and their angular distribution parameter (β) show a pronounced second maximum in the vicinity of the 3d3/2 threshold. In dipole approximation the β parameter governs the differential cross section from linearly polarized light impinging on unpolarized atoms via the following formula: σ 1 dσ (θ) = [1 + β(3 · cos2 θ − 1)] dΩ 4π 2 The extra features are not reproduced by uncorrelated numerical calculations (e.g. Hartree-Fock).

cross setion (a.u.)

3d3/2

3d5/2

750

745 740 735 730 ionization energy (eV)

725

Figure 1: Cs 3d photoelectron spectrum and angular distribution fitted to the experimental data for 3d5/2 Individual photoelectron spectra are recorded for four different angles of linear polardσ ization using a fixed-in-space electron analyzer. The derived dΩ (θ) is fitted with a angular distribution function using β, σ and the instrumental angle as free parameters to acquire β. See [4] for more details.

82

1 3d5/2 RPAE 3d3/2 RPAE 3d5/2 experiment

relative cross section (a.u.)
0 730

740

750

760

770

photon energy (eV)

Figure 2: Photoelectron yield measured for Cs 3d5/2 , calculated for both spin-orbit doublets
2

1

β
0

3d5/2 RPAE 3d3/2 RPAE 3d5/2 experiment 3d3/2 experiment

-1 730

740

750

760

770

photon energy (eV)

Figure 3: Experimental and predicted β data The interchannel coupling effects are clearly confirmed, while the predicted energy position of the features diverges from the experimental data. Also the deviation is different for the photoelectron yield and the β parameter. This is remarkable as the interference effect should be quite sensitive to even small variations in any channel. We are thankful for the support of the BESSY staff. This work was in part funded by the Deutsche Forschungsgemeinschaft. [1] A. Kivim¨ki, U. Hergenhahn, B. Kempgens, R. Hentges, et al. Phys. Rev. A, 63 a 012716 (2000). [2] M. Y. Amusia, L. V. Chernysheva, S. T. Manson, A. Z. Msezane, et al. Phys. Rev. Lett., 88 093002 (2002). [3] M. Y. Amusia, A. S. Baltenkov, L. V. Chernysheva, Z. Felfli, et al. J. Phys. B, 37 937 (2004). [4] K. Godehusen, H.-C. Mertins, T. Richter, P. Zimmermann, et al. Phys. Rev. A, 68 012711 (2003).

83

Mass-ltered cobalt clusters on non-magnetic surfaces
J. Bansmann, P. Imperia1 , A. Kleibert, F. Bulut2 , M. Getzla2 , C. Boeglin3 , A. Barla3 , K.H. Meiwes-Broer Institute of Physics, University of Rostock, D-18051 Rostock 1 Hahn-Meitner-Institut, D-14109 Berlin Institute of Applied Physics, University of Düsseldorf, D-40225 Düsseldorf 3 IPCMS Strasbourg, 23 rue du Loess, F-67033 Strasbourg, France supported by DFG via BA 1612/3-1 (in the DFG priority call 1153). It is well-known that small clusters deposited on surfaces exhibit magnetic properties that strongly dier from those for thin lms and bulk systems [1,2]. Analogously, even large clusters with about several thousands of atoms have enhanced orbital moments although only the outer two layers of the clusters are inuenced [3-5]. Moreover, the electronic structure of the interface also has an inuence on the magnetic properties of the clusters that cannot be neglected. Large mass-ltered Fe and Co clusters deposited on Ni(111) lms show dierent magnetic properties when compared to deposition on iron and cobalt lms [6]. Hybridization eects of the electronic states between clusters and underlying substrate / lm can modify these cluster-specic properties signicantly. Up to now we have deposited Fe and Co clusters on ferromagnetic lms in order to magnetize the clusters remanently by exchange interaction with magnetic lm. For the investigations presented here, we have chosen an Au(111) single crystal as a substrate. Cobalt lms on Au(111) show spin-reorientation transitions both as a function of lm thickness and temperature [7,8]. Mass-ltered cobalt clusters with a size of about 9 nm have been deposited in-situ onto the clean surface and afterwards investigated with XMCD using the superconducting magnet systems of the Strasbourg group. For this purpose the small and transportable arc cluster ion source (briey described in [9]) has been adapted to the preparation chamber of the magnet system. The source was developed for producing mass-ltered metal clusters in the size range from 5 nm to 15 nm that can be deposited in-situ on surfaces without fragmentation. The mass ltering process is carried out in an electrostatic quadrupole deector, cf. g. 1. The measurements

2

Figure 1: Schematic drawing of the arc cluster ion source ACIS and the mass-ltering unit.

84

Figure 2: Photoabsorption (upper part) and XMCD spectra (lower part) from an 9 nm clusters Co deposited on clean Au(111). have been carried out at the UE46-beamline at dierent temperatures (40 K and 300 K). An external magnetic eld of 4 T has always been applied along the direction of the incoming photon beam. For the individual measurements, the photon polarisation has been reversed. The XMCD data taken in total electron yield clearly show the typical absorption features at the Co 2p edge with dierent intensities in the two peaks when switching the circular polarisation from σ + to σ − , cf. g. 2. The upper panels show the spectra recorded at 40 K (left) and 300 K (right) at an angle of 40◦ taken for opposite photon helicities together with the background from the clean Au(111) sample. Please note the strong feature in the clean gold spectrum at low temperature which is located directly at the Co 2p absorption edge. The lower panels display the corresponding XMCD intensity dierence. Clearly, the area of the Co 2p1/2 peak is smaller at 300 K indicating a reduced spin moment when compared to 40 K. By integrating the XMCD curve, one has access to the ratio of orbital to spin moment which is being dicussed in the following. The ratio of orbital (mL ) to spin moment (mS ) presented in g. 3 signicantly depends on the angle of incidence (and thus the magneization direction), the values are dierent in normal emission and 40◦ o-normal. The lower data set (40 K) indicates a preferential magnetization orientation in perpendicular magnetization direction whereas the increasing value for 300 K is a clear hint for an in-plane magnetization direction. The spin moment of the cobalt clusters, however, is independent of the direction of

85

Figure 3: Ratio of orbital (mL ) to spin moment (mS ) for 9 nm Co nanoparticles on Au(111) at 300 K and 40 K for dierent angles of the magnetization direction with respect to the surface normal. The solid green line denotes the hcp bulk cobalt value. magnetization after correcting for the dipole term. Thus, it does alter this ratio. Of course, the spin moments are dependent on the temperatures, they vary from mS ≈ 1.65 µB at 40 K to 1.4 µB at 300 K. The easy magnetization direction is usually oriented along the maximum value of the orbital moment. These results clearly underline the importance of the electronic structure of the substrate on the magnetic properties of deposited clusters. It is not only the size of the clusters that determines the spin and orbital moments, one has also to take into account the interaction with the surface. Further investigations on dierent surfaces are planned in order to get a better understanding of the interaction between deposited clusters and the surface of the substrate.

References: [1] K.W. Edmonds et al., J. Magn. Magn. Mat. 220, (2000). [2] J.T. Lau et al., Phys. Rev. Lett. 89, 057201 (2002). [3] J. Bansmann and A. Kleibert, Appl. Phys. A 80, in press (2005). [4] J. Bansmann et al., Surface Science Reports 56, 189 (2005). [5] A. Kleibert, Disseration, Univ. Rostock (2006). [6] J. Bansmann et al., Appl. Phys. A 82, 73 (2006). [7] R. Sellmann et al., Phys. Rev. B 64, 054418 (2001). [8] Weller et al., Phys. Rev. Lett. 75, 3752 (1995). [9] R.P. Methling et al., Europ. Phys. J. D 16, 173 (2001).

86

High-Resolution Spectroscopy of Core-Excited Organic Van der Waals-Clusters R. Flesch, I.L. Bradeanu, J. Plenge, T. Arion, and E. Rühl Institut für Physikalische Chemie, Universität Würzburg, Am Hubland, 97074 Würzburg Molecular van der Waals clusters reveal in many cases well-resolved core-to-valence transitions in the regime of inner-shell excitation. Spectral changes occur as a result of cluster formation. These are investigated by using high-resolution near-edge spectrsocopy. It has been earlier shown that spectral shifts, homogenous and inhomogeneous line broadening, as well as variations of the Franck-Condon structure occur as a result of cluster formation [1]. We have studied the C 1s excitation regime in homogenous clusters of pyridine (C5H5N) and fluorobenzene (C6H5F). These serve as model systems of aromatic molecules. The experiments have been carried out at the UE-52-SGM beamline at BESSY. Clusters are produced by utilizing an adiabatic seeded beam expansion through a nozzle (d=50 µm), where argon is bubbled through the liquid samples. The skimmed jet is tranferred into the ionization region of a time-of-flight mass spectrometer where it is crossed by dispersed undulator radiation. Cations are separated and detected in a time-offlight mass spectrometer. Photoion yields of fragment ions reflecting properties of the bare molecule and clusters are recorded simulateneously as a function of the photon energy. Fig. 1: Comparison of the C 1s → π* transition in As a result, spectral shifts and changes pyridine (solid line) and homogenous pyridine clusters (dashed line); seed gas Ar, stagnation pressure: 1.5 bar. in spectral line shapes are accurately determined (cf. [1]). Fig. 1 shows the photoion yield of molecular pyridine (solid line corresponding to the C4H4+-yield) and pyridine clusters (dashed line corresponding to the (C5H5N)2+-yield) in the C 1s → π*regime. The transition of the bare molecule is split due to a chemical shift induced by nitrogen in the heterocyclic system. Ortho (o) carbon sites absorb at higher energy than the meta (m) and para (p) carbon sites (see Fig. 2). It is inferred from Fig. 1 that the overall shape of the band is similar in molecules and clusters and there is excellent agreement with earlier work on

87

molecular pyridine (cf. [2]). However, there are significant changes between molecular and the cluster spectra: (i) The onset of the C 1s → π*- transition near 284.3 eV is redshifted by 105 meV in clusters relative to isolated molecules; (ii) the o-C 1s → π* transition is significantly broadened in clusters, but not shifted in energy; (iii) a broadening of all spectral features is found in clusters; (iv) the vibrational fine structure at the high-energy side of the band is somewhat less distinct in the case of clusters. We note that similar results are obtained for fluorobenzene, furan, and pyrrole. Spectral redshifts of core-to-valence transitions in clusters have been observed before, but these are significantly smaller in clusters containing small molecules [1] or homoaromatic species, such as benzene [4]. Detailed investigations have been performed using ab initio calculations, similar to related work on benzene clusters [4], where the GSCF3program is used [5]. We have assumed that a reasonable model structure of the dimer is the antiparallel isomer, which is calculated to be the most stable one. The results are shown in Fig. 2, where a comparison is made between the calculated molecular and dimer v=0transition. The calculated transition energies are boradened by Voigt-profiles in order to compare the results to the experimental spectra in Fig. 1. Clearly, the sites next to the nitrogen site (oFig. 2: Comparison of the calculated C 1s → π* (v=0)- carbon) show a negligible redshift, transition in pyridine (solid line) and the pyridine dimer (dashed line). The calculated energy shifts (vertical lines whereas that of the m- and p-carbon correspond to calculated oscillator strengths) are broadened sites is substantially enhanced. This is in by Voigt profiles (Gaussian width: 40 meV, Lorentzian full agreement with the experimental width: 110 meV). results and underlines the previously observed site-specific energy shifts in clusters containing aromatic molecules [4]. References 1 A. A. Pavlychev, R. Flesch, and E. Rühl, Phys. Rev. A 70, 015201 (2004). 2 C. Kolczewski, R. Püttner, O. Plashkevych, H. Ågren, V. Staemmler, M. Martins, G. Snell, A. S. Schlachter, M. Sant’Anna, G. Kaindl, and L. G. M. Petterson, J. Chem. Phys. 115, 6426 (2001). 3 I.L. Bradeanu, R. Flesch, N. Kosugi, A.A. Pavlychev, E. Rühl, manuscript in preparation (2006). 4 I.L. Bradeanu, R. Flesch, N. Kosugi, A.A. Pavlychev, E. Rühl, Phys. Chem. Chem. Phys., submitted for publication (2005). 5 N. Kosugi, Theoret. Chim. Acta 72, 149 (1987).

88

Electronic Ni 3d surface states in NiO/HOPG nanostructures
L. Soriano,1 M. Abbate,2 I. Preda,1 S. Palacín,1 A. Gutiérrez,1 J.F. Trigo,3 A. Vollmer4 and P.R. Bressler 4 Departamento de Física Aplicada C-XII, Universidad Autónoma de Madrid, Cantoblanco E-28049 Madrid, Spain 2 Departamento de Física, Universidade Federal do Paraná, Caixa Postal 19091, 81531-990 Curitiba PR, Brazil 3 Departamento de Energía, CIEMAT, Avda. Complutense 22, E-28040 Madrid, Spain 4 BESSY, Albert Einstein Strasse15, D-12489 Berlin, Germany This work deals with the study of the electronic structure of the NiO nanostructures formed at the early stages of growth of NiO on highly oriented pyrolytic graphite (HOPG). Our main aim is the study of nanostructured NiO systems where possible surface effects are enhanced by the large surface to volume ratio of the nanostructures. In fact, early studies [1] on 3 nm NiO nanoparticles with unique catalytic properties revealed a splitting of the unoccupied Ni eg states, as shown by the O 1s XAS spectra. This splitting was interpreted as the result of the lack of the apical oxygen at the NiO surface and the large surface to volume ratio of the nanoparticles. On the other hand, the early stages of growth of NiO/HOPG are known to produce planar NiO islands along the graphite steps as shown by Atomic Force Microscopy (AFM) images.[2] Such a particular arrangement of NiO nanostructures is expected to exhibit similar surface effects as in the NiO nanoparticles. Therefore, the study of this system seems to be well justified. NiO was deposited on HOPG by reactive evaporation from a pure Ni filament in the preparation chamber. The HOPG substrate was cleaved in air just before being introduced in the preparation chamber. Then, it was thermally annealed in UHV at 300oC to remove any possible surface contamination. Reactive evaporation was performed in an oxygen atmosphere (5×10-5 Torr), with the substrate kept at room temperature. The oxygen gas was aimed directly to the sample using a narrow pipe to enhance the oxidation efficiency. The evaporation rate was maintained low enough to study the early stages of NiO growth in more detail. After each XAS analysis, the substrate was introduced in the preparation chamber for the successive evaporations. XAS measurements were performed at the PM4 plane grating monochromator in the BESSY II storage ring (Berlin). This experiment requires a high photon flux in order to detect an acceptable signal from the small amount of NiO. The optical arrangement of this monochromator was set to optimize both, high photon flux and resolution. The estimated overall resolution was better than 100 meV at 530 eV. The spectra were collected in the total electron yield detection mode. In order to observe possible dichroism effects, the spectra were measured at normal and grazing angles with respect to the incident light. The spectra were normalized to the I0 current, measured from a clean gold sample, to correct for the beam current. The NiO coverage was calculated from the O 1s XAS intensities following conventional methods. Since the growth of NiO on HOPG is not in a layer-by-layer mode, the estimated coverages should be understood as the equivalent material to form a monolayer. Fig. 1 shows the O 1s XAS spectra of the NiO overlayers for (a) low and (b) large coverages. For large coverages, the spectrum is in very good agreement with previous spectra published for bulk NiO. This shows that a thin film of stoichiometric NiO can be grown on HOPG at room temperature using this growth method. The unoccupied density of electronic
1

89

states (UDOS) of O p character obtained by ab-initio band- structure calculations for antiferromagnetic NiO, are also shown in Fig. 1(c). The spectrum for low coverages (0.5 ML) presents significant differences with respect to that of large coverages, i.e., bulk NiO. The most relevant change in the spectra concerns the eg region, which is shown in more detail in the inset of Fig. 1. The well defined eg peak in bulk NiO is split in two broad and unresolved peaks for 0.5 ML of NiO/HOPG. The Ni 3d states in NiO are split by the octahedral crystal field produced by the O into the t2g and eg sub-bands. However, the lack of the apical O at the surface of NiO breaks the symmetry and results in a pyramidal crystal field. This effect produces the additional splitting of the eg sub-bands observed for low coverages, where the relative weight of the surface states is much larger than for bulk NiO.

Fig. 1: O 1s XAS spectra for low (a) and large (b) coverages of NiO on HOPG

Fig. 2: Near edge region of the O 1s XAS spectra for (a) low, (b) large coverages of NiO on HOPG and 3 nm NiO anoparticles.

To corroborate this theory, we have calculated the O 1s XAS spectra using cluster model calculations in octahedral and pyramidal symmetries. Fig. 2 shows the near-edge region of the experimental spectra of: (a) large and (b) low coverages of NiO/HOPG. The spectra have been fitted using Lorentzian curves at the positions given by the calculations together with other typical functions in the fittings of XAS spectra to simulate the background and the tails of the higher energy structures. The octahedral calculation for bulk NiO shows a single line (short dash) corresponding to transitions to eg states. The pyramidal calculation for surface NiO presents two lines (short dots) corresponding to the x2-y2 and z2 final states. As shown in Fig 3(a) the agreement with the spectrum of bulk NiO is excellent. To fit the spectrum of the NiO sub-monolayer, shown in Fig. 2(b), not only the two surface components have been used but also a small contribution of the bulk component has been included.

90

Fig. 2(c) shows the near edge region of the O 1s XAS spectrum of 3 nm NiO nanoparticles fitted with the surface components. The excellent agreement with the calculations confirm the interpretation of the O 1s XAS spectrum of NiO nanoparticles made in Ref. 1, and strongly suggests that the same splitting mechanism is operating in the case of (0.5 ML) NiO/HOPG. Fig. 3 shows the O 1s XAS spectra of 0.5 ML NiO/HOPG taken at (a) grazing and (b) normal photon incidence. The spectra exhibit a clear dependence (dichroism) with the relative polarization of the incident light. The normal incidence spectrum, with the electric field E parallel to the surface, resembles that of bulk NiO with a relatively broader single eg peak (short dash). In particular, the distinct splitting due to surface effects (short dots) is very week Fig. 3: Near edge region of the O 1s XAS in this spectrum. On the other hand, in the spectra of 0.5 Ml of NiO/HOPG taken at: grazing incidence spectrum, with the a) grazing and b) normal incidences. electric field E perpendicular to the surface, the double peak (x2-y2 and z2) structure (short dots) is clearly observed. The dichroism effect shows that the splitting affects mostly the states perpendicular to the surface, and supports the idea that the splitting is related to the absence of the apical oxygen at the surface. It is worth noting that, in this case, the z2 orbital forms a dangling bond perpendicular to the surface. This might be related to the enhanced catalytic activity of the larger surface/bulk ratio NiO nanostructures. In summary, in this work, we have studied the electronic structure of the NiO nanostructures formed at the early stages of growth (0.5 ML) of NiO on HOPG. The results have been compared to those of and 3 nm NiO nano-particles. The Ni 2p XAS spectra of the NiO planar islands confirm that Ni atoms are present in the high spin Ni2+ form. On the other hand, the O 1s XAS spectra show exactly, as in the NiO nanoparticles, a splitting of the eg band which is explained as due to the lack of the apical O atoms at the surface. Acknowledgments: We want to thank the support of the CICYT of Spain under contract BFM2003-03277 and the European Union through the R II 3.CT-2004-506008 contract. We also thank the staff of BESSY by technical support.
References:

[1] L.Soriano, M.Abbate, J. Vogel, J.C. Fuggle, A. Fernández, A.R. González-Elipe, M. Sacchi and J.M. Sanz, Chem. Phys. Lett. 208, 460 (1993). [2] C. Morant, L. Soriano, J.F. Trigo and J.M. Sanz, Thin Solid Films, 317, 59 (1998).

91

Inner-Shell Excitation of Free Nanoparticles H. Bresch,1 B. Langer,1,2 R. Lewinski, 1 P. Brenner,1 R. Flesch,1 C. Graf,1 T. Martchenko,3 O. Ghafur,3 M.J.J. Vrakking,3 T. Leisner,4 B. Österreicher,4 and E. Rühl1
Institut für Physikalische Chemie, Universität Würzburg, Am Hubland, 97074 Würzburg 2 Max-Born-Institut, Max-Born-Str. 1A, 12489 Berlin 3 FOM Institute AMOLF, Amsterdam, The Netherlands 4 Fakultät für Mathematik und Naturwissenschaften, Fachgebiet Experimentalphysik II, TU Ilmenau
1

Electronic properties of free nanoparticles are investigated in a continuous particle beam using monochromatic synchrotron radiation in the soft x-ray regime. This approach of sample preparation delivers continuously fresh nanoparticles for core level excitation experiments, where any contact of the particles to a substrate is avoided. This approach goes beyond related work, where elastic light scattering has been studied in the vacuum ultraviolet regime [1,2]. It is also complimentary to single, trapped nanoparticles that have been studied by synchrotron radiation in the soft x-ray regime [3,4]. The specific advantage of the present approach is, that properties of the isolated nanoscopic matter cannot be modified as a result of charging or radiation damage. The experimental setup is schematically shown in Fig. 1. It consists of a particle source (atomizer), where either solutions or suspension of pre-made particles are sprayed into a controlled gas phase at normal pressure. The solvent is evaporated in a diffusion dryer. Subsequently, the sample can be massselected in a differential mobility analyzer (DMA), so that monodisperse particles are introduced into the experimental chamber, where synchrotron radiation interacts with the particle beam. Alternatively, preferably in the case of pre-made particles from chemical syntheses of well defined size, the entire particle beam is used without primary Fig. 1: Schematic diagram of the experimental setup. mass selection. The particle beam is efficiently focussed by an aerodynamic lens, where the beam size is ∼0.5 mm in the ionization region. The lens also works as a first differential pumping stage, which is followed by two other differential pumping stages. As a result, the particles are excited and ionized in a high vacuum surroundings, as shown in Fig. 1. Note that this is a widely applicable approach for sample preparation of free nanoscopic matter of diameters larger than 50 nm in the gas phase, where the sample is typically transferred within ca. 5 - 10 s from the solution into the ionization region, which includes size-selection.

92

We used for first experiments a total electron detector, which is part of a photoelectron imaging spectrometer. This detector also allows us to measure the kinetic energy of photoelectrons emerging from nanoparticles after core level excitation. We have performed the following experiments: (i) total electron yield spectra of mass-selected nanoparticles in the regime of inner-shell absorption edges; (ii) total electron yields of pre-made nanoparticles without primary mass-selection. The occurrence of aggregates of nanoparticles is avoided, as monitored by a condensation nuclei counter; (iii) photoelectron spectra of nanoparticles at selected excitation energies using the imaging spectrometer. Fig. 2 shows as one typical example the near-edge spectrum of sodium sulfate (Na2SO4 ⋅ 10 H2O) nanoparticles in the O 1s-regime. These particles were not size-selected and the particle size distribution peaks at ∼170 nm. We observe in the total electron yield distinct resonances, which come from the oxygen sites that are bound in the sulfate moiety as well as water moieties that are bound as crystal water. We have modelled the O 1s absorption using the GSCF3 ab initio approach [5] along with the known crystal structure of this salt [6]. It becomes evident that the features in the 533-535 eV regime are due sulfate, whereas the maximum near 537 eV is assigned to crystal water. Additional experiments have been performed on NaCl-nanoparticles. These indicate nearedge features at the Cl 2p- and Na 1sFig. 2: O 1s-excitation regime of sodium sulfate edges, which are comparable to crystalline nanoparticles as recorded by the total electron yield. NaCl. Size-selection indicates that there are no size effects in electronic structure in the regime >50 nm. Experiments at the O 1s-edge indicate that no water is kept in NaCl-crystals, which are prepared in situ by primary aerosol formation and subsequent drying. This is unlike the results on sodium sulfate, where the water is bound in the crystal lattice. Furthermore, we have performed first experiments on semiconductor nanoparticles.
This work is supported by the Bundesministerium für Bildung und Forschung (BMBF) grant no.: 05 KS4WW1/7 and the Deutsche Forschungsgemeinschaft (SFB 410-TP C8). References 1 J.N. Shu, K.R.Wilson, M. Ahmed, S.R. Leone, C. Graf, E. Rühl, J. Chem. Phys. 124, 034707 (2006). 2 J.N. Shu, K.R.Wilson, A.N. Arrowsmith, M. Ahmed, S.R. Leone, Nano Lett. 5, 1009 (2005). 3 M. Grimm, B. Langer, S. Schlemmer, T. Lischke, W. Widdra, D. Gerlich, U. Becker, E. Rühl, AIP Conf. Proc. 705, 1062 (2004). 4 M. Grimm, B. Langer, S. Schlemmer, T. Lischke, W. Widdra, D. Gerlich, U. Becker, R. Flesch, E. Rühl, Phys. Rev. Lett., in print (2006). 5 N. Kosugi, Theor. Chimica Acta 72, 149 (1987). 6 H.A. Levy, G.C. Lisensky, Acta Cryst. B 34, 3502 (1978).

93

Investigation of Trapped Liquid Microdroplets containing Nanoparticles by Soft X-Rays C. Graf, a B. Langer,a,b S. Dembski,a and E. Rühla
a

Institut für Physikalische Chemie, Universität Würzburg, Am Hubland, 97074 Würzburg b Max-Born-Institut, Max-Born-Str. 2a, 12489 Berlin

The approach of confining simple model particles in an electrodynamic trap has been demonstrated to be suitable to investigate nanoscopic particulate matter without any contact to a substrate [1]. This requires the use of slightly charged particles, where the particle injection process facilitates electrical charging. More recent work has shown that the charging mechanisms of differently charged, trapped nanoparticles can be derived from such experiments [2]. These works were limited to SiO2-model particles and served as proof of principle experiments. The results presented in this work are motivated by the progress of colloidal chemistry that allows us to prepare structured nanoparticles in the condensed phase. We have prepared numerous structured core-shell nanoparticle systems containing II-VIsemiconductors of variable size as well as multi-core particles, which are embedded in different environments, such as silica [3] or polymers. We report here first results that build on earlier work, where liquid microparticles were successfully stored in an electrodynamic trap, where the liquid dispersions were injected into the trap via a Piezo nozzle [4]. The required charging is accomplished by the injector nozzle, similar to recent work [5]. The volatile solvent H2O is readily evaporated, so that finally the trapped liquid particle consist of a liquid polymer droplet (dimethyl siloxane - (60% propylene oxide, 40% ethylene oxide) block/graft-copolymer (polyalkylene oxide modified silicone oil) obtained from ABCR) with quantum-size semiconductor nanoparticles. This microdroplet is trapped in a high vacuum surroundings. This allows experiments using monochromatic synchrotron radiation in the soft x-ray regime for element-selective excitation, probing the local electronic structure of the absorbing elements. The present experiments were carried out on nanoparticles containing a CdSe core (2.8 nm in diameter) that is surrounded by a thin ZnS shell (thickness: ∼0.5 nm), where cysteine is used for stabilization and dispersion in the polymer droplet (see Fig. 1(a)). The investigated particles contain a larger number of such core-shell systems (10-3 mol/ℓ), so that element-selective charging experiments can be carried out. The aim of the present work is to detect via particle charging element-selectively electronic properties of the particle core. This was not easy to accomplish, because of the small size of the cores. The element-selective charging current is derived from changes in particle charge as a function of the excitation energy near the Cd 3d-absorption edge, where the first derivative of the charging curves is used. Note that this quantity is fairly easy to derive for solid nanoparticles, where the constituents represent the major composition of the particles, as

94

evidenced in the case of SiO2 particles [1,2]. In the case of small core-shell systems this is more difficult to accomplish, since (i) the absorbing cadmium sites are located in the center of the particles and (ii) the semiconductor nanoparticles are located not only on the surface of the polymer droplet that is trapped. We show as one example recent results from energy scans recorded near the Cd 3d-edge, where the charging current yields distinct maxima that are ascribed to the near-edge structure of these trapped, polymer oil embedded core-shell nanoparticles. The near-edge structures are clearly split as a result of spin-orbit splitting. Earlier work on deposited CdSe nanocrystals indicates that there are size effects in near-edge structure [6]. The present results indicate that the features shown in Fig. 1(b) occur in the same energy regime as in ref. [6]. The present results require further analysis and theoretical modelling in order to derive the extent of changes in electronic structure, which is induced by the local surrounding in polymer embedded core-shell systems.
Charging Current [arb. units] 405

CdSe/ZnS nanoparticle:

Polymer/ water droplet

CdSe ZnS Cysteine Polymer droplet Injection Evaporation

410 415 420 Photon Energy [eV]

(a) Fig. 1:

(b)

(a) Schematic diagram of the CdSe/ZnS nanoparticles and the evaporation of the primarily formed solution droplet; (b) Charging current of a trapped polymer droplet containing CdSe/ZnS nanoparticles in the Cd 3d-excitation regime.

Financial support by the DFG (Sonderforschungsbereich 410 (TP C8)) is gratefully acknowledged. References 1 M. Grimm, B. Langer, S. Schlemmer, T. Lischke, W. Widdra, D. Gerlich, U. Becker, R. Flesch, and E. Rühl, AIP Conf. Proc. 705, 1062 (2004). 2 M. Grimm, B. Langer, S. Schlemmer, T. Lischke, W. Widdra, D. Gerlich, U. Becker, R. Flesch, and E. Rühl, Phys. Rev. Lett., in print (2006). 3 C. Graf, S. Dembski, A. Hofmann and E. Rühl, Langmuir, submitted for publication (2006). 4 C. Graf, B. Langer, S. Dembski, R. Lewinski, and E. Rühl, BESSY Annual Report 2004, p. 67. 5 M.A. Hamza, B. Berge, W. Mikosch and E. Rühl, Phys. Chem. Chem. Phys. 6, 3484 (2004). 6 K. S. Hamad, R. Roth, J. Rockenberger, T. van Buuren, and A. P. Alivisatos, Phys. Rev. Lett. 83, 3474 (1999).

95

Gold nanoparticles deposited on sulphide minerals: A combined SR-XPS and STM/STS study
Yu. Mikhlina, A. Romanchenkoa, L. Makhovab, and R. Szarganb
a b

Institute of Chemistry and Chemical Technology SB RAS, Krasnoyarsk,660049, Russia Wilhelm-Ostwald-Institut fur Physikalische und Theoretische Chemie, Universitat Leipzig, D-04103 Leipzig, Germany

The chemical state and properties of noble metals deposited on metal sulphides is important for several fields. Particularly, so-called “invisible” gold in sulphide minerals forms mainly nanometer-scale metallic particles [1], which are expected to exhibit still poorly studied specific characteristics due to size and surface effects. The reduction of aqueous precious metal complexes is of interest also for concentrating these elements from hydrometallurgical solutions and for preparation of composite nanomaterials. Here, we applied synchrotron radiation XPS together with scanning tunneling microscopy and spectroscopy (STM/STS) to characterize gold species spontaneously deposited on natural metal sulphides under uniform conditions. An unusually strong suppression of electronic tunneling on gold nanoparticles was found both in XPS and STS. Single crystals (pyrite, FeS2, galena, PbS) or research grade polycrystalline mineral samples (arsenopyrite, FeAsS, chalcopyrite, CuFeS2, pyrrhotite, Fe7S8, sphalerite, (Zn,Fe)S) were polished at silicon carbide paper and then were cleaned by wet filter paper. Some specimens were fractured in air just before microscopic studies or conditioning in aqueous solutions. In etching experiments, the samples were treated in 1 M HCl, 0.5 M H2SO4, 1 M HCl + 0.4 M FeCl3, or 0.5 M H2SO4 + 0.2 M Fe2(SO4)3 solutions without stirring at 50±2 °C, rinsed quickly with a respective cold dilute acid and then with distilled water. Gold was deposited on the minerals from unstirred solutions of HAuCl4 (pH 1.5 or 3) at room temperature; the 10-4 M solution and 10 min exposure were used unless otherwise stated. The reacted samples were rinsed with water and allowed to desiccate in air before examination. Photoelectron spectra were measured at the Russian-German beamline equipped with a VGCLAM 4 analyzer. The photon energy was usually 1000 eV, and the pass energy was 20 eV. Binding energies were corrected for electrostatic charging using the C 1s peak (285.0 eV). The spectra were fitted by convolution of Lorentzian and Gaussian functions using a Unifit program [2]. AFM, STM and STS investigations were performed using a Solver P47 device (NT-MDT, Russia) under ambient conditions in dry air. The tips used in STM/STS measurements were mechanically cut 90% Pt - 10% Ir wires; positive bias was defined as a positive voltage on the sample with respect to the tip. The uptake of gold was semi-quantitatively characterized by S/Au ratios obtained from the broad XPS scans along with AFM and STM data. The quantities of gold deposited during 10 min contact of a mineral with HAuCl4 media depend strongly on the preliminary treatment of a mineral, being lower for fracture surfaces. The quantities increase if the surfaces were moderately oxidized, in particular, as a result of polishing or exposure of minerals to atmosphere for several days. However, the oxidative etching in Fe3+-bearing media retards the cementation of gold in many times. Figure 1 shows a series of Au 4f spectra from various samples. The peaks of dispersed metal are usually shifted to higher binding energies in comparison with that of a polycrystalline gold plate taken as a standard (83.6 eV). The lines have diverse widths and shapes due to contributions from a few gold species. The spectra are better fitted with three line. The component with binding energy of 83.6 eV can be assigned to bulk gold, and that shifted by 0.5-1 eV to a higher BE is associated with ultrasmall metal particles. A share of Au+-S species (BE of 85 eV or higher) is minor in all the cases; it somewhat increases with decreasing the total amount of precipitated gold. The very broad Au 4f line observed in the case of sphalerite could be well fitted with similar three components.

96

Alternatively, the large band width can be explained in terms of non-uniform electrostatic charging of the surface of the wide gap semiconductor. In contrast, the Au 4f band for gold on galena is rather narrow and shifted by almost 1 eV. The uptake of gold onto pyrite surfaces is roughly proportional to the exposure duration; the Au 4f7/2 binding energy decreases with the growth of the surface concentration of gold, approaching the BE of bulk Au0. S 2p spectra from minerals after the deposition of gold demonstrate increased FeS2 30 min intensities at binding energies of about 163.5 eV, suggesting the formation of FeS2 10 min polysulfide anions within the reacted surface layers enriched in sulfur due to FeS2 2 min oxidation by Au(III) species. Only small amounts of sulfur-oxygen species are ZnS present. Unfortunately, it was difficult to distinguish a weak contribution from PbS sulphur bonded to Au atoms. The changes of corresponding XPS spectra of metals CuFeS2 (FeCl3) were not rather informative, and X-ray absorption Fe L-, Cu L-, S L-edge spectra CuFeS2 recorded in the TEY mode agreed in general with the above conclusions. Fe7S8 (Fe2(SO4)3) STM examination found that a Fe7S8 (H2SO4) considerable part of the surface of galena after 10 min conditioning in the 10-4 M Fe7S8 HAuCl4 solution is bespread with rather uniform gold particles of about 5 nm in 0 Au diameter (Fig. 2a). A number of particles are placed above fairly loose first particle 89 88 87 86 85 84 83 82 layer, while uncovered PbS spots up to Binding energy, eV several hundred nm in size still remain. Gold on the surfaces of pyrrhotite (Fig. 2b) Fig. 1. Au 4f spectra from polycrystalline gold 0 -4 and chalcopyrite represent nanoparticles (Au ) and polished minerals reacted with 10 M from 3-5 nm to 30 nm in size, either isolated HAuCl4 solution for 10 min; for pyrite the or arranged in submicrometer composite reaction times are shown near the spectra. Some specimens were treated before the gold islands. STM images from pyrite show that deposition in solutions marked in brackets. gold islands grow both in the number and size from 5 nm to about 50 nm with increase in the deposition time. The majority of the islands are composed of nearly 10 nm clusters (Fig. 2c), which become harder packed over the reaction. The nanoscale inhomogeneity of the mineral surfaces arisen as a result of their oxidation and disordering was also observed. Tunneling spectra measured above bare metal sulphide surfaces (Fig. 3) commonly display a conductance gap that is typical for semiconducting minerals, while the plots acquired above gold nanoparticles revealed a positive correlation between the current magnitude and the diameter of Au NPs. The tunneling currents increase as the nanoparticles associate to form composite islands and it disappears for dense gold films and sponges, approaching the curves from bulk metal. The suppressed electronic tunneling seems to cause both the increase in Au 4f binding energies in XPS and the decline of currents in STS as compared with the bulk metal. These phenomena are attributable to Coulomb blockade effects, which take place if the electrostatic energy, E = e2/2C, is larger than the thermal energy, with an insulating layer existing between a nanoparticle and substrate [3,4]. The temporal charging of a nanoparticle over the

97

photoionization process in XPS decreases kinetic energies of the photoelectrons by their interaction with the photoholes formed at core levels in final state [5,6]. The diameter d of a gold nanoparticle should be less than 20 nm to be under the E > kT requirement at room temperature and the effect scale obeys to 1/d law [6]. The effects have been found on gold nanoparticles less than 5 nm due to leakage currents and electron trapping in a majority of previous studies. In this study, the suppression of electronic tunneling was detected for clusters as large as 20-30 nm, implying special properties of the gold - metal sulphide systems, in particular, a dielectric coating on the oxidized metal sulphides or/and Au0 NPs. Moreover, a submonolayer of adsorbed sulphur may retard the tunneling by modifying the surface electronic structure. The interaction between metal nanoparticles and supporting materials is known to influence the XPS binding energies too [5,6]. The linkage between the STS and XPS results give more credibility to the Coulomb blockade hypothesis, although it is possible that several factors have simultaneous effect.

a

b

c

Fig. 2. STM images from the surfaces of (a) PbS (100) (ISP=0.1 nA, VB=-0.1 V), (b) monoclinic pyrrhotite (ISP=0.2 nA, VB= -0.4 V), and (c) pyrite (ISP=0.6 nA, VB=0.1 V) after interaction with 10-4 M HAuCl4 solution for 10 min.
40

a

b

20 0 -20 -40 -0.8 -0.4 0.0 0.4

1 2 3
0.8
-0.8 -0.4 0.0 0.4

1 2 3
0.8

Fig. 3. Tunneling spectra from PbS (a) and FeS2 (b) samples reacted in 10-4 M HAuCl4 for 10 min measured above: (a) bare PbS surface (1), a separated ∼5 nm Au0 particle (2), and a composite formed by Au NPs (3), ISP=0.1 nA, VB=-0.1 V; (b) bare FeS2 surface (1), ∼12 nm Au0 particle (2), associated Au NPs (3). ISP=0.6 nA, VB=-0.1 V.

Tunneling current, nA

Voltage, V

Voltage, V

This work was supported by the bilateral program “Russian-German Laboratory at BESSY”. We thank the staff of BESSY and the Russian-German Laboratory for technical assistance.

1. C. S. Palenik, S. Utsunomiya, M. Reich, S. E. Kesler, L. Wang, and R. C. Ewing, Am. Mineral., 89, 1359 (2004). 2. R. Hesse, T. Chassé, and R. Szargan, Fresenius’ J. Anal. Chem., 365, 48 (1999). 3. M.-C. Daniel and D. Astruc, Chem. Rev., 104, 293 (2004). 4. T. Ohgi and D. Fujita, Surf. Sci., 532–535, 294 (2003). 5. J. Radnik, C. Mohr, and P. Claus, Phys. Chem. Chem. Phys., 5, 172 (2003). 6. H.-G. Boyen, A. Ethirajan, G. Kästle, F. Weigl, P. Ziemann, G. Schmid, M. G. Garnier, M. Büttner, and P. Oelhafen, Phys. Rev. B, 94, 016804 (2005).

98

C. Boeglin1, M. Pilard1, O. Ersen1, S. Cherifi2, B. Carvello2, F. Scheurer1, P. Imperia3, B. Muller1 1 IPCMS, Groupe Surfaces-Interfaces, 23, rue du Loess, Strasbourg CNRS-UMR7504, 2 Laboratoire Louis Néel, Grenoble 3 HMI - Berlin Oxide nanostructures show fascinating new electronic and magnetic properties that suggest successful implementation in many devices of technological interest. In magnetic spin-valves the pinning of a ferromagnetic (FM) transition metal layer is provided by an antiferromagnet (AFM) such as NiO. The insertion of AFM NiO thin films in such complex spin-valve structures makes the characterization of their magnetic properties much more difficult. Realistic models for the magnetic exchange coupling at the FM / AFM interfaces suggest an atomic-level description of these interfaces. The way we decided to study this interfaces is to perform epitaxial growth of NiO on ferrimagnetic Fe3O4 (100) in the ultra-thin film and small cluster limit. The NiO AFM material is still a challenging by itself because of the complexity of epitaxial growth process and the non-trivial behaviour of the size dependent antiferromagnetic order. A complete work has been performed in our group in order to optimize the growth of NiO films and clusters on metals [1-3]. Since the magnetic properties of these AFM/FM bilayers are strongly related to the interface properties (interdiffusion, oxido-reduction in the metallic FM/AFM oxide layers respectively…) [1-4] we decided to stabilize the FM/AFM interface using a controlled oxidation of the metallic layer at the interface. The Fe3O4 template prevents the reduction of NiO and allows a direct measurement of the induced coupling into the Fe layer at the interface. The measurements were performed at the UE 46 beam line at the Bessy synchrotron storage ring using polarized X-rays and the superconducting 7 Tesla magnet belonging to the IPCMS Strasbourg. The experimental environment of the sample in the superconducting 7 Tesla magnet (fig.1) allows us to prepare in-situ layers in UHV and to apply a magnetic field up to 7 Tesla at a temperature between 4 K and 300 K. The sample can be rotated in the field and with respect to the incident photon beam using two angular rotations, the incidence angle θ and the azimuth angle φ. This allows us to align a specific crystalline axis of the sample along the incident polarized photon beam, usually set parallel to the applied magnetic field. The complex sample rotation (θ, φ) reduces the possibility to work in the very low temperatures ranges of the cryostat (1 – 4 K). In the framework of size dependent magnetic structures as well as magnetic coupling of AFM/FM spin valve systems we used X-ray magnetic circular and linear dichroïsm XMC(L)D in order to correlate the size dependent magnetism with the structure and topography obtained by LEED-STM. The main goal of our project is to study the magnetic coupling at the interface between a thin antiferromagnetic (AFM) NiO oxide layer and a ferrimagnetic (FM) Fe3O4 (100) layer using chemical and magnetic sensitivity of XMC(L)D. Moreover, magnetic field-dependent XMCD measurements at the FeL3 edge allows to measure the chemical resolved magnetic hysteresis loops in order to probe the presence of exchange bias field and the NiO thickness dependence. Starting from an ex-situ Fe3O4 (100) prepared and fully characterized template, an in-situ annealing procedure has been performed in order to obtain a clean and ordered Fe3O4 (100) surface. Comparing the XAS spectra recorded at the FeL2,3 edges to XAS spectroscopic fingerprints of different iron oxides, we could define the chemical state of our sample before

MAGNETIC DOMAINS STRUCTURES IN ULTRA-THIN NiO/Fe3O4(100).

99

performing the NiO growth. A complete angular dependent magnetic characterization of this ferrimagnetic template layer was performed previously to any deposit leading to reference data (Fig2). The measurements were performed along two azimutal orientations of the sample, the in-plane easy axis [100] and the in-plane hard axis [110]. A full set of incidence angles was considered along each direction in order to define the orientations of the hard and easy axis and the magnetocrystalline anisotropy before any magnetic coupling with the ultrathin NiO layer (fig.3). Our quantitative data show that the magnetic moment in Fe3O4 films is similar to bulk whereas its magnetic exchange coupling with ultra-thin NiO (1Å – 50Å) top-layer brings up new properties induced by the structural anisotropy of the interface. The relatively short sampling depth (about 2 nm) of the XMC(L)D technique, combined with the small thicknesses of the NiO enables sensitivity to the magnetic order in both materials in the vicinity of the interface. Thus we can explore the in-plane spin orientation with respect to crystallographic directions as a function of NiO thickness. We reveal the influence of the structural anisotropy in the NiO films on the magnetic properties at the interface with the Fe3O4 film. We show that the strength of the step-induced magnetic anisotropy at the NiO / Fe3O4 interface leads to a progressive and thickness dependent reduction of the spin and orbital magnetic moment in the Fe3O4 (100) near interface. XMCD quantitative data shows strong modifications of the spin and orbital magnetic moments in the Fe3O4 film as well as strong magnetic anisotropies induced by the structural anisotropy of the grown NiO layer. The magnetic coupling between antiferromagnetic (AFM) NiO and ferrimagnetic (FM) Fe3O4 layers has been studied in the ultra-thin film limit. Performing circular and linear magnetic dichroïsm we describe the size dependent magnetic coupling at the NiO/ Fe3O4 interface. The strength of the step-induced magnetic anisotropy at the NiO / Fe3O4 interface leads to a progressive reduction of the spin and orbital magnetic moment in the Fe3O4 (100) near interface. We reveal the influence of the crystallogrophic and structural anisotropy in the NiO films leading to exchange induced specific magnetic properties in the Fe3O4 (100) layer.

[1] S. Stanescu, C. Boeglin, A. Barbier, J.P. Deville, Phys. Rev. B 67 (2003) 035419, [2] S. Stanescu, C. Boeglin, A. Barbier, J.P. Deville, Surface Science 549 (2004) 172-182. [3] PhD Thesis S. Stanescu – Univ. Strasbourg 2002 [4] H. Ohldag et al. PRL 86 (2001) 2878, H. Ohldag et al. Phys. Rev. Lett. 87 (2001) 247201

100

Fig. 1 : Sample holder with two rotation axis (incidence angle θ and azimuth angle φ) and temperature ranging from 4 K to 300 K. The applied magnetic field can be set to a maximum value of 7 Tesla.

0.30

12Å NiO / Fe3O4 (100)
Fe L2,3 edges, 7T, T=200 K

0.08

0.25

I / Io (arb. units)

0.20

σ+ σ0.06

M perp E M // E

0.04

M perp E

I/Io ( arb. units)

Intensity (arb.units)

0.15

0.02

M // E

0.10

XAS

863

864

865

866

867

868

869

870

0.05

0.04

Energy (eV)

XMCD
0.00

-0.05

0.02

-0.10

-0.15

840
700 710 720 730 740

850

860

870

Energy (eV)

Energy (eV)

Fig. 2 : (a) X-ray absorption spectra (XAS) taken at the Fe L2,3 edges with two oppositely circularly polarized light and the dichroism signal (XMCD) obtained at 200 K in an applied field of 6.5 Tesla. (b) XAS taken at the Ni L2,3 edges with vertical and horizontal polarized light in normal incidence in the remanent state and at 200 K .

101

XMCD @ FeL2,3

spin moment (µ B / f.u.)

4.0 3.5 3.0 2.5

Fe3O4(100)

[100] [110]
12 Å NiO/Fe3O4(100)

8 nm NiO (PLD)/Fe3O4(100)

2.0

0

10

20

30

40

50

60

70

80

90

Angle θ°

Fig. 3 : Angular evolution of the spin magnetic moment for clean Fe3O4(100) before and after NiO growth.

102

Ion spectroscopy on a free water cluster jet
Silko Barth,1 Volker Ulrich,1 Sanjeev Joshi,1 Axel Reinköster,2 and Uwe Hergenhahn1a 1 Max-Planck-Institut für Plasmaphysik, Boltzmannstr. 2, 85748 Garching 2 Fritz-Haber-Institut der Max-Planck-Gesellschaft, Faradayweg 4-6, 14195 Berlin The investigation of water clusters can help to understand the properties of the hydrogen network in liquid water, which still challenges theoretical and experimental physical chemists. We have constructed a molecular beam source for water clusters to carry out experiments at BESSY. For this project, in 2005 two weeks of beamtime at a dipole beamline and two weeks of single bunch beamtime at an undulator were alloted. Experimental: Water clusters can be formed by expanding a jet of water vapour from a heated reservoir into vacuum. Our device is based on this principle, resembling work of other research groups [1]. The reservoir has a volume of 12 ml and it was heated up to 94°C. The hole nozzle with a diameter of 50 µm was operated at a slightly higher temperature of 102°C. Our vacuum setup consists of an expansion chamber (p = 1.6e-3 mbars) extending into the main vacuum chamber (p = 3e-6 mbars). The two volumes are separated from each other by a conical skimmer. The vacuum in the expansion chamber was maintained by two turbomolecular pumps with a total pumping speed of 390 l/s, while in the main chamber two turbomolecular pumps (1200 l/s) and a cryopump (900 l/s) evacuated the system. Clusters were ionized by synchrotron radiation of the TGM4 and the UE52/SGM beamlines. Ions were detected by a linear time-of-flight mass spectrometer with an acceleration region of 6 mm and a drift tube of 80 mm length (see also [2]). The extraction field was pulsed with a fixed frequency of 12.5 kHz for experiments in the multi-bunch mode. In single-bunch mode, alternatively it was possible to operate the instrument with static extraction fields and to measure the ion time-of-flight relative to the synchrotron radiation pulse.
M18+

10

4

M17+

hν = 39 eV

Intensity (events)

10

3

M28+

M32+

M37+

M55+

M16+

M73+

M91+

M109+

M1+

M127+

10

2

10

1

0

M14+

M2+

2

4 Time of flight (µs)

M145+ M163+ M181+ M199+ M217+ M235+ M253+

6

8

Figure 1: Mass spectrum of our water cluster jet. The largest observed fragment, 253 amu, pertains to the parent cluster (H2O)15.

Results: A typical mass spectrum obtained in multi-bunch mode is shown in Fig. 1. Only protonated water cluster ions can be detected, which is the behaviour observed in experiments with multi-photon laser ionization and electron impact ionization [3]. Due to the large
a

Mailing address: IPP, c/o BESSY, Albert-Einstein-Str. 15, 12489 Berlin, E-Mail: uwe.hergenhahn@ipp.mpg.de

103

geometry change between the neutral and the ionized cluster, after ionization the clusters stabilize by emission of a neutral OH group. Near-edge X-Ray Absorption (NEXAFS) spectra were recorded along the O 1s ionization edge. The partial yield for selected fragments is shown in Fig. 2. The yield curves for the molecular fragments agree with published data [4]. The first two pre edge excitations have been assigned to 1s excitations into the 4a1 and 2b2 antibonding orbitals. All features at higher energy have Rydberg character.
+

(H2O)4H (H2O)3H (H2O)2H (H2O)H H2O OH
+ +

max. 2811

+

max. 4057

+

max. 7940

Intensity (arb. u.)

+

max. 14255

max. 16069

+

max. 1.12e+05

O

max. 1.46e+05

O

2+

max. 8867

H

+

max. 360331

530

535 540 545 Photon Energy (eV)

550

Figure 2: Partial ion yield curves of a free water cluster jet. Measurements were made with a time-of-flight analyser using a static extraction field in single bunch mode. The ionization threshold is 539.9 eV for molecular water (arrow) and about 1.4 eV below that for water clusters [1].

The yield curves for the cluster fragments show a marked dissimilarity with the molecular ones. Indeed, qualitatively they agree much better with published X-Ray absorption data for ice [5]. Comparing with the curves published in [5], we can say that the yield curves from the protonated cluster fragments we observe are in-between the data for surface and bulk ice. A feature typical for the NEXAFS curve of bulk ice is the shape-resonance like enhancement around 541 eV, a typical feature for NEXAFS of surface ice – and for liquid water – is the pronounced pre edge peak around 534.5 eV. Similar ion yield scans for the lightest two cluster fragments have been performed by Björneholm et al. [6]. We can obtain additional insight on the nature of these signals by comparing partial yield curves recorded with static and pulsed extraction fields. Due to the fixed, rather low
104

frequency of the pulsed field, in this case only fragments with a low or vanishing kinetic energy release are sampled. This comparison is shown in Fig. 3. While the curves coincide over most of the data range, the shape resonance-like amplification appears to be less pronounced for pulsed extraction fields.
(H2O)4H static pulsed
+

2.5 2.0 1.5 1.0 0.5 0.0

0.8 0.6 0.4 0.2 0.0

530

535 540 545 Photon Energy (eV)

550

Figure 3: Comparison of the yield curve for the fragment with mass 73 amu measured with a static and a pulsed extraction field. Pulsed extraction fields with two different voltages were used in the experiment (black: 1240 V, blue 310 V).

More experimental work is needed to clarify the parentage of the fragments shown in Fig.s 2, 3 however. Preliminary ion-ion coincidence data recorded as part of this work show that above the O 1s ionization edge massive fragmentation of the clusters sets in as can be expected. We have demonstrated that experimental ion spectroscopy on a water cluster beam can contribute to the understanding of this fascinating and most important substance. Partial funding by DFG project He 3060/3-1,2 is gratefully acknowledged. References
[1] C. Bobbert and C. P. Schulz, Eur. Phys. J. D 16, 95 (2001); G. Öhrwall, R. F. Fink, M. Tchaplyguine, L.
Ojamäe, M. Lundwall, R. R. T. Marinho, A. N. d. Brito, S. L. Sorensen, M. Gisselbrecht, R. Feifel, T. Rander, A. Lindblad, J. Schulz, L. J. Saethre, N. Martensson, S. Svensson, and O. Björneholm, J. Chem. Phys. 123,

054310 (2005).
[2] [3] A. Reinköster, S. Korica, G. Prümper, J. Viefhaus, K. Godehusen, O. Schwarzkopf, M. Mast, and U. Becker, J. Phys. B 37, 2135 (2004). P. P. Radi, P. Beaud, D. Franzke, H.-M. Frey, T. Gerber, B. Mischler, and A.-P. Tzannis, J. Chem. Phys. 111, 512 (1999).

[4]
[5] [6]

M. N. Piancastelli, A. Hempelmann, F. Heiser, O. Gessner, A. Rüdel, and U. Becker, Phys. Rev. A 59, 300 (1999).
P. Wernet, D. Nordlund, U. Bergmann, M. Cavalleri, M. Odelius, H. Ogasawara, L. A. Näslund, T. K. Hirsch, L. Ojamäe, P. Glatzel, L. G. M. Pettersson, and A. Nilsson, Science 304, 995 (2004). O. Björneholm, F. Federmann, S. Kakar, and T. Möller, J. Chem. Phys. 111, 546 (1999).

105

Intensity (Pulsed Field, arb. u.)

Intensity (Static Field, arb. u.)

1.0

Multiplet splitting in X-ray Absorption Spectroscopy of Small, Size-selected Transition Metal Clusters on Alkali Metal Surfaces
J.T. Lau1 , L. Glaser2 , S. Hankemeyer2 , M. Martins2 , T. M¨ller1 , and W. Wurth2 o
2

Technische Universit¨t Berlin, IAPF, PN3-1, Hardenbergstraße 36, 10623 Berlin a Universit¨t Hamburg, Institut f¨r Experimentalphysik, Luruper Chaussee 149 22761 Hamburg a u e-mail: tobias.lau@tu-berlin.de

1

Electron localization and correlation are important factors determining the electronic structure of 3d transition metals [1], which retain some of their atomic properties even as itinerant bulk metals [2]. In free 3d transition metal atoms, Coulomb interaction of the 2p core hole with the localized 3d electrons leads to multiplet splitting of the L2,3 absorption lines [3]. In bulk transition metals, valence electrons are partially delocalized. In addition, multiplet splitting might be masked by line broadening. The issue of electron localization/delocalization and correlation in the transition from single atoms to bulk metals can ideally be addressed in the investigation of size-selected transition metal clusters. So far, however, the 3d multiplet structure in L2,3 X-ray absorption of deposited small metals clusters has not been observed neither on metal nor on HOPG substrates [4–7]. We report here on the first study of the multiplet structure of small chromium clusters deposited on thin potassium films on Cu(100). Potassium films were chosen as substrates, since 3d transition metal impurity atoms on alkali metal films show highly localized atomic configurations, as was conclusively shown in XAS for iron, cobalt, and nickel atoms [8].
Cr4 / K / Cu(100)

X-ray absorption [arb. units]

Cr3 / K / Cu(100)

Cr1 / K / Cu(100)

Cr atom (Arp et. al.)

565

570

575 580 585 photon energy [eV]

590

595

Figure 1: Chromium L2,3 X-ray absorption spectra of small chromium clusters on thin K/Cu(100) films measured at 30 K.

For sample preparation, a Cu(100) single crystal was cleaned by repeated sputter and anneal cycles. Thin potassium films were prepared in UHV by evaporation of potassium from alkali metal dispensers (SAES getters). To obtain high quality thin films, these samples were annealed to the onset of potassium desorption [9]. From thermal desorption spectroscopy, we estimate the potassium film thickness

106

to be about 10–15 monolayers. Chromium clusters were produced in a sputtering source [10] and deposited onto the potassium films by a soft-landing technique, using krypton as a buffer layer. For krypton desorption after cluster deposition, the sample was heated to 100 K. All sample preparation was carried out in UHV with a base pressure of less than 1 × 10−10 mbar. X-ray absorption (XA) spectra were taken in total electron yield mode. All spectra shown here were normalized to incoming photon flux. For the chromium spectra, a linear background has been fitted to the pre-edge region and was subtracted to account for a substrate absorption signal.
X-ray absorption [arb. units]

Crn / K / Cu(001) annealed

565

570

575 580 585 photon energy [eV]

590

595

Figure 2: Chromium L2,3 X-ray absorption spectra of small chromium clusters on thin potassium films after annealing to 320 K. The spectra are qualitatively similar for Cr1 , Cr3 , and Cr4 .

The L2,3 X-ray absorption spectra of Cr1 , Cr2 , and Cr4 on K/Cu(100), depicted in figure 1 clearly show a multiplet structure at the L3 edge, proving that atomic-like line shapes can indeed be observed for small metal clusters on suitable metal substrates. Three different components can be distinguished at the L3 edge of these clusters. In the corresponding spectrum of free chromium atoms [11], also shown in figure 1, three intense lines at the L3 edge and two intense lines at the L2 edge carry the main intensity. Although a detailed analysis of the experimental data is still in progress, comparison of the spectra in figure 1 suggests that the same components might also be visible in small clusters. However, they are not resolved at the L2 edge of the clusters. The intensity at the L2 edge is much higher for the clusters than for the atom. For the clusters, the branching ratio seems to be rather bulk like, as can be seen by comparison with electron energy loss spectra [12].
X-ray absorption [arb. units]

Crn / K / Cu(001)

before annealing

after annealing

290

292

294 296 298 300 photon energy [eV]

302

304

Figure 3: Potassium L2,3 X-ray absorption spectra of Crn / K / Cu(100) before and after annealing to 320 K. The spectra are qualitatively similar for Cr1 , Cr3 , and Cr4 . No baseline has been subtracted.

Annealing the cluster samples to 320 K, just below the onset of potassium desorption, leads to visible changes in the XA spectra. After annealing, the spectra for all three cluster sizes studied are qualitatively very similar. An averaged spectrum is shown in figure 2. A preliminary analysis suggests that peak intensities change upon annealing, while peak positions seem to remain fixed. The main intensity at the L3 edge is now concentrated in the medium energy line. Since all clusters yield the same spectra, annealing most likely results in cluster disintegration and chromium atom diffusion into

107

the potassium film. This might be the result of high atomic mobilities in the potassium film close to the desorption temperature. The effect of annealing on the potassium underlayer can be seen in figure 3. Here, potassium L2,3 absorption spectra of the potassium films before and after annealing to 320 K are compared. Again, all spectra are qualitatively similar for all clusters studied. Before annealing, the spectra show a slight double peak structure in the spin-orbit split absorption lines. After annealing, the spectral features are narrower, as was also observed for CO/K/Ni(100) [13]. Here, the splitting of the lines was explained by crystal field splitting and loss of metallic character of the potassium film. A thorough analysis of the Cr/K/Cu(100) spectra is currently in progress. A slight decrease in intensity of the potassium L2,3 XA spectra probably indicates potassium desorption upon annealing. The slight double peak structure visible in potassium L2,3 XA might result from the krypton desorption procedure, where the sample is temporarily heated to 100 K and adcluster diffusion might already occur. In summary, we could observe for the first time atomic-like absorption lines of small transition metal clusters on metal substrates. Further investigations will yield a more detailed understanding of the evolution of electronic properties in small metal clusters. We gratefully acknowledge technical support by BESSY staff members as well as stimulating discussions with P. Gambardella, P. Zimmermann and A. F¨hlisch. o

References
[1] G. van der Laan, J. Electron Spectrosc. Relat. Phenom. 117–118, 89 (2001). [2] M. Magnuson, N. Wassdahl, A. Nilsson, A. F¨hlisch, J. Nordgren, and N. M˚ o artensson, Phys. Rev. B 58, 3677 (1998). [3] M. Martins, K. Godehusen, T. Richter, P. Wernet, and P. Zimmermann, J. Phys. B: At. Mol. Opt. Phys. 39, R79 (2006). [4] J. T. Lau, A. Achleitner, and W. Wurth, Surf. Sci. 467, 834 (2000). [5] J. T. Lau, A. F¨hlisch, R. Nietuby`, M. Reif, and W. Wurth, Phys. Rev. Lett. 89, 057201 (2002). o c [6] K. Fauth, S. Gold, M. Heßler, N. Schneider, and G. Sch¨tz, Chem. Phys. Lett. 392, 498 (2004). u [7] M. Reif, L. Glaser, M. Martins, and W. Wurth, Phys. Rev. B 72, 155405 (2005). [8] P. Gambardella, S. S. Dhesi, S. Gardonio, C. Grazioli, and C. Carbone, Phys. Rev. Lett. 88, 047202 (2002). [9] W. Frieß, Ph.D. thesis, Technische Universit¨t M¨nchen, 1995. a u [10] J. T. Lau, A. Achleitner, H.-U. Ehrke, U. Langenbuch, M. Reif, and W. Wurth, Rev. Sci. Instrum. 76, 063902 (2005). [11] U. Arp, K. Iemura, G. Kutluk, T. Nagata, S. Yagi, and A. Yagishita, J. Phys. B: At. Mol. Opt. Phys. 28, 225 (1995). [12] J. Fink, T. M¨ller-Heinzerling, B. Scheerer, W. Speier, F. U. Hillebrecht, J. C. Fuggle, J. Zaanen, and u G. A. Sawatzky, Phys. Rev. B 32, 4899 (1985). [13] J. Hasselstr¨m, A. F¨hlisch, R. Denecke, and A. Nilsson, Phys. Rev. B 62, 11192 (2000). o o

108

Study of chemically synthesized bimetallic nanoparticles by XMCD
O. Margeata, D. Ciuculescua, C. Amiensa, B. Chaudreta, P. Lecanteb, K. Fauthc, M. Kawwamc, G. Schützc
a: LCC-CNRS, 205, route de Narbonne, F-31077 Toulouse b: CEMES-CNRS, 29, rue Jeanne Marvig BP 4347 F-31055 Toulouse c: Max-Planck-Institut für Metallforschung Heisenbergstr. 3 D-70569 Stuttgart

Soft ferromagnets are of interest for many applications; especially in the nanosized regime they could be relevant to reduce the size of inductors for on chip electronic devices. Size control of NiFe nanoparticles has recently been achieved in our group via the combined use of organometallic precursors and organic ligands (such as long chain carboxylic acid and amine mixtures) during a liquid phase synthesis process[1]. The particles have been characterized by transmission Electron Microscopy (TEM-Figure 1), Wide Angle X-Ray Scattering (WAXSFigure 2) and Extended X-Ray Absorption Fine Structure (EXAFS) measurements, and magnetic measurements by SQuID (Figure 3). A typical TEM image is displayed Figure 1, showing particles of mean size 2.4nm, well sprayed on the carbon foil of the TEM grid. Interestingly, the structure of these particles is different from that determined for the bulk phase of identical composition as evidenced by WAXS (Figure 2). This structure is surprisingly close to one adopted by Mn in the bulk (β-Mn), which means a larger distribution of metal-metal distances and a globally more compact arrangement. This effect can be related either to surface stress arising from the high surface/core ratio of atoms or to a strong effect of the chemical environment used to control the synthesis as already observed for Fe[2], Co[3] or CoRh[4] systems. Given this exotic structure and the well known sensitivity of the iron magnetic moment on structural details, one can expect that the magnetization in the nanoparticles might deviate considerably from bulk material properties. Furthermore a fit of the ZFC/FC curve leads to an anisotropy value of 3.105 J/m3, i.e. almost 200 times more than in the bulk. Accordingly we can expect a large increase of the orbital moment. This can be related to both size effect and structural changes and raises the question of the atomic distribution in the particle. We have thus studied the magnetic properties with element sensitivity by XMCD at the L2,3-edges of both nickel and iron atoms to determine their spin and orbital contributions to the magnetic properties of the NiFe nanoparticles.
80

D = 2.8 nm (σ = 0.3 nm)

Part. Numb.

60

40

20

Figure 1 : TEM image of NiFe particles and related size histogram
1 2 3 4 5

0

0

Diameter (nm)

Fe-Ni exp.
FDR (u.a.)

fcc model

cc model
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8

Figure 2 : Experimental (red) Radial Distribution Function (RDF) obtained by WAXS investigation and comparison to RDF computed from fcc (blue) or cc (green) models, and from a polydetrahedral arrangement such as in β-Mn (doted black line)

r (nm)

109

200

100

θ = -7,5K
0

0

100

200

T (K)

Figure 3 : ZFC/FC curves on diluted NiFe particles (blue) and interacting NiFe particles (red).

M (a.u.)

0

50

100

M-1

150

200

250

300

T (K)

Experiments were carried out on line PM-3. The nanoparticles were deposited as dense layers on silicon substrate in a glove box and transferred into the measurement chamber via a transfer container kept under high vacuum. XAFS spectra were first recorded at both edges (Figure 4), first on as prepared samples, and after voluntary exposure of the samples to air. First, typical absorptions are observed, characteristic of non oxidized nickel and iron atoms. It is noteworthy that in the presence of traces of oxygen, oxidation is observed at iron edge, while nickel XAFS spectrum remains unchanged. After further exposure of the sample to air, nickel atoms start to oxidize. At this stage, the iron content of the nanoparticles if fully oxidized. However, some nickel atoms remain in the metallic state. These results are indicative of a partial chemical segregation in the particles, their surface being enriched in iron while nickel atoms are mainly buried in the core of the particles. This has been further confirmed by Mössbauer studies (Figure 5). Indeed the best fit of the Mössbauer spectra evidence as major contributions: iron atoms (C4: 20%), with high IS and hyperfine field, corresponding to segregated iron atoms at the nanoparticles surface, a NiFe mixed phase (C2 and C3 : 63%) and iron atoms in a nickel rich environment (C1: 15%). These results are once again consistent with a nickel rich core surrounded by an iron rich surface, with a progressive enrichment in iron atoms from the core to the surface of the nanoparticles.
Sample signal norm. by ring current (arb. units)

Ni L3,2

Sample signal norm. by ring current (arb. units)

Fe L3,2

840

860

880

700

710

720

730

740

Photon energy (eV)

Photon energy (eV)

Figure 4: Effect of progressive oxidation on XAFS spectra at Ni and Fe L2,3 edges

Figure 5: Mössbauer spectrum recorded at 4K and the 5 contributions to the fit.

Magnetisation loops recorded both at low temperature and close to room temperature at both Fe and Ni edges show that iron and nickel are strongly coupled despite the core-shell structure of the nanoparticles.[5] XMCD signals were then recorded at various temperatures (Figure 5, T=85K, B = 2.5T) and the magnetic spin and orbital contributions were calculated [6]. Results are reported Table 1. At 14K the value of the magnetic moment per atom is 1.52 µB, which allows to estimate the value at saturation: µ =1.58 µB, in excellent agreement with the value determined previously from SQUID measurements (1.69 +/- 0.05 µB at 2K, 5T).[1]
Sample signal norm. to ring current (10-14A/mA)
4,5

T = 85 K

Sample sign. norm. to ring current (10 A/mA)

7,0

T = 85 K

4,0

-14

6,0

3,5

5,0

0,0

0,0

Figure 6: XMCD signal recorded at 85K at both edges

-0,2

Ni L3,2
840 860 880 900 920

-0,5

Fe L3,2
680 700 720 740 760 780

Photon energy (eV)

Photon energy (eV)

110

Fe Ni µS µL µL/µS µS µL µL/µS 14K 1.8 0.11 0.06 1.0 0.13 0.13 85K 1.8 0.14 0.08 0.95 0.09 0.09 268K 1.6 0.08 0.05 0.93 0.02 0.02 Table 1: Ni and Fe spin and orbital magnetic moments as deduced from XMCD measurements (14K, 2.5T) These results evidence a strong polarisation of nickel atoms and large orbital contributions as compared to the bulk metal. However, whereas this is the direct consequence of size reduction or chemical segregation inside the particles, is an open question. Acknowledgements: We thank C.-H. Fischer and I. Lauermann (both HMI) for access to the CISSY glove-box system, and the permission to adapt it for our transfer purposes. Thanks also to A. Vollmer and M. Neeb (BESSY) for the vacuum transfer chamber. This project was supported by the EU. Contract number : RII 3 CT-2004-506008.

[1] (a) O. Margeat et al. Synthèse et propriétés de nanoparticules de FeNi obtenues par voie organométallique. Oral presentation Journée Grand Sud-Ouest, SFC, Oct. 2004, Toulouse. (b) Synthesis of NiFe Nanoparticles : a new structure, new physical properties. O. Margeat, C. Amiens, B. Chaudret, M. Respaud, P. Lecante, in preparation [2] (a) C. Amiens et al. Magnetic Properties of Iron Nanoparticles of New Crystal Structure. Oral presentation, MRS, Boston Nov; 2004. [3] F. Dassenoy, M.-J. Casanove, P. Lecante, M. Verelst, E. Snoeck, A. Mosset, T. Ould Ely, C. Amiens, B. Chaudret, Experimental Evidence of Structural Evolution in Ultra Fine Cobalt Particles Stabilized in Different Polymers. From a Polytetrahedral Arrangement to the Hexagonal Structure. J. Chem. Phys. 2000, 112, 81378145. [4] D. Zitoun, M. Respaud, C. Amiens, B. Chaudret, A. Serres, M.-J. Casanove, M.-C. Fromen, P. Lecante, Magnetic Enhancement in Nanoscale CoRh Particles, Phys. Rev. Lett., 2002, 89(3), 37203. [5] O. Margeat, D. Ciuculescu, C. Amiens, B. Chaudret, P. Lecante, K. Fauth, M. Kawwam, G. Schütz, XMCD Studies of FeNi Nanoparticles Synthesized by an Organometallic Approach, Poster, User meeting, Bessy 2005 [6] O. Margeat, D. Ciuculescu, C. Amiens, B. Chaudret, P. Lecante, K. Fauth, M. Kawwam, G. Schütz, manuscript in preparation.

111

Electron correlation in Rare Earth Phosphate Nano-particles studied by NEXAFS and Resonant inelastic X-ray scattering
E. Suljoti, A. Pietzsch, M. Nagasono, A. Föhlisch and W. Wurth Institut für Experimentalphysik, Universität Hamburg, Luruper Chaussee 149, 22761 Hamburg
Lanthanide phosphate nanocrystals (LaPO4 – YbPO4), capped by organic ligands were chemically prepared in liquid-phase synthesis [1]. These nanocrystals have a mean-size of 5-6 nm, high crystallinity and a good solubility in different organic solutions as well as a high luminescence quantum yield, which makes them useful in a variety of applications. As a consequence, it is interesting to study their electronic structure, as a function of lanthanide-ion, cluster size and interface between cluster and ligands. In order to investigate the unoccupied density of states of these clusters we have recently performed near edge xray absorption fine structure spectroscopy (NEXAFS). The Lanthanide nanocrystals were dissolved in 1% solutions of Methanol, and spin-coated on p-type (100) silicon wafers. The Si surfaces were hydrogen-passivated by HF etching to remove the surface oxide. Scanning electron micrographs (SEM) of the spin coated layers showed an uniform size distribution of the nanocrystals in a densely packed layer on the Si surface. Using the example of ErPO4 nanocrystals, we observe significant differences in the Er N4,5 –edge (4d) NEXAFS depending on the detection mode we employ (Figure 1). In Total Electron Yield (TEY) we measure different spectral weights in comparison to the Partial Fluorescence Yield (PFY) (0-30eV energy loss), where only a subset of decay channels is monitored.

Figure1. NEXAFS of a monolayer ErPO4 colloidal nanocrystals spin coated on a Si (100) wafer. Significant differences in Total Electron Yield and Partial Fluorescence Yield occur.

112

Figure 2. Resonant inelastic X-ray scattering in ErPO4 nanocrystals for different excitation energies according to Fig. 1. Different valence electron-hole final states depending on the excitation energy are created In the previous part of our investigation, we have established that all the lanthanide-ions in the rare earth phosphate nanocrystals, including Er are in a ‘3+’ configuration. To investigate the electron correlation, indicated by the NEXAFS spectra (Fig. 1), we have performed resonant inelastic X-ray scattering (RIXS), with highly selective resonant and detuned excitation. In Figure 2 the ErPO4 RIXS are summarized. As the inelastic scattering process leads to a valence excited final state (1h,1e), different excitation energies emphasis different electron/hole final states. The focus of our investigation is now to determine deviations between nano-scale crystals and macroscopic crystals, which differ in terms of the surface contributions and eventually in their intrinsic geometric structure. We are thankful for the nanocrystals by Prof. M. Haase und K. Rücker. We acknowledge support by the DFG Graduiertenkolleg ‘Spektroskopie an lokalisierten atomaren Systemen, Felder und lokalisierte Atome – Atome und lokalisierte Felder’ and the SFB 508 ‘Quantenmaterialien – laterale Strukturen, hybride Systeme und Cluster –‘ . [1] O. Lehmann, H. Meyssamy, K. Kömpe, H. Schnablegger and M. Haase, J. Phys. Chem. B 107 (2003) 7449.

113

Magnetic properties of CoPt clusters and nano particles
L. Glaser , M. Martins∗ , M. Wellh¨fer∗ , W. Wurth∗ , V. Alesandrovic† , H. Weller† , C. Boeglin‡ o
Universit¨t Hamburg, Institut f¨ r Experimentalphysik, Luruper Chaussee 149, D-22761 Hamburg a u † Universit¨t Hamburg, Institut f¨r Physikalische Chemie, Grindelallee 117, D-20146 Hamburg a u ‡ Institut de Physique et Chimie des Materiaux de Strasbourg, 23 rue du Loess BP43, 67034 Strasbourg
∗

∗

Small metal clusters show a strong size dependence of their physical properties. Element specificity and the ability to investigate low target densities makes x-ray magnetic circular dichroism (XMCD) an ideal tool for the exploration of the magnetic structure of low coverage cluster systems [1–4]. In the context of possible applications in magnetic storage media, alloys of 3d metal atoms as Co or Fe with heavy elements like Pd or Pt are promising candidates. Therefore we have started to investigate the magnetic properties of CoPt systems in the size range from very small, mass selected clusters made of several atoms up to nano particles with a diameter of a few nm. The experiments on the mass selected clusters have been performed using our UHV cluster source, which is described in detail in [5]. The small clusters are generated by high energy ion bombardment (Xe+ ) of a cobalt/platinum target. After mass separation using a magnetic dipole field, the clusters are decelerated and deposited onto a thin, out of plane magnetized Fe/Cu(100) surface utilizing krypton as buffer gas in a soft landing scheme. XMCD measurements have been performed under UHV conditions at a base pressure below 3 · 10−10 mbar. The samples have been prepared in situ under UHV conditions and the dichroism spectra have been recorded using the total electron yield (TEY), i.e. the sample current. The measurements have been performed at beamline UE52/1-SGM in a normal incidence geometry. The nano particles have been prepared chemically [6] and are deposited on a Si wafer as a substrate. Different methods, as spin and dip coating have been applied to create nano particle thin films with a coverage up to one monolayer. The XMCD measurements on the nano particles have been performed in a strong magnetic field up to 7 T using a super conducting magnet at the beamlines UE46-PGM and UE56/2-PGM/1 also in a normal incidence geometry.
0.6 0.5

B=7T

Photo absorption (arb. units)

Left

0.4 0.3 0.2 0.1 0 XMCD −0.1 −0.2 −0.3 760 770 780 790 800 810 820 Right

Photon energy (eV)
Figure 1: Typical XMCD spectrum of a CoPt3 nano particles sample taken with a magnetic field B=7 T at the Co L2,3 edges. The inset shows an SEM image of the sample. Figure 1 shows a typical XMCD spectrum of a CoPt3 sample at the Co L2,3 edges. The inset shows a SEM image of the sample. For possible future applications the chemical stability and the aging of the nano particles is highly important. Figure 2 shows the spectra of two different CoPt3 samples of different age and size. In this case the 7.6 nm sample has been measured approx. 6 month after the synthesis of the nano particles, whereas the 5.0 nm sample is only 2 weeks old. 1

114

Clearly in the older 7.6 nm sample a fine structure can be observed, due to oxidation of the Co. Also the newer sample shows some fine structure, however, to a much lower extend. The observed
8

6

Magnetic Field 6T Samples at 150K
little oxidization
left circular scan right circular scan

Photoabsorptionintensity (arb. u.)

4

2

5.0nm

0 8

6

strong oxidization
4

2

7.6nm
760 770 780 790 800 810

0

Photonenergy in eV

Figure 2: Absorption spectra taken from two samples (5.0 nm, new and 7.6 nm, old) with circular polarized light. The “old” sample shows a strong oxidation. fine structure has no influence on the dichroism spectrum, which has also the shape as in figure 1. In figure 3 the spin and orbital moments of different samples, which have been calculated using the XMCD sum rules [7], are depicted. Whereas the orbital moment is not affected by the aging (oxidation) the spin moment drops by 30% from ≈ 2.3µB for the new samples down to ≈ 1.6µB for the older samples.
New
2.5

Old

Samples at 150K Magnetic field 6T

2.0

magnetic moment in mb

1.5

1.0

spin moment orbital moment

0.5

0.0

290305c 220305b

010405b 100305a

200_SC 170

69_50b 170SC

145SC 145

Sample

Figure 3: Magnetic spin and orbital moments of different CoPt3 nano particles. To study the influence of the Pt atoms on the Co atoms in detail we have measured the XMCD spectra of mass selected Con Ptm clusters; a typical mass spectrum using a 25% Co/75% Pt alloy target is shown in figure 4. In the right part of figure 4 a typical absorption and XMCD spectrum is shown for deposited CoPt2 clusters. Due to the low coverage of only 3% of a monolayer, the photoabsoprtion and the XMCD signal a very weak. This makes the ongoing analysis of the spectra complicated, due to the strong EXFAS oszillation from the thin Fe film.

2

115

Pt3 Co Co 1 Co 2 Xe 1 Pt

Clustersignal Clustersignal (10−fold)

Clustersignal (arbitrary units)

Co Pt 2 / Fe / Cu(100)
total electron yield (arb.u.) left right

Pt1 Co

Pt4 Co Pt5 Co

XMCD

Pt1 Co 2

Pt2 Co

Pt1 Co 3 Pt2

Pt2 Co 2 Pt3
500 600 700 800 900 1000 1100 1200

100

200

300

400

760

770

780

790

800

810

820

Clustermass (atomic units)

photon energy in eV

Figure 4: Left: Mass spectrum obtained by our cluster source using a 25% Co, 75% Pt alloy target. Right: X-ray absorption and XMCD spectrum measured for deposited CoPt2 clusters.

The authors would like to thank the BMBF for financial support through grant 05KS4GUB/6 and the BESSY staff for their support during the beamtimes.

References
[1] F. Gambardella, A. Dallmeyer, K. Maiti, M. Malagoli, W. Eberhardt, K. Kern, and C. Carbone, Nature 416, 301 (2002). [2] F. Gambardella, S. Dhesi, S. Gardonio, C. Grazioli, P. Ohresser, and C. Carbone, Phys. Rev. Lett. 88, 047202 (2002). [3] P. Gambardella, S. Rusponi, M. Veronese, S. S. Dhesi, S. Rusponi, M. Veronese, S. S. Dhesi, P. H. Dederichs, K. Kern, C. Carbone, et al., Science 300, 1130 (2003). [4] J. T. Lau, A. F¨hlisch, R. Nietubyc, M. Reif, and W. Wurth, Phys. Rev. Lett. 89, 057201 o (2002). [5] J. Lau, A. Achleitner, H.-U. E. amd U. Langenbuch amd M. Reif, and W. Wurth, Rev. Sci. Instr. 76, 063902 (2005). [6] E. Shevchenko, D. Talapin, A. Rogach, A. Kornowski, M. Haase, and H. Weller, J. Am. Chem. Soc. 124, 11480 (2002). [7] G. van der Laan, J. Electron Spectrosc. Relat. Phenom. 101-103, 859 (1999).

3

116

ASAXS investigation of gold nanoparticles in glass
1

A. Hoell, 1I. Zizak, 2D. Tatchev, 1S. Haas, 3M. Eichelbaum, 3K. Rademann, 1J. Banhart 1 Hahn-Meitner-Institut, Department of structural research, D-14109 Berlin 2 Institute of Physical Chemistry, Bulgarian Academy of Science, BG-1113 Sofia 3 Humboldt-Universität zu Berlin, Institut für Chemie, D-12489 Berlin

Colorless glasses doped with very few Au atoms become characteristically ruby-colored after annealing because of the formation of gold colloids. The absorption of light is caused by an exitation of collective oscillations of gold valence electrons, called surface Plasmon resonance. The wavelength of the Plasmon depends on size, shape, topology and the dielectric environment of the metal clusters [1, 2]. Some years ago the nonlinear optical properties of gold nanoparticles became accessible. Therefore it is a challenge to control locally the size, and shape of gold clusters in glasses. The nucleation process of gold clusters can be activated by using synchrotron radiation [3]. Nanosized gold clusters were obtained in glasses of composition 70SiO2-20Na2O10CaO (mol %) which contains 0.01 mol% of dispersed Au. The gold nanoparticles were grown during annealing at 550oC up to 30 min in regions previously irradiated by 32 keV Xrays. These glasses exhibit interesting nonlinear optical properties that depend on the size and the size distribution of the nanoparticles. Therefore, the aim is to estimate the particle size distributions and their homogeneity by ASAXS.
40
30 minutes 20 minutes 10 minutes & unannealed glass

2D-detector [thousends counts]

30

20

1.0 Transmission

10

0.5

0 15

20

25

30

35

40

0.0 45

Sample holder position [mm]
Figure 1: After annealing for 30 minutes the irradiated area scatters significantly stronger than the non-irradiated and also stronger than the shorter annealed samples. In comparison the transmissions in the graphs below show no effect.

In Fig. 1 the integral detector intensities of three different annealed samples are shown as the function of the sample positions. Do to the small amount of gold the transmissions of the irradiated and non irradiated regions are the same showing that the compositions are unchanged. The differences of the intensities from the irradiated and non irradiated area collected by the 2D detector suggest that the growth behavior of gold nanoparticles is faster in irradiated regions. As shown in Figure 2 the intensity of the scattering at small angles increases with the annealing time, from 10 to 30 min, due to the process of nucleation and growth of gold nanoparticles. The strong anomalous effect near the Au absorption edge, LIII, is presented in Figure 2. 1
117

1 Intensity [a.u.]

Annealing time 10 min 20 min 30 min

0.1 X-ray energy 11873 eV 0.1 Q [nm ]
Figure 2: Scattering curves for samples annealed for different time periods at 550oC, measured in the preirradiated regions.
-1

1

Annealing time 30 minutes 1 Intensity [a.u.]

0.1

X-ray energy 11790 eV 11873 eV 11904 eV 11915 eV 1 Q [nm ]
-1

0.1

Figure 3: The scattering curves measured at different energies show very clear anomalous scattering effect. Ek(LIII)=11919 eV

A nice outcome is depicted in Figure 3. The Guinier radius of the particles is independent of the x-ray energy. This proves that the particles are homogeneous. After background subtraction, the scattering curves were fit with spherical particle model by the maximum entropy method. Figure 4 shows the differential volume fraction size distributions for two annealing times. It is seen that the particle size and volume fraction grow with the annealing time. These results represent the first anomalous small angle x-ray scattering experiment performed with the new ASAXS device installed on the 7T-WLS beamline at BESSY.

2
118

-13 Annealing time 30 minutes X-ray energy 11790 eV 11873 eV 11904 eV 11915 eV

-14 ln(I(Q))

-15
Energy; Rg

-16

-17

11790 eV; 2.32 nm 11873 eV; 2.30 nm 11904 eV; 2.29 nm 11915 eV; 2.31 nm

Particle radius 2.98+0.01 nm

0.2

0.4
2

0.6 Q [nm ]
-2

0.8

1.0

1.2

Figure 3 The Guinier radius is independent on the x-ray energy. This means that the particles are homogeneous.

1.4 1.2 dw/dR [a.u.] 1.0 0.8 0.6 0.4 0.2 0.0 1.0 1.5 2.0 2.5 3.0 R [nm]

Annealing time 20 min 30 min

3.5

4.0

4.5

5.0

Figure 4 Differential volume fraction size distributions obtained from the scattering curves by the maximum entropy method.

[1] U. Kreibig, M. Vollmer, Optical properties of Metal Clusters, Springer, Berlin, 1995, p.13. [2] R.H. Doremus, A.M. Turkalo, J. Mat. Science 11 (1976) 903-907. [3] M. Eichelbaum, K. Rademann, R. Müller, M. Radke, H. Riesemeier, W. Görner, Angew. Chem. Int. Ed. 44 (2005) 7905.

3
119

Photoelectron spectroscopy on size-selected Cu-clusters on Si N. Ferretti, B. Balkaya, A. Vollmer, M. Sperling, M. Neeb, and W. Eberhardt BESSY mbH, Albert-Einstein Str. 15, D-12489 Berlin-Adlershof

XPS and NEXAFS spectra of deposited mass-selected Cu-clusters of up to 10 atoms have been measured at the Optics-BL PM4 and compared with the respective bulk spectra. The cluster were produced by a magnetron sputter cluster source and mass-selected prior to deposition by a magnetic sector field. A typical Cu-mass spectrum, as measured behind the exit slit of the magnetic mass filter, is shown in the adjoining figure. After passing the mass filter an individual cluster mass was softly landed (~1 eV/atom) onto a biased Sisubstrate (1x1cm2). A cluster coverage of 13x1012 cluster (~100pA) was reached within 30 minutes for each sample. During

deposition the cluster current was recorded on the sample by an electrometer.

Afterwards the deposited samples were transfered from the cluster machine to the photoemission chamber SURICAT at PM4 by a battery driven mobile UHV suitcase. A pressure of < 5x10-9 mbar was maintained during deposition and transfer.

NEXAFS (Cu L3-edge) and XPS (Cu 2p3/2,1/2) spectra are shown in the adjoining figures. The absorption spectra have been recorded by measuring the photocurrent on the sample via an amperemeter. The typical signal-to-background signal was 1-10
Photocurrent
1,0

NEXAFS (Cu L3-edge) of Cun-clusters on Si

0,8

pA

(monochromator

exit

slit

0,6

setting=13.5 µm). For a better statistics 5-10 scans have been accumulated for each cluster sample; 20-30 scans for each XPS spectrum (Epass=50eV). The L3-edge is clearly shifted to higher absorption energy for all cluster by 1-1.5

0,4

Bulk Cu Cu3 / Si-wafer Cu7 / Si-wafer Cu10 / Si-wafer

0,2

0,0

930 931 932 933 934 935 936 937 938 939 940 941 942 943

Photon energy / eV

eV. Similarly, the XPS threshold, as taken from the maximum of the Cu-peak 2p3/2, shifts to

120

higher binding energy by 0.5-0.9 eV as compared to the bulk. The positive shift results from a reduced final-state screening due to the smaller number of atoms in the clusters and the missing metallicity. Another distinct difference between the bulk and cluster NEXAFS spectra is the missing fine structure beyond the L3 edge for energies >936 eV (and beyond L2 absorption feature, not shown here) as these are usually explained by van Hove singularities of a fcc-crystal structure [1] which are absent in the cluster.

A close inspection of all cluster spectra
XPS of Cu5 on Si, hν=1143 eV
Cu bulk Cu5 / Si-wafer

shows that the absorption edge slightly oscillates as function of the cluster size. The absorption energy for clusters with an odd number of atoms is 0.1-0.2 eV

Cu 2p3/2

Photoelectron counts

Cu 2p1/2

higher than this of the even numbered clusters. This can be explained by an alternating open/closed shell structure as also shown by laser photodetachment of

180

185

190

195

200

205

210

215

220

Kinetic energy / eV

free Cu-cluster anions [2]. Each time a new valence shell is excited from the 2p

core-level an increase in energy is observed. Each additional Cu-atom contributes with a single s-electron to a shell which, in case of non-degeneracy, is filled by two electrons. This leads to an up and down of the absorption energy by each additional atom. Furthermore a closed d-shell is not only indicated by the shell structure but also by the energy of the L3absorption edge which is located in the continuum, e.g. above the 2p XPS threshold, similar to the bulk. Upon oxidation, as experimentally shown for an oxidized Cu10-sample, the first absorption maximum moves below the XPS threshold due to hybridisation of the oxygen with the d-shell. From this an open d-shell results which is located below the XPS threshold as in bulk CuO. At a cluster size of 10 atoms the XPS and L-edge absorption threshold start to converge towards the respective bulk value which has to be further examined for larger clusters in the future (>10 atoms). The difference between the absorption maximum and the XPS threshold is distinctly enhanced by ~1 eV as compared to the bulk which is a clear indication of a semiconductivity of small Cu-cluster on Si. Furthermore the individual character of the cluster spectra verifies that the Cu-cluster are softly landed on the substrate without fracturing and that the clusters do not agglomerate into islands. [1] H. Ebert, J. Stöhr, S. Parkin, M. Samant, A. Nilsson, Phys. Rev. B 53, 16067 (1996). [2] C. Y. Cha, G. Ganteför, W. Eberhardt, J. Chem. Phys. 99, 6308 (1993).

121

Wave-Vector Conservation upon Hybridization of 4f and Valence-Band States Observed in Photoemission Spectra of Ce Monolayer on W(110)
D. V. Vyalikh1, Yu. Kucherenko1, 2, S. Danzenbacher1, Yu. S. Dedkov1, C. Laubschat1, and S. L. Molodtsov1,
1 2

Institut fur Festkorperphysik, Technische Universitat Dresden, D-01062 Dresden, Germany Institute of Metal Physics, National Academy of Sciences of Ukraine, UA-03142 Kiev, Ukraine
Angle-resolved resonant photoemission data for a hexagonally ordered monolayer of Ce on W(110) are presented. The spectra reveal a splitting of the 4f 0 ionization peak around a point in k space where a degeneracy with a valence-band state is expected. The phenomenon is described within a simple approach to the periodic Anderson model. It is found that the Ce 4f state forms a band and hybridization predominantly occurs between the 4 f and the valence-band states at the same wave vector.

The interaction of localized 4f states with itinerant conduction-band states leads to a series of correlation phenomena that have attracted considerable interest in the last few decades. The hybridization may lead to such intriguing phenomena as noninteger f occupation, heavy fermion behavior, or even a breakdown of Fermi-liquid properties [1]. A typical and well studied system is Ce metal where hybridization of the trivalent 4f 1(5d6s)3 with 4f 0(5d6s)4 and 4f 2(5d6s)2 configuration is responsible for the isostructural - phase transition related to a volume collapse of 15% [2]. This hybridization is reflected in photoemission (PE) by a characteristic double-peaked structure of the 4f spectral function, where in addition to the ionization peak at around 2 eV binding energy (BE) expected for an unhybridized 4f 1 ground state ( 4f 0 peak), a second feature appears at the Fermi energy (EF) reflecting the Kondo-Suhl resonance with a final-state f occupancy close Figure 1. Resonance PE spectra of a Ce to 1. For the phase the relative intensity of monolayer on the W(110) taken at the Ce 4d 4f threshold. Lower panel: normal emission PE the latter with respect to the ionization peak is spectra for different coverages, -like Ce (A) and much larger as for the phase reflecting the -like Ce (B). Upper panel: Angle-resolved PE increased hybridization of the ground state [3]. spectra for -like Ce taken along - direction of This effect may quantitatively be de- the surface Brillouin zone. scribed in light of the GunnarssonSchonhammer approach [4] to the single-impurity Anderson model [5] (SIAM) that considers the interaction of an isolated 4f 1 impurity at energy f with surrounding valence-band (VB) states via hopping processes. A natural weakness of SIAM is, however, that it ignores completely the effects of translation symmetry of structurally ordered solids. Consideration of the latter leads to the periodic Anderson model (PAM). In PE hybridization effect should be reflected by changes of the f signal as a function of a wave-vector (k vector), and, in fact, weak dispersion [6]. A possibility to check it is to study the behavior of the spectral function at points in k space where the unhybridized f state is en-

122

ergetically degenerate with a VB state. Here, the PAM predicts a splitting of the ionization peak into two components corresponding to symmetric and antisymmetric linear combinations of the electron states. An ideal candidate for such an experiment, however, is a Ce monolayer on W(110): As was shown by Gu et al. [7], Ce atoms form an ordered hexagonal overlayer on W(110) with interatomic distances shrinking continuously as Ce is added from 9% larger than that in -Ce to 3% smaller than that in the phase. PE spectra reveal a respective increase of the Kondo-peak intensity with decreasing interatomic distances as expected for - transition. The important point here is that since the overstructure (Ce monolayer) is incommensurate and reveals a different symmetry as compared to the substrate surface, only weak electron interactions with the substrate without a pronounced k dependence are expected, and the spectra will be governed by the intrinsic properties of the two-dimensional Ce layer. In the present contribution we show by means of angle-resolved resonant PE experiments at the Ce 4d 4f excitation threshold that the 4f derived spectral function reveals a splitting of the ionization peak just around a point in k space where a band of sd character crosses the BE position, f, of the unhybridized 4f 1 state. The data are analyzed within the approach to PAM [8] and excellent agreement between the theory and the experiment is obtained. The experiments were performed at BESSY using radiation from U125/1-PGM-1 undulator. PE spectra were acquired with a VG CLAM-4 analyzer. The overall energy resolution accounting for thermal broadening was set to 150 meV and angular resolution better than 1o was used. A structural Ce monolayer was grown on clean W(110) substrate kept at room temperature. All spectra were taken at h =121 eV, corresponding to the Ce 4d 4f resonance, were the electron emission from Ce 4f states considerably dominates over that from 5d states of tungsten that are close to a Cooper minimum of the photoionization cross section at this photon energy. The lower panel of figure 1 shows PE spectra taken in normal emission geometry for two different surface densities of Ce atoms on the W(110) substrate. In both cases the corresponding LEED patterns indicate a hexagonal structure of the Ce layer. The spectrum denoted as A corresponds to the lowest Ce coverage where a hexagonal overstructure in the LEED pattern related to a respective arrangement of Ce atoms is still observed. The shape of this spectrum is typical for -like Ce. Further Ce deposition leads to decreasing spacing between Ce atoms, and the shape of the spectrum is changed to that of -like Ce, with strongly enhanced intensity of the peak near EF. The spectrum denoted as B corresponds to a Ce Figure 2. Calculated PE spectra of an -like Ce(111) monolayer for different emission anmonolayer with interatomic spacing close to that gles (given in the figure). Inset: the energyof bulk -Ce. band structure for the - direction. The upper panel of figure 1 displays a set of angle-resolved PE spectra recorded along the - direction of the surface Brillouin zone for the hexagonal -Ce monolayer. One can see that the peak at EF, remains practically unchanged in both intensity and line shape. However, the same is not true for the ionization peak. When going away from normal emission, it splits in, at least, two components. This energy splitting is maximal for polar emission angles between 2o and 3o. For polar angles lar123

ger than 5o this splitting disappears and the shape of the PE spectra shows no significant differences to the one of normal emission. For interpretation of the obtained data we consider first the valence-band structure of the hexagonal -Ce(111) monolayer. Assuming that the incommensurate hexagonal Ce overlayer undergoes only weak interactions with the W(110), we have calculated by LMTO method [9] the electronic structure of a free atomic monolayer with interatomic distances of 3.42 A that are equal to those in bulk -Ce. The results of the calculations for the - direction are shown in the inset in figure 2. As follows from geometry of our experiment, for an excitation energy of 121 eV the K point is reached at a polar emission angle 12.6 o. Thus the k points in the given symmetry direction could be characterized by respective polar angles. The bottom of VB is found at the point at a BE of 2.4 eV. When going away from point the valence band shows a parabolic dispersion, and its angular momentum character becomes smoothly changed. The energy position of the unhybridized 4f level in Ce metal is about 1.5 eV [10]. Therefore, assuming the 4f states to create a dispersionless band at this energy, this f band should cross the parabolic VB close to a k point corresponding to 3o. Thus, in the region of this point hybridization effects between 4f and VB states are expected. Our following analysis concerning, how these effects influence the 4f emission, is the approach based on a simplified PAM applied recently to CePd3 [8]. The calculated PE spectra are shown in the figure 2. In inspection of the spectra one can conclude, that their shape is changed with an increase of the emission angle. The ionization peak split into two maxima that diverge from each other by up to about 1 eV, This behavior is in excellent agreement with that observed in the experimental PE spectra. Thus, the splitting of the peak at 2 eV BE in the angle-resolved spectra of the Ce monolayer on W(110) may be ascribed to the interaction of the 4f states with the parabolic VB that leads to the typical picture of two hybridized energy bands. This work was supported by the Deutsche Forschungsgemeinschaft, SFB 463, Projects TP B4 and TP B16 and BMBF Project N o 05-SF8OD1/4. We are grateful to R. Follath for the expert operation of the U125/1-PGM-1 undulator beamline.

[1]. P. Fulde, Electron Correlations in Molecules and Solids (Springer, Heidelberg, 1995). [2]. D. C. Koskenmaki and K. A. Gschneider, in Handbook on the Physics and Chemistry of Rare Earths, edited by K. A. Gschneider and L. Eyring (Elsevier Science, Amsterdam, 1978), Vol. 1, p. 337. [3]. D. Wieliczka, J. H. Weaver, D. W. Lynch, and C. G. Olson, Phys. Rev. B 26, R7056 (1982). [4]. O. Gunnarsson and K. Schonhammer, Phys. Rev, Lett. 50, 604 (1983); Phys. Rev. B 28, 4315 (1983). [5]. P. W. Anderson, Phys. Rev. 124, 41 (1961). [6]. A. B. Andrews, J. J. Joyce, A. J. Arko, Z. Fisk, and P. S. Riseborough, Phys. Rev. B 53, 3317 (1996); S. L. Molodtsov, J. Boysen, M. Richter, P. Segovia, C. Laubschat, S. A. Gorovikov, A. M. Ionov, G. V. Prudnikova, and V. K. Adamchuk, Phys. Rev. B 57, 033104 (2005). [7]. C. Gu, X. Wu, C. G. Olson, and D. W. Lynch, Phys. Rev. Lett. 67, 1622 (1991). [8]. S. Danzenbacher, Yu. Kucherenko, M. Heber, D. V. Vyalikh, S. L. Molodtsov, V. D. P. Servedio, and C. Laubschat, Phys. Rev. B 72, 033104 (2005). [9]. O. K. Andersen, Phys. Rev. B 12, 3060 (1975). [10]. Yu. Kucherenko, S. L. Molodtsov, M. Heber, and C. Laubschat, Phys. Rev. B 66, 155116 (2002).

124

OPTICAL SPECTRA OF TRIGLYCINE SULFATE CRYSTALS IN THE RANGE OF 410 EV B. Andriyevsky a), N. Esser b), A. Patryn a), C. Cobet b), W. Ciepluch-Trojanek a), M. Romanyuk c)
a) Faculty of Electronics and Computer Sciences, Technical University of Koszalin, Śniadeckich Str. 2, PL-75453, Koszalin, Poland b) ISAS – Institute for Analytical Sciences, Department Berlin, Albert-Einstein-Str. 9, 12489 Berlin, Germany c) The Ivan Franko National University of L’viv, Kyryla i Mefodiya Str. 8, UA-790005, L’viv, Ukraine The funding source and grant number: EU, R II 3.CT-2004-506008

Theoretical and experimental studies of the electron energy characteristics and optical spectra for triglycine sulphate crystal (TGS), (NH2CH2COOH)3⋅H2SO4, in the ferroelectric phase have been presented in this paper. Triglycine sulphate crystal (TGS), (NH2CH2COOH)3⋅H2SO4, is ferroelectric below 322 K belonging to the monoclinic space group P21, and above the transition temperature it becomes paraelectric with the center symmetry monoclinic space group P21/m [1-3]. Different kinds of chemical bonds are characteristic for the crystal: strong covalent-and-ion bonds in the ions NH3+CH2COO-, NH3+CH2COOH, and SO42-, and more weaker ion and hydrogen bonds between these quasi-molecular complexes. Because of low symmetry of TGS crystal the essential anisotropy of the optical functions in the fundamental absorption range hν > 5,1 eV is expected. Calculations of the band energy dispersion E(k), density of electron states (DOS), and dielectric functions ε(ω) of the TGS crystal have been performed for the first time using the CASTEP (CAmbridge Serial Total Energy Package) first principal code based on the density functional theory [4]. Measurements of the dielectric properties of TGS crystals have been done by spectroscopic ellipsometry using the synchrotron-ellipsometer [5] attached to the 3mNIM-1 off-Rowland circle normal incidence monochromator of the Berlin electron storage ring (BESSY II). The magnitude of the monochromator exit slit gave the possibility to resolve spectral changes of approximately 0.02 eV in the range of 7.3 eV. The complex reflectance ration ρ of the TGS samples was measured with an incidence angle of about 68° from 4.0 to 9.9 eV and converted to the pseudodielectric function <ε> = <ε1> + i<ε2> via the two-phase (substrate ambient) model [6]. Measurements were done for two types of sample’s: (1) the samples with a cleaved surface perpendicularly to the Y-direction of TGS crystal, and (2) the samples with a mechanically polished surfaces perpendicularly to the Y- and Z-direction. The finish polishing of TGS samples was done with a paste of 1 - 3 µ diamond grains. Most of the energy states of TGS crystal are of low dispersion in E(k). The dispersion of E(k) is generally greater for the conduction band states than that for the valence states. One of the peculiarities of the TGS band structure is flatness of the bottom states of the conducting bands. The energy band gap of the TGS crystal is indirect and corresponds to optical transitions between Γ [0 0 0] and D [-½ 0 ½] points of the Brillouin zone. The magnitude of this value Egi = 4.65 eV is close to the experimental one Egi(e) = 4.97 eV obtained from the optical absorption study [7]. The results on DOS for the crystal studied are presented in Fig. 1, 2. The upper part of the valence band (-3.0 ÷ 0.5 eV) is mainly (95%) of p-character. In the valence band energy range, 10.0 ÷ -3.0 eV, the part of p-states is equal to about 70%, whereas the part of p-states is about 20% in the range -23.0 ÷ -11.0 eV (Fig. 2).

125

30
30 total s+p total s total p

p-total p-SO4 p-glycine 20 DOS [arb. un.]
DOS [arb. un.]

20

total O total C total H total S total N

DOS [arb. un.]

10

20

10

0 -10

-5

0

E [eV]

5

10

15

10

0

-20

-15

-10

-5

E [eV]

0

5

10

15

0 -10

-5

0

E [eV]

5

10

15

Fig. 1. Densities of electron states (total, s, and p) of TGS crystal at ferroelectric phase.

Fig. 2. Densities of electron p-states (total, SO4, and glycine) of TGS crystal at ferroelectric phase. In the insertion, densities of electron states (total, s, and p) for atoms O, C, H, S, and N of the crystal.

The lower part of the conducting band (4.0 ÷ 6.0 eV) is also mainly of p-character (80%). An analysis of DOS in Fig. 2 reveals that both glycine and SO4 groups give input to the density of p-states in the upper part of the valence band (-3.0 ÷ 0.5 eV), whereas the lower part of the conducting band (4.0 ÷ 6.0 eV) is formed predominantly (98%) by the states of three glycine groups. Additional analysis has revealed that the DOS energy dependencies in this region for three different glycine groups are relatively shifted. This testifies the antibonding character of the corresponding electron states. An analysis of the DOS with regard to local bonding has revealed that the predominant part of DOS in the range of -3.0 ÷ 0.5 eV (92%) is associated with the p-states of oxygen (Fig. 2). In particular, the highest valence energy states at E = -0.2 eV are formed by the oxygen of SO4groups and glycine II group without “short hydrogen” near oxygen. The lower part of the conducting band (4.0 ÷ 6.0 eV) is formed by carbon (53%), oxygen (30%), and hydrogen (16%). In the range of 6.0 ÷ 11.0 eV, the part of the hydrogen electron states is equal to about 45%. One of the clear peculiarities of the conduction band DOS is also its mixed character related to the character (s-, and p-type) and origin (chemical elements) of electron states. The pseudo-dielectric functions <ε> = <ε1> + i<ε2> of the TGS crystal experimentally obtained for different geometries of the relative orientation “light - sample” are presented in Fig. 3. The dielectric functions ε2(ω) calculated using the CASTEP code and experimentally measured using a synchrotron radiation are presented in Fig. 4. The experimentally obtained ε2(ω) are characterized by a clear spectral band in the range of 6.6 ÷ 8.2 eV and a increase in the range of 9.0 ÷ 10.0 eV (Fig. 3,4). The spectra of ε2(ω) depend significantly on the orientation of the sample surface. The most pronounced peak of ε2(ω) is observed at the photon energy ω = 7.3 eV for geometry 1, whereas this band is absent for geometry 3 (Fig. 3). The magnitudes of ε1(ω) at ω = 4 eV (Fig. 3) agree satisfactorily with the square of the refractive indices of TGS crystals [9] mentioned in the introduction. The pseudo-dielectric functions obtained from cleaved and polished TGS crystal surfaces with the same orientation are equal within 15%. This means that the roughness degree of the polished surface does not substantially influence the pseudo-dielectric functions of the crystal in the investigated spectral range. Comparative analysis of the experimental and theoretical line shape of ε2(ω) of the TGS crystal gives best agreement assuming a scissor factor of 0.9 eV (Fig. 4).

126

5 4 3
ε1, ε2

Geom. 1 Geom. 2 Geom. 3 Geom. 4

4

3
ε2

theory, E||X, scis.=0.9 eV theory, E||Y, scis.=0.9 eV theory, E||Z, scis.=0.9 eV exper., Geom. 1

2 1 0 4 5 6 7
ω [eV]

2

1

0
8 9

4

5

6

7

ω [eV]

8

9

10

Fig. 3. Optical spectra of real ε1(ω) and imaginary ε2(ω) parts of pseudo dielectric permittivity of TGS crystals for different characteristic geometries and state of reflecting surface: Geom. 1 – Y-cut cleaved, E || X mainly; Geom. 2 – Y-cut cleaved, E || Z mainly; Geom. 3 – Z-cut polished, E || Y mainly; Geom. 4 – Zcut polished, E || X mainly.

Fig. 4. Theoretical spectra of the imaginary part ε2(ω) of dielectric permittivity of TGS crystals for the cartesian directions X, Y, Z and scissor factor 0.9 eV, and experimental spectrum of pseudo dielectric permittivity ε2(E) for the characteristic geometry 1 (Y-cut cleaved, E || X mainly).

Taking this into account, one can state that the strong spectral band of ε2(ω) with a maximum at ω = 7.3 eV corresponds to direct optical transitions at the Γ-, Y-, B, and E-points between the highest valence band (-1.07 ÷ -0.25 eV), and conduction bands between 5.4 and 6.5 eV. These transitions are associated with the valence p-states of oxygen and, predominantly, with the conducting states of hydrogen and carbon (Fig. 2). Whilst these valence p-states of oxygen are flat and therefore are of localized type, the corresponding conducting states of hydrogen and carbon are of more delocalized character. The large anisotropy of ε2(ω) in the region of ω = 7.3 eV and peculiarities of the placement of crystal’s fragments in the unit cell support the suggestion that the mentioned valence p-states of oxygen are associated with SO4 groups, whereas the conducting states of hydrogen and carbon are mainly belong to the glycine I. A similar analysis of the experimental and theoretical dielectric functions ε2(ω), band dispersion E(k), and densities of states of TGS crystal evidences an assignment of the experimental maxima in ε2(ω) at 8.35 eV and 9.55 eV to the transitions between the oxygen valence p-states, probably partially delocalized, and lower lying delocalized states of the conducting band. Small maxima in the theoretical dielectric functions of ε2(ω) in the range of 5.4 ÷ 7.0 eV (Fig. 4) correspond to weak optical transitions probably between the electronic states of the glycine groups being of the localized antibonding type in the range of 4.4 ÷ 5.4 eV. These transitions are in good agreement with a position of the long-wave edge of optical absorption of TGS crystals at ω ≥ 5.2 eV [7-9].
[1] [2] [3] [4] [5] [6] [7] [8] [9] S. Hoshino, I. Okaya, R. Pepinsky, Phys. Rev. 115 (1959) 323. M. I. Kay, R. Kleinberg, Ferroelectrics 5 (1973) 45. S. R. Fletcher, E. T. Keve, A. C. Skapski, Ferroelectrics 14 (1976) 775. V. Milman, B. Winkler, J. A. White, C. J. Pickard, M. C. Payne, E. V. Akhmatskaya, R. H. Nobes, Int. J. Quant. Chem. 77 (2000) 895. T. Wethkamp, K. Wilmers, N. Esser, W. Richter, O. Ambacher, H. Angerer, G. Jungk, R.L. Johnson, M. Cardona, Thin Solid Films 313-314 (1998) 745. R. M. A. Azzam, and N. B. Bashara, Ellipsometry and Polarized Light: North-Holland Personal Library, Amsterdam, 1987, paperback ed. A. Abu El-Fadl, Physica B: Physics of Condensed Matter 269 (1999) 60. N. A. Romanyuk, B. V. Andriyevsky, I. S. Zheludev, Ferroelectrics 21 (1978) 333. N. A. Romanyuk, A. M. Kostetsky, B. V. Andriyevsky, Physics of Solid State 19 (1977) 3095.

127

Molecular orientation of substituted phthalocyanines: influence of the substrate roughness
H. Peisert1, I. Biswas1, L. Zhang1, M. Knupfer2, M. Hanack3, D. Dini3, D. Batchelor,4 T. Chassé1 1 University of Tübingen, IPC, Auf der Morgenstelle 8, 72076 Tübingen, Germany 2 Leibniz Institute for Solid State Research Dresden, P.O. Box 270116, D-01171 Dresden 3 University of Tübingen, Inst. Organ. Chem., Auf der Morgenstelle 18, D-72076 Tübingen 4 Universität Würzburg, Am Hubland, D-97074 Würzburg, Germany Recently it was shown that the orientation of organic molecules directly at metal interfaces cannot be deduced from the orientation of the same molecule in thin film systems [1]. In this report we study the influence of the substrate roughness on the growth – we compare the orientation of several alkyl-substituted Phthalocyanine (Pc) molecules on relatively ill defined, technically relevant polycrystalline substrates and on single crystalline model substrates. We focus on the relation between the molecular orientation in the first layers and in thin films using polarization dependent X-ray absorption spectroscopy. The measurements were performed at the UE 52-PGM beamline. The energy resolution was set to 80 meV at photon energy of 400 eV. The absorption was monitored indirectly by measuring the partial electron yield using a channeltron with an applied bias voltage of 360 V (N 1s edge). The raw data are corrected by the energy dependent photon flux and by an additional linear background due to energy dependent cross sections for the absorption. The spectra were normalized to have the same absorption edge step height well above threshold.

a)

(t-but)4PcZn on Au(100)

b)
thickness: (sub-) monolayer

(t-but)4PcMg on Au(100) thickness: ~ 6 nm

π* σ*
Normalized electron yield (arb. units)

θ = 90°

π* σ *
θ = 90°

θ = 70°

θ = 70°

θ = 50°

θ = 50°

θ = 30°

θ = 30°

θ = 10°
θ = 10°

400

410

420

430

400

410

420

430

Photon energy (eV)

Photon energy [eV]

Fig. 1 N1s excitation spectra of (t-But)4PcMg grown Au(100): Different angular dependences are observed for a 10 nm thick film (a) and for a film thickness of 1-2 nm (b). The lower energy features (E < 402 eV) represent the π* resonances, whereas those features above 402 eV are related to the σ* resonances. The sketch illustrates the experimental geometry for p-polarized light.

128

As an example, in Fig. 1 we compare N1s excitation spectra for different coverages of (t-But)4PcMg on Au(100) as a function of the incidence angle θ of the linearly polarized synchrotron radiation. The geometry of the measurement is depicted in the inset of Fig. 1, normal incidence corresponds to θ = 90° and grazing incidence to θ = 10°. At grazing beam incidence of p-polarized light (i.e. the electric field vector is parallel to the beam – sample normal plane) for flat lying sp2-hybridised carbon systems, such as phthalocyanines, a maximum intensity is expected for dipolar transitions into groups of individual π* states and the opposite behavior is expected for the 1s σ* resonances. For the (sub)-monolayer coverage in Fig. 1a we observe a very clear angular dependence of the π* resonances (E < 402 eV). This means, that (t-But)4PcMg forms well defined layers, with the molecules lying flat on the Au(100) substrate surface. In contrast, for the 6 nm thick film of (tBut)4PcMg on Au(100) in Fig. 1b no clear angle dependence can be detected, therefore we suppose that the molecules are almost disordered. A similar behavior was observed for the other substituted Pc’s, a quantitative analysis of the angle-dependence of the N1s-π* resonance intensity for different coverages of 1,4-(Dec)8PcZn, 2(3)-(t-But)4PcMg and PcH2 on single crystalline gold is presented in Fig. 2. The data points for the ultrathin films and for the thicker film of the unsubstituted PcH2 follow the expected curve for lying molecules, whereas almost no intensity changes are observed for thicker films of substituted Pcs. The exception of PcH2 is not surprisingly, similar to other unsubstituted Pc’s on single crystalline substrates [3] the initial lying adsorption geometry may be locked in and is also observed in thicker films. In both cases, for substituted and unsubstituted Pc’s, it seems that the moleculesubstrate interaction causes an ordered growth within the first layers. For substituted Pc’s however, the relatively weak molecule-molecule interaction cannot enable an ordered growth on single crystalline substrates in thicker films. On the other hand, even on relatively illdefined surfaces an ordered growth for Pc molecules (partly with small substituents) was observed – the orientation in the thin films was “standing” on the substrate surface [1]. Therefore the question arises whether or not an ordered molecular growth (with a possible change of the orientation) can be obtained for thicker coverages on well defined single crystalline substrates. For this experiment we have chosen octasubstituted 1,4-(But)8PcZn with a relatively small substituent as organic, and GeS(001) as a very flat substrate with a weakly interacting (van-der-Waals) surface. Compared to PcH2 in (But)8PcZn the separation between the aromatic cores is larger and thus the molecule-molecule interaction is decreased. In this case, not the polar angle of the sample, but the direction of the E vector of the synchrotron radiation

Fig. 2 Angle dependence of the intensity of the π* resonances of (t-But)4PcMg, (Dec)8PcZn and PcH2 on Au(100). The integrated intensity of the structures < 402 eV was used for the analysis. The expected intensity profiles for standing and lying molecules are indicated by dotted and dashed lines, respectively. A preferred orientation (lying molecules) can be only observed within the first organic layers and for thicker films of the unsubstituted PcH2.

1.0 Normalized N1s-π* intensity 0.8 0.6 0.4
~ 6 nm (t-but)4PcMg

0.2 0.0 0°

20-30 nm (dec)8PcZn ~ 8 nm (H)2Pc ~ ML (t-but)4PcMg ~ ML (dec)8PcZn

standing
20° 40° 60°

lying
80°

Angle of incidence θ

129

a)
s-polarized

b)
N 1s edge
(E to sample surface)

(But)8 PcZn on GeS(001)
c)
s-polarized
s-polarized

N 1s edge

Normalized electron yield (arb. units)

thickness: ~2 nm
p-polarized (E to sample normal)

thickness: ~8 nm
p-polarized

thickness: > 40 nm
p-polarized

390

400

410

420

430

440

390

400

410

420

430

440

390

400

410

420

430

440

Photon energy (eV)

Photon energy (eV)

Photon energy (eV)

Fig. 3 Comparison of N1s excitation spectra for 1-2 nm (a), of 8 nm (b) and of more than 40 nm (c) thick films of 1,4-(But)8PcZn grown on GeS(001) at grazing incidence using differently polarized light. Whereas an orientation of the molecules parallel to the substrate surface is observed for the first layers, the spectra indicate an almost disordered growth for higher coverages and finally standing molecules in thick films.

was changed at grazing beam incidence, as illustrated in the insets of Fig. 3. For a ~2 nm coverage of 1,4-(But)8PcZn on GeS(001) we observe for p-polarized light in Fig. 3a (lower panel) a high intensity of N 1s π* transitions and a low intensity of N 1s σ* transitions, the opposite behavior is observed for s-polarized light (lying molecules). In contrast, the spectra for 8 nm thick film for p- and s-polarized light (Fig. 3b) show only weak differences, which agree well with the result above on substituted Pc/Au(100) interfaces. Therefore, one could conclude that also 1,4-(But)8PcZn grows almost disordered on GeS(001) at room temperature, except of the first 1-3 layer(s). However, for a very thick film of about 40 nm the corresponding XAS spectra (Fig. 3c) show an opposite intensity variation compared to Fig. 3a. Thus, the geometry changes from lying for the first layer(s) via disordered to standing for thicker films. The change from an ordered growth to a disordered growth after a few layers can be understood assuming a favored maximal interaction at the interface. In this case the aromatic cores of the substituted Pc molecule are lying directly on the interface and the (sterically demanding) substituents disable a strong interaction between aromatic cores in subsequent layers. Possible adsorption sites on the already covered single crystalline surface are energetically similar and a disordered growth occurs. A condition for an ordered growth of standing molecules seems to be a certain degree of roughness, at least on the scale of the size of the molecule, otherwise an initial lying adsorption geometry may be locked in – as observed above for the unsubstituted PcH2 on Au(100). The roughness may originate from the substrate surface itself or from disordered intermediate layers on flat substrates. The favored growth of standing molecules over lying molecules can be discussed on a microscopic scale as well as phenomenologically in terms of different surface energies [4]. For valuable discussions and technical assistance we thank H. Kuhlenbeck, M. J. Cook, I. Chambrier, S. Pohl, Ch. Jung, W. Neu and R. Hübel. Financial support by BESSY is gratefully acknowledged.
[1] H. Peisert, I. Biswas, L. Zhang, M. Knupfer, M. Hanack, D. Dini, M.J. Cook, I. Chambrier, T. Schmid, D. Batchelor, T. Chassé, Chem. Phys. Lett. 403 (2005) 1. [2] Y. Karzazi, X. Crispin, O. Kwon, J.L. Bredas, J. Cornil, Chem. Phys. Lett. 387 (2004) 502. [3] H. Peisert, T. Schwieger, J. M. Auerhammer, M. Knupfer, M. S. Golden, J. Fink, P. R. Bressler, and M. Mast, J. Appl. Phys. 90 (2001) 466. [4] H. Peisert, I. Biswas, L. Zhang, M. Knupfer, M. Hanack, D. Dini, M. J. Cook, I. Chambrier, T. Schmid, D. Batchelor, T. Chassé, Surf. Sci. (accepted).

130

Optical Properties of UN and UPtGe Single Crystals between 1 and 32 eV
M. Marutzky, U. Pelzer, H. Schröter, S. Weber, and J. Schoenes Institut für Physik der Kondensierten Materie, Technische Universität Braunschweig, Mendelssohnstr. 3, 38106 Braunschweig, www.ipkm.tu-bs.de BMBF 05ES3XBA/5 The uranium monopnictides have been intensively studied since the 1980ies [1]. One topic which have been discussed was the degree of localization of the 5f-states in the monopnictides. In fact, the 5f-states are localized in the heavier uranium monopnictides like USb, and when the pnictide is getting lighter, the 5f-states are becoming more and more delocalized. In consequence, the 5f-states in uranium mononitride are claimed to be relatively itinerant. UN is an antiferromagnet with TN = 50 K and crystallizes in the rocksalt structure like the other uranium monopnictides. While the optical and magneto-optical properties of most of the uranium pnictides are known [2], studies on single crystalline UN have been lacking until now. We have continued the systematic study of the uranium pnictides with the investigation of the optical and magnetooptical properties of UN [3]. In contrast to UN, the antiferromagnet UPtGe has a complicated orthorhombic crystal structure which is discussed for a long time. Nowadays, a noncentrosymmetric EuAuGe structure with two different uranium sites with different magnetic moments is favoured [4]. The magnetic structure of UPtGe is also interesting because UPtGe orders in an incommensurable cycloidal spin structure at TN = 50 K. It was found that the magnetic and electrical properties are anisotropic [5]. We have investigated the optical and magneto-optical anisotropy [6] and related them to UPtGe’s electronic structure. This study can test a crystalorientation dependent calculation of the band structure of UPtGe. The determination of the optical functions were done mainly by ellipsometric spectroscopy. Ellipsometric spectroscopy provides the full complex optical functions without requiring a Kramers-Kronig transformation, and surface features do affect less the results than other forms of optical spectroscopy. Therefore, ellipsometric spectroscopy is a powerful, nondestructive tool to analyze the electronic structure of materials. The measurements between 1 at 4.5 eV were performed at our home ellipsometer and for higher photon energies at the VUV-ellipsometer at BESSY II. We had deduced a scheme of the electronic structure of UN (Fig. 1) from magneto-optical measurements in the energy range from 1 to 5 eV and optical measurements from 1 to 10 eV, partially done at the 3m-NIM monochromator at BESSY II [3, 7]. The scheme in Fig 1 predicts a transition 2p(N) → 6deg(U) with a transition energy of 11.1 eV which is beyond the energy range of our previous measurements. In order to prove this expected transitions, the optical studies were expanded with the VUV ellipsometer at the TGM4 monochromator. The results are shown in Fig. 2. Unfortunately, it was not possible to get reliable results between 10 and 15 eV because of the low intensity at the TGM4 monochromator. Between 15 and 32 eV this was possible, and the measured slope of the optical conductivity is in accordance with studies on UN polycrystals [8]. In the high energy range a broad peak is visible which is probably a superposition of two absorptions of the transitions 6p(U) → 6dt2g(U) and 6p(U) → 6deg(U).

131

Fig 1: Scheme of the DOS of UN [3].

5f(U) -> 6dt2g(N)

optical conductivity σ1 [10 s ]

2p(N)6d(U) -> 5f(U)

5,0 4,5 4,0 3,5 3,0 2,5 2,0 1,5 1,0 0,5 0,0

2p(N) -> 6deg(U)

UN single cristal
6p(U) -> 6dt2g(U) 6p(U) -> 6deg(U)

15

-1

2p(N) -> 6dt2g(U)

0

2

4

6

8 10 12 14 16 18 20 22 24 26 28 30 32

energy [eV]
Fig 2: Real part of the optical conductivity of single cristalline UN. The structures in the spectrum are assigned to electronic transitions. The measurements between 1 and 10 eV were performed at the 3m-NIM, those between 15 and 32 eV at the TGM 4-monochromator. The blue line is the expected slope of the optical conductivity.

In order to investigate the optical anisotropy of UPtGe single crystals, we had made measurements on this material with the polarization vector orientated in respect to the crystal direction. This is depicted in Fig 3. The data between 1 and 10 eV once again were collected at the 3m-Nim monochromator, the data at higher energies at the TGM 4. In the energy range between 10 and 15 eV, the same problem with low intensity occurred. Obviously, the optical conductivity of UPtGe shows large anisotropic behaviour. The peaks at 1 eV and at 4 eV arise under participation of f-states, as we know from the Kerr-spectra [6]. Recent calculations [9] let suggest that the structure at 4 eV exists due to 5d(Pt) → 5f(U) transitions and those at 19 and 25 eV are due to 6p(U) →5d(Pt) transitions. The measurements of the optical properties between 15 and 32 eV at the VUV ellipsometer have expanded our studies on the electronic structure of UN and UPtGe single crystals. Of course, in order to get a full determination of the electronic structure, efforts must be made to close the gap between 10 and 15 eV and to proceed this studies to even higher energies than 32 eV.

132

}

4,0

optical conductivity σxx [10 s ]

3,5 3,0 2,5 2,0 1,5 1,0 0,5 0,0 2 4

UPtGe

σaa (E || a) σbb (E || b) σcc (E || c)

15

-1

6

8 10 12 14 16 18 20 22 24 26 28 30 32

energy [eV]
Fig 3: Real part of the optical conductivity of UPtGe with polarization vector parallel to the a, b, and c-axis.

We are indepted to the group of N. Esser, especially C. Cobet and M. Rakel for the opportunity to make measurements at the VUV ellipsometer, the staff at BESSY II, and the BMBF for financial support. [1] J. Schoenes, in Handbook Physics and Chemistry of the Actinides (1984) [2] J. Schoenes, Phys. Rep. 66 (1980), 187-212 [3] M. Marutzky, U. Barkow, J. Schoenes, and R. Troć, J. Magn. Magn. Mat. 299 (2006), 225-230 [4] D. Mannix, S. Coad, G.H. Lander, and J. Rebizant, Phys. Rev. B 62 (2000), 3801-3810 [5] R. Troć, J. Stępień-Damm, C. Sułkowski, A.M. Strydom, Phys. Rev. B 69 (2004), 094422 [6] M. Marutzky, U. Barkow, S. Weber, J. Schoenes, and R. Troć, in preparation [7] U. Barkow, M. Marutzky, A.-M. Cârsteanu, D. Menzel, A. Navratil, H. Schröter, S. Weber, J. Schoenes, C. Cobet, and R. Troć, BESSY annual report (2003), 155-157 [8] E.T. Arakawa, and M.W. Williams, J. Nucl. Mat. 41 (1971), 91-95 [9] M. Samsel-Czekała, and R. Troć, unpublished

133

Low-energy electrodynamics of superconducting diamond
S. Lupi1, M. Ortolani1,2, L. Baldassarre1, U. Schade2, P. Calvani1, Y. Takano3, M. Nagao3, T. Takenouchi4, and H. Kawarada4 Coherentia-INFM and Dipartimento di Fisica, Università La Sapienza, Rome, Italy Berliner Elektronenspeicherring-Gesellshaft für Synchrotronstrahlung, Berlin, Germany 3 National Institute for Materials Science, Tsukuba, Japan 4 School of Science and Engineering, Waseda University, Tokyo, Japan.
2 1

Diamond, with its extraordinary mechanical properties, excellent thermal conductivity, and large insulating gap between the valence and the conduction band, is in principle much more attractive than silicon as a semiconductor. Therefore the electric properties of artificial diamond films, doped by the inclusion of acceptors or donors, are being extensively explored in view of a possible, future diamond-based electronics. In this framework, it has been discovered recently that boron-doped diamond can also become a superconductor [1] below critical temperatures Tc well above the liquid helium temperature [2]. The strongly covalent bonds, the high concentration of impurities, and the high energy of its optical phonon in comparison with the Fermi energy [3], make diamond much different from elemental superconductors where the BCS theory of superconductivity holds. In September 2005 we measured the reflectivity R(ω) of a superconducting diamond film down to sub-Terahertz frequencies by use of Coherent Synchrotron Radiation (CSR) produced at BESSY in the low-alpha multibunch mode (stored current 10-20 mA). The CSR beam was extracted at the infrared beamline IRIS and sent to a Michelson interferometer for Fourier-Transform Infrared Spectroscopy (FT-IR). The intensity of this unique CSR source in the sub-THz range (0.1-1 THz corresponding to 3 < ω < 30 cm-1) is three orders of magnitude higher than conventional broadband sources. It becomes therefore possible to analyze in detail the low-energy electrodynamics of superconductors, being ω < 50 cm-1 ( ω < 5 meV) the typical relevant energy range for the opening of the gap in the density of states at Fermi level. The energy gap manifests itself in the sub-THz response as total reflection (R(ω)=1) with an edge at ω = 2∆ (see Fig. 1-a). The diamond film was grown by Chemical Vapour Deposition technique. Boron-doping up to around 3% was achieved by exposition to a Boron-rich atmosphere as described in Ref. 2.The superconducting Tc of the film was 6 K. We measured the reflectivity change trough the superconducting transition from 15 K to 2.6 K (Fig. 1-a). We then extracted the value of the superconducting gap 2∆(T) by means of a fitting procedure based on the BCS theory. We find 2∆(T)/ ~ 12 cm-1 which gives 2∆/kBTc ~ 3, making the BCS assumption consistent wit the data. The full reflectivity spectrum from the THz to the visible range was then measured at 10 K, and combined with the sub-THz data to obtain the optical conductivity σ1(ω) at all measured T´s. The inspection of this latter quantity in the sub-THz range the energy range where the infrared conductivity is depleted to build up the zero-frequency collective mode (Fig. 1-b). At the lowest T, the zero-absorption (σ1(ω) = 0) range corresponds again to ω < 2∆ The difference in the frequency-integrated σ1(ω) between the normal and the superconducting state allows one to calculate the field penetration depth λ(T), reported in Fig. 1-c. Furthermore, the spectrum of the charge-lattice interaction α2F(ω) (not shown) indicates a coupling between the charge carriers and the optical phonon branch around 120 meV, in agreement with recent theoretical calculations [4]. Our results consistently indicate that

134

diamond, in spite of the above cited peculiarities, behaves as a BCS superconductor in the "dirty" regime (Γ>>∆).

Figure 1. Reflectivity (a), optical conductivity (b), and field penetration depth (c) of a heavily boron-doped, superconducting, diamond film 3-µm thick, with Tc = 6 K. In c), the inverse square of the penetration depth, obtained from missing area in b) (experimental points), is reported vs. T/Tc and compared with its behavior for a BCS superconductor in the dirty limit [5], normalized to zero temperature (grey line).

[1] E. A. Ekimov et al., Nature 428, 542-545 (2004). [2] Y.Takano et al., Appl. Phys. Lett. 85, 2851-2853 (2004). [3] K.-W. Lee and W.E. Pickett, Phys. Rev. Lett. 93, 237003 (2004). [4] H.J. Xiang et al., Phys. Rev. B 70, 212504 (2004). [5] M. Dressel and G. Grüner, Electrodynamics of Solids Ch. 5, 7, 14 (Cambridge University Press, Cambridge, UK, 2002).

135

Two-gap superconductivity in MgB2 probed in the infrared
D. Di Castro1, M. Ortolani1,2, P. Calvani1, N. Zhigadlo3, S. Kazakov3, J. Karpinski3, U. Schade2, H. Keller4. Coherentia - INFM and Dipartimento di Fisica, Universitá La Sapienza, Rome, Italy Berliner Elektronenspeicherring Gesellschaft für Synchrotronstrahlung, Berlin, Germany 3 Eidgenössische Technische Hochschule, Zürich, Switzerland 4 University of Zürich, Switzerland
2 1

Since when MgB2 was found to be a superconductor with a remarkable Tc of 40 K, its electronic properties have been intensively studied. The metallic conduction of MgB2 takes place in two distinct electronic bands: the quasi-2D σ-band, formed by the hybridized spxpy B orbitals, and the 3D π-band, made of pz orbitals [1]. The different parity between the π− and σ-bands suppresses the impurity interband scattering, giving rise to the most intriguing feature of the MgB2 superconductor: multi-gap superconductivity. Indeed, the strong electron-phonon coupling in the σ-band is responsible for a value of the energy gap ∆σ, which opens below Tc at the Fermi level, much larger than the π-band gap ∆π (about 7 and 2 meV respectively). In a standard BCS superconductor, the far-infrared reflectivity R(ω) in the superconducting state is close to 100% for ω < 2∆, because of total radiation screening from the supercurrent flowing in a surface sheet, whose thickness is determined by the field penetration depth Λ. The normal state R(ω) < 100% is then recovered even at T << Tc for ω > 2∆, since the radiation with energy ω larger than the Cooper pair energy 2∆ cannot be screened. The latter property makes R(ω) a useful quantity to determine the gap value, since below Tc it will increase for ω < 2∆. If one now turns to the two-gap case, which is relevant for our work, one may ask whether the radiation with 2∆π < ω < 2∆σ would be screened or not, since normal electrons, which are unpaired due to radiation-induced Cooper-pair breaking in the πband, will coexist with the surface supercurrent generated by the pairs in the σ-band. This question has not yet been addressed theoretically to our knowledge. Information on the gap value of MgB2 from Fourier-Transform Infrared Spectroscopy (FT-IR) measurements has been obtained up to now on films, single crystal mosaics and pellets. [2,3,4]. However, we decided to measure the intensity reflected by a single crystal, not a mosaic or a film, in order to be able to compare the data with quantitative models of the electrodynamic in-plane response, based on the Bardeen-Cooper-Schrieffer (BCS) theory. MgB2 single crystal have a typical size of 300 µm. Few reports of far-infrared ( < 100 cm−1) FT-IR experiments exist on such small crystals, since the available surface area for the reflection (300x300 µm2) is 50 to 1000 times smaller than that of a film. The loss of signal intensity can be recovered by the use of infrared synchrotron radiation. Thanks to the high photon flux at the infrared beamline IRIS at BESSY, we could obtain reliable data for the reflectivity ratio R( , T = 4.2 K)/R( , T = 45K) (hereafter Rs/Rn) in the frequency range 30 < < 100 cm−1 with a spectral resolution of 2 cm−1 and a remarkable uncertainty of ±0.5%. We studied two single crystals of Mg(B1−yCy)2, one with y = 0 (pure, Tc = 38 K) and one with y = 0.083 (C-doped, Tc = 32 K). The effect of C-doping is a moderate decrease of Tc, ∆σ and ∆π, together with a strong increase in the carrier scattering rates Γσ, Γπ, which leads to a higher critical field [5], a quantity of interest for applications. The Rs/Rn of the two crystals is reported in Fig. 1-a. In the Rs/Rn of the C-doped sample one finds a cut-off around 60 cm−1, higher than that of the pure sample (40 cm−1). The data can be compared with the prediction

136

of two different BCS-based models for the infrared conductivity of a two-gap superconductor (Fig. 1-b,c). In case 1, we assume that the radiation is screened for any ω < 2∆σ; we have input 2∆π = 34 and 28 cm−1 2∆σ = 104 and 68 cm−1 for the pure and C-doped sample respectively, in good agreement with the photoemission data of Ref. 6. In case 2 we assume instead that the screening from the σ-band supercurrent takes place only for ω < 2∆π, and not for 2∆π < ω < 2∆σ. This is done by using a single effective gap value ∆ = ∆π = 34 and 28 cm−1 for the pure and C-doped sample respectively. The result of the analysis of Fig.1 is the following: case 1 can account for the data from the Cdoped sample but not for those from the pure sample. On the other hand, case 2 may explain the data from the pure sample. The increase of Γσ due to substitutional impurities drives the superconductor to the dirty limit (Γσ >> ∆σ) [4]. In the dirty limit condition, the field penetration depth Λ is much larger than in the clean-limit condition (Γσ < ∆σ), which is verified in the pure sample. Further infrared studies on high-quality MgB2 films are needed to clarify the role of the σ-band carriers in the increase of the critical field by C-doping [5,6]. In conclusion, the far-infrared data in Fig. 1 confirm the clean-limit for the pure MgB2 sample and the dirty-limit for the C-doped sample with y = 0.083, and provide new insights in the understanding of two-gap superconductivity. A paper with the results described here has been submitted to Physical review B.

[1] A.Y. Liu, I.I. Mazin, and J. Kortus, Phys. Rev. Lett. 87, 087005 (2001). [2] J. J. Tu et al., Phys. Rev. Lett 87, 277001 (2001). [3] A. Perucchi et al., Phys. Rev. Lett. 89, 097001 (2002). [4] M. Ortolani et al., Phys. Rev. B 71, 172508 (2005). [5] S. M. Kazakov et al., Phys. Rev. B 71, 024533 (2005). [6] S. Tsuda et al., Phys. Rev. B 72, 064527 (2005).

137

L-edge x-ray magnetic circular dichroism in transmission and total electron yield of Co2 Cr1−x Fex Al Heusler alloy films
M. Kallmayer,1 H. Schneider,1 G. Jakob,1 K. Kroth,2 H. Kandhpal,2 U. Stumm,2 S. Cramm,3 and H. J. Elmers1
Institut f¨r Physik, Johannes Gutenberg-Universit¨t Mainz, u a Staudingerweg 7, D-55099 Mainz, Germany 2 Institut f¨r Anorganische Chemie und Analytische Chemie, u Johannes Gutenberg-Universit¨t Mainz, Duesbergweg 10-14, D-55099 Mainz, Germany a Institut f¨r Festk¨rperforschung, Forschungszentrum J¨lich GmbH, D-52425 J¨lich, Germany u o u u
1

3

Invaluable fundamental quantities for the understanding of electronic properties of matter are the spin and orbital magnetic moment. Using X-ray magnetic circular dichroism (XMCD) in the X-ray absorption spectroscopy (XAS) element-specific moments can be derived from two magnetooptical sum rules [1]. A transmission experiment provides straightforward determination of the X-ray absorption coefficient. The transmission signal is free of artifacts [2], e.g. self-adsorption of the X-ray intensity in the sample, complicating a quantitative evaluation of magnetic moments, or surface properties being different from bulk properties. However, for transmission experiments the samples must not be thicker than a few hundred nanometer. Therefore, the frequently applied experimental methods include a measurement of the total electron yield (TEY) and total fluorescence yield (TFY), which can be applied to thick samples, too and provide a signal proportional to the X-ray absorption. TEY measures the x-ray absorption indirectly via the photoemitted electrons originally stemming from the Auger relaxation of the 2p-core hole [3]. Saturation effects can be accounted for if the X-ray penetration depth is known [4]. TEY is surface sensitive to the limited escape depth of the low-energy electrons (about 25 ˚). TFY measures the A X-ray photons stemming also from the core-hole relaxation. Since the penetration depth of the outgoing photons is similar to the incident photons self-absorption changes the measured spectra more significantly for TFY than for TEY. The information depth of TEY is about 2 nm and it would be very helpful to have in addition information on the bulk properties of the sample. Here, we report on an experimental method allowing the measurement of XMCD in transmission [5]. The method can be applied to epitaxial films and allows to determine the X-ray penetration depth of the incoming photons needed for the self absorption correction. We measured XMCD simultaneously with TEY and in transmission for epitaxial Co2 Cr1−x Fex Al films, which are interesting for their potential half-metallic ferromagnetic properties [6]. Upon comparing both signals, any uncertainties on the degree of magnetic saturation of the sample or the polarization of the X-ray beam could be avoided. We show that the magnetic moments calculated from the surface sensitive TEY are 17% smaller than the corresponding moments determined from the transmission signal. In both cases the magnetic moment of Cr is much smaller than the expected value as already found for bulk samples [7]. We prepared thin films of the Heusler compound Co2 Cr1−x Fex Al with a B2 structure on a-plane (11¯ 20)Al2 O3 by magnetron sputtering [8]. The films were capped in-situ by 6 nm Al in order to prevent oxidation. The XAS experiments were performed at the UE56/1 - SGM beamline. The incident photon flux was monitored by a Au net. TEY was measured via the sample current (see Fig. 1). The sample was shielded by a conducting tube on a positive bias voltage (100 V) in order to collect all electrons. For the X-ray absorption in the transmission geometry the photon flux transmitted through the thin Heusler films was detected via X-ray luminiscence in the Al2 O3 substrate [9, 10]. The light intensity in the visible wavelength range (VIS) escaping at the substrate edge was measured by a GaAs - photodiode. Using a reference sample with an ultrathin Co layer on a 250 nm thick Mo seed layer that is not penetrated by Xrays we verified that no X-ray fluorescence light from the sample surface was detected by the photodiode. An external magnetic field of 1.6 Tesla, that is sufficiently large to saturate the sample magnetization was applied perpendicular to the film surface. Fig. 2 shows the incident-photon-flux-normalized transmission XAS spectra of a 110 nm thick Heusler + − alloy film taken with the magnetic field applied parallel (IPD , solid curve) and antiparallel (IPD , dashed curve) to the circular polarization of the incident photons. The reference spectra Iref was assumed to increase linearly with the photon energy, normalized at the pre-edge region of the corresponding element (equivalent to an infinitely large penetration depth) [2]. This reference spectra is needed in order to calculate the relative absorption cross sections from the transmission spectra using the equation

138

2

I0,

±

ITEY e-

h

IPD

FIG. 1: Cartoon of the experimental setup used for the x-ray absorption experiment in transmission and TEY.

0.2 I PD (pA/nA) 0.1 0.0 0.5 Cr

I PD I PD I PD Fe
+ + -

+

I PD Co I PD
+ -

+

I PD
+ -

-ln(I PD /I ref)

-

0.0 0.0

x 0.5

( - )

+

-

-0.2
S,eff

<0.1

B

S,eff

-0.4 580 600

r

=2.64 =0.02

B

S,eff

r

=0.89 =0.08

B

700 720 740 780 800 820 Photon Energy (eV)

FIG. 2: Absorption spectra obtained by transmission measurements of a 110 nm thick Co2 Cr1−x Fex Al Heusler alloy films grown on a-Al2 O3 and capped by 6 nm Al. The top row shows the photodiode current normalized to the Au-net reference current and the assumed linear increasing reference signal Iref of the sample. In the center row we plot the absorption µ and the step function. The bottom row shows the XMCD spectra.

µ± (hν) = −ln[I± (hν)/Iref (hν)]/d, where d is the thickness of the film. The simultaneously measured TEY spectra normalized to the incident photon flux are shown in Fig. 3. After subtracting the background signal the XAS spectra were multiplied by a constant factor in order + − to achieve (ITEY + ITEY ) = µ+ + µ− at the L3 maximum. A quick inspection shows that the dichroism + − signal in the TEY (ITEY − ITEY ) (Fig. 3) is smaller than in the case of the transmission signal (Fig. 2). As observed for all three elements the step jump between pre- and post-edge intensity is larger for the case of TEY compared to the transmission XAS, indicating a reduction of the number of d-holes at the interface. The observed reduction can partly be attributed to a temperature effect, i.e. spin wave excitations with anti-nodes at the surface. Low-temperature experiments (100 K) indicate an increase of the Co XMCD - signal by 8 % compared to the value obtained at 300 K. Previous experiments on interface magnetism of pure elements indicate that significant reduction of magnetization occurs only in the topmost 1-2

139

3
I TEY (arb. units) Cr 0.5 I I
+

Fe

I I

+

Co

I I
x 0.5

+

0.0 0.0

(I -I )

+

-0.2
S,eff

<0.2

B

S,eff

r

=1.90 =0.05

B

S,eff

r

=0.74 =0.11

B

-0.4 580 600 700 720 740 780 800 820 Photon Energy (eV)

FIG. 3: Absorption spectra obtained by TEY simultaneously measured as the data in Fig. 2.

atomic layers [11]. In view of the limited surface sensitivity of TEY our observation can be explained only by an unrealistic complete quenching of magnetization in the topmost 2 layers. The most probable explanation is therefore a disturbed atomic structure of the Heusler alloy at an extended interface region. It cannot be excluded that an interdiffusion of the Al capping layer with the Heusler alloy took place. This interpretation is supported by the observation of a decrease of the number of d-holes at the interface. The authors would like to thank for financial support from the Deutsche Forschungsgemeinschaft (EL172/12-1).

[1] B.T. Thole, P. Carra, F. Sette, and G. van der Laan, Phys. Rev. Lett. 68, 1943 (1992). [2] C.T. Chen, Y.U. Idzerda, H.-J. Lin, N.V. Smith, G. Meigs, E. Chaban, G.H. Ho, E. Pellegrin, and F. Sette, Phys. Rev. Lett. 75, 152 (1995). [3] J. St¨hr, J. Magn. Magn. Mater. 200, 470 (1999). o [4] R. Nakajima, J. St¨hr, Y. U. Idzerda, Phys. Rev. B 59, 6421 (1999). o [5] M. Kallmayer, H. Schneider, G. Jakob, K. Kroth, H. Kandpal, U. Stumm, S. Cramm, H. J. Elmers, will appear in Appl. Phys. Lett. (2006). [6] C. Felser, B. Heitkamp , F. Kronast , D. Schmitz , S. Cramm , H. A. D¨rr , H.-J. Elmers , G. H. Fecher , u S. Wurmehl , T. Block , D. Valdaitsev , S. A. Nepijko , A. Gloskovskii , G. Jakob , G. Sch¨nhense, and W. o Eberhardt, J. Phys.: Condens. Matter 15, 7019 (2003). [7] H. J. Elmers, G. H. Fecher, D. Valdaitsev, S. A. Nepijko, A. Gloskovskii, G. Jakob, G. Sch¨nhense, S. o Wurmehl, T. Block, C. Felser, P.-C. Hsu, W.-L. Tsai, Phys. Rev. B 67, 104412 (2003). [8] G. Jakob, F. Casper, V. Beaumont, S. Falk, H. J. Elmers, C. Felser, H. Adrian, J. Magn. Magn. Mater. 290-291, 1104 (2005). [9] M. Kirm, G. Zimmerer, E. Feldbach, A. Lushchik, Ch. Lushchik, and F. Savkhin, Phys. Rev. B 60, 502 (1999). [10] D. J. Huang, C. F. Chang, J. Chen, H.-J. Lin, S. C. Chung, H.-T. Jeng, G. Y Guo, W. B. Wu, S. G. Shyu, C. T. Chen, J. Electrr. Spectr. Rel. Phenomen. 137, 633 (2004). [11] U. Gradmann, in Handbook of Magnetic Materials, 7 ed. by K. H. J. Buschow, Elsevier, Amsterdam (1993).

140

Free-Layer Dynamics of a Synthetic Spin Valve With Antiparallel Pinning
F. Wegelin, A. Krasyuk, D. Valdaitsev, S. A. Nepijko, H. J. Elmers, and G. Schönhense Johannes Gutenberg-Universität Mainz, Institut für Physik, D-55128 Mainz, Germany I. Krug, C. M. Schneider Institut für Festkörperforschung IFF-6, Forschungszentrum Jülich GmbH, D-52425 Jülich, Germany

Applying a biasing magnetic field on a thin micron-sized permalloy layer leads to suppression of domain wall creation and thus to the suppression of Landau-Liftschitz flux-closure pattern formation. The magnetization’s dynamical behavior of such a magnetically pinned and uniformly magnetized platelet differs from an unbiased particle because the pinning field acts as an additional local bias field. Such an intrinsic biasing field generating a uniform magnetic ground state can be realized in a spin valve by depositing a multilayer stack of an antiferromagnetic layer (AF) and several ferromagnetic layers (FM) with varying coercitivity which are separated by non-magnetic spacers (NM) such as Cu or Ru. An advanced spin valve structure, designed to maximally exploit the GMR (giant magnetoresistive) effect [1], is depicted in Fig. 1a. The in-plane resistance and ∆R/R in dependence of an applied field is shown in Fig. 1b. At the interface between the bottommost CoFe (FM) and the PtMn layer (AF) a unidirectional exchange anisotropy is established by cooling down the stack from above Neél temperature while simultaneously applying a field of 1 Tesla. A second CoFe layer of equal thickness is separated from the first by a Ru layer, whose thickness has been adjusted to evoke a strong antiferromagnetic coupling between the two CoFe layers. This way, their magnetic moments cancel out. Due to the strong antiferromagnetic coupling the CoFe/Ru/CoFe sub-stack is magnetically inert against the field magnitudes applied in this experiment. The magnetically soft CoFe/NiFe free layer is separated from the CoFe/Ru/CoFe substack by a Cu layer providing a weak antiferromagnetic coupling (in its first antiparallel maximum). The experiment has been performed in a pump-probe setup [2] at BESSY II, Berlin. The pumping is realized by generating an Oerstedt field of several mT in the coplanar waveguide on which the magnetic platelets have been deposited. Ultra-short soft Xray pulses achieved from synchrotron at the Ni L3 absorption edge are used for probing stroboscopically while imaging the x-ray magnetic circular dichroism (XMCD) with photoemission electron microscopy (PEEM) with sub-100 nm lateral resolution. A digital electronic delay allows stepwise shifting of the time between pump and probe pulse. The time resolution of 15 ps is not limited by the x-ray pulse length (3 ps) but by the electronic jitter of the trigger pulse (12 ps).
Ta NiFe/ CoFe Cu CoFe Ru CoFe PtMn magnetic free layer cap FM NM FM NM FM AF

J

J – electric pulse Hp – pulse field

hν Hp

Hexch

Hexch – pinning field via unidirectional exchange anisotropy hν – incident Xray pulses

Fig. 1a: GMR spin valve multilayer stack

b: Measurement of the MR-effect. ∆R/R = 14%.

c: Schematic view of experimental setup.

141

strip line signal output [V]

0,4 0,3 0,2 0,1 0,0

pulsewidth 500 ps 200 ps 100 ps

large rectangle norm. grey scale int.

small rectangle norm. grey scale int.

From the magnetic response a critical damping can be concluded.
1000 2000 3000

time [ps]

Fig. 2a: Buckling domain structure within antiparallel pinned spin valve

b: Magnetic response of the two rectangular platelets (upper two plots) on the magnetic field (lower) estimated by the output signal of the strip line. Pulse widths: 100 ps, 200 ps and 500 ps (red, black, blue, respectively).

The magnetic platelets investigated, are 15 µm x 10 µm (upper feature in Fig. 2a) and 10 µm x 5 µm (lower feature in Fig. 2a) in dimension. The microstripline measures 20 µm in width and has been periodically pumped with electric pulses of several 100 ps pulse width. Both rectangular platelets exhibit an almost uniform magnetization state which cannot be observed in unbiased permalloy platelets [3]. A densely packed system of interacting low-angle Neél walls becomes visible which stabilizes itself and forms a buckling state (Fig. 2a). The pulse field Hp causes an in plane rotation of the magnetization whose ground state orientation is initially parallel to the strip line due to the exchange anisotropy field Hexch (Fig. 1c). This becomes clear by analyzing the grey scale values within the regions of the spin valve platelets. As the magnetic field pulse propagates through the stripline, the magnetization rotates coherently out of its initial ground state orientation and falls back into it after the pulse has passed (Fig. 3). Since this critically damped oscillation takes place in absence of magnetic ringing as has been observed in permalloy monolayer platelets, fast magnetic switching is possible [4]. The investigated multilayer stack has already been implemented successfully as read elements for fast magnetic bits’ stray field readout in hard disc drives.

Fig. 3: Selected images (time in ps) of a sequence of one period of 2 ns measured with time increment of 15 ps.

Funded by BMBF (03 N 6500 „Nanocentre“). We thank the staff of BESSY for excellent cooperation.
[1] Giant Magnetoresistance in soft ferromagnetic Multilayers. B. Dieny, V.S. Speriosu, S.S.P. Parkin, B.A. Gurney, D.R.Wilhoit, D. Mauri, Phys. Rev. B 43, 1297 (1991). [2] Time-resolved Photoemission Electron Microscopy of magnetic Field and Magnetisation Changes. A. Krasyuk, A. Oelsner, S. A. Nepijko, A. Kuksov, C. M. Schneider, G. Schönhense, Appl. Phys. A 76: 863-868, 2003. [3] Self trapping of Magnetic Oscillation Modes in Landau Flux-Closure Structures. A. Krasyuk, F. Wegelin, S.A. Nepijko, H.J. Elmers, G. Schönhense, M. Bolte, C.M. Schneider, Phys. Rev. Lett. 95, 207201 (2005).

142

X-ray photoconductivity due to trap-sensitive relaxation of hot carriers
Wilhelm Ostwald Institute for Physical and Theoretical Chemistry, Universität Leipzig, Linnéstr. 2, D-04103 Leipzig, Germany

I. Konovalov, L. Makhova, D. Wett, K.-H. Hallmeier, R. Szargan

R. Mitdank

Department of Physics, Humboldt University of Berlin, Newtonstr. 15, D-12489 Berlin, Germany

Until recently, there was no possibility to directly relate the structural and the electrical manifestations of defects in semiconductors. This relation was usually concluded indirectly by comparing a first-principles calculation of the density of electronic states of the defect structures with the measured density of states.1 An unambiguous identification of their origin requires a direct experimental method. Recently, a new direct experimental method was proposed which can support the identification of defects in multinary semiconductors.2 This method has been successfully applied to (MDS) structures, and it requires either an structure or a diode for measurements of the capacitance. Unfortunately, the utilization of metal-dielectric-semiconductor structures multiplies the possibilities of the defect formation, unless a simple p-n homojunction is used, because of the utilization of several different materials in contact. We were able to observe a similar temperature dependent influence of the soft X-ray illumination on the ac photoconductivity of a thin ZnO film without any special barrier structure. If the photon energy is not enough for photons to be transferred from the core level into another level above the Fermi level, there is no change of the conductivity because the possible final states are normally filled. If the photon energy is large enough for a transition into the conduction band, the conductivity changes immediately after the light pulse. In the intermediate energy region, the ionized donors become filled, and they contribute to the conductivity in the band only after their thermal activation. The latter is a slow and temperature-dependent process, so that the change of the electrical conductivity lags with respect to the excitation. The change of the phase of the photoconductivity signal should be especially pronounced at a temperature at which the electron emission rate is equal to the synchrotron frequency of 1.25 MHz. At lower temperatures, the signal magnitude due to emission of carriers from defect states decreases, whereas at higher temperatures the phase lag gets smaller according to a faster emission rate. An obvious “spectral” measurement records the amplitude and phase of the photoconductivity signal at a given (optimal) temperature versus the photon energy in the vicinity of the correspondent X-ray absorption edge. Another “DLTS-like” mode is a measurement of the amplitude and phase versus the temperature at a suitable photon energy. It turned out that only the DLTS-like measurement mode is useful, because the signal in the spectral mode suffers from a significant energy-dependent interference originating from the temperature-independent photoelectron emission current. This fundamental problem is related to the fact that much of the excitation energy is wasted in photoelectron emission processes, which are also capable of causing electrical currents in the sample. Unfortunately, all the measurements of the photoconductivity in the spectral mode performed so far, were dominated by the energy dependent photoelectron signal to such an extent, that no possibility to separate both contributions from each other is in view at the moment. The samples were ZnO films deposited onto 1x1 cm2 sapphire substrates using pulsed laser deposition. Sample 1 was an undoped ZnO film, deposited at 16 Pa O2 at 615 °C, while Sample 2 was a MgO doped (~ 0.05 % MgO according to X-ray fluorescence analysis) ZnO film, deposited at the same temperature, but at 0.03 Pa O2. A sapphire substrate is necessary for cooling during the measurement. The 0.8x0.8 mm2 area of interest was separated by scribing four isolating scratches from the middle of the sample area to its sides. The contacting of the samples has been performed at the corners using evaporated Au contacts. The electrical resistance of the samples was in the range 5-15 kOhm at room temperature. During the lock-in photoconductivity measurement, a synchrotron light source (BESSY) in the single bunch mode supplied also a 1.25 MHz bunch synchronous electrical reference signal. A liquid nitrogen cooler and a resistive heater were used for controlling the temperature of the sample in the range of 75…430 K. The temperature of the sample holder was measured using a PT100 sensor, a DT400 sensor was used for comparison. The sample was dc biased by 9 V, so that the current in the sample was proportional to its electrical conductivity and changed periodically with the bunch repetition frequency. After amplification and filtering, the signal was lock-in detected using a quadrature multiplying detector (Gilbert cell type). Two phase components of the signal were digitized and presented in form of amplitude and phase. Unfortunately, the phase information was not useful so far because of a relatively large temperature-dependent phase shift introduced by the stray capacitance in combination with the large

1
143

resistance of the sample. The dark resistance of the sample represents a compromise between the background conductivity and the cutoff frequency of the input circuit. An optimal signal magnitude can be achieved, if most of the applied voltage drops over the illuminated volume of the sample. The ac current measurements in the spectral mode (not shown here) could be performed not only at the sample area, but also at an arbitrary conductor connected to the input of the amplifier. In the last case, the signal used to be identical with the total electron yield measured simultaneously by a channeltron detector. However, an additional ac signal of a comparable amplitude, which we relate to photoconductivity, emerged as soon as the illumination spot was located within the narrow current path of the ZnO film. During the measurements versus temperature, the signal to noise ratio could be improved at the expense of the spectral resolution. Similarly to the synchrotron DLTS measurements, we performed excitation around two energies, 15 eV below and 15 eV above the absorption edge. The exit slit of the beamline monochromator (U41PGM) was set to 3 mm, resulting in an estimated spectral linewidth of about 30 eV. In order to estimate the temperature dependence of the photoelectron current, we measured the temperature dependence of the sample current with the beam directed at a solid ZnO contact, while the beam did not illuminate the narrow current path of the film, but one of the contact areas of the film. We recorded rather small continuous variations of the photoelectron (PE) signal with temperature, probably related to a drift of the equipment (Fig. 1, solid line). Typical measurements of the ac photoconductivity versus temperature on both samples are also shown in Figure 1. A constant background was subtracted from the curves at energies above each absorption edge to account for the additional photoelectron current. The measurements were performed going from the cold state to the hot one (both graphs above) and vice versa (below). The features appear shifted by 50 K with respect to each other along the temperature axis at lower temperatures, obviously due to a 25 K delay related to the limited thermal conductivity of the sample, its holder and the contact between them, in spite of a long measurement time of about two hours per curve. The excitation energy was 15 eV below the O 1s and Zn 2p3 absorption edges (solid squares), as well as 15 eV above it (open circles). The shape of the curves is generally similar, but it differs significantly from that with a misaligned beam. Two specific temperature regions can be distinguished: shallow traps visible between 100 and 200 K, having at least two components, and deeper traps at temperatures above 300 K. This finding is in good agreement with known properties of intrinsic traps in ZnO. Using the 0.20 hν = 1035 eV literature data,3,4 we calculated the temperatures at 6 -1 hν = 1005 eV which an emission rate of 1.25×10 s (light pulse cold to hot repetition frequency) was expected for various 0.15 impurities. E1 trap is expected to manifest itself at a temperature of 93 K, E2 at 193 K, E3 at 302 K Zn L edge and E4 at 339 K. These temperatures match both hν = 545 eV 0.5 regions where changes of the conductivity were hν = 515 eV observed. Extrinsic traps related to Al diffusion are possible, but were not considered here. Traps 0.4 nal of the lower temperature region (E1, E2) were PE sig O K edge significantly less sensitive to the variation of the 0.4 excitation energy around the Zn L edge than hν = 1035 eV hν = 1005 eV around the O K one. The trap of the higher hot to cold temperature region (E3 or E4) is sensitive mostly to the variation around the Zn L edge rather than O 0.2 Zn L edge K one. This trap is present in sample 2, but it was not detected in sample 1, deposited at a higher hν = 545 eV oxygen pressure. It shows a reverse behavior with 0.6 hν = 515 eV less signal while being excited above both absorption edges, and a larger signal magnitude O K edge 0.4 while excited below the edges. In order to interpret the acquired data it 100 200 300 400 has to be clarified, why the change of the carrier temperature (K) concentration in the bands, which is being induced directly by a core-specific excitation, is related to Figure 1. Amplitude of x-ray photoconductivity the trap levels. This question is in focus of the variation versus temperature. current investigation, but it also arises when
signal amplitude, corrected by mirror current

2
144

considering synchrotron DLTS.2 Indeed, when using an excitation spectrum of 30 eV width, the excitation of electrons from core levels is hardly done into the defect states directly, but mainly into the conduction band, or even in vacuum, depending on whether the excitation energy is slightly or much larger than the x-ray absorption edge. Nevertheless, investigations by Fujioka2 as well as our own ones show that the carrier concentration in the band, influenced in such a way, is very defect sensitive. In the following, we propose a tentative explanation of this phenomenon. The inner shell electrons which are being excited are strongly localized near the core. The band states, however, are of delocalized nature and they are also located in the space between the atom frames. Each state in the (conduction) band represents a standing electron wave in the whole crystal. However, the electron cloud is distributed non-uniformly, so that nodes and antinodes are located periodically in space. According to Bloch’s theorem, the electron wavelength in a crystal is equal or a multiple of the lattice constant. The locations of the antinodes of a plain electron wave, where the electron is most likely to be found, are therefore the same within each of that lattice cells, where the antinodes are present. Consequently, the probability of electron transitions from the localized core levels of an atom to the delocalized band states of the crystal is larger for that band states which antinodes are in the next vicinity of the atoms in question. The electronic states in a band of a multinary compound can be sorted according to their localization at various atoms in the unit cells resulting in the local partial density of states distribution. If we consider trap states, they are more or less localized around their defects, and the deep states are particularly strongly localized. The local nature of the excitation of core electrons makes clear, why the relaxation of hot carriers through the defect states is so probable: the excited electron at one atom (e. g. at a lattice node) arises at the equivalent locations of the corresponding electron wave in many other unit cells (at equivalent lattice nodes), where the defect density of states is likely to be localized. The usual relaxation of a hot carrier through delocalized random band states is less probable in the case of excitation of a core electron, so that relaxation through the defect states with the same localization becomes more probable. Such a transition does not require a significant change of the electron localization within the unit cell. In addition, intraband transitions are often symmetry-forbidden. Therefore, the relaxation of a localized core electron, which was excited into the depth of the conduction band at an atom, is more likely to happen again into a localized (atomic, defect) state at one of the equivalent lattice sites, being not necessarily the same site and being even not necessarily occupied by the same atom sort, if substitutional defects are present. Although all the conduction band states are spread throughout the crystal, the electron which was just excited into the state, tends to conserve its former localization and prefers to transit into a similar locality again. In ZnO, the density of E1 and/or E2 trap states, if identified correctly, is therefore localized in the vicinity of oxygen sites, whereas that of E3 and/or E4 in the vicinity of Zn sites. An oxygen vacancy consists of Zn frames, so that localization of its electrons is at Zn atoms. Sample 2 was deposited at a low oxygen pressure and it is likely to contain more oxygen vacancies than Sample 1. It shows a distinct E3/E4 feature, as related to Zn sites, whereas Sample 1 deposited at high oxygen pressure does not show this feature at all. Moreover, from this point of view we were even able to satisfactorily explain the reverse excitation behavior above 300 K, which was described in the beginning. Since transitions from Zn 2p3 to 4p conduction band states are symmetryforbidden, transitions at higher energies are possible into Zn s and d states above. Both E3 and E4 traps are donors and are filled in ground state. If related to Zn, they are Zn 4s states. Transitions from s and d states of the conduction band are symmetry-forbidden. Only transitions from Zn 2p3 directly into 4s donor traps at lower excitation energies are allowed and observed. In conclusion, the influence of trap-sensitive relaxation of hot carriers on the ac x-ray photoconductivity in semiconductors were discovered. We were able to separate the defect-induced x-ray resonant photoconductivity signal of thin ZnO films from artifacts. All the results could be qualitatively explained using symmetry selection rules and local transition probabilities. BMBF supportort 05 ES3XBA/5 is acknowledged. We thank Dr. R. Pickenhain, Holger von Wenckstern, and Prof. M. Grundmann for a discussion. S. B. Zhang, Su-Huai Wei, A. Zunger, and H. Katayama-Yoshida, Phys. Rev. B 57, 9642 (1998). H. Fujioka, T. Sekiya, Y. Kuzuoka, M. Oshima, H. Usuda, N. Hirashita, and M. Niwa, Appl. Phys. Lett. 85, 413 (2004). 3 F. D. Auret, S. A. Goodman, M. Hayes, M. J. Legodi, H. A. van Laarhoven, and D. C. Look, Appl. Phys. Lett. 79, 3074 (2001). 4 F. D. Auret, S. A. Goodman, M. J. Legodi, W. E. Meyer, and D. C. Look, Appl. Phys. Lett. 80, 1340 (2002).
2 1

3
145

Mapping the electron correlation in two-electron photoemission
F.O. Schumann, C. Winkler, G. Kerherve and J. Kirschner Max-Planck Institut für Mikrostrukturphysik, Weinberg 2, 06120 Halle, Germany 1. Experiments at CP-NIM Introduction Due to the Coulomb interaction it is energetically unfavorable for electrons to be close to each other. Additionally, the Pauli principle demands that electrons with parallel spin can not be at the same location. Averaging over both spin directions still gives a reduced probability of finding two electrons at the same location. A more elaborate theoretical treatment defines a pair correlation function g(r,r').1 This describes the probability to find an electron at coordinates r when a second is located at r'. The key result is that g is essentially constant except for small distances r-r'< a few Å where g adopts smaller values. The spatial extent of this region is called exchange-correlation (xc) hole and describes the length scale over which the correlation between electrons is relevant. A recent theoretical publication suggests that k mapping of the double photoelectron emission (DPE) intensity opens an avenue of imaging the xc-hole.2 We will discuss our results obtained by DPE on a NaCl(100) surface. The experiments were conducted under UHV conditions featuring a novel time-of-flight spectrometer at beamline CP-NIM during single bunch time.3 A central collector accepts electrons only within a solid angle of ~0.02 sr, the detected electron we may term as "fixed electron". A resistive anode serves as the second detector which allows for a spatial resolution of the impact position. Electrons within a solid angle of ~1 sr are registered which we term as "free electron". A coincidence circuit allows the determination of the individual flight times. Results In fig.1 we plotted the 2D energy distribution of coincidence electron pairs upon excitation with 34 eV photons. The energy of the "fixed electron" is labelled with E1 whereas the "free electron" has the energy E2. We observe the onset of DPE when the sum energy E1+E2 equals ~14.6 eV where DPE becomes energetically possible. More insight can be obtained if we take advantage of the lateral resolution of the set-up. In a first step we select only those coincidences for which the energies E1and E2 are fixed. In other words, we pick a point in the 2D energy distribution shown in fig.1. In order to obtain sufficient statistics we actually select an energy window ~ 0.8 eV around the respective energies. This has been indicated by the square boxes in fig.2 labelled a) and b). We can now proceed and plot the coincidence

146

Fig.1 : The 2D energy distribution of coincidence electron pairs from a NaCl(100) surface is plotted. The photon energy was 34 eV. The energy E1 (E2) refers to the "fixed electron" ("free electron") The dashed diagonal line marks the onset of pair emission, which occurs for a sum energy of ~14.6 eV. The square boxes labelled a) and b) indicate the events used to generate the 2D momentum plots displayed in fig.2. intensity as a function of the in-plane momentum k|| of the "free electron". We have selected two different regimes within the 2D energy distribution highlighted in fig.1 by the black squares. In the case a) we are right at the onset of pair emission. Case b) describes the situation if emission below the highest occupied level is possible. In fig.2 we display the resulting momentum distributions. In fig.2 a) the energies are E1=5.5 eV and E1=9.5 eV (region a) in fig.2). We clearly observe that the region k||=0 (outside the "blind spot") is surrounded by a region of diminished intensity. The intensity increases for larger k|| values and reaches a maximum for k|| ~0.55 Å-1 and then falls off rapidly towards the edge of the channelplate. A dramatically different situation is depicted in fig. 3b) where we select E1=5.5 eV and E1=7.5 eV. Now the ring of enhanced intensity is essentially gone. Energetically the sum energy E1+E2 has been reduced from 15 eV to 13 eV. This energy difference allows for emission of a deeper laying valence band electron or inelastic scattering losses if the electrons originate from the top of the valence band. Our results demonstrate the importance of inelastic scattering which is very effective in destroying the hole shown in fig.2 a).

147

Fig.2: The 2D k|| distribution for two different energy pairs from a NaCl(100). In panel a) we have selected E1=5.5 eV and E2=9.5 eV. Whereas in panel we have chosen E1=5.5 eV and E2=7.5 eV,. In c) we show the result from a LiF(100) surface exited with 30.7 primary electrons. The energies are E1=7.5 eV and E2=9.5 eV} Discussion It can be shown within the dipole approximation that a product of single particle wave functions yields a zero DPE intensity. However, due to the correlation/interaction such a product of wave functions is not correct and a nonzero DPE intensity results when going beyond the single particle picture.5 Therefore we can explain the momentum distribution in fig.3 a) as a consequence of the xc-hole. Such a notion is corroborated by a more thorough calculation by Fominykh et al. on the double photoemission of Cu(100).2 They computed the in-plane momentum distribution (of the "free electron") similar to the plots shown in fig.2 for a photon energy of 42 eV and found that it exhibits a reduced intensity until k|| adopts a value of 1.4 Å-1. At this point the intensity rises sharply by roughly an order of magnitude. Shortly thereafter the intensity returns quickly to a small value. The ring of enhanced intensity has a diameter of 2.8 Å-1 and a width of ~ 0.2 Å-1 The important outcome of the theoretical work is that the reduced intensity is a manifestation of the xc hole. Furthermore it was found that the DPE intensity also displayed the crystallographic symmetry of the surface. For NaCl we find the diameter of the reduced intensity region to be ~1.1 Å-1 if the energy of the "free electron" is 9.5 eV, this diameter is significantly smaller than the theoretical value for Cu. Whether this difference is due a comparison between different materials (noble metal versus insulator) is not clear. In that case we may take this as a hint of a material dependence. We emphasize that the size of the xc-hole has been determined from the diameter of the maximum intensity ring, which is near the edge of the detector, hence it is possible that the ring is even larger. This view is supported by the observation that the diameter increases with increasing energy E2 of the "free electron" (from 0.9 to 1.3 Å-1 for 7.5 to 13 eV), because the covered momentum space of the detector becomes larger. According to theory the xc-hole shrinks if

148

E2 is increased.2 Nevertheless, our key observation of a region of reduced intensity due to the xc-hole remains valid and we quote the value for E2=9.5 eV. In this context we would like to point out that we have performed a similar series of experiments on a LiF(100) surface albeit excited by a primary electron gun.3 As an example of the resulting momentum distributions we display in fig.3c) the situation for a primary energy of 30.7 eV. The energies E1 and E2 are 7.5 eV and 9.5 eV, respectively. With this selection the sum energy has the highest possible value and only pair emission without any inelastic scattering of the pair is possible. Furthermore we have chosen the same value of E2 as used in fig.2 a) in order to facilitate direct comparison. We immediately notice that there is no qualitative difference between fig. 2a) and b). The study on LiF also showed that inelastic scattering destroys the region of reduced intensity similar to the plot in fig.2 b). Although two different materials have been studied (NaCl versus LiF) their electronic properties are very similar. Hence we conclude that DPE and (e,2e) experiments give qualitatively similar results despite the fact that the underlying mechanisms bear some significant differences.

2. Experiments at TGM 4 Additionally to the spectrometer employed at CP-NIM we started to study the xc hole at metal surfaces at the beamline TGM 4. Since the work function of a metal (typically ~5eV) is smaller than the bandgap of NaCl we expected the energy resolution to be a limiting factor. An obvious solution is an increase of the flight path, while maintaining the angular acceptance angle. This has been achieved by using an elliptical mirror together with the proven detectors also in use with the spectrometers in operating at CP-NIM. With the elliptical mirror set-up we investigated the momentum distribution of the coincidence intensity from a Cu(100) surface. Despite the fact that the spectrometer was working properly we encountered a too low count rate. Our recent results on the momentum distribution of LiF and NaCl have shown that the size of the xc-hole was just within the acceptance range of the detector. This fact suggested that the size of the xc hole of Cu could be larger than expected. This hypothesis was confirmed in a second experimental run, where a spectrometer with an acceptance angle of 2π in one dimension was employed. 1. P. Fulde, Electron Correlations in Molecules and Solids, Springer Series in Solid State Sciences, Vol. 100 (Springer, Berlin, 1993). 2. N. Fominykh, J. Berakdar, J. Henk, and P. Bruno, Phys. Rev. Lett. 89, 086402 (2002). 3 F.O. Schumann, J. Kirschner, and J. Berakdar, Phys. Rev. Lett. 95, 117601 (2005). 4 F.O. Schumann, C. Winkler, G. Keherve, and J. Kirschner, Phys. Rev. B 73, 041404(R) (2006). 5 J. Berakdar, Phys. Rev. B 58, 9808 (1998).

149

Momentum dependence of the mass renormalisation in lightly underdoped Bi2Sr2CaCu2O8
I. Santoso1, A. Mans1, W. K. Siu1, S. de Jong1, Y. Huang1, R. Follath2, P. Bressler2 M. S. Golden1
1

Van der Waals-Zeeman Institute, Universiteit van Amsterdam, Valckenierstraat 65, 1018 XE Amsterdam, The Netherlands 2 BESSY GmbH , Albert Einsteinstrasse 15, 12489 Berlin, Germany

The understanding of the mechanism of high temperature cuprate superconductors (HTSC) remains probably the fundamental challenge for solid state science. Much progress has been booked but there is still disagreement as to the nature of the bosonic mode mediating pairing in the HTSC. The two main contenders are currently spin fluctuations and phonons. Angle-resolved photoemission offers a very direct handle on the effects of coupling of the electronic system to a bosonic mode in terms of the observation of renormalisation effects in the band structure. In the case of coupling, there is a deviation of the 0.1 energy position of the many-body, interacting Bare Band Renormalised Band band from the position of the bare (noninteracting) band close to the Fermi level in ARPES 0.0 spectra. This leads to two branches in the dispersion relation, as illustrated in Fig. 1. -0.1 Through characterising the energy, momentum, ImΣ(κ,ω) doping and temperature dependence of these -0.2 renormalisation effects, we can hope to be able to pin down the identity of the bosonic mode -0.3 involved. Most ARPES data have dealt with the investigation -0.2 -0.1 0.0 0.1 0.2 of the renormalisation effects along the nodal k (zone diagonal) direction [1,2], there are fewer Figure 1: Cartoon of mass studies of the antinodal ([π,0]) region of k-space renormalisation [3,4]. The latter region is technically more tricky, as in Bi2212-based systems one needs to explicitly resolve the c-axis bilayer splitting, as well as studying modulation-free crystals (usually achieved by doping Pb for Bi during crystal growth). Here we report studies of the mass renormalisation of the Fermi surface states in modulation-free, untwinned underdoped (Pb,Bi)2Sr2CaCu2O8 (TC=84K). The crystals were grown in Amsterdam using the travelling solvent floating zone technique and show excellent cleavage surfaces, as can be seen from the LEED image shown in the left Figure 2: Left: LEED image of the sample showing the panel of Fig. 2. The systematic absences (indexed in an orthorhombic cell) and a systematic absences lack of modulation-induced reflections. Left: the sharp highlighted with the red superconducting transition at 84 K from SQUID data arrows point to the fact that there is a glide mirror plane
E(k)

150

parallel to the real-space bc plane. This arises from distortions of the atomic positions away from the ideal tetragonal structure and gives rise to the infamous shadow Fermi surface [5]. The focus of our study is on the investigation as a function of Fermi surface angle (Φ), where 0º means the antinodal (AN) direction and 45º the nodal (see also Fig. 3C).

Figure 3: (A) ARPES intensity plots as a function of energy and wave vector for each Fermi surface angle at T =28 K (upper panel) and T = 120 K (lower panel). Left to right = AN towards N. (B) Zoom of same data. (C) Typical EDC fit (left) and MDC fit (right) at kF , used for extracting the band dispersion. (D) Fermi surface cartoon, indicating location of the cuts in k-space.

151

The ARPES experiments were carried out at the U125/1-PGM beamline using the IFWDresden group's SCIENTA SES100-based end station. Spectra were taken using a photon energy of 38 eV in order to give maximal contrast between the two c-axis split bilayer bands. The analyser slit is aligned parallel to the (π,π) – (π,-π) direction. In order to minimise the impact of photon-induced changes on the spectra, we took a different spot on the crystals for each Fermi surface angle measurement. In Fig. 3A, we show the collection of data from AN towards the nodal recorded in the superconducting state at T = 28 K (upper panel) and T =120 K (lower panel). One can clearly distinguish the two bilayer-split bands both in the low and high temperature data. To follow the band dispersion, we plot the peak position by fitting the EDC and MDC cuts through the 2D datasets. The EDC dispersion (overlaid in red) is more sensitive to the presence of the gap compared to the MDC dispersion (overlaid in blue). The MDC's are fitted using a simple Lorentzian, while the EDC is fitted using a Lorentzian multiplied by the FermiDirac distribution, including a scaled background taken from k < kF . In the superconducting state (T = 28 K), the data show clearly a momentum dependence of mass renormalisation. The bonding bilayer split band is strongly renormalised at the antinode (leftmost data panel in Fig. 3A and B). This can be seen by the dramatic flattening of the dispersion (increase in mass) close to EF compared to the steeper dispersion at higher binding energies. For a Fermi surface angle of ca. 25° (rightmost data panels), the change in velocity at low energies is significantly less. There is also a clear temperature dependence of the renormalisation effects. For T>TC, and particularly away from the antinode above TC, the renormalisation effects become quite weak (see for example the rightmost panel of Fig. 3B). This focus of the strongest renormalisation effects in the region of the (p,0) point and their clear reduction in magnitude for T>TC would speak in favour of the magnetic fluctuation scenario and against the phonon scenario discussed in the introduction. The next step, of course, is a quantitative evaluation of the renormalisation effects (for example of the bonding band) as a function of FS location and temperature. For 'classical' superconductors, the T 0 for FM and Jinter < 0 for AFM coupling. Earlier experiments have shown that ∆TC,Ni oscillates as a function of the spacer thickness in the same manner as the IEC does [4,8]. The first interpretations of this behavior have been carried out within a mean field theory (MFT) which turn out to be insufficient [5]. Collective spin excitations have to be taken into account in a more advanced theoretical description. For the present study this has been done within a Green’s function theory (GFT). For the details about the calculations, please see Ref. [1].

2

DTC,Ni / TC,Ni

1

1 2 3 4 5 6

Ni ML ML ML ML ML ML

0

0

|J

inter |

200

(meV/atom)

400

Figure 4: Relative temperature shift ∆TC,Ni /TC,Ni of the Ni magnetization as a function of the strength of the IEC |Jinter | for different dNi [1].

to the temperature of the maximum of the

For the further analysis the characteristic Ni temperatures of the two cases need to be determined, ∗ i.e. TC,Ni for the full trilayer and TC,Ni for the corresponding Cu/Ni/Cu(100) bilayer system. This is done with the help of a standard magnetization curve which has been obtained from various measurements of 3−5 ML Ni/Cu(100) films [9]. Fitting this standard curve to MNi (T ) of a Cu/Ni/Cu(100) sample yields TC,Ni with an accuracy of a few kelvin ∗ (dashed line in Fig. 3). Determining TC,Ni of the trilayer is less evident due to the presence of the tail. Therefore, the same standard magnetization curve is fitted to the data points of MNi (T ) in the trilayer (dotted line in Fig. 3). The point at which the fitted standard curve meets the x-axis is identified ∗ ∗ with the temperature TC,Ni . This TC,Ni is identical susceptibility [5, 10].

171

M (kA/m)

Before the desired combination of the two known effects Jinter (dCu ) and ∆TC,Ni (dNi ) can be accomplished, it is essential to express ∆TC,Ni /TC,Ni as a function of Jinter . In earlier works a linear dependence has been derived in a MFT [4, 8]. However, since MFT does not satisfy the description of the temperature shift ∆TC,Ni , it is appropriate to find an improved description of ∆TC,Ni (Jinter ) within the GFT. For the present study, the relative temperature shift ∆TC,Ni /TC,Ni has been calculated as a function of |Jinter |. As Fig. 4 shows the dependence is clearly nonlinear. The influence of the spin fluctuations, visu(a) alized in the relative shift ∆TC,Ni /TC,Ni , de1 pends on the one hand on dNi . On the other hand it depends on the strength of the IEC, i.e. |Jinter |. Combining both variables with 0 the help of the relation given in Fig. 4, the 3D plot as anticipated at the beginning can 1 0.5 3 be established. The result is a curved sur5 face of ∆TC,Ni /TC,Ni = f (dNi , dCu ). Figure 5 6 8 10 2 4 dCu (ML) dNi (ML) shows this final result of the calculation to(b) gether with the experimental findings. The 1 experimental data are shown as full dots and in the projection to the dNi –dCu plane (open 0 circles). The results are sorted according to the thickness of the Co film: dCo = 2 ML in 1 Fig. 5 (a) and dCo = 3 ML in Fig. 5 (b). The 1 3 zero plane is given by Jinter = 0 with a shift 2 ∆TC,Ni = 0. The illustration has been chosen 5 2 4 6 such that regions with a parallel alignment dCu (ML) dNi (ML) of the magnetizations of the two FM layers Figure 5: Two-parameter plot of the rela- (J inter > 0, above the zero plane) are distintive temperature shift ∆TC,Ni /TC,Ni (dNi , dCu ) for guished from the ones with an antiparallel Co/Cu/Ni/Cu(100) trilayers as a function of the Ni film thickness dNi and the thickness dCu of the alignment (Jinter < 0, below the zero plane). In conclusion, fair agreement between the Cu spacer layer with (a) dCo = 2 ML and (b) experimental results and the calculations even dCo = 3 ML [1]. in the simplified model show the importance of spin fluctuations in the coupled trilayers especially at thin thickness. Furthermore, the desired combination of the two dependencies on dNi and dCu and their simultaneous consideration has been accomplished. This work was supported by the BMBF (05 KS4 KEB/5).
FM AFM FM AFM

DT C,Ni / T C,Ni

DT C,Ni / T C,Ni

References
[1] A. Scherz, C. Sorg, M. Bernien, et al. , Phys. Rev. B 72, 054447 (2005). [2] P. Bruno and C. Chappert, Phys. Rev. Lett. 67, 1602 (1991). [3] J. Lindner and K. Baberschke, J. Phys.: Condens. Matter 15, S465 (2003). [4] A. Ney, A. Scherz, P. Poulopoulos, et al. , Phys. Rev. B 65, 024411 (2002). [5] P. J. Jensen, K. H. Bennemann, P. Poulopoulos, et al. , Phys. Rev. B 60, R14994 (1999). [6] H. Wende, C. Sorg, M. Bernien, et al. , Phys. Stat. Sol. (b) 243, 165 (2006). [7] C. Sorg, A. Scherz, H. Wende, et al. , Phys. Scripta T115, 638 (2005). [8] G. Bayreuther, F. Bensch, and V. Kottler, J. Appl. Phys. 79, 4509 (1996). [9] P. Poulopoulos and K. Baberschke, Lecture Notes in Physics, Springer 580, 283 (2001). [10] U. Bovensiepen, F. Wilhelm, P. Srivastava, et al. , Phys. Rev. Lett. 81, 2368 (1998).

172

Induced magnetism in O as a surfactant for Fe, Co, and Ni films
C. Sorg1 , N. Ponpandian1 , R. Q. Wu2 , M. Bernien1 , K. Baberschke1 , and H. Wende1
1

Institut f¨r Experimentalphysik, Freie Universit¨t Berlin, Arnimallee 14, D-14195 Berlin, Germany u a 2 Department of Physics and Astronomy, University of California, Irvine, California 92697, USA

Surfactant assisted growth of nanoscale structures Ni on surfaces is a well established technique to improve Cu (100) Cu (100) Cu (100) the growth mode of 3d ferromagnetic films on sinor or or Cu (110) Cu (110) Cu (110) gle crystalline substrates toward a more layer-by-layer one [1]. Tailoring the growth modes for these ultrathin +O +Ni films is crucial since the magnetic properties are highly Figure 1: Schematic illustration of the sensitive to minimal structural changes: If the nearest oxygen surfactant assisted growth of Fe, neighbor distance varies by 0.03-0.05 ˚ only, the mag- Co, and Ni films on Cu(100). A netic anisotropy energy may change by 102 − 103 . The preparation procedure is schematically presented in Fig. 1 and explained in detail in Refs. [2–4]: At first, atomic oxygen induced is adsorbed onto the clean Cu(100) crystal. Then the Fe, Co, and Ni films are prepared onto the reconstructed surface. All three metals grow more layer-bylayer than without O up to > 15 ML. The O atoms always “float” on top of the ferromagnetic films. An interesting question is, if the surfactant oxygen affects the magnetic properties of the films. It was demonstrated that the magnetic anisotropy energy of Ni is significantly enhanced using this surfactant, mainly because of a decrease in the magnitude of the surface anisotropy [5–7]. Via the element-specific X-ray magnetic circular dichroism (XMCD) technique we were able to identify an induced moment in the surfactant oxygen [4, 8]. In the present work ab initio calculations of the O K edge XMCD were carried out to understand the spectroscopic fine structures from a fundamental point of view.
Norm. XAS / XMCD (arb. units)

X-ray absorption spectroscopy (XAS) and XMCD measurements were carried out at the undulator beamline UE56/2-PGM2 on 3 ML Fe, 4 ML Co, and 15 ML 4 Ni films. The advantage of measurements in the soft Co L2 edge O K edge X-ray range is that the O K edge as well as the L2,3 2 edges of the 3d ferromagnets Fe, Co, and Ni are lo0 cated in this regime. Both the induced magnetism in the surfactant and the magnetic properties of the ferro´30 -2 magnetic films can be probed in the same experiment 540 790 530 780 800 810 as shown in Fig. 2 for the case of Co. A clear XMCD E (eV) signal at the Co L2,3 edges is determined but also a tiny Figure 2: X-ray absorption coefficient and XMCD at the O K edge and Co L2,3 XMCD signal at the O K edge can be revealed. Due to edges for the surfactant grown Co film. the excellent performance of the beamline this oxygen signal of about 8 % normalized to the small oxygen edge-jump (jump ratio Jr =6 %) can be revealed. The appearance of this signal shows that indeed the Co film induces a magnetic moment in the surfactant oxygen. However, one has to keep in mind, that with XMCD at K edges only the orbital moment µL is probed [8]. The sharp structure in the XAS located at 530 eV originates
6
+ m mXMCD

Co L3 edge

173

1.0

0

0.5

Normalized XMCD (arb. units)

hn hn

-0.05

Normalized XAS (arb. units)

0 1.5 1.0 0.5 0 1.0

Fe

hn

3 ML Fe

0

-0.05

Co

hn

4 ML Co

0

Figure 3: Angular-dependent NEXAFS (left) and XMCD (right) at the O K edge of the 3d ferromagnets grown on Cu(100) with oxygen as a surfactant.

0.5

-0.05

x3
hn

0

Ni
530

15 ML Ni
530 535 540

E (eV)

550

570

525

E (eV)

from hybridized 2pz 3d states [9]. Since the XMCD is observed exactly at the same position this is a first indication that the 2pz states of the oxygen are magnetically polarized. For the investigation of systematic trends we carried out these measurements for Fe, Co and Ni films. We start the discussion with the analysis of the angular-dependent NEXAFS spectra presented in Fig. 3 (left). The NEXAFS of Co was recorded with linearly polarized X-rays, the E-vector aligned perpendicular to the [011] direction. In the case of Ni and Fe the isotropic XAS which is the average of the XAS recorded with circular polarized light of the two helicities is shown. The clear angular dependence of the spectra for all the 3d films show that (i) no bulk-like oxide with the 3d elements is formed and (ii) the oxygen atoms “float” to the top of the surface. If the oxygen atoms were incorporated into the film, the angular dependence would be negligible because of the high symmetry. In contrast, on the surface the individual oxygen orbitals can be investigated. The spectra show quite similar trends for the three films: A sharp peak located at 530 eV displays the hybridized oxygen 2pz – 3d metal states. Therefore, this peak is prominent at grazing X-ray incidence and decreases at normal X-ray incidence. The second structure at 538 eV originates from transitions to the hybridized O 2pxy orbital – 3d metal 4sp bands. It is strong only in the normal incidence geometry where ∆m = ±1. For pz states m = 0 and hence the pz state does not respond to normal incident light with circular polarization. Therefore, the opposite angular dependence is observed in comparison to the 530 eV structure. The broad peak at 550 eV stems from the scattering of the photoelectron at the nearest neighbor 3d metal atoms. Since quite similar spectral features and angular dependencies are determined as seen in Fig. 3 (left) we conclude that the local geometry and the local bonding of the surfactant oxygen atom to the 3d metal atoms is alike for the three films. Turning to the dichroic spectra shown in Fig. 3 (right) we find similar trends: All the XMCD spectra exhibit a sharp dichroic contribution at 530 eV. This negative contribution reveals that the induced orbital moment is aligned parallel to the spin and orbital moments of the ferromagnetic films. Since no XMCD signal is determined at 538 eV we conclude that for all the films the O 2pz are magnetically polarized. Density functional calculations for O adsorption on the fcc Fe(100), Co(100), and Ni(100) surfaces were conducted using the thin-film full potential linearized augmented plane wave (FLAPW) method [4]. Both the local density approximation (LDA) and the generalized gradient approximation (GGA) were adopted to describe the exchange-correlation interaction. The optimized distances between the oxygen adatom and its nearest Fe, Co, and Ni neighbors

174

are 1.85-1.86 ˚ from LDA calculations and 1.93 ˚ from GGA calculations. From the surface A A extended X-ray absorption fine structure (SEXAFS) at the O K edge of 15 ML Ni grown with O surfactant on Cu(100) we determine a nearest neighbor distance Rnn = (1.85 ± 0.03) ˚ A of the O atoms to the Ni atoms of the topmost layer [4]. Since the LDA results agree with our SEXAFS data very well they are employed for other comparisons henceforth. The experimental determination of the local structure is quite helpful since it turns out that the size of the magnetic moments of the Ni atoms on the surface and also the induced moment in the surfactant oxygen sensitively depend on this distance. The calculated spin magnetic moments projected 0.05 2 into the O muffin-tin sphere (r = 0.74 ˚) are 0.053µB , A 0.132µB , and 0.053µB in O/Fe(100), O/Co(100), and O/Ni(100), respectively. Their orbital moments are very 0 0 small, 0.0024µB , 0.0047µB , and 0.0021µB . The O adFe atoms significantly reduce the magnetization of the Co -2 surface 3d atoms. Particularly, the spin magnetic moNi -0.05 ment of the surface Ni atom is only 0.26µB , much 2 smaller than those of the interior Ni atoms, 0.66 − 0.02 0.69µB . It is important to note that the spin and orbital magnetic moments of both O and surface Fe, Co, 0 0 and Ni atoms strongly depend on the relaxation of the oxygen atom. Experiment -0.02 The calculated XAS and XMCD spectra for the nor-2 Theory mal incidence geometry are presented in Fig. 4, accom530 540 550 E (eV) panied by the experimental data for O/Ni(100). For more details about the calculations, see Ref. [4]. In- Figure 4: Top: Calculated XAS and triguingly, all the spectroscopic features, including the XMCD of the O adatom on fcc Fe(100) peak structures and the XMCD/XAS ratio, are sat- (dash-2-dotted line), Co(100) (dashed isfactorily reproduced. Note also that there is no free line), and Ni(100) (dotted line) films for the normal incidence geometry. Bottom: scaling factor between experiment and theory. This inMeasured (solid line) and calculated (dotdicates that the model used in calculations represent ted line) XAS and XMCD of the O the atomic arrangements in experimental samples very adatom on the Ni(100) film at normal inwell. Despite the fact that the spin and orbital mag- cidence. netic moments of O are small, the XMCD signals are sizable at the threshold for all three systems. Concluding, the excellent signal-to-noise-ratio for XMCD measurements at the UE56/2PGM2 allows for a determination of the induced magnetism at 0.5 ML O on surfactant grown Fe, Co, and Ni films. Calculations reproduce the measured spectra quite well and complement the experimental findings by providing, for instance, the magnetic moments. This work was supported by the BMBF (05 KS4 KEB/5). References
[1] M. Farle, Surf. Sci. 575, 1 (2005). [2] R. N¨nthel, T. Gleitsmann, P. Poulopoulos, et al. , Surf. Sci. 531, 53 (2003). u [3] R. N¨nthel, J. Lindner, P. Poulopoulos, and K. Baberschke, Surf. Sci. 566-568, 100 (2004). u [4] C. Sorg, N. Ponpandian, M. Bernien, et al. , Phys. Rev. B. 73, in print (2006). [5] J. Lindner, P. Poulopoulos, R. N¨nthel, et al. , Surf. Sci. 523, L65 (2003) . u [6] J. Hong, R. Q. Wu, J. Lindner, et al. , Phys. Rev. Lett. 92, 147202 (2004). [7] T. Nakagawa, H. Watanabe, and T. Yokoyama, Phys. Rev. B 71, 235403 (2005). [8] C. Sorg, N. Ponpandian, A. Scherz, et al. , Surf. Sci. 565, 197 (2004). [9] F. May, M. Tischer, D. Arvanitis, et al. , Phys. Rev. B 53, 1076 (1996).
Normalized XAS (arb. units)

Normalized XMCD (arb. units)

175

Measuring the kernel of time-dependent density functional theory with X-ray absorption spectroscopy of 3d transition metals
A. Scherz1,2 , E.K.U. Gross1 , H. Appel1 , C. Sorg1 , K. Baberschke1 , K. Burke3 and H. Wende1
1

Institut f¨r Experimentalphysik, Freie Universit¨t Berlin, Arnimallee 14, D-14195 Berlin, Germany u a 2 SSRL, 2575 Sand Hill Road, Menlo Park, California 94025, USA 3 Department of Chemistry and Chemical Biology, Rutgers University, 610 Taylor Rd, Piscataway, NJ 08854, USA
E3
(3d)

e3

Ground-state density functional theory (DFT) is wellestablished for atoms, molecules, and solids. But groundw1, fs,1 W1, f1 state DFT produces only a one-particle picture of the w2, fs,2 W2, f2 electronic transitions in matter, neglecting interactions E (2p ) between excitations. The spectroscopic properties of e E (2p ) matter in the X-ray regime are substantially governed by dynamical many-body effects involving the creation e of a localized core hole [1–6]. While GW calculations Figure 1: Schematic illustration of the and the Bethe-Salpeter equation can be used [4], these DPA model. The model describes the shifts of the excitation energies (un- are computationally demanding. The simpler and less correlated ωi and correlated Ωi ) and expensive methodology of TDDFT is now being develthe changes in corresponding oscillator oped for these effects [5]. We analyze this approach to the X-ray absorption of strengths fi in the presence of an excited core hole [1]. itinerant systems like the L2,3 absorption of 3d transition metals (TMs), i.e., exciting a photoelectron from the localized 2p core states into the 3d band [1]. L2,3 X-ray absorption spectra (XAS), especially of early 3d TMs, suffer from core-hole correlation effects [2]. Schwitalla and Ebert [3] applied TDDFT linear response theory to calculate the XAS of the 3d TMs. Using a local approximation to the frequency-dependent exchange-correlation (XC) kernel, as proposed by Gross and Kohn [7], they qualitatively reproduced the trend of the branching ratios. However, whenever DFT is applied in a new regime, a difficult question arises: Are the existing functional approximations sufficiently accurate in this new regime? And how does one separate XC errors from those due to the practical approximations needed for realistic calculations? The true value of DFT is in constructing one XC approximation that covers many situations, in order to build-in knowledge of the underlying physics. Our approach here is different, and is based on the philosophy of Ref. [8]. That work examined the TDDFT response when excitations are not strongly coupled to each other. A useful series was developed in the strength of the off-diagonal matrix elements, relative to the frequency shifts induced by diagonal terms. The leading term yields the single-pole approximation [9], which has proven very useful in understanding TDDFT corrections to the one-particle picture. It even yields an immediate estimate of the XC kernel, but only if excitations are well-separated, a criterion rarely realized in practice [8]. However, the same philosophy applies to cases of two levels strongly coupled to one another, but weakly coupled to the rest of the spectrum. We call this the three-level or double-pole approximation (DPA), cf. Fig 1. Moreover, the L2,3 absorption of 3d TMs provides an ideal example of two transitions much closer to each other than the rest of the spectrum. With this in mind, we experimentally measured the branching ratios and level splittings of the 2p3/2 (L3 ) and 2p1/2 (L2 ) core states, and now deduce off-diagonal matrix
2
3/2

2

1

1/2

1

176

elements of the unknown XC kernel [1]. Since we can also compare with the one-particle KohnSham (KS) spectrum, we can also deduce the diagonal matrix elements. We find that, despite the large deviation of branching ratios from their single-particle values, the off-diagonal matrix elements, a measure of core-hole interaction, are not large, and explain why. Thus the DPA to TDDFT explains the observed shifts and oscillator strengths, and also provide benchmarks for future XC kernel approximations. We believe this is the first experimental measurement of a matrix element of the XC kernel of TDDFT. In the L2,3 XAS of the 3d TMs, the description of the L L Ti electron core-hole interaction may be simplified by the assumption that the relativistic spin-orbit coupling (SOC) 1 0 in the 3d band states (∼ 0.05 eV) is small compared to 460 470 that of the core states (several eV) and can be neglected. V This means that the oscillator strengths fj of these lev1 els are all about equal, as their KS orbitals are essen0 510 520 530 tially identical. Since, in this limit, the absorption area Cr is proportional to the oscillator strength, weighted statistically according to the manifold of the j = 3/2 and 1 0 j = 1/2 subshells, the branching ratio of the KS system 580 590 L is BKS = A3/2 /(A3/2 + A1/2 ) ≡ 2/3, where Aj is the area Fe L under the peak of the j-th subshell. Here we replace all 1 dipole-allowed transitions ωjk from a particular absorp0 710 720 730 tion edge into the 3d band by a single particle transition, Photon Energy (eV) as illustrated in Fig. 1. In Fig. 2, we show our experimental Figure 2: The experimental isotropic absorption spectra (solid line) at the isotropic XAS for the 3d TM with almost empty 3d bands L2,3 edges are shown for the early 3d taken from Fe/TM/Fe sandwiches with TM = Ti, V, Cr TMs Ti, V, and Cr versus Fe. The and bulk-like Fe. The data were recorded at the UE56edge jumps are normalized to unity 1/PGM beamline at BESSY (for details, see Ref. [10]). for direct comparison. The continuum The edge jumps are normalized to unity. From these specin the experimental spectrum is simu- tra and their absolute energy dependence, the excitation lated by a two-step function as shown energies Ωq=1 at the L3 edge and Ωq=2 and the L2 edge are for Fe (dashed-dotted line). The treatdetermined. For the quantitative analysis of the branching ment of the core hole red-shifts the independent particle spectrum (dot- ratio B, we very carefully determined the L2,3 absorption ted line) and changes the statistical areas Aj . This determination has the advantage that B branching ratio in the correlated spec- becomes independent of the different L3 and L2 lifetime trum (dashed line) as revealed by the broadening and experimental resolution. Note, that the DPA model [1]. proper experimental intensity is given by the area and not by the height of the resonance. To determine the correct area of the L3 and L2 resonances the continuum contribution is removed (e.g. gray line for Fe in Fig. 2). Since the 2p SOC decreases towards lower atomic numbers the deconvolution is more complicated for the early 3d TMs Ti, V, and Cr because of the strong L2,3 overlap. The areas have been fitted using the Fe absorption spectrum as a background simulation underneath the L2 edge. In the case of Fe the L2 absorption is approximately half of the L3 peak, in agreement with the KS prediction. However, the branching ratios for the other 3d elements differ significantly from this. In particular, Ti has an L2 peak that is even larger than its L3 absorption. Thus, the experimental branching ratios cannot be interpreted in terms of KS orbitals, suggesting strong electron core-hole interactions. In the language of TDDFT, there must be significant offdiagonal matrix elements in Casida’s equations, describing the influence of the electron corehole interaction on the L2,3 XAS. (If only diagonal elements are considered, the eigenvalues are shifted but the eigenvectors are not rotated, and the oscillator strengths retain their KS values [8].) However, a fully numerical solution of the equations is not needed, as we know there are
3 2 3 2

XAS Intensity (arb. units)

177

KS Table 1: Excitation energies in eV obtained from KS calculations (ωi ) and from experiment (Ωi ), experimental branching ratio B and matrix elements Kij [1]. The experimental error of Ωi is below 10−3 , the one of B in the order of 1 %.

3d 22 23 24 26

TM Ti V Cr Fe

KS ω1 460.8 519.1 580.3 711.3

KS ω2 467.5 527.7 590.3 724.6

Ω1 455.4 513.6 575.1 706.7

Ω2 461.0 520.4 583.6 719.5

B 0.47 0.51 0.56 0.70

K11 -2.57 -2.65 -2.55 -2.29

K22 -3.34 -3.73 -3.40 -2.55

K12 0.54 0.54 0.47 -0.25

only two dominant transitions, so the electron core-hole interaction can be analyzed within the DPA model. The detailed calculations within the DPA model are given in [1] and the results are presented in Table 1, the corresponding theoretical DPA spectra are shown in Fig. 2. What can we learn from this elementary analysis? Our chief result is that the deviation from the KS branching ratio does not imply large off-diagonal matrix elements of K, i.e., large core-hole correlation. In fact, the off-diagonal elements are all about 1 eV or less, compared to diagonal elements of 5-7 eV. Moreover, from Ti-Cr, it is almost constant. The deviation from the KS branching ratio is simply level (or in this case, transition) repulsion, as the two transitions near one another. Thus the shifts are simply interpreted as diagonals of K, while the branching ratios are a sensitive determinant of off-diagonal elements. The success of DPA shows that very little effort beyond a ground-state DFT calculation is needed to compute these spectra in TDDFT. One only needs to integrate a given approximation to the XC kernel for the two diagonal matrix elements, and one off-diagonal. In summary, we have used TDDFT to understand the XAS of 3d transition metals by deriving a double-pole approximation. The main features observed in the experiments can easily be explained by assuming that the spectrum is dominated by two strongly coupled poles via the 2p − 3d core hole interaction. This shows that, for the beginning of the 3d series, the reduced 2p-SOC is responsible for the strong variation of the branching ratio, not strong interactions between the transitions. Our analysis does not replace a full TDDFT calculation of X-ray absorption spectra. Rather, for the very specific case of spectral regions dominated by two poles it provides, on the one hand, a transparent picture of the changes of spectral weights in particular for the early 3d TMs, and on the other, a straight-forward route to testing approximate XC kernels against experimental data. The work was supported by BMBF (05 KS4 KEB/5), the EXC!TING Research and Training Network of the EU, the NANOQUANTA network of excellence, the US DOE (DE-FG0201ER45928) and NSF (CHE-0355405). References
[1] A. Scherz, E.K.U. Gross, H. Appel, C. Sorg, K. Baberschke, H. Wende and K. Burke, Phys. Rev. Lett. 95, 253006 (2005). [2] J. Fink et al., Phys. Rev. B 32, 4899 (1985); J. Zaanen, et al., ibid. 32, 4905 (1985). [3] J. Schwitalla and H. Ebert, Phys. Rev. Lett. 80, 4586 (1998). [4] E.L. Shirley, Phys. Rev. Lett. 80, 794 (1998). [5] A.L. Ankudinov, A.I. Nesvizhskii, and J.J. Rehr, Phys. Rev. B 67, 115120 (2003). [6] O. Wessely, M.I. Katsnelson, and O. Eriksson, Phys. Rev. Lett. 94, 167401 (2005). [7] E.K.U. Gross and W. Kohn, Phys. Rev. Lett. 55, 2850 (1985). [8] H. Appel, E.K.U.Gross, and K. Burke, Phys. Rev. Lett. 90, 043005 (2003). [9] M. Petersilka, U.J. Gossmann, E.K.U. Gross, Phys. Rev. Lett. 76, 1212 (1996). [10] H. Wende, Rep. Prog. Phys. 67, 2105 (2004).

178

ARPES study of low dimensional surface states on the Pt(110) surface
M. Minca, E. Dona, S. Penner, V. Brouet*, A. Menzel, and E. Bertel Institut für Physikalische Chemie, Universität Innsbruck, Innrain 52A, A-6020 Innsbruck (Austria) *Laboratoire de physique des Solides, Université Paris-Sud, 91405 Orsay (France)

The photoemission intensity is described by two factors, the hole spectral function Ah(k,ω) and its modulation by the photoemission process, usually referred to as photoemission matrix element (ME) effects. ME effects might be strong [1] due to e.g. interferences from various atomic sites in the final state [2,3]. Thus, for a satisfactory interpretation of the ARPES spectra it is necessary to distinguish the effects of spectral function and photoemission matrix element. Although calculations in the framework of the one-step model of photoemission (see e.g. [2,3,4]) are able to capture experimentally observed features, the general understanding of ME effects and their dependence on dimensionality, geometry and temperature is still relatively poor. Recently, we found on clean Pt(110) low dimensional electronic states at the Fermi-energy EF which show an unusual intensity variation depending on temperature and adsorbates at 21.2eV [5,6]. The intense photoemission peak around 100meV below EF at the X point of the (1x1) surface Brillouin zone is derived from the backfolding of a surface resonance at S due to the 1x2 missing row reconstruction. In the following we discuss photoemission intensity changes of this peak which can be induced by adsorbing hydrogen into the so-called β2-adsorption-state on the Pt(110) surface. The 1x2 overstructure is not affected by hydrogen absorption, whereas the interlayer distance between the outermost missing rows and the second layer is substantially changed from 1.15A to 1.25A [7], the bulk interlayer distance being 1.38A. Since β2-hydrogen adsorption changes the geometry by increasing the first layer distance by 9%, this model system provides a unique possibility to disentangle different factors determining photoemission intensities.

clean Pt(110) norm. Int. [arb. un.] 3 2
hν=18eV 19

20 21 22 23 24 26

28

5 norm. Int. [arb. un.] 4 3 2 1 0 H/Pt(110) clean Pt(110)

1 0 10 12 14 16 18 20 22 24

13

14

15

16

kinetic energy[eV]

kinetic energy[eV]

Fig.1: Left: Comparison of EDCs at different photon energies for the clean surface. Right: Zoom into the intensity rise around 20eV photon energy, comparing H/Pt(110) and clean Pt(110). Gaussian envelopes (dotted, black is shifted by -150meV and multiplied by 0.68) are given to guide the eye.

This study was performed at the 10m-NIM beamline using the SURICAT experimental setup. The spectra were taken at the X -point of the surface Brillouin zone at T=300K (above the hydrogen desorption temperature) for the clean surface and at T=140K for β2-H/Pt(110). The analyzer resolution (PE=10eV) was 30meV, much better than the natural width of the surface resonance of roughly 100meV. All the spectra have been normalized to the lowest background signal as observed in wide energy range (12eV) spectra (see Fig1, left). Fig 1 shows a comparison of energy distribution curves (EDCs) at various photon energies for the two different surfaces. Several observations can be made: (a) the position of the peak at EF does not depend on photon energy and (b) is the same for both surfaces. (c) for all photon

179

energies, the intensity of the peak is smaller for the clean surface. As has been shown before [6,7], this is not due to the difference in the Fermi-distribution at 140K and 300K. (d) the intensity oscillates with photon energy where (e) the behaviour of intensity versus photon energies is very similar for both surfaces. Observations (c) and (e) show that photoelectron diffraction by the outermost layer cannot be the reason for the influence of hydrogen and the intensity oscillation in general: In the simple diffraction model, the 9% change in geometry between first and second layer should change the kinetic energy for constructive interference for clean and H/Pt(110) in the final state by more than 1.5eV. As shown in Fig.1 (right) by a zoom into the photon energy range from 17.5 to 21eV, this is clearly not observed. Furthermore, the spacing between intensity the maxima (d) at roughly 15, 20, and 30eV (not shown) amounts to a few eV, which in reciprocal space corresponds only to a small fraction of the Brillouin zone in the direction perpendicular to the surface (k⊥). Observations (a), (d) and (b) corroborate earlier assignments of the peak at EF [5,6] and hydrogen adsorption sites [7]: Generally, a surface state or a surface resonance should not show any sizeable dispersion (a) with momentum k⊥. Nevertheless, a surface state or resonance extends coherently over a few layers and thus some bulk k⊥ components will dominate in an expansion in bulk Bloch states. Consequently, the remainder (a “propensity rule”) of the k⊥-conservation rule in the bulk is expected to modulate (d) the photoemission intensities upon variation of photon energy. The symmetry of the surface resonance wave function [5,6] and the H adsorption site [7] are consistent with (b) no shift in energy after hydrogen adsorption: The H1s orbital is situated in the short bridge site (symmetry gerade) and cannot interact with the quasi-1D Tamm d-type surface resonance wave function at the X -point (ungerade). On the basis of the present analysis, a simple diffraction model in the final state, such as the assignment of the peak at X to a final state umklapp from the S point, can be definitely excluded. However, a clear separation of subtle ME effects (see the shift of 150meV suggested by the Gaussian envelopes drawn in Fig1, right) and changes in the initial state effecting Ah(k,ω) is not possible. This issue will be addressed in a more detailed analysis. Acknowledgements: We thank A. Vollmer and G. Reichardt for their excellent support. Financial support by the European Community and the Austrian Science Found (FWF) through contracts R II 3-CT2004-506008 and S 9004-N02, respectively, is gratefully acknowledged. References : [1]. M.C. Asensio, J. Avila, L. Roca, A. Tejeda, G.D. Gu, M. Lindroos, R.S. Markiewicz, and A. Bansil, Phys. Rev. B 67, 014519 (2003). [2]. M. Lindroos, S. Sahrakorpi, and A. Bansil, Phys.Rev. B 65, 054514 (2002). [3]. R. Matzdorf, R. Panagio, G. Meister, A. Goldmann, Ch. Zubrägel, J. Braun, and G. Borstel, Surf. Sci. 352-354, 670 (1996). [4]. R. Eder and H. Winter, Phys. Rev. B 70, 085413 (2004). [5]. A. Menzel, Zh. Zhang, M. Minca, Th. Loerting, C. Deisl, E. Bertel, New Journal of Physics 7, 102 (2005). [6]. A. Menzel, Zh. Zhang, M. Minca, Th. Loerting, C. Deisl, E. Bertel, J. Phys. Chem. Sol., in press (2005). [7]. Z. Zhang, M. Minca, C. Deisl, T. Loerting, A. Menzel, E. Bertel, R. Zucca and J. Redinger, Phys. Rev. B 70, 121401(R) (2004).

180

Schottky barrier formation and interdiffusion at CdTe/metal interfaces.
B. Späth, J. Fritsche, A. Klein, W. Jaegermann Cadmium telluride is a promising photovoltaic material for thin-film solar cells. It has a near ideal band gap for photovoltaic conversion efficiency of 1.49eV [1] and a high optical absorption coefficient. The processes used to make all the films, which compose the cell, are quite simple and fast. Small-area CdTe cells with efficiencies of more than 15% have been developed [2, 3]. Ohmic contacts without electrical losses are needed to fully exploit the potentials of CdTe solar cells. Formation of low resistance and stable back contacts has been a research issue for many years[4]. Best contacts are typically obtained with Cu-containing contacts materials. It is generally believed that diffusion of Cu into CdTe leads to an increase of p-type doping of the CdTe substrate, enabling contact formation. In previous studies CdTe/metal contacts have shown significant limitations due to a formation of a Schottky barrier of approx 0.9eV independent of the deposited metal. An important factor for the Schottky barrier formation seems to be a reaction of the metal with the CdTe, the resulting compounds and the formation of elementary Cd. The formations of CdTe/Cu contacts were investigated with XPS measurements. A stepwise deposition of Cu onto CdTe was carried out at room temperature and 250°C. After every deposition step XPS measurements were made. The SoLiAS UHV system attached to the TGM7 monochromator offers all the experimental requirements needed. It combines a high-resolution spectrometer and a preparation chamber. The Te4d-, Cd4d- and valence band spectra for different Cu deposition times at room temperature are shown in Fig. 1. As expected with increasing Cu coverage, the Te4d- and Cd4d- emissions decrease while the structure of the valence band spectra from CdTe changes to the structure of the valence band of Cu. After a deposition time of 4000s seconds no Cd emissions remain in the spectra, but a small Te4d-emission can be seen. After 90s Cu deposition a slight shift of 0.13eV towards higher binding energies can be seen in the Te4dand Cd4d-emission, due to a shift of the Fermi level in the CdTe substrate. From this follows a Schottky barrier of ΦB=0.85±0.1eV. After 270s Cu deposition an additional Cd4d-emission is detectable at lower binding energies, due to the appearance of elementary Cd0. This is the result of a reaction of Cu with CdTe with the formation of Cu2Te and Cd0: CdTe + 2Cu = Cu2Te + Cd. After an annealing step at 400°C the Te4d –emission increase and the structure of the valence band changes but no Cd-emission is detectable. This indicates the formation of Cu2Te. In Fig. 2 the Te4d-, Cd4d- and valence band spectra for different Cu deposition times at 250°C are shown. A shift of 0.14eV towards higher binding energies in the Te4d- and Cd4demission is detectable after 90s Cu deposition resulting in a Schottky barrier of ΦB=0.85±0.1eV. Also a decrease of the Te4d- and Cd4d-emission can bee seen but no additional emission of Cd0 is detectable. Presumably, the Cd0 diffuses into the CdTe layer or evaporates at this temperature. From the beginning of the deposition of Cu the valence band spectra does not show the structure of elementary Cu but it looks similar to the post heated surface from the first experiment. At 250°C substrate temperature the same Schottky barrier is formed like at room temperature and the Cu reacts with the CdTe: CdTe + Cu = Cu2Te + Cd. The performed experiments clearly show the interface reaction of CdTe with Cu. The Fermi level at the CdTe/Cu interface is again stabilized at ~ 0.85eV, which is close to the calculated defect level of interstitial Cd.
181

Fig.1: Te4d, Cd4d and valence band spectra from a CdTe surface with increasing Cu coverage at room temperature and an additional post heating.

Fig.2: Te4d, Cd4d and valence band spectra from a CdTe surface with increasing Cu coverage at 250°C substrate temperature. References 1. Fritsche, J., et al., Band energy diagram of CdTe thin film solar cells. Thin Solid Films, 2002. 403: p. 252-257. 2. Britt, J. and C. Ferekides, Thin-Film CdS/CdTe Solar-Cell with 15.8-Percent Efficiency. Applied Physics Letters, 1993. 62(22): p. 2851-2852. 3. Aramoto, T., et al., 16.0% efficient thin-film CdS/CdTe solar cells. Japanese Journal of Applied Physics Part 1-Regular Papers Short Notes & Review Papers, 1997. 36(10): p. 6304-6305. 4. Kraft, D., Präparation und Charakterisierung von Dünnschichtmaterialsystemen für die Rückkontaktbildung bei polykristallinen Dünnschichtsolarzellen, ed. Dissertation. 2003, Technische Universität Darmstadt.

182

Spin-conservation upon hybridization: A spin-resolved resonant photoemission study of 1ML Ce/Fe(110)
Yu. S. Dedkov,1 M. Fonin,2 S. L. Molodtsov,1 U. Rüdiger,2 and C. Laubschat1
1 2

Institut für Festkörperphysik, Technische Universität Dresden, 01062 Dresden Germany , Fachbereich Physik, Universität Konstanz, 78457 Konstanz, Germany

The electronic properties of rare-earth (RE) transition-metal (TM) compounds are strongly dependent on the properties of the RE 4f states that maintain their atomic-like character in the solid state and become exchange-coupled via the RKKY-interaction. Hereby, light REs couple usually antiferromagnetically with respect to the TM ions while heavy REs couple ferromagnetically. On the other hand, magnetic ordering may compete with the Kondo-effect that can be described by an electron hopping interaction in the light of the Anderson model and leads to mixed-valent behavior, heavy-fermion properties and in extreme cases even to a breakdown of the Fermi-liquid picture [1,2]. Particularly Ce, the first element of the RE series, is subject of such kind of phenomena. As a function of temperature or pressure, Ce metal undergoes an isostructural phase transition from the paramagnetic - to the almost nonmagnetic -phase that is accompanied by a volume collapse by more than 15%. The latter indicates a decrease of 4f occupation caused by increased hybridization of the 4f states with the surrounding valence-bands (VB). In CeTM compounds, the 4f hybridization is usually strong due to the large VB DOS and the systems behave so-called -like . Most direct insight into the electronic structure of these systems is obtained by means of photoemission (PE): In PE spectra, Ce 4f hybridization is reflected by a characteristic double-peak structure consisting on a peak at around 2 eV bindingenergy (BE) that may be viewed as the 4f 0 configuration expected from the photoionization of the Ce 4f 1 ground state, and a spin-orbit split feature at the Fermi energy, EF, that is a pure hybridization effect and reproduces the 4f 1 ground state. This phenomenon may quantitatively be described within the single-impurity Anderson model using the BE of the unhybridized 4f 0 state, , and a hybridization parameter, , as adjustable parameters [3]. The latter describes the hopping interaction between the 4f state and the VB states represented by the non-4f derived density of states (DOS) calculated in the light of a LDA bandstructure calculation. Introduction of the on-site Coulomb repulsion energy, Uff, allows additionally to describe effects of doubleoccupation of 4f states caused by (a) (b) 2 admixtures of 4f configurations Fig. 1. LEED images of (a) to the ground and final states. Fe(110) surface and (b) after No spin-resolved PE study on a deposition of 1 ML of Ce on it. ferromagnetically ordered CeTM (c) Surface crystallographic structure of 1 ML Ce/Fe(110) compound has been reported so far. system obtained after simulation In the present contribution we of corresponding LEED patterns. report on first results of such a Rhombus and rectangular shows study on Ce/Fe(110). Thin film Fe and Ce unit cells for samples were prepared in-situ by corresponding (110) planes of bulk materials. (i) deposition of 5 nm Fe on a W(110) substrate, followed by (ii) (c) thermal annealing in order to

183

achieve an epitaxial Fe layer with (110) orientation, and (iii) deposition of a monolayer Ce. As a result, a sharp LEED-pattern was obtained showing a ( 2 2 2 ) (in analogy with squared structure) overstructure with respect to Fe(110) that might be explained by an arrangement of Ce and Fe atoms as shown in Fig. 1. Spin-resolved resonant PE experiments at the Ce 4d 4f absorption threshold were performed using a POIBOS 150 electron-energy analyzer combined with mini-Mott detector for spin-analysis from SPECS and synchrotron radiation from the U 125 undulator beamline of BESSY II. The samples were magnetized by a magnetic coil and PE spectra were taken in remanence. The upper pannel of Fig. 2 shows spin-resolved off-resonant PE spectra taken at 112 eV photon energy that are dominated by emissions from the exchange-split Fe 3d bands. As for pure Fe, the majority-spin band (triangle up) is shifted towards higher BE and is almost occupied, while the minority-spin band is shifted towards lower BE Fig. 2. (a) Spin-resolved off-resonance photoand is cut by the Fermi energy in a region of high electron spectra of the Ce/Fe(110) system. (b) Ce DOS. The lower panel of Fig. 2 shows respective 4f partial contribution obtained after subtraction Ce 4f spectra, taken on-resonance at 121 eV and (inset) corresponding spin polarization. photon energy. Residual Fe 3d contributions were removed by subtracting the off-resonance spectra, properly normalized with respect to the beam current. The observed asymmetry indicates that the Ce 4f states are indeed mainly antiferromagnetically coupled with respect to the Fe spins. Both majority and minority spin components reveal the characteristic double peak structure expected for an -like system, interestingly, however, the relative intensity of the Fermi-level peak with respect to the 4f 0 feature of the minority spin component is somewhat larger than the one of the majority spin component pointing to an increased hybridization of the minority spin channel. This could be explained by the fact that for the minority spin direction the Fe 3d DOS at the Fermi level is higher than the one of the opposite spin direction leading to enhanced hybridization if spinconservation upon hopping is assumed. In the energy region between the 4f 0 and 4f 1 peaks, a third, but weaker feature becomes visible that changes its BE position with spin orientation and reflects probably a Ce 5d signal. In order to confirm this assumption layer-resolved bandstructure calculations are under way. We would like to acknowledge help and useful discussions of BESSY staff during experiment. The work was supported by the BMBF, contract no. 05 KS4OD1/4, and the SFB 463, TP B4. [1] H. R. Kirchmayr and C. A. Poldy, in Handbook on the Physics and Chemistry of Rare Earths ed. K. A. Gschneidner, Jr. and L. Eyring, vol. 2 (North-Holloand, Amsterdam, 1978). [2] L. Degiorgi, Rev. Mod. Phys. 71, 687 (1999). [3] R. Hayn et al., Phys. Rev. B 64, 115106 (2001).

184

X-ray Emission Investigation of the O Kα Band in H-RUB-18 and in α-quartz
O.Yu. Khyzhuna,1, T. Strunskusa , H. Giesb, Ch. Wölla Lehrstuhl für Physikalische Chemie I, Ruhr-Universität Bochum, Universitätsstraβe 150, D44780 Bochum, Germany b Lehrstuhl für Kristallographie, Institut für Geologie, Mineralogie und Geophysik, RuhrUniversität Bochum, Universitätsstraβe 150, D-44780 Bochum, Germany 1 Permanent address: Frantsevych Institute for Problems of Materials Science, National Academy of Sciences of Ukraine, 3 Krzhyzhanivsky street, UA-03142 Kiev, Ukraine
a

1. Introduction Highly crystalline α-quartz and the crystalline layered silicic acid H-RUB-18 were used as reference systems for a ongoing study of microporous and mesoporous silica materials. The possible bonding neighbourhoods of oxygen are similar in all materials, but their quantity varies in the different silicas. Ideally as in α-quartz, oxygen is bonded to 2 silicon atoms. In defects, on the other hand oxygen is bound to one Si atom and one H atom. In H-RUB-18 oxygen is bound to one silicon and one proton in crystalline order. This allows to study the influence of the presence of silanol units (Si-OH) upon the electronic structure in silica materials. 2. Experimental In the present study the following samples were used: 1) pure α-quartz, α-SiO2; Dörentrup (Sauerland); space group (SG): P3221; lattice parameters: a = 4.92 Å, c = 5.41 Å, γ = 120°; 2) the crystalline layered silicic acid H-RUB-18; SG: I41/amd; the structure of HRUB-18 can be represented as a sequence of pseudo tetragonal silicate layers with an intra-layer repeat unit a = 7.38 Å and inter-layer distances of c = 7.44 Å.1 Resonant SXE and near-edge X-ray absorption fine structure (NEXAFS) measurements at the O K edges were carried out using undulator beam-line U41-PGM. The spectra were measured with a high-resolution Rowland-mount grazing-incidence grating Xray emission spectrometer Scienta XES 300 (Gammadata, Sweden) equipped with a twodimensional detector. The exit slit of the beam-line during the RSXE measurements was set to 100 μm, yielding a total resolution (combined from synchrotron excitation and spectrometer energy resolution) of approximately 1.1 eV. This value was determined by measuring the full width at half maximum (FWHM) of the elastic peaks in the energy region at the O K edge. The NEXAFS spectra were recorded in the fluorescence yield (FY) mode. No changes in the resonant SXE and NEXAFS spectra were observed as a function of exposure time to the synchrotron radiation. All the measurements were made at room temperature. 3. Results and Discussion Figure 1 represents the NEXAFS O 1s spectra of α-quartz and H-RUB-18. Energies of photons used in the present work for excitation of the resonant SXE O Kα spectra are marked
185

2 by arrows above the NEXAFS spectra. It is apparent from Figure 2 that the energy positions of the O K onsets differ slightly for the compounds studied and some minor changes are characteristic for the shapes of the NEXAFS O 1s spectra just above the top of the O K absorption bands positioned at about 539-542 eV.
NEXAFS O 1s Intensity (arbit rary units) D H-RUB-18 C B A
α -Quart z

Figure 1. NEXAFS O 1s spectra of pure α-quartz H-RUB-18; abscissa is the excitation photon energy. Energies of photons employed for excitation of the resonant soft X-ray emission spectra presented in Figure 2 are marked by arrows above the NEXAFS O 1s spectra.

530

540 550 Photon energy ( eV)

560

In Figure 2 we present a series of the resonant SXE O Kα spectra recorded for αquartz and H-RUB-18 when employing different excitation energies as marked in Figure 1. The O Kα bands excited with the photon energy corresponding to the top of the broad O K absorption bands (point D on the NEXAFS O 1s spectra presented in Figure 1) look quite similar for the two compounds. For the shape of the O Kα band in the two compounds studied, in addition to the main peak “d”, the presence of a pronounced feature “b” and two shoulders, “a” and “c”, is characteristic.
α-Q uar tz

(I )
Intensi ty (arbi trary u nits)

H-RUB-1 8

(I I)

Intensi ty (arbi trary u nits)

A ( 5 34.0 eV)

A

B ( 53 5.0 eV) C ( 53 6.5 eV) b D ( 54 0.0 eV) 512 a c

d

B C D 5 12 516 520 52 4 Photon energy (eV) 528 532

51 6 520 524 Photon ener gy ( eV)

528

5 32

Figure 4. Resonant soft X-ray emission O Kα bands of (I) α-quartz and (II) H-RUB-18 excited with photon energies indicated by the arrows A–D in the NEXAFS O 1s spectra presented in Figure 1. When tuning the excitation energy across the slope of the O K absorption edge just below the O K absorption band, a small increase of the relative intensity of the feature “b” of the O Kα band (with respect to the intensity of its maximum “d”) is observed and the valley between the sub-band “b” and the main portion of the O Kα band vanishes (cf. curves D and

186

3 C in Figure 2). When tuning the excitation photon energy in the energy region corresponding to the position of the O K onset, the valley vanishes completely. The shape of the resonant SXE O Kα bands excited with the photon energies of 534 eV and 535 eV employed in the present experiments differ significantly. The similarity of the SXE O Kα spectra excited with the photon energy taken above the O K absorption band (curves D in Figure 2) we can explain by the fact that in all the studied compounds the chemical bonding is determined mainly by the interaction between silicon and oxygen atoms. Surprisingly, the SXE spectra reveal that, the presence of the silanol units in H-RUB-18 does not affect the shape of the SXE O Kα bands that is characteristic for α-quartz. The relative intensities of the fine-structure features of the SXE O Kα bands (as compared to the intensity of the maximum “d” of the band) in the above compounds are quite similar if the spectra are excited with photon energies above the top of the O K absorption bands. Currently, ab initio quantum mechanical calculations are in progress in order to help to understand this result. Changes of the shape of the O Kα bands start to become pronounced when exciting with the photon energies close to the position of the onset of the O K thresholds. The resonant SXE O Kα spectra reveal the appearance of the Raman-like inelastic peak with an energy loss of −11 eV as can be seen from Figures 2. The changes of the SXE O Kα bands when employing photon excitation energies close to the O K onset energy in H-RUB-18 and in αquartz is explained by superposition of the inelastic peak “B′” with the remaining part of the O Kα band. The rather big differences in relative intensities of the inelastic peak and of the resonantly excited SXE O Kα bands in the two materials are not well-understood at present and additional studies are in progress. 4. Conclusions We have measured NEXAFS O 1s and resonant SXE O Kα spectra for a series of excitation energies at the O K threshold of a crystalline layered silicic acid H-RUB-18 and pure α-quartz. Significant differences were expected for the O Kα spectra for the two compounds since in α-quartz the chemical bonding is determined only by Si–O bonds whereas H-RUB-18 contains a large fraction of silanol groups. Surprisingly, the nonresonant SXE O Kα spectra for the two compounds look very similar and only the resonant spectra (when core electrons are excited by photons to unoccupied states close to the position of the Fermi level) reveal specific changes. The influence of the silanol groups on the shape of the SXE O Kα band in silica materials is not well understood in detail at present and ab initio quantum mechanical calculations are in progress to gain more insight into this matter. Acknowledgements The authors thank support by Prof. R. Szargan (University of Leipzig) and the BESSY staff (especially Ch. Jung and M. Mast). Travelling costs for synchrotron measurements provided by BMBF through grant 05 ES3XBA/5 are gratefully acknowledged. Reference [1] Borowski, M.; Marler, B.; Gies, H. Z. Kristallogr. 2002, 217, 233.

187

The Molecular Orientation of DNA Bases on H-passivated Si(111)(7x7) investigated by means of Near Edge X-ray Absorption Spectroscopy
S. Seifert, G. Gavrila, Y. Suzuki, D.R.T. Zahn Institut für Physik, Technische Universität Chemnitz, D-09107 Chemnitz, Germany W. Braun BESSY GmbH, Albert-Einstein-Str. 15, D-12489 Berlin, Germany In recent years DNA bases have been discussed as promising candidates for electronic applications, such as (bio-)organic field effect transistors [1] or molecular nano-wires [2]. The strong anisotropy of charge transport within molecular crystals makes the knowledge of the molecular orientation of the DNA bases crucial for any device design. We therefore performed a systematic Near Edge X-ray Absorption Fine Structure (NEXAFS) spectroscopy study of layers of the DNA bases adenine, cytosine and guanine on H-passivated Si(111). The measurements were performed using the Multi User Stage for Angular Resolved Photoemission (MUSTANG) experimental station at the Russian German beam line at BESSY. Experiment: The substrates were cut from a n-type, highly phosphor doped (resistivity 7.5 Ω/cm) silicon(111) wafer supplied by SilChem GmbH. The samples were annealed by direct current (DC) heating under ultra high vacuum (UHV) conditions (base pressure < 3.10-10mbar) up to 750°C-800°C. The natural oxide was removed by several DC flushes of 20s duration up to 1100°C-1300°C. The substrates were then cooled down slowly to preserve the (7x7) reconstruction. In order to prevent a reaction of the DNA bases with the substrate, the surface was passivated in situ by exposure to (2.0±0.5) Langmuir atomic hydrogen.The dosis was chosen carefully and should be just enough to saturate the dangling bonds of the Si(111)(7x7) without etching the surface. This method of passivation leads to considerably lower surface roughness than achieved by a wet-chemical cleaning and passivation treatment of Si(111).The DNA base layers were deposited by organic molecular beam deposition (OMBD). The nominal layer thickness was monitored by a quartz micro balance. The (NEXAFS) spectra were recorded in the partial electron yield mode in the region of the secondary electron background (at a kinetic energy of 10eV). The angle of incidence, Θ, between the incident light and the sample surface was varied between of 22°-115°. The measured data were corrected for the photon flux by division of the spectra by the electron yield of the clean H-Si(111)(7x7)

Fig.1: The carbon K-edge NEXAFS spectra of a 10nm adenine (left) cytosine (midle) and guanine (right) on H-Si(111)(7x7) as a function of the angle of incidence, Θ.

188

sample and the synchrotron ring current, the background was subtracted and the spectra were normalized to the absorption step edge at 320eV. The angular dependent carbon K-edge NEXAFS spectra of 10nm thick layers of adenine, cytosine and guanine on H-Si(111)(7x7) are shown in Fig.1. Peak assignment: The prominent features at the low excitation energies correspond to transitions from the carbon 1s core levels into the lowest unoccupied molecular orbitals (LUMO, LUMO+1 ...). The validity of this assignment can be checked by comparison the measured spectra with theoretical calculations employing density functional theory (DFT) for the single DNA base molecules (B3LYP functionals; 6-311G+(d,p) basis sets). For these calculations the commercial software package GAUSSIAN03 [3] was used. The contribution of different excitation sites (i.e. the carbon atoms) are treated separately. In molecules as small as the DNA bases the core hole created at the excitation site has a strong impact on the molecular orbitals and has to be considered in the calculation. This is done by the introduction of an extra charge to the core of the excited atom by replacing it with its Z+1 equivalent (i.e. nitrogen). The DFT calculation is then performed for the positively charged molecule, in order to keep the number of electrons constant. Afterwards, the calculated unoccupied molecular orbitals are decomposed into the contributing atomic orbitals*. The contribution of a single carbon atom to the π*-resonances in the NEXAFS spectrum is mainly contained in the contribution of its antibonding 2pz atomic orbital (were the z-axis is perpendicular to the molecular plane) to the unoccupied molecular orbitals. The excitation energy necessary for a transition into these states is calculated by subtracting the 1s core level binding energy of the particular carbon atom (measured by core level photoemission spectroscopy) from the calculated eigen energies. The contributions of all the carbon atoms within the molecule are averaged and broadened with Voigt functions of 0.3eV FWHM. In Fig.2 the calculated curves are compareed to the measured spectra. For adenine and cytosine, the Z+1 apporoach leads to very good agreement between the simulation and the measurement. The resemblance is not as good in the case of guanine, but still the π*-nature of the peaks at the lowest excitation energies becomes obvious. The relatively large shift, which had to be introduced, to match the peaks with the highest intensities is due to the fact, that the extra positive charge on the molecule is overestimating the core hole effect on the molecular orbitals.

Fig.2: In the Z+1 approximation the DFT calculations are performed after substitution of a carbon (Z=6) atom by a nitrogen (Z=7) atom. The spectra are derived by assuming vertical transition between the atomic 1s and 2p z orbitals. The spectra were shifted by ∆EC = 4.36eV, ∆EA = 2.86eV and ∆Eg = 2.54eV towards higher excitation energies.
*

The atomic orbitals were calculated from the Gaussian’03 output, using the AOMix program [5,6], which employs a Mulliken population analysis

189

Fig.3: The dependence on the incidence angle (Θ) of selected π∗-transition peaks in the NEXAFS spectra of adenine (left), cytosine (middle) and guanine (right). By variation of α and C to optimize the match between measurement and equation (1) the molecular orientation of each DNA base is obtained.

Molecular orientation: For the quantitative analysis of the angular dependence of the NEXAFS spectra at first the π*-transition peaks were fitted using Voigt functions. Because of the spherical symmetry of the initial state (a 1s orbital), the transition dipole moments of these resonances are oriented parallel to the final state, π*-orbitals, which are oriented perpendicular to the molecular plane. In this case, the dependence of the resonance intensity on the angle of incidence, under the condition of threefold (or higher) substrate symmetry is given by [7] 1 2 1− P 2 2 2 2 I =C [ P cos  cos  sin  sin  sin ] 2 2 1

where P is the degree of polarization and C a normalization factor. Θ is the angle of incidence and α the angle between the transition dipole moments and the surface normal (or the molecular tilt angle). The angle Θ and the polarization factor are known quantities, which only leaves the molecular orientation and the normalization constant C unknown. These parameters can be determined by curve fitting the above equation to the relative intensities of the π*-resonances in the NEXAFS spectra of adenine, cytosine and guanine. The fitted curves are presented in Fig.3. The average molecular tilt angles of the DNA base molecules with respect to the substrate surface are determined to be: αA = (24°±3°); Acknowledgments: αc = (40.5°±1.5°); αG = (25.8°±1.3°) members for their

The authors would like to thank the BESSY staff

assistance during the beam times (especially Mike Sperling for the technical support). We also acknowledge the financial support granted by the BMBF (FK MUSTANG 05KS40C1/3, FK 05KS1OCA1). References:
[1] G. Maruccio, Field effect transistor based on a modified dna base. Nano Letters 3, 479 (2003) [2] D. Porath, A. Bezryadin, S. de Vries, and C. Dekker, Direct measurement of electrical transport through dna molecules. Nature 403, 635 (2000) [3] Gaussian 03, Revision C02, M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, J. A. Montgomery, Jr., T. Vreven, K. N. Kudin, J. C. Burant, J. M. Millam, S. S. Iyengar, J. Tomasi, V. Barone, B. Mennucci, M. Cossi, G. Scalmani, N. Rega, G. A. Petersson, H. Nakatsuji, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O. Kitao, H. Nakai, M. Klene, X. Li, J. E. Knox, H. P. Hratchian, J. B. Cross, V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R. E. Stratmann, O. Yazyev, A. J. Austin, R. Cammi, C. Pomelli, J. W. Ochterski, P. Y. Ayala, K. Morokuma, G. A. Voth, P. Salvador, J. J. Dannenberg, V. G. Zakrzewski, S. Dapprich, A. D. Daniels, M. C. Strain, O. Farkas, D. K. Malick, A. D. Rabuck, K. Raghavachari, J. B. Foresman, J. V. Ortiz, Q. Cui, A. G. Baboul, S. Clifford, J. Cioslowski, B. B. Stefanov, G. Liu, A. Liashenko, P. Piskorz, I. Komaromi, R. L. Martin, D. J. Fox, T. Keith, M. A. Al-Laham, C. Y. Peng, A. Nanayakkara, M. Challacombe, P. M. W. Gill, B. Johnson, W. Chen, M. W. Wong, C. Gonzalez, and J. A. Pople, Gaussian, Inc., Wallingford CT, 2004. [5] Gorelsly, S. 'AOMix program, revison 5.92'. http://www.sg-chem.net/ [6] Gorelsly, S. & a.b.p. Lever. J Organomet. Chem. 635, 187 (2001) [7] J. Stöhr and D. Outka, Determination of molecular orientations on surfaces form the angular dependence of near-edge x-rayabsorption fine structure spectra. Phys. Rev B 36, 7891 (1987)

190

Molecular Interactions in Organic Ultra Thin films Studied by VUV Spectroscopic Ellipsometry
O. D. Gordan, D. Lehmann, S. Hermann, D. R. T. Zahn
Institut für Physik, TU Chemnitz, 09107 Chemnitz, Germany

S. Silaghi, C. Cobet, N. Esser
ISAS - Institute for Analytical Sciences, D-12489 Berlin

W. Braun
BESSY GmbH, 12489 Berlin, Germany

The ellipsometric studies at BESSY makes it possible to investigate the electronic transitions of organic layers in the Vacuum Ultra Violet (VUV) range. Especially for small molecules, like DNA base molecules which have the HOMO-LUMO transitions at energies higher than 4 eV [1], the access to the VUV range is essential. While the synchrotron radiation extends the range beyond the capabilities of commercial ellipsometers, we proved that the lower wavelength of the light allows the dielectric function of ultra-thin organic films to be investigated on the sub-nanometer scale [2]. Therefore in this work we report the changes in the Vacuum Ultra Violet (VUV) spectra upon increasing thickness for two DNA base molecules: cytosine and guanine. The molecular structure of these two molecules is presented in figure 1.

cytosine guanine Figure 1. Molecular structures of cytosine (left) and guanine (right) projected in the (x, y) plane with z perpendicular on the molecular plane Vacuum Ultra Violet (VUV) Spectroscopic Ellipsometry (SE) measurements were performed in situ for optical characterization of ultra-thin films of cytosine and guanine. The layers were prepared by organic molecular beam deposition (OMBD) on hydrogen passivated silicon H-Si(111). The measurements were performed at BESSY at the 3m-NIM 1A beam line using a rotating-analyser ellipsometer operating in the 4-9.5 eV range with an energy step of 0.025 eV. MgF2 Rochon prisms were used as polarizers (for details see [3-5]). Figure 2 shows the imaginary part of the measured effective dielectric function <ε2> for H-Si(111) and for a 0.35 nm cytosine layer on H-Si(111). As can be seen in figure 2 clear changes in the ellipsometric data can be observed even for this coverage. While interpreting the ellipsometry data for very thin overlayers on a substrate (less than 10 nm) is rather difficult, the access to the VUV at BESSY makes it possible to achieve a higher separation in the experimental ∆ values even for very small

191

changes in the refractive index [6]. In this case the dielectric function of the overlayer can be extracted using a first order approximation [6].

Figure 2. The measured effective <ε2> of the 0.35 cytosine sample compared with the <ε2> of the H-Si(111) substrate. According to Aspnes [6], the measured effective dielectric function <ε> can be approximated by the formula presented above. As εs is the measured dielectric function of the substrate and d is the layer thickness, the quadratic equation can be easily solved to find the dielectric function of the layer εL. The solution of the above equation for the 0.35 cytosine layer is presented on the left side of figure 3.

Figure 3. Left - Comparison between the imaginary part of the dielectric functions of the bulk cytosine and 0.35 nm cytosine layer. Right - TD-DFT calculation of the excited states for a single cytosine molecule. For comparison the imaginary part of the dielectric function of the ultra-thin layer is plotted together with the imaginary part of the dielectric function of bulk cytosine. A detailed description of the experimental conditions and the ellipsometric model used to determine the bulk values can be found in ref. [7, 8]. While in the low energy range (below 7 eV) the dielectric function of the ultra-thin layer has similar shape like the bulk one, in the high energy range a clear splitting at the positions indicated by

192

arrows can be observed. The splitting can be attributed to the reduced intermolecular interactions between the cytosine molecules in the overlayer. This hypothesis is supported by theoretical calculations [8] of the excited states using time dependent density functional theory (TD-DFT). The computation was performed for an isolated molecule using Gaussian software [9]. The convolution of the excited states using Gaussian functions with 0.2 eV broadening is plotted on the right side of figure 3.

Figure 4. Left – the measured effective dielectric function of H-Si(111) and guanine layers of 0.4, 0.6 and 0.8 nm. Right - Comparison between the dielectric functions of bulk guanine and 0.4 nm guanine layer. A similar study was performed for guanine layers. The solution of Aspnes formula presented in the right part of figure 4 yields in this case a dielectric function of a 0.4 nm guanine layer very similar to the bulk one. This is probably related to an island growth mode of the guanine. The authors gratefully acknowledge the BMBF project 05 KS4KTB/3 and BMBF project 05 ES3XBA/5 and the Deutsche Forschungsgemeinschaft, Graduiertenkolleg 829 “Akkumulation von einzelnen Molekülen zu Nanostrukturen“. [1] S. D. Silaghi, Yu J. Suzuki, O. D. Gordan, C. Himcinschi, G. Salvan, M. Friedrich, T. U. Kampen, D. R. T. Zahn C. Cobet, N. Esser, W. Richter, W. Braun – BESSY Jahresbericht (2003) 209 [2] O.D. Gordan, C. Himcinschi, Yu J. Suzuki, G. Salvan, D. R. T. Zahn, C. Cobet, N. Esser, W. Richter, W. Braun – BESSY Jahresbericht (2004) 170 [3] R. L. Johnson, J. Barth, M. Cardona, D. Fuchs, and A. M. Bradshaw – Rev. Sci. Instrum. 60, 2209, (1989). [4] J. Barth, R. L. Johnson, and M. Cardona, - Handbook of Optical Constants of Solids II, edited by E. Palik, Academic Press, New York, (1991) [5] T. Wethkamp, K. Wilmers, N. Esser, W. Richter, O. Ambacher, H. Angerer, G. Jungk, R. L. Johnson, and M. Cardona - Thin Solid Films 313-314, 745, (1998) [6] D. Aspnes – Spectroscopic Ellipsometry of Solids, Chap 15, Optical Properties of Solids-New Developments, ed B.Seraphin, North Holland 1976
[7] Y. Suzuki, O.D. Gordan, S.D. Silaghi, D.R.T. Zahn, A. Schubert, W.R. Thiel, C. Cobet, N. Esser, W. Braun – Appl. Phys. Lett., 87 (2005) 214101

[8] S. Silaghi- PhD. Thesis, Chemnitz (2005) http://archiv.tu-chemnitz.de/pub/2005/0077/index.html [9] M. J. Frisch et al. - Gaussian 98, Version 5.2; Gaussian, Inc.: Pittsburgh, PA, 2001

193

Electronic Structure of Diluted Magnetic Semiconductor Zn1-xCoxO
M. Fonin,1 E. Biegger,1 Yu. S. Dedkov,2 L. Burova,3 A. R. Kaul,3 and U. Rüdiger1 Fachbereich Physik, Universität Konstanz, 78457Konstanz, Germany 2 Institut für Festkörperphysik, Technische Universität Dresden, 01062 Dresden, Germany 3 Department of Chemistry, Moscow State University, Moscow 119899, Russia Diluted magnetic semiconductors (DMSs) produced by doping of transition metal ions into nonmagnetic semiconductors which combine charge and spin degrees of freedom in a single material are promising candidates for the next generation spintronic devices [1]. Additionally, wide gap DMSs, among them doped ZnO and SnO2, combine their electrical conductivity and ferromagnetism with optical transparency, thereby opening up the possibility of new device concepts. Since the prediction by Dietl et al. [2] of the Curie temperature (TC) of Mn-doped ZnO exceeding room temperature (RT), diluted magnetic semiconductors on the basis of ZnO attract strong research interest. However, despite many efforts the experimental situation remains highly controversial. For nominally identical systems reports of high TC coexist with reports that exclude intrinsic ferromagnetism [3]. Moreover, the origin of the ferromagnetism in doped magnetic ZnO and related DMS materials is still under strong debate [3]. Until now only two studies were performed on the electronic properties of Codoped ZnO [4,5]. Wi et al. [4] performed x-ray absorption (XAS) and photoelectron spectroscopy (PES) on bulk Co-doped ZnO samples prepared by solid-state reaction method which did not exhibit ferromagnetic behavior. XPS and XAS measurements showed that the Co states are divalent (2+) under a tetrahedral symmetry giving clear evidence of properly substituted Co ions into ZnO lattice. On the basis of these findings it was concluded that ferromagnetic properties cannot be produced when Co ions are properly substituted for Zn sites and the origin of the ferromagnetism in Co-doped ZnO should have extrinsic nature [4]. Recently, Kobayashi et al. [5] performed a combined PES and XMCD spectroscopic study of ferromagnetic Zn1-xCoxO films prepared by a pulsed laser deposition technique. XMCD measurements on the Zn1-xCoxO films showed a multiplet structure characteristic for the Co2+ ion tetrahedrally coordinated by oxygen suggesting that ferromagnetism is caused by the substituted Co2+ ions at the Zn site in the ZnO lattice and is therefore intrinsic. In the present study a combined photoelectron spectroscopy (PES) and near edge x-ray absorption fine structure spectroscopy (NEXAFS) of ferromagnetic as well as paramagnetic Zn1-xCoxO films was performed in order to determine the electronic structure associated with the Co ions in the ZnO host lattice. PES as well as NEXAFS experiments were carried out at RT at the RGBL-PGM beamline at the BESSY II storage ring. The UHV system located at the Russian-German Laboratory (base pressure of 1×10-10 mbar) was equipped with a 127° CLAM4 analyzer. The total energy resolution in the XPS measurements was set to 150 meV. The position of the Fermi energy was determined form the valence-band spectrum of a polycrystalline Au foil in the electrical contact with the sample. All spectra were normalized to the incident photon flux. NEXAFS spectra were collected in the total electron yield mode and normalized to the maximum intensity. The energy resolution in the NEXAFS experiments was set to 100 meV. High quality 100-200 nm thick Zn1-xCoxO films with different concentrations of Co (x=0.05;0.1) were prepared by magnetron sputtering on Al2O3(0001) substrates. The substrate temperature during the growth was maintained at about 700 K. Two sets of samples were prepared: samples in the first set were deposited in 10-3 mbar Ar (samples S1) with short postannealing at 700K, samples in the second set were deposited in 10-3 mbar Ar/O2 (1:1) gas mixture with subsequential annealing in O2 (1bar) at 1000 K (samples S2). After introducing
1

194

Fig. 1. In (a) experimental Co L2,3 NEXAFS spectra of two Zn0.95Co0.05O films (S1 and S2) are presented. Inset shows the corresponding Co 2p core level spectra. (b) Valence band PES spectra of Zn0.95Co0.05O film (S2) obtained at photon energies marked by numbers in the corresponding NEXAFS spectrum .

the samples into the UHV chamber at BESSY II their surface were cleaned by Ar+ sputtering under grazing angle (E=800 V, p=5×10-6 mbar). X-ray diffraction confirmed that all films crystallized in the wurtzite structure and no impurity or secondary phases were observed. Magnetic properties of the Zn1-xCoxO films were characterized by means of a Quantum Design superconducting quantum interference device (SQUID) from 0 up to 5 T in the temperature range of 5-300 K. In S1 samples ferromagnetism with Tc above the room temperature (RT) was found with magnetic moments of about 0.05-0.1 µB/Co at RT increasing to 0.2 µB/Co at 5 K. In S2 samples were paramagnetic at RT and only at temperatures below 50 K a weak ferromagnetic signal was observed. Fig. 1 (a) compares the Co L2,3 NEXAFS spectra of two Zn0.95Co0.05O samples of the S1 and S2 sets. The peak positions as well as the line shape [features 2,3,4,5,6 in Fig. 1 (a)] of both Co 2p NEXAFS spectra are almost identical and are in good agreement with those measured before for Co-doped ZnO [4,5]. The obtained NEXAFS spectra look similar to that of CoO and are quite different from those of Co metal as well as of Co3+ indicating that Co ions in both samples are present in divalent Co2+ state. Moreover, calculations on the basis of atomic multiplet theory performed for Co2+ ions tetrahedrally coordinated by oxygen was shown to yield the best fit for the presented experemental spectra [5]. Hence, Co ions in both Zn0.95Co0.05O samples are divalent and tetrahedrally coordinated by oxygen ions. Careful analysis of the S1 and S2 sample spectra shows that features 5 and 6 in the S2 spectrum have slightly larger intensities than those of S1 which may be attributed to better Co incorporation in the sample. The inset in Fig.1 (a) shows the core-level Co 2p XPS spectra of both (S1 and S2) Zn0.95Co0.05O samples taken at hν=1000 eV. The spectra are similar to that of CoO confirming again the presence of Co2+ ions in Co-doped ZnO samples. Thus NEXAFS and XPS show that in both samples Co atoms are properly incorporated in the ZnO lattice by substitution of Zn sites.

195

Fig. 2. On-resonance (hν=774 eV) and off-resonance (hν=777 eV) valence band photoemission spectra of Zn0.95Co0.05O: (a) sample S1 and (c) sample S2. Corresponding difference curves showing the Co 3d PDOS are presented for both samples in (b) and (d), respectively.

Fig. 1 (b) and (c) shows representative valence band photoemission spectra of the S1 and S2 Zn0.95Co0.05O samples as a function of photon energy including the Co 2p − 3d coreexcitation region. The valence band spectra of Zn0.95Co0.05O are similar to that of ZnO showing a sharp peak at about 11 eV of the binding energy (BE) corresponding to Zn 3d states as well as a broad feature between 3 and 8 eV of BE corresponding to O 2p band. Almost no photoemission intensity was observed in the EF region confirming the insulating nature of the Zn0.95Co0.05O films. Thus the charge carrier mediated ferromagnetism based on the RKKY interaction can be ruled out. By substracting the off-resonance spectrum from the on-resonance one Co 3d partial density of states (PDOS) can be obtained. Fig. 2 represents 2p3/2 − 3d on-resonance (hν=774 eV) and off-resonance (hν=777 eV) valence band photoemission spectra of Zn0.95Co0.05O: (a) sample S1 and (c) sample S2. The corresponding difference spectra for sample S1 as well as for sample S2 are shown in (b) and (d), respectively. Here, the intensities of the off-resonance spectra at 7.5 eV were normalized to those of the on-resonance spectra. Both difference curves are almost identical with a sharp peak at about 3.5 eV of BE which is in good agreement with previous studies [4,5]. This results show that in ferromagnetic as well as in paramagnetic Co-doped ZnO samples the Co 3d states are located near the top of the valence band. Concluding, in both ferromagnetic (S1) and paramagnetic (S2) samples Co ions are divalent (Co2+) and are tetrahedrlally coordinated by oxygen. Ferromagnetism in the S1 samples is not due to the charge carriers as no photoemission intensity was observed at EF. The origin of different magnetic behaviors of the samples (S1 and S2) can be due to ZnO lattice distortion or defects as well as variations of local Co concentration leading to Co-rich phases. This work was supported by SFB 513. The authors (M. F. and E. B.) would like to thank BESSY II for financial support. 1. 2. 3. 4. 5. H. Ohno, Science 281, 951 (1998). T. Dietl, H. Ohno, F. Matsukura, J. Cibert, and D. Ferrand, Science 287, 1019 (2000). J. M. D. Coey, M. Venkatesan, and C. B. Fitzgerald, Nature Mater. 4, 173 (2005). S. C. Wi et al., J. Appl. Phys. 84, 4233 (2004). M. Kobayashi et al., Phys. Rev. B 72, 201201(R) (2005).

196

Electronic structure of organic thin films on metal surface: NTCDA/Ag(111) Azzedine Bendounan, Frank Forster, Achim Schöll, David Batchelor, Eberhard Umbach, and Friedrich Reinert Experimentelle Physik II, Universität Würzburg Am Hubland, D-97074 Würzburg, Germany Today many electronic devices contain organic components such as organic light emitting diodes (OLEDs) and organic field electronic transistor (OFETs). Particularly, large conjugated planar aromatic molecules are well suited for such applications, because they present intermolecular π-bonds in which direction the electron mobility is relatively high. On the other hand, the study of the electronic structure of organic/metal interfaces represents a key point to control and improve the properties of these components. We report here on a new approach to study the bonding in thin organic films deposited on metal surface by a combination of two experimental techniques: resonant photoelectron spectroscopy (R-PES) and near-edge X-ray adsorption fine structure (NEXAFS) spectroscopy using the high-brilliance third-generation synchrotron beamline UES2-PGM at BESSY. We present a method which enables us to identify the origin of different molecular orbitals observed in the valence band measured by PES with respect to the carbon bondings in the molecule. NEXAFS has particularly been used to determine the molecular orientation, i.e., substantial spectroscopic changes are observed depending whether the molecules are oriented parallel or upright to the substrate surface [1]. Photoemission spectroscopy has also been applied to investigate the electronic properties, e.g., the electron-vibron coupling [2,3]. As an ideal model system to explore the possibilities of this method, we have chosen one monolayer of 1,4,5,8-naphthalene tetra-carboxylicacid dianhydride (NTCDA) on Ag(111), since it has already been studied well by NEXAFS and PES. The NTCDA films were evaporated in situ by organic molecular beam deposition from a Knudsen cell on clean Ag(111) substrate. The monolayer can easily be obtained by deposition of a thick film followed by a subsequent desorption of the multilayers upon annealing at about 385 K. At this temperature only the first monolayer remains adsorbed on the metal, due to the strong bonding at the interface. This bonding forces the molecules also to adopt a parallel and flat orientation along the Ag(111) substrate [1]. In Fig.1, we present resonant photoemission data on 1 ML NTCDA on Ag(111) measured with p-polarized synchrotron light. Panel (a) displays the PES intensity map, where the photon energy was tuned through the Carbon K-edge. This intensity map is dominated by the features between E B=3.8 eV and 8 eV, which correspond to the d-bands of Ag. In the region with lower binding energy between the Ag d-bands and the Fermi level, two spectroscopic structures are observed. The intensity of these two structures strongly depends on the photon energy as it is clearly observed in the energy distribution curves (EDCs) (panel b). The structure indicated by IV at E B=2.6 eV represents the highest occupied molecular orbital (HOMO). The second structure indicated by III appears at higher energy EB=3.5 eV and is assigned as HOMO-1. A variation of the photon energy induces characterising changes in the photoelectron intensity. As illustrated in panel (c), scans at constant binding energy (constant initial state (CIS) spectra), and the NEXAFS spectrum show the presence of four pronounced resonance features. The resonances marked by (C) and (D) in the CIS spectra were assigned to the excitations of C-1s electrons of the naphthalene core of the molecule to

197

different unoccupied molecular orbitals [3], whereas resonances (A) and (B) are associated mainly to the excitation of the carbon in the anhydride groups [1].

(a)

(d)

(c)

Fig.1: Resonant photoemission data on 1 ML NTCDA on Ag(111) measured with ppolarized light. Panel (a) gives the photoelectron intensity versus photon energy and binding energy. Panel (b) displays energy distribution curves (EDC) obtained at the photon energy corresponding to on-resonance and off-resonance of the NTCDA molecule. Panel (c) shows constant initial state (CIS) spectra obtained at different binding energy together with a NEXAFS spectrum.

198

A comparison of the EDCs with the CIS data, we can assign the molecular orbital peaks observed in the photoemission spectrum to which bonding within the molecule are origin. At a photon energy corresponding to the C1s adsorption resonance of the naphthalene core, the HOMO-1 structure in the EDC spectrum shows high intensity compared to the HOMO. The opposite behaviour is obtained at a photon energy corresponding to the C1s adsorption of the anhydride group. This indicates that the HOMO-1 is due to the carbon bonding of the naphthalene core whereas the HOMO peak seems to be from the carbon bonding of the anhydride group of the molecule. This interpretation is consistent also with the evolution of the spectroscopic structures labelled I and II in Fig. 1, which appear, as the HOMO peak, with high intensity at photon energy corresponding to C1s resonance of the anhydride groups. Additionally, we observe in Fig.1, another structure with binding energy varying linearly with the photon energy. This structure, which has a constant kinetic energy, is described as a constant final state (indicated by CFS in the intensity map of Fig.1). Acknowledgments: We would like to thank BESSY staff for the helpful support. References: [1]: D. Gador, C. Buchberger, R. Fink, and E. Umbach, Europhys. Lett. 41, (1998), 231. [2]: A. Schöll, Y. Zou, D. Hübner, Th. Schmidt, R. Fink, and E. Umbach, BESSY – Highlights (2001), 14-15. [3]: A. Schöll, Y. Zou, L. Kilian, D. Hübner, D. Gador, C. Jung, S. G. Urquhart, Th. Schmidt, R. Fink, and E. Umbach, Phys. Rev. Lett. 93, (2004), 146406.

199

Electronic structure of filled and intercalated single wall carbon nanotubes
T. Pichler1, H. Shiozawa1, H. Rauf1, M. Knupfer1, M. Kalbac1, S. Yang1 L. Dunsch1, B. Büchner1, D. Batchelor2, and H. Kataura3
1 2

IFW-Dresden, P.O. Box 270116, D-01171 Dresden, Germany Universität Würzburg, BESSY II, Albert-Einstein-Straße 15 D-12489 Berlin, Germany 3 Nanotechnology Research Institute, AIST, Tsukuba 305-8562, Japan

In the recent years the functionalization of single-wall carbon nanotubes (SWCNTs) via intercalation of atoms and filling of the molecules was in the focus of research on molecular nanostructures since it gives rise to fascinating electronic properties as a consequence of the charge transfer and orbital hybridisation within the quasi-one-dimensional nanospaces. For instance, alkali metal intercalation is a rout for modifying the electronic property of the SWCNTs via electron doping [1,2]. Fullerene-filled SWCNTs, known as peapods, is also being of great interest because of their unique structural and electronic properties [3,4], as well as potential application for the nanoscale devices. Specifically, the endohedral fullerene peapods are though to have a great potential to functionalize single-wall carbon nanotubes in a controlled manner because of a variety of the endohedral species in contrast to empty fullerenes [5]. Especially, high-energy spectroscopy such as xray absorption and photoemission has been shown to be a very useful tool to investigate the elementand site-selective electronic structures of the endohedral fullerenes and their peapods since it offers a direct probe of the valency of the encaged metal ions in the endohedrals. Recent photoemission, x-ray absorption and electron energy-loss spectroscopic studies on endohedrals have given great contributions to the understanding of a nature of the endohedrals [6-11]. In this project we have performed high-energy spectroscopy using synchrotron radiation to explore the advanced electronic structures of endohedral fullerenes, as well as functionalized single-wall carbon nanotubes. In-situ functionalization of SWCNTs was carried out via potassium intercalation and filling with the endohedral fullerenes Dy3N@C80(I), a recently synthesized novel trimetal nitride fullerene [12]. The influence of the one-dimensional environment on the electronic structure of encaged endohedrals as well as the impact of the fullerene accommodation on the electronic properties of the carbon nanotubes were investigated in comparison to the electronic structure of the pristine endofullerenes using high-resolution x-ray absorption and photoemission spectroscopy as probe.

C1s

π∗

Intensity (arb. units)

C1s

σ∗
Dy3 N@C80

Intensity (arb. units)

Dy3N@C80

Peapods - SWCNT

Peapods SWCNT

SWCNT

Peapods

EELS (SWCNT) 285 290 Photon Energy (eV) 295

20

15

10

5

0

Binding energy (aV)

Fig. 1. C1s absorption spectrum of SWCNTs, Dy3N@C80 and the endohedral peapod together with electron energy loss spectrum (EELS) of SWCNTs.

Fig. 2. Photoemission spectra of Dy3N@C80, SWCNTs and their peapods measured at 400 eV photon energy. The difference spectrum between the peapods and Dy3N@C80 is also plotted.

200

The photoemission experiment was carried out at CRG BUF beamline UE 52 PGM, BESSY II, using a hemispherical photoemission electron energy analyser, SCIENTA SES 200. The x-ray absorption spectrum was obtained by measuring the drain current of the sample. The experimental resolution and Fermi energy were determined from the Fermi edge of a clean Au film. All spectra were recorded with an overall energy resolution better than 50 meV. The base pressure in the experimental setup was kept below 3 x 10–10 mbar. From high-resolution (better than 30 meV) absorption spectroscopy on pristine SWCNTs at C1s excitation edge, we have successfully observed a fine structure in the unoccupied SWCNT desnsity of states. This is complementary to the results of our previous lab based photoemission and electron energy-loss spectroscopic results conducted at the IFW-Dresden [1-3]. In addition, high-energy valence-band photoemission spectroscopy was used to investigate the metal valency of trimetal nitride fullerene Dy3N@C80. The valence-band photoemission spectrum measured at 400 eV photon energy clearly shows the trivalent Dy 4f multiples as main structures and weak divalent Dy multiplets in lower energy side as shown in Fig. 2. The spectral shapes are in good agreement with those observed in x-ray photoemission using Al Kα radiation [11], and points to the intermediate valency of Dy in the fullerene cage. A comparison of the Dy3N@C80 spectrum with atomic calculations estimates the effective valency of Dy ions inside fullerene cage to be 2.9. This value is much bigger than 2.4 of Sc in Sc3N@C80, but slightly smaller than 2.9 of Tm in Tm3N@C80. Considering the Lanthanide contraction from Dy to Tm and the Sc 3d orbitals much delocalised compared to the rare-earth 4f orbitals, we found that the effective metal valency in the endohedrals depends on the size of the metal ions as well as the orbital overlap between the metals and the fullerene cage. As a next step, Dy3N@C80 was encapsulated in the SWCNT. The filling factor and possible changes of the cluster molecular electronic properties were analysed using x-ray absorption spectroscopy as well as high-energy and resonance photoemission spectroscopy [5]. The C1s absorption spectrum of the endohedral fullerene peapods is similar to that of the SWCNTs. The vHs peaks are observed at the same energies as those of the SWCNTs. As observed in Fig. 2, the highenergy valence-band photoemission spectrum of the endohedral fullerene peapods exhibits prominent structures corresponding to the trivalent Dy 4f multiples in contrast to the broad spectrum of the pristine SWCNT. The Dy 4f spectrum derived by subtracting the SWCNT spectrum from the Dy3N@C80 peapods spectrum also exhibits trivalent Dy 4f character. A more detailed analysis of the Dy valency in the peapods is given by the resonance photoemission study across Dy 4d-4f edge. Figure 3 shows the Dy 4f 150 155 160 spectra extracted from the resonance Photon energy (eV) photoemission spectra of the endohedral fullerene peapods and SWCNT. The photon energies of on-resonance (161 eV) and offresonance (149 eV) were determined from the xray absorption spectrum across Dy 4d-4f edge as x1 4f9 4f 8 plotted in the inset of Fig. 3. The x-ray absorption spectrum shows the trivalent Dy x 50 4f10 4f 9 multiplets similar to those of Dy metal. The Dy 4f spectrum is very well reproduced with the 15 10 5 0 Binding energy (eV) photoemission multiplets of the trivalent Dy 4f states. The divalent Dy multiples have almost no Fig. 3. Dy 4f spectrum (above) and calculated contribution to the spectrum. This fact spectrum (below) together with the photoemission demonstrates the Dy valency to be close to 3.0. multiplets (vertical bars). The inset shows x-ray Compared to the Dy valency of 2.9 in the absorption spectrum of the peapods. The arrows indicate the on-resonance and off resonance photon pristine Dy3N@C80, the fact might indicate the energies. charge transfer from the SWCNT to the
off on

201

Intensity (arb. units)

endohedral fullerenes within the endohedral fullerene peapod. To summarize, we demonstrated that a combination of high-resolution photoemission and x-ray absorption across the resonance edges of the composite elements is feasible to analyse the partial electronic structures of the Dy3N cluster and carbon cages, as well as to estimate the filling factor of the endohedral fullerene peapods. This methodology for determination of the filling ratio and investigation of the partial electronic structures can be applied generally to characterize such molecular nanostructures composed of different molecules and atoms, and to testify the principle of low-dimensional physical properties within such carbon nanostructures. References
[1] [2] [3] [4] T. Pichler, X. Liu, M. Knupfer, and J. Fink, New Journal of Physics 5, 156 (2003). H. Rauf, T. Pichler, M. Knupfer, J. Fink, and H. Kataura, Phys. Rev. Lett. 93, 096805 (2004). H. Rauf, H. Shiozawa, T. Pichler, M. Knupfer, B. Büchner, and H. Kataura, Phys. Rev. B 72, 245411 (2005). H. Shiozawa, H. Ishii, H. Kihara, N. Sasaki, S. Nakamura, T. Yoshida, Y. Takayama, T. Miyahara, S. Suzuki, Y. Achiba T. Kodama, M. Higashiguchi, X. Y. Chi, M. Nakatake, K. Shimada, H. Namatame, M. Taniguchi, and H. Kataura, Phys. Rev. B 73, 075406 (2006). [5] H. Shiozawa, H. Rauf, T. Pichler, M. Knupfer, M. Kalbac, S. Yang, L. Dunsch, B. Büchner, and D. Batchelor, submitted to Phys. Rev. B. [6] T. Pichler, M. S. Golden, M. Knupfer, J. Fink, U. Kirbach, P. Kuran, and L. Dunsch, Phys. Rev. Lett. 79, 3026 (1997). [7] L. Alvarez, T. Pichler, P. Georgi, T. Schwieger, H. Peisert, L. Dunsch, Z. Hu, M. Knupfer, J. Fink, P. Bressler, M. Mast, and M. S. Golden, Phys. Rev. B 66, 035107 (2002). [8] T. Pichler, Z. Hu, C. Grazioli, S. Legner, M. Knupfer, M. S. Golden, J. Fink, F. M. F. de Groot, M. R. C. Hunt, P. Rudolf, R. Follath, Ch. Jung, L. Kjeldgaard, P. Br\"uhwiler, M. Inakuma, and H. Shinohara, Phys. Rev. B 62, 13 196 (2000). [9] X. Liu, M. Krause, J. Wong, T. Pichler, L. Dunsch, and M. Knupfer, Phys. Rev. B 72, 085407 (2005). [10] M. Krause, X. Liu, J. Wong, T. Pichler, M. Knupfer, and L. Dunsch, J. Phys. Chem. A 109, 7088 (2005). [11] H. Shiozawa, H. Rauf, T. Pichler, D. Grimm, X. Liu, M. Knupfer, M. Kalbac, S. Yang, L. Dunsch, B. Büchner, and D. Batchelor, Phys. Rev. B 72, 195409 (2005). [12] S. Yang and L. Dunsch, J. Phys. Chem. B 109, 12320 (2005).

202

Electronic structure of ultrathin MnO films in comparison to MnO bulk samples studied by soft x-ray emission spectroscopy and x-ray absorption spectroscopy
M. Nagel, L. Zhang, H. Peisert, T. Chassé Institut für Physikalische und Theoretische Chemie, Universität Tübingen, Auf der Morgenstelle 8, 72076 Tübingen, Germany Recently it was discussed that ultrathin oxide films on metals may have unique physicalchemical properties due to the mutual interactions between the oxide and the metal such as image potential screening and hybridizational effects [1]. In ultrathin films, the Coulomb interaction U and the anion charge transfer energy ∆ may change significantly compared to the corresponding bulk value. Using the MgO/Ag(100) interface as a model system, it was TEY-XAS: Mn L2,3 demonstrated that Coulomb and charge-transfer energies in oxide layers deposited on a highly polarizable medium like a metal, are reduced from their bulk values, by as much as 1.8 eV and 2.5 eV, respectively [1]. The properties of nonMnO on Ag(001) correlated oxides are mainly determined by electron delocalisation and hybridizational effects. In this report we study the electronic properties of the strongly correlated MnO on Ag(100).

a)

XAS and XES measurements were carried out at the U41 PGM beamline using ROSA endstation equipped with a SCIENTA XES 300 spectrometer and EA10 hemispherical analyser. The absorption was monitored by measuring the total electron yield. The raw data are corrected by the energydependent photon flux. Additional XPS reference measurements were performed with a Phoibos 100-MCD 5 analyzer (SPECS GmbH, Germany) and a Mg Kα x-ray source. The MnO thin film was prepared by sputtering and annealing of a 45 Å thick Mn3O4 film. The sputtering by argon ions causes a reduction of Mn3O4 whereas the subsequent annealing to 400°C produces stoichiometric MnO thin films. The estimated film thickness deduced from photoemission peak intensities was about 6 Å. Additional experiments in the lab applying the same preparation procedure have shown, that the annealing leads to an island formation combined with a relaxed MnO lattice (see also [2]). The island formation was proved by LEED and by XPS signal intensities.

Total electron yield [arb.units]

MnO(001) single crystal

640

650

660

670

680

Photon energy [eV]

Intensity [a.u.]

b)

XPS: Mn2p

MnO film @ Ag(001) MnO(001)

665

660

655 650 645 640 Binding energy [eV]

635

Fig. 1 Characterization of the prepared MnO monolayer: The oxidation state was checked a) by the shape of the XAS spectra and b) by the satellite structures in Mn 2p XPS spectra.

203

Fluorescence intensity [a. u.]

Mn L2,3 (3d->2p) hν=716 eV

638.2eV

648.1eV

MnO(001)

MnO film @ Ag(001)

Fig. 2 Comparison of Mn L2,3 emission spectra. black: 6 Å MnO on Ag(001) (film) red: MnO(001) single crystal (bulk). The different distance of both components is due to interfacial phenomena discussed in the text.

645.9eV

620

630 640 650 Photon energy [eV]

660

In Figure 1 we show XAS and XPS spectra of the prepared 6 Å MnO film. The Mn 2p absorption spectra are very similar for the single crystal and for the thin film all main features are present in both spectra. XAS is very useful for the identification of Mn species, since spectra for other oxidation states are clearly different [3]. An additional suitable tool for the analysis of the Mn oxidation state is the satellite structure and the spin-orbit coupling of the Mn 2p XPS spectrum. Since satellites are absent for Mn2O3 and for Mn3O4, the observed features at 645 eV and 650 eV (Fig. 1b) point clearly to MnO in each case. The preparation and characterisation of ordered MnO films is described in detail in Ref. [4]. In Fig. 2 we compare Mn L2,3 emission spectra for the thin film and for the bulk (MnO(001) single crystal). The Mn L2,3 emission consists of the to main peaks L3 and L2 separated by the spin-orbit coupling. According to the dipole selection rules (∆l = ±1 and ∆j = 0 bzw. ±1), the L3 peak incorporates two main contributions, namely the transitions 3d5/2 2p3/2 (higher photon energy) and 3d3/2 2p3/2 (lower photon energy), whereas the L2 peak is only the 3d3/2 2p1/2 transition. In addition, contributions from 4s 2p can be expected, however the intensity of these transitions is very weak [5]. Beside intensity variations possibly caused by a different ability for self-absorption and fluorescence with increasing film thickness, we observe clearly a change of the energetic position in the spectra, in particular the distance between the main components is decreased significantly by 2.2 eV for the 6 Å MnO film on Ag(001) compared to bulk MnO. In general, different reasons for this observation are possible which will be discussed in turn below: The oxidation state of an element changes the size of the spin orbit coupling which affects directly the energy of x-ray emission lines. As discussed above, the composition was checked by XPS and XAS, therefore this explanation can be ruled out. An effect of the geometric structure can be expected which affects the position of energetic levels. For instance, for very small particle sizes a splitting of electronic levels may occur. Furthermore, the small shoulder on the low photon energy side of the thin film XES spectrum may suggest the occurrence of two different species. The octahedral ligand symmetry for manganese atoms in MnO is reduced at the surface. Consequently a contribution of such species to the spectrum should be enhanced in ultrathin films. Alternatively an interface strain leading to a tetragonal distortion of the first layers could
204

-

-

-

explain additional contributions. As drastic changes are not observed in Mn2p XPS or in XAS spectra such explanations seem unlikely. The spin-orbit splitting of Mn 2p may be different for the thin film and for the bulk. This can be also ruled out, since no differences were observed in the corresponding XPS spectra. The energetic position of the Mn 3d valence states may be different for the thin film and for the bulk. Compared to Mn 2p the measurement of the Mn 3d levels in XPS is more complicated. In the thin film system the corresponding features overlap with the intense substrate Ag 4d features as well as with O 2p contributions from the film itself. Measurements of valence band difference spectra were carried out in the lab but the results were not unambiguous. But changes of this feature may explain the observed results.

In particular at interfaces, the Mn 3d levels can be affected by different mechanisms. First of all, the Mn 3d shell of Mn2+ is only half occupied, together with the influence of the oxygen ligands this results in a rich multiplett structure of the spectra. In interface systems the Coulomb interaction U and the anion charge transfer energy ∆ may change significantly compared to the corresponding bulk value (see above). Close to the metal surface, the creation of a positive (negative) charge is accompanied effectively by the simultaneous creation of a negative (positive) image charge in the metal, so that the ionisation energy (IE) is reduced and the electron affinity (EA) increased each by the image charge energy (Eimage), and consequently U is reduced by 2*Eimage [1]. A similar behaviour is expected for ∆, i.e. the charge transfer energy from O2p to a metal cation (Mn 3d). The evaluation of valence band spectra of bulk MnO shows a similar contribution of d4 and d5L final states [6]. Consequently a change in the charge transfer energy ∆ in a thin film on a metal substrate as discussed above should alter the photoemission spectrum of the valence band. The Mn L2,3 XES spectrum represents the same final state as the manganese contribution to the valence band XPS. Therefore a change in ∆ should be clearly seen in the XES spectrum, too. In addition, hybridizational effects between orbitals of the substrate and the oxide affect the electronic structure of transition metal oxides at metal surfaces [1]. A hybridisation of O 2p with the Ag4d / Ag 5sp band may affect indirectly the Mn 3d levels. In summary, we have shown that the properties of correlated oxides can be altered at interfaces to a highly polarizable medium. For valuable discussions and technical assistance we thank R. Szargan, D. Wett, Ch. Jung and W. Neu. The financial support from BESSY is gratefully acknowledged.
[1] a) S. Altieri, L. H. Tjeng, G. A. Sawatzky, Thin Solid Films 400 (2001) 9. b) S. Altieri L. H. Tjeng, F. C. Voogt, T. Hibma, G. A. Sawatzky,, Phys. Rev. B 59 (1999) R2517. [2] F. Müller, R. de Masi, D. Reinicke, P. Steiner, S. Hüfner and K. Stöwe, Surf. Sci. 520 (2002) 158. [3] B. Gilbert, B. H. Frazer, A. Belz, P. G. Conrad, K. H. Nealson, D. Haskel, J. C. Lang, G. Srajer, and G. De Stasio, J. Phys. Chem. A 107 (2003) 2839-2847. [4] M. Nagel, L. Zhang, H. Peisert, T. Chassé, microchimica acta (submitted). [5] J.A. Bearden, Rev. Mod. Phys. 39 (1967) 78-124. [6] F. Parmigiani, L. Sangaletti, J. Electron. Spectrosc. Relat. Phenom. 98–99 (1999) 287–302.

205

1

Self-Organized Nanostructure on Si(111): Accumulation Layer on a Step Bunched Surface K. Skorupska, M. Lublow, M. Kanis, H. Jungblut, H. J. Lewerenz Division of Solar Energy, Interface Engineering Group, Hahn- Meitner.Institut, Glienicker Str. 100, 14109 Berlin, Germany Wet chemical treatments with alkaline subsequently, samples were rinsed in solutions are of central importance for deionized water, dried in N2 and transferred into the UHV system for outminiaturization on the nanometre scale. gassing and SRPES measurements. Despite widespread use of alkaline etching, Float Zone n-Si(111) with specific several aspects of silicon dissolution are still not fully understood. Detailed resistivity 5 Ωcm and 0o nominal miscut knowledge of (electro)chemical dissolution served as working electrode, Ag+ / AgCl steps in combination with a refined control was used as references electrode, a Pt of the (electro)chemical parameters can cylinder was used as counter electrode. provide new strategies for directed Connections were made to an EG&G 326 nanostructuring. potentiostat. We report here on nanostructure formation All electrolytes are prepared from and a free standing accumulation layer on ultrapure chemicals, dissolved in deionized n-Si(111) after cathodic polarization in 2M water (18 Ωcm), and purged with nitrogen NaOH thus forming a two-dimensional (5.0). The pH of solutions is controlled electron gas (2DEG). 2DEGs are of with a calibrated pH-meter and ultrapure considerable interest for basic research H2SO4. (Quantum Hall effect) and applications, Electrochemical H-termination comprises e.g., high electron mobility transistors. oxidation in potassium hydrogen phthalate Experiments were done at SoLiAs, the solution under illumination and combined chemistry/ultra high vacuum potentiostatic etching in dilute ammonium (UHV) analysis system, were synchrotron fluoride solution of pH=4.0 and pH=4.9 radiation photoelectron spectroscopy respectively. (SRPES) at the U49/2 beamline at Bessy II was performed. -1.1 0 Electrochemical experiments were done in current vs time voltage vs time -1.2 an N2 atmosphere-purged three-electrode -20 system, directly attached to the UHV -40 -1.3 apparatus [1] thus protecting samples from -60 -1.4 ambient air contamination. -80 Wet chemically H-terminated specimen -1.5 -100 were inserted via a load lock into the UHV apparatus and then (using the internal -120 -1.6 0 25 50 75 100 125 time / s sample transport path system) placed in the glass sphere, where electrochemistry was performed, from the machine side. Fig.1 Dark current and applied potential Electrochemical conditioning was done vs. time for n-Si(111) in 2M NaOH. Scan using an electrolyte drop, applied by the velocity 5mVs-1, open-circuit potential -1.2 capillary with reference electrode. V (SCE). Electrical contact is obtained by lowering Figure 1 shows electrochemical the Pt cylinder working electrode towards conditioning in 2M NaOH (experiment the electrolyte drop. For fast interruption was performed in the dark). The potential of the electrochemical processing, this has been altered from OCP until a current drop was blown off by an N2 jet;
current / µAcm potential / V (SCE)
-2

206

2 of - 100 µA is obtained; at this stage the potential is held for 40s. In this specifically designed experiment, the competition between chemical etching and electrochemical reactions, both of them appearing at silicon kink side atoms at Si(111) (1x1):H can be studied. Figure 2 shows the reaction schemes for chemical and electrochemical processes [2]. For the chemical reaction route, two possible sites for hydrolysis can be identified: the Si-Si back bond or the Si-H dangling bond. All reaction paths, however, lead to dissolution of the kink side atom that leaves the Si surface with underlying H-termination. The electrochemical reaction consumes two electrons from the valence band leading, through formation of a radical and hydrogen evolution, to the initial surface state. No silicon atom is dissolved in this process.

Fig.2 Representation of chemical and cathodic electrochemical reaction paths in alkaline solution in the dark; after refs. [3,4] SRPES experiments were made for two surface sensitivities for the Si 2p core level (Fig.3): excitation energies of hν=150 eV and hν=585 eV, corresponding to λesc = 4 and 15 Å escape depth, respectively, were used. For deconvolution, a Shirley-type background has been subtracted and Lorenz- Gaussian lines have been applied. Five components are found: a dominant silicon bulk signal at 99.7 eV and signals shifted by -0.3, 0.2, 0.5 and 0.8 eV with regard to the bulk line. Their intensity decreases with increasing escape depth, but the ratio between them remains unaltered for both photon energies.

Fig.3 Si 2p lines obtained by SRPES on nSi(111) for excitation energies of 150 and 585 eV. The contribution shifted by 0.2 eV is attributed to silicon bonded to one hydrogen atom (≡Si-H) on (1x1):Hterminated Si(111) [5]. Smaller peaks (0.5 and 0.8 eV) are attributed to =Si-H2 and =Si-H-OH [6,7] respectively. The presence of silicon atoms bonded to H and to an OH group suggests that in the reaction sequence, a nucleophylic substitution reaction (SN2) takes place where the formation of an activated intermediate is important for chemical etching [2]. Further inspection of the SRPES data reveals a shift of the Si 2p core level of about 0.2 eV towards lower binding energy for higher excitation energy (i.e. lager escape depth). The high surface sensitivity x-ray photoelectron valence band spectrum shows that the energetic position of the valence band w.r.t. the Fermi level is extrapolated to be located at 1.06 eV (data not shown here)[1]. From the band diagram in figure 4 we can learn that an accumulation condition at the surface is responsible for the observed 0.2eV shift in the SRFE spectra. The effective surface electron concentration, ns=3x1018 cm-3, is calculated from the energetic distance of

207

3 0.06 eV (~2kT) between Fermi level and conduction band The difference of 0.2 eV in electrostatic potential in the accumulation layer can be calculated according to Thomas-Fermi theory for the two electron escape depths used in our experiment with the calculated surface electron concentration ns=1018cm3 (data not shown here) [1] showing excellent agreement with the measured SRPES data. be seen, the Si surface developed step bunching. In the figure, accumulated steps are visible that are about 0.3 µm wide and from 9 to 12 BL high. The phenomenon of step bunching was recently also observed also by others [8] but no indication of a 2DEG was reported. a

b

Fig.4 Energy band diagram of n-Si(111) for (a) flat band situation, (b) the electronic surface condition for a stepbunched structure obtained after cathodic polarization in alkaline electrolyte. For surface nanotopography analysis exsitu experiments using contact mode atomic force microscope using Si3N4 tips were performed. The two images in figure 5 represent a Si(111) H-terminated surface, (top) as starting condition, where atomically flat terraces are visible; their height is a multiple of 3.14 Å which is the height of one BL (bilayer). The bottom part of figure 5 shows the surface after electrochemical experimentation. As can

Fig.5 AFM-CM experiment for (a) Hterminated n-Si(111); starting condition and (b) the same surface subjected to electrochemical conditioning as described above. SRPES data deconvolution showed that almost 30% of surface species are non (111) species. Figure 5 and the dissolution model show that the sample topography is built by the surface of {111} terraces but also by the side walls of bunched steps which are likely made of (100) oriented facets where the non-(111) species are localized.

References: 1. K. Skorupska, M. Lublow, M. Kanis, H. Jungblut, H. J. Lewerenz, Appl. Phys. Lett. 87 (2005) 1. 2. K. Skorupska, M. Lublow, M. Kanis, H. Jungblut, H. J. Lewerenz, Electrochem. Comm. 7 (2005) 1077. 3. P. Allogue, V. Costa-Kieling, H. Gerischer, J. Electrochem. Soc. 140 (1993) 1018. 4. Th. Baum, D.J. Schiffrin, J. Electroanal. Chem. 436 (1997) 239. 5. H. J. Lewerenz, M. Aggour, C. Murrell, J. Jakubowicz, M. Kanis, S. A. Campbell, P.A. Cox, P. Hoffmann, H. Jungblut, D. Schmeißer, J. Electroanal. Chem. 540 (2003) 3. 6. S. R. Sundaram et al., J. App. Phys. 60 (1996) 2530. 7. H. J. Lewerenz, M. Aggour, C. Murrell, M. Kanis, H. Jungblut, J. Jakubowicz, P.A. Cox, S. A. Campbell, P. Hoffmann, D. Schmeißer, J. Electrochem. Soc. 150 (2003) E185. 8. S. P. Garcia, H. Bao, M. Hines, Phys. Rev. Lett. 9 (2004) 166102-1.

208

A high-resolution NEXAFS investigation of the PEDOT:PSS/pentacene system
M. B. Casua, P. Cosseddub, D. Batchelora, A. Bonfigliob, and E. Umbacha
b

Experimentelle Physik II, Universität Würzburg, Am Hubland, D-97074 Würzburg Dept. of Electrical and Electronic Engineering, University of Cagliari, Piazza d'Armi, I-09123 Cagliari

a

Organic electronic devices offer an interesting alternative to inorganic semiconductor electronics due to low-cost deposition methods, flexible substrates, and simple packaging [1, 2, 3]. The organic molecules can be vapour deposited under vacuum, spin coated, dip coated or printed on the proper substrate. All these techniques are relevant for low cost electronics. However, the optimisation of the devices requires a strict control of the morphological and electronic properties of the active medium. Thus, studies of the morphology, growth, and structure of organic thin films are the subject of very intense investigations. In addition, the deep knowledge of the substrate/organic interface plays a role of paramount importance also in device applications. A particular class of interfaces is represented by the organic/organic ones. Their knowledge has a strong technological relevance when an organic film, working as active medium in a device, is deposited on a thin film, also organic, working as electrode. Poly(3,4-ethylenedioxythiophene):poly(styrenesulfonate) (PEDOT:PSS) is commonly used as electrode in various organic devices [4], and pentacene is one of the most widely used organic active media due to its physical, chemical and morphological properties. In particular its high mobility and the possibility to grow highly oriented thin films lead to a strong improvement of the device performances [5, 6]. It is easily understandable that the coupling of two such materials, especially toward the realisation of “all plastic” electronics, can be considered an important step in device engineering. In this framework, we improved an organic field effect transistor (OFET) [7], substituting gold with PEDOT/PSS for making the contacts. We have observed a strong improvement of the device performance in the linear part of the Id-Vd curve. This indicates a lowering of the series resistance effect, one of the most important optimisation parameters at present, since mobilities in organic semiconductors have reached high values. These results demand a better understanding of the interface PEDOT:PSS/pentacene, giving the motivation for the present work. In this report, we present highly-resolved near-edge x-ray absorption fine structure (NEXAFS) spectroscopy measurements taken on pentacene thin films of different thickness deposited on a spin coated PEDOT:PSS substrate. The main goal of our investigation is focused on the determination of the pentacene molecular orientation. In other words, one of the principal needs is understanding if pentacene can be grown with a certain degree of order also when the substrate is extremely rough and soft like PEDOT:PSS. In addition, the understanding of how the interface PEDOT:PSS/pentacene looks like in terms of morphology is another important aspect. The measurements were performed at the beamline UE52-PGM at BESSY. This beamline is characterized by a plane grating monochromator. The photon energy ranges from 100 to 1500 eV, with an energy resolving power of E/∆E= 10500 at 401 eV (cff=10, 10 µm exit slit). The main chamber (base pressure 2x10-10 mbar) is equipped with a standard twin anode x-ray source, a SCIENTA SES200 electron energy analyser, and a home-made partial electron yield detector. PEDOT:PSS thin films (~ 50-100 nm) were spin-coated on Si wafers from a commercially available aqueous solution (Baytron P, 1:20). Thin films of pentacene were prepared by organic molecular beam deposition (OMBD) in-situ using strictly controlled evaporation conditions. The deposition (rate: ~3 Å/min) was monitored with a quadrupole mass spectrometer. The nominal thickness was determined by a previously calibrated evaporator and by using the attenuation of the x-ray photoemission (XPS) substrate signal after pentacene deposition on a Ag(111) single crystal. We carried out NEXAFS measurements in the total electron yield (TEY) mode in grazing incidence (60°). In order to investigate the molecular orientation in the films we took advantage of the dependence of the NEXAFS spectra on the polarisation of the incident radiation [8]. Hence, we measured the spectra by using both in plane (p-pol) and out of plane (s-pol) polarised synchrotron radiation, tuning the polarisation by means of the undulator. Finally, the spectra were normalised using the I0 current and the substrate signal [9]. The energies have been carefully calibrated according to reference 9. Figure 1 shows the C-K NEXAFS spectra obtained from 84 Å (a), and ~100 ML (b) of pentacene on PEDOT:PSS. The spectra were taken in grazing incidence for p- (black curves) and s- (red curves)

209

polarization. Two main groups of π∗-resonances, 1 and 2, dominate the 282-288 eV photon energy range. These features are due to transitions from C1s levels of non-equivalent carbon atoms into the lowest unoccupied molecular orbital (LUMO) and LUMO+1 [10]. It is worth to mention that the energy resolving power of the UE52 beamline gives the opportunity to perform highly-resolved measurements. Thus the spectra exhibit the presence of features that were not resolved in previous works. For example, it is possible to observe the clear presence of two components in feature A (Fig 2), due to contributions from two different carbon sites in the pentacene molecule [10]. The spectral features exhibit a strong polarisation dependence, indicating a high degree of orientational order in the films, and allowing us to determine the molecular orientation from the observed dichroic behaviour [8]. In both cases the molecules adopt an upright standing position of their axis with respect to the substrate. This is clearly seen since the π∗-resonances show their strongest intensities for the spectra taken for s-polarisation of the incident radiation (i.e. E vector parallel to surface). The calculated molecular orientation is 80° for the 84 Å film, while is very close to a perfect upright position for the thicker film. However, from the morphological point of view the two situations are different. While in the case of the 100 ML film the XPS signal from the substrate is not visible, the substrate signal is still clearly visible in the case of the 84 Å. Taking the calculated molecular orientation into account, and assuming that the molecules are NEXAFS C1s standing with their long axis up, 84 Å are equivalent to ~ 5 ML. For this film 2 thickness, the XPS substrate signal should be completely attenuated by the 84 Å pentacene deposition, when layer-bylayer growth occurred. In our case, the 1 presence of a strong sulphur XPS signal p-pol coming from PEDOT: PSS (not shown s-pol here) indicates that pentacene follows an island growth mode. 280 290 300 310 320 Comparing the spectra obtained from Photon Energy (eV) the two films with different thickness, we can observe a change of the spectral shape and of the ratio of the relative NEXAFS C1s intensities when looking at the two main ~ 100 ML groups of π∗-resonances, 1 and 2. This 2 indicates that pentacene molecules interact with PEDOT:PSS, thus perhaps causing a different short range order and mutual interaction which then changes 1 with thickness. p-pol A point that is still not completely clear s-pol regards the possibility that, due to the high PEDOT:PSS roughness, pentacene 280 290 300 310 320 islands could grow embedded in the Photon Energy (eV) matrix formed by PEDOT:PSS at least up to a nominal thickness of around 100 Fig. 1: C1s NEXAFS spectra obtained from 84 Å (a) and 100 Å. This aspect is important not only for ML (b) of pentacene on PEDOT:PSS deposited under different the obvious importance of a basic preparation conditions. The spectra were taken in grazing characterisation of this system, but also incidence for p- (black curve) and s- (red curve) polarisation. because of its technological relevance. As a matter of fact, the device characteristics depend also on the respective charge injection barriers at the interfaces. This can be influenced by the interface morphology in terms of intermixing of the two organic materials. This aspect is still under investigation and requires a careful interpretation of the details characterising the PEDOT:PSS XP spectra as well as the use of complementary techniques like atomic force microscopy to get a picture of the pentacene island distribution. In summary, our results show that pentacene thin films on PEDOT:PSS are characterised by up-right standing molecules. It worth to point out that the NEXAFS signal is averaged over the area sampled
TEY (a. u.) TEY (a. u.)

210

by the incident spot (~50 x 100 µm2) and that the obtained molecular orientation is an average value. Nevertheless, in the present case, due to the strong dichroic behaviour, the calculated values give a clear indication of the real molecular arrangement in the films and of a high orientational order.

NEXAFS C1s D E TEY (a. u.) B A C F

NEXAFS C1s 2 1 ML

TEY (a. u.)

1

p-pol s-pol

~100 ML 282 284 286 288 Photon Energy (eV) 290

280

290 300 Photon Energy (eV)

310

Fig. 2: π∗-region enlarged view of the NEXAFS spectra taken for the 100 ML film.

Fig. 3: C1s NEXAFS spectra obtained from a monolayer of pentacene on PEDOT:PSS, taken in grazing incidence for p- (black curve) and s- (red curve) polarisation.

A consequent crucial question is: do the molecules also stand up-right in the first layer? To give an answer we have performed NEXAFS measurements on a pentacene monolayer in order to investigate the molecular orientation of the first pentacene layer when deposited on PEDOT:PSS. The spectra are shown in Fig 3. According to the previous explanation, it is immediately clear that the molecules, also in this case, take a position such that their axis is perpendicular to the substrate. The dichroism is very pronounced indicating once more a high order already at the monolayer level. The calculated molecular orientation is 77°. This report is only a first step in order to understand the details of the organic/organic PEDOT:PSS/pentacene interface by using synchrotron radiation techniques. Much more experimental and theoretical work needs to be done to answer the demanding open questions. Acknowledgements The authors would like to thank the BESSY staff, Dr. Th. Schmidt, Universität Würzburg, and S. Pohl, Fritz-Haber-Institut, Berlin, for the beamline support, Dr. P. Imperia, Hahn-Meitner-Institut, Berlin, for support while measuring at BESSY, and Dr. N. Koch, Humboldt Univestität, Berlin. Financial support by the DFG through the OFET-Schwerpunktprogramm Um 6/8-1 + Um 6/8-2, by the BMBF under contract 05KS4WWC/2, and by the Italian Ministry of Education and Research under the FIRB program is gratefully acknowledged. References [1] S. R. Forrest, Nature 428 (2004) 911. [2] G. Horowitz, J. Mater. Res. 19 (2004) 1946. [3] J. R. Sheats, J. Mater. Res. 19 (2004) 1974. [4] M. M. de Kok, M. Buechel, S. I. E. Vulto, P. van de Weijer, E. A. Meulenkamp, S. H. P. M. de Winter, A. J. G. Mank, H. J. M. Vorstenbosch, C. H. L. Weijtens, and V. van Elsbergen, Phys. Stat. Sol. (a) 201 (2004) 1342. [5] F. Garnier, A. Yassar, R. Hajlaoui, G. Horowitz, F. Dloffre, B. Servet, S. Ries, and P. Alnot, J. Am. Chem. Soc. 115 (1993) 8716. [6] C. D. Dimitrakopoulos P. R. L. Malefant, Adv. Mater. 14 (2002) 99. [7] A. Bonfiglio, F. Mameli, and O. Sanna, Appl. Phys. Lett. 82 (2003) 3550. [8] J. Stöhr and D. A. Outka, Phys. Rev. B 36 (1987) 7891. [9] A. Schöll, Y. Zou, Th. Schmidt, R. Fink, and E. Umbach, J. Electron. Spectrosc. 129 (2003) 1. [10] M. Alagia, C. Baldacchini, M. G. Betti, F. Bussolotti, V. Carravetta, U. Ekström, C. Mariani, and S. Stranges, J. Chem. Phys. 122 (2005) 124305.

211

Conduction band properties of III-nitrides characterized by synchrotron ellipsometry on core level excitations
C. Cobet1 , M. Rakel1,2 , R. Goldhahn3 , and N. Esser1
1

ISAS- Institute for Analytical Sciences, Department Berlin, Albert-Einstein-Str. 9, D-12489 Berlin 2 Institut f¨r Festk¨rperphysik, TU Berlin, Hardenbergstr. 36, D-10623 Berlin u o 3 Inst. f. Physik, Zentrum f. Mikro- und Nanotechnologien, TU Ilmenau, D-98684 Ilmenau

Group III nitrides such as Ga-, Al- and In-nitride and compounds are a class of semiconductors with one of the highest technological potential today. But in contrast to the recent advances in epitaxial growth and technology, the understanding of fundamental material properties is surprisingly poor. Many questions concerning the electronic band structure are still open. A prominent example is the ongoing discussion about the exact value of the fundamental band gap of InN. A mayor problem for theoretical calculations is the correct treatment of In4d/Ga3d-core electrons, which slightly interact with the nitrogen 2s valence electrons. In GaN this interaction is much weaker than in InN. The s-d-coupling is difficult to handle, because it results also in an additional p-d repulsion effect [2]. This coupling is may be responsible for the very low band gap of InN. The determination of the relative energetic positions between the valence and conduction bands could provide already helpful information [7]. But a complete experimental analysis of electronic transitions (excitations) is mandatory for testing and validating theoretical results. Common methods are either photoemission experiments, which determine e.g. the valence band DOS with respect to the Fermi level, or optical reflection and absorption measurements, which reveal interband electronic transitions in the visible and VUV spectral range. We use the synchrotron light at BESSY II in order to access the optical properties of GaN and InN in the far VUV spectral range by spectroscopic ellipsometry [4, 11]. This measurement technique is based on the determination of the polarization of light before and after the reflection on the sample surface under a certain angle of incidence. The measured amplitude and phase differences can be translated subsequently into the dielectric function (DF) or the commonly used refractive and absorption coefficient. Above 18 eV the DF of GaN and InN is dominated by optical transitions between the Ga3d/In4d core states and the Figure 1: Imaginary part of the measured DF of hexagonal p-like unoccupied (conduction) electron and cubic GaN (solid) in comparison to the calculated PDOS bands. Transitions to conduction states of the p-like empty states around the Ga-atoms (patterned) 1 with s-character are not allowed due to [10]. The Ga3d transition structures are denoted with DIII − 5 DIII according to the notation of Cardona et.al. [3]. the dipole selection rules. However, the localized d-states have a sharp binding energy without significant dispersion in the entire Brillouin zone. Thus, these d-levels can be used to explore the site specific density of empty states in a similar manner as known from X-ray absorption on the N1s core level edge [8]. The stable crystal structure of III-nitrides is the hexagonal wurtzite lattice where the hexagonal closed packed III-nitride layers are stacked in an ABAB... sequence. Thus, the electronic and optical properties reveal an extraordinary axis, the c-axis, parallel to the stacking sequence. All other perpendicular directions remain degenerated. The metastable cubic zincblende crystal structure of GaN was successfully stabilized

212

in thin films on appropriate cubic substrates [1]. The electronic as well as optical properties of this cubic material are assumed to be isotropic. However, the integrated density of conduction band states should be almost identical in both crystal structures. This is, in fact, observed in our ellipsometric measurements on a wurtzite and zincblende GaN sample. Figure 1 shows the imaginary part of the DF in the spectral range of Ga3d transitions. The DF is proportional to the number of excited electrons and should directly relate to the partial density of states (PDOS) with dominant p-character in the conduction bands. Both line shapes are almost the same. The lower overall amplitude and the broadening of transition features are related to the lower crystal quality of the cubic sample. We compare these measurements in figure 1 also to the calculated PDOS of p-like states on the Ga site (patterned) of cubic GaN. As assumed, the DF is approximately reproduced by the calculated PDOS although excitonic effects as well as k and energy dependent variations of matrix elements could influence the DF. The DF of wurtzite GaN was determined on a c-plane sample, where the c-axis is perpendicularly oriented to the surface. In this orientation we have determined the ordinary DF, which relates to excitations with the electric field vector perpendicular to the c-axis. In order to ac- Figure 2: Imaginary part of the ordinary and extraordinary cess also the extraordinary DF we use M- DF of GaN between 18 and 27eV measured on M-plane GaN plane GaN [0100] sample, where the c- [1-100]. axis lies in the surface plane. By measuring in the two high symmetry orientations we can determine both dielectric tensor components which are presented in figure 2. According to these measurements we observe a reasonable shift to lower energies and a slightly different line shape of the extraordinary DF while the ordinary component is almost identical to the measurements on the c-plane sample. This shift could be explained by a distortion of the chemical bondings along the c-axis. Polarization effects as well as a charge accumulation at interfaces along the extraordinary crystal axis are already known in III nitrides. Figure 3: Imaginary part of the measured DF of hexagonal Figure 3 shows a comparison between InN (solid) in comparison to the calculated PDOS of the pthe imaginary part of the ordinary DF of like empty states around the In-atoms (patterned) [5]. The In4d transition structures are denoted with 1 DIII −6 DIII . wurtzite InN and the calculated PDOS of All these valence band characteristic features split up due to p-like InN conduction band states. Be- a spin orbit splitting of the In4d-states, which is notable e.g. tween 16 and 28 eV the DF is again dom- in the 2 DIII and 2 DIII + ∆d structure. inated by excitations of In4d core-states. The line shape of the DF and the assumed PDOS are nearly identical to the respective GaN measurements but shifted by about 2 eV to lower energies. This shift correlates to the lower band gap of InN in comparison to GaN. Recent photoemission experiments report a binding

213

energy of the In4d5/2 electrons of 17.4±0.1 eV with respect to the valence band maximum [9]. On the other hand, we can determine the onset of In4d transitions to conduction bands at about 18 eV (fig. 3). The comparison of both measurements further support the new results about a InN band gap at 0.68 eV [6]. We gratefully acknowledge the BESSY assistance and the financial support of the BMBF under the contract number 05 KS4KTB/3.

[1] D. J. As, D. Schikora, and K. Lischka. Phys. Stat. Sol. (c), 0(6):1607–1626, 2003. [2] B. Bouhafs, F. Litimein, Z. Dridi, and P. Ruterana. Phys. Stat. Sol. (b), 236(1):61–81, 2003. [3] M. Cardona, W. Gudat, E. E. Koch, M. Skibowski, B. Sonntag, and P. Y. Yu. Phys. Rev. Lett., 25(10):659– 661, 1970. [4] N. Esser, M. Rakel, C. Cobet, W. G. Schmidt, W. Braun, and M. Cardona. phys. stat. sol. (b), 242(13):2601– 2609, 2005. [5] F. Fuchs, 2004. Institut f¨r Festk¨rpertheorie und -optik, Friedrich-Schiller-Universit¨t Jena, Fr¨belstieg 1, u o a o 07743 Jena, Germany, priv. com. [6] R. Goldhahn, P. Schley, A. T. Winzer, G. Gobsch, V. Cimalla, O. Ambacher, M. Rakel, C. Cobet, N. Esser, H. Lu, and W. J. Schaff. phys. stat. sol. (a), 203(1):42–49, 2006. [7] W. R. L. Lambrecht, B. Segall, M. Yoganathan, W. Suttrop, P. Davaty, W. J. Choyke, J. A. Edmond, J. A. Powell, and M. Alouani. Phys. Rev. B, 50(15):10722–10726, 1994. [8] K. Lawniczak-Jablonska, T. Suski, I. Gorczyca, N. E. Christensen, K. E. Attenkofer, R. C. C. Perera, E. M. Gullikson, J. H. Underwood, D. L. Ederer, and Z. Liliental Weber. Phys. Rev. B, 61(24):16623– 16632, 2000. [9] L. F. J. Piper, T. D. Veal, P. H. Jefferson, C. F. McConville, F. Fuchs, J. Furthm¨ller, F. Bechstedt, Hai u Lu, and W. J. Schaff. Phys. Rev. B, 72:245319, 2005. [10] W. G. Schmidt, 2004. Theoretische Physik, Univ. Paderborn, Warburger Str. 100, 33098 Paderborn, Germany, priv. com. [11] H. G. Tompkins and E. A. Irene. Handbook of Ellipsometry. William Andrew Publishing, Norwich, 2005.

214

Spectroscopy of ordered V2 O5 and MoO3 layers on Au(111)
S. Guimond, Y. Romanyshyn, H. Kuhlenbeck, H.-J. Freund Fritz Haber Institute of the Max Planck Society, Chemical Physics Department Faradayweg 4-6, 14195 Berlin
This work has been supported by the Deutsche Forschungsgemeinschaft through its Sonderforschungsbereich 546, ’Transition metal oxide aggregates’

V2 O5 and MoO3 are oxidation catalysts which means that they are catalytically active for reactions involving transfer of oxygen. For this reason vanadium oxides as well as molybdenum oxides have been the topic of experimental and theoretical studies. V2 O5 and MoO3 can be prepared as small single crystals by chemical transport methods and thanks to their layered structure they can easily be cleaved. However, the fabrication of the crystals is involved and time-consuming, and the handling and cleaving of the small and mechanically sensitive single crystals is not very comfortable. Besides this the crystals are sensitive to beam damage which means that cleavage may be required often which is a problem in view of the small crystal size. These complications may be avoided by the use of thin films. Due to the high oxygen content of the oxides, oxygen pressures are required which are not suitable for UHV systems. Therefore we have used a high pressure cell for the oxidation procedure. Au(111) was chosen as a substrate since gold is insensitive to elevated oxygen pressures. Both types of oxides were prepared by evaporation of a certain amount of the respective metal onto the Au(111) substrate followed by oxidation in the high pressure cell at a pressure of about 50 mbar with the sample held at 400 ◦ C. The vapor pressure of both oxides is rather high already at low temperature so that evaporation during the oxidation process played a role. ˚ LEED patterns are shown in figure 1. The c(4×2) pattern observed after oxidation of 0.5 A of Mo can not be explained by scattering at regular surfaces of MoO3 . This pattern has also been reported by Biener et al [1]. For V2 O5 a number of different superstructures is observed at low coverages: a (1.25×3.75) structure, a (1.2×1.2)R30◦ structure, a structure with an unit cell ˚ ˚ of 3.6 A×9 A, and some intensity due to regular V2 O5 (001). The LEED patterns of the oxide ˚ layers obtained after oxidation of a metal layer with a thickness of 5 A may be explained by assuming that the surface is covered with rotationally disordered V2 O5 (001) and MoO3 (010) crystallites, respectively. The significant background intensity points towards the existence of an appreciable density of defects, which is at least partly due to damage induced by the electron
Au(111), 50 eV 0.5Å Mo, 50 eV 5Å Mo, 50 eV

0.5 Å V, 50 eV

5 Å V, 50 eV

Figure 1: LEED patterns obtained after deposition of different amounts of vanadium and molybdenum onto Au(111) with subsequent oxidation in an atmosphere of 50 mbar of O2 at 400 ◦ C.

215

V2O5/Au(111) XPS O1s+V2p hν=630 eV
V2p3/2 517.2 V2p1/2

O1s

232.7
530.1

232.6

MoO3/Au(111) XPS Mo3d hν=630 eV

MoO3/Au(111) XPS O1s hν=630 eV
530.5

530.6

529.9

517.05 2ÅV 515.2 (interf.?)

70°

70°
0°

0.5 Å V

70°

70° 232.4

0°

530.3

516.4

529.4

5 Å Mo, 0°

231.4

0.25 Å V

0° 520 525 530 Binding energy [eV] 535 228

510

515

230

232 234 236 Binding energy [eV]

5 Å Mo, 0° 70° 0.5 Å Mo, 0° 238 240 242

530.2

516.5

529.4 70°

70°
0.5 Å Mo, 0°

526

528 530 532 Binding energy [eV]

534

Figure 2: Core level photoelectron spectra of V2 O5 /Au(111) (left) and MoO3 /Au(111) (center and right) as a function of the deposited amount of the respective metal. Detection angles of ◦ 0 and ◦ 70 with respect to the surface normal were employed.

beam of the LEED optics. The oxidation state, the homogeneity, and the structure of the oxide layers have been studied with photoelectron spectroscopy of the core levels (figure 2) and the valence band (figure 3), and NEXAFS (figure 4). The V2p and Mo3d binding energies of the thick films correspond well to reported values for V2 O5 (≈517 eV [2]) and MoO3 (≈232.7 eV [3]), indicating that the high pressure oxidation indeed leads to the formation of V5+ and Mo6+ oxides. The peaks do not exhibit obvious fine structure, indicating that homogeneous phases were formed. This is also indicated by the corresponding valence band spectra shown in figure 3: for both oxides the intensity in the gap is negligible for the thick films which indicates that lower oxidation states are not present in noticeable amounts. The core level binding energies for the thin films are somewhat smaller. This is not necessarily an indication of a lower oxidation state since the XPS final state core holes may be effectively
V2O5/Au(111) valence band hν=121 eV

MoO3/Au(111) valence band hν = 121 eV

2ÅV

70° 0°

5 Å Mo 0°
1ÅV 70° 0°

70°

0.5 Å V

70° 0°

0.5 Å Mo 0° 70°
-2

0.25 Å V -2 0 2 4 6 8 10 12 Binding energy [eV] 14

70° 0° 16

0

2

4 6 8 10 12 Binding energy [eV]

14

16

Figure 3: Valence band photoemission spectra of V2 O5 /Au(111) (left) and MoO3 /Au(111) (right) as a function of the deposited amount of the respective metal. Detection angles of ◦ 0 and ◦ 70 with respect to the surface normal were employed.

216

V2p3/2 70°

V2p1/2

V2O5/Au(111) NEXAFS sample current
V2O5(001) O1s 2ÅV V2O5(001) 2ÅV

MoO3/Au(111) Mo3p NEXAFS sample current
MoO3 single crystal, (010)-pol

MoO3/Au(111) O1s NEXAFS sample current
MoO3 single crystal, (010)-pol 5 Å Mo, 70°

0°

5 Å Mo, 70°

70° 0° 70° 0° 70° 0° 515 520 525 530

1ÅV 1ÅV 0.5 Å V

MoO3 single crystal, (100)+(001)-pol

MoO3 single crystal, (100)+(001)-pol 5 Å Mo, 0° 0.5 Å Mo, 70°

5 Å Mo, 0° 0.5 Å Mo, 70°

0.5 Å V 0.25 Å V 0.25 Å V 535 540 380

0.5 Å Mo, 0° 390 400 410 420 430 440 525

0.5 Å Mo, 0° 530 535 540 545 550 555 560

Photon Energy [eV]

Photon energy [eV]

Photon energy [eV]

Figure 4: NEXAFS spectra of V2 O5 /Au(111) (left) and MoO3 /Au(111) (center and right) as a function of the deposited amount of the respective metal. Light incidence angles of ◦ 0 and ◦ 70 with respect to the surface normal were employed. The MoO3 single crystal data have been taken from reference 3.

screened by Au(111) substrate electrons in the case of the thin layers which would also lead to a reduced binding energy as compared to the thick film case. The observation that the metal levels and oxygen levels are shifted by about the same energy into the same direction may be viewed as an indication that the shift is a final state effect since the O1s binding energies usually depend only weakly on the oxidation state. In the case of Mo3d level of the thin MoO3 layer a shoulder is observed at lower binding energy. This may be due to different types of molybdenum atoms in the oxide layer or to shake-up processes which may come up due to the increased screening. The observation that the intensity ratio does not depend on the detection angle is somewhat in favor of the second explanation. Figure 4 exhibits NEXAFS spectra obtained at light incidence angles of 0◦ and 70◦ for V2 O5 and MoO3 layers with different thicknesses. One point to note is that the Mo3p, V2p and O1s spectra of the thick films depend considerably on the light incidence angle. This demonstrates that the films are ordered as also indicated by the LEED images in figure 1. The thick film data are compared with single crystal spectra in figure 4 showing that the V2 O5 [001] and the MoO3 [010] directions are oriented along the surface normal. As shown by LEED (figure 1), the other directions which are in plane exhibit rotational disorder with some preferential azimuthal orientations. These observations would fit to a film consisting of well ordered crystallites exposing the (001) (V2 O5 ) or (010) (MoO3 ) surfaces towards the vacuum. The crystallites exhibit different azimuthal orientations, likely due to a weak dependence of the interaction with the Au(111) substrate on the azimuthal orientation.

References
[1] M. M. Biener, J. Biener, R. Schalek, and C. M. Friend, J. Chem. Phys. 121, 12010 (2004). [2] J. Mendialdua, R. Casanova, and Y. Barbaux, J. Electron Spectrosc. Relat. Phenom. 71, 249 (1995). [3] M. Sing, R. Neudert, H. von Lips, M. S. Golden, M. Knupfer, J. Fink, R. Claessen, J. M¨ cke, u H. Schmitt, S. H¨ fner, B. Lommel, W. Aßmus, Ch. Jung, and C. Hellwig, Phys. Rev. B 60, u 8559 (1999).

217

Surface sensitive X-ray magnetic circular dichroism (XMCD) measurements of Al2O3 capped magnetite (Fe3O4) for implementation in a magnetic tunnel junction Marc D. Sacher, Volker Höink, Jan Schmalhorst and Günter Reiss Thin Films and Nanostructures,Department of Physics, University of Bielefeld, P.O. Box 100131, 33501 Bielefeld, Germany Magnetic tunnel junctions (MTJs) consisting of two ferromagnetic electrodes (FM) separated by an ultra-thin insulator (barrier) have a large application potential in magnetoelectronics. The tunneling magnetoresistance amplitude is defined as TMR = (2P1P2) / (1-P1P2) by Julliere1 with P1,2 the spin polarization of electrode 1 and 2. Half-metallic ferromagnets with only one occupied spin channel at the Fermi energy and thus 100% spin polarization are of greatest interest. There are several materials in current discussion, e.g. Heusler alloys2 and magnetic oxides3. But up to now there is no working MTJ with electrodes showing a spin polarization of 100% at RT. The half metallic magnetic oxide with the highest TC of 858K is the ferrimagnet magnetite (Fe3O4) being a good candidate for future applications in spintronic devices at high temperatures. Although Versluijs measured a magnetoresistance of 85% in a nanocontact between Fe3O4 crystals4, only a few percent TMR have been measured in a plane Fe3O4-MTJs up to now5. Magnetite grows in the spinell structure, which is a combination of tetrahedral (Td) and octahedral (Oh) lattices sites. Berdunov showed by STM investigations the large influence of surface defects (unoccupied lattices sites) of magnetite on the spin polarization and thus the TMR value6. In a MTJ the same can be assumed for the interface between barrier and Fe3O4. Thus the parameters for the creation of the Al2O3 barrier by deposition of metallic aluminum and subsequent oxidation must be adjusted carefully. With X-ray magnetic circular dichroism (XMCD) in surface sensitive total electron yield mode it is possible to detect even very small changes in stoichiometry of magnetite. This can be done with respect to the very well understood characteristic peakform in the XMCD asymmetry of magnetite7. The excitation of electrons at different lattice sites leads to the characteristic shape of the XMCD spectra at the Fe-L2,3 edge. In the spectra measured for a magnetite single crystal the sites related to Fe2+Oh : Fe3+Td : Fe3+Oh have a correlation of the peak-height of roughly -2 : 1 : -1,5. Even small changes of the stoichiometry of the Fe3O4 result in a significant change of the shape of the spectrum. For additional iron (oxygen) the Fe3+Oh (Fe2+Oh) peak is enlarged. We used the “BESSY Polarimeter” at beamline UE56/2-PGM utilizing 90% right elliptically polarized X-rays at the Fe-L2,3-edge. For these measurements the samples were saturated with an alternating field of about 300Oe. The TEY measurements were done in the remanent state. The measurement in the remanent state ensures equal detection efficiencies of the secondary electrons for parallel and antiparallel alignment of the remanent sample magnetization and the photon spin. The angle of incidence of the X-rays was θ = 30° with respect to the surface. The investigated layer stack consists of 25nm copper conduction line and about 70nm Fe3O4. The magnetite is grown by alternating deposition of a 2nm thick iron layer and subsequent ECR-oxidation (tOx=200s, U=-25V). The free parameters of this ECR oxidation method are the oxygen-ion dose (controlled with the oxidation time tOx) and the ion-energy (controlled with the acceleration voltage U). A detailed description of the oxidation method can be found elsewhere8. After deposition the Fe-O layer is in-situ annealed to 450°C for one hour to reorganize the iron and oxygen atoms to Fe3O4. After cooling down the layer stack is capped with an alumina layer with a thickness of 1,8nm, a typical tunneling barrier in a magnetic tunnel junction. The alumina is also prepared by deposition of metallic aluminum and subsequent oxidation.

218

We investigated the interface between the magnetite and the barrier with XMCD, with a special focus on the different stoichiometry of the magnetite at this Fe3+Td interface. This strongly depends on the oxidation parameters of the Fe-O layer 3+ itself, as well as on the interaction with Fe Oh the adjacent alumina layer at the 2+ interface. In the case of a not fully Fe Oh oxidized aluminum layer, one expects a reduction of the magnetite layer. For a strong overoxidized alumina layer additional oxygen atoms might diffuse into the magnetite and change the FIG 1: Annealed and as grown sample without capping compared to an Al2O3 capped sample. The capping stoichiometry of the Fe3O4 to Fe2O3. results in a reduction of the magnetite. An XMCD measurement of a not annealed sample (tOx=200s, U=-25V) without a capping layer is shown in Figure 1 (black line). Because the Fe3+Oh is obviously larger than the Fe2+Oh, the surface is slightly overoxidized. Furthermore the magnetic moment is rather small. This overoxidation can be reduced by a loss of oxygen and a reorganization during an annealing to 500°C. The spectrum of the annealed sample shows the expected XMCD curve for magnetite (red line). The preparation parameters of this sample are used for all further investigations of the influence of the adjacent alumina on the magnetite/alumina interface. The influence of a standard alumina tunneling barrier as often used in MTJs8 on the XMCD signal can also be seen in Fig. 1 (green line). A strong reduction of the magnetite was found due to the deposition of the metallic Al after annealing. The subsequent oxidation of the aluminum layer did not reverse this effect. In order to build a MTJ with magnetite and alumina, both effects described above were combined. The results of an XMCD measurement made at samples, which were annealed at temperatures ranging from 100°C up to 500°C for one hour after the deposition of the Al2O3-capping, are shown in Fig. 2. The deposition of the capping layer without a previous annealing step results again in a reduction of the Fe-O, which can be observed as an enlargement of the magnetic moment. For temperatures up to 400°C a further reduction of the Fe-O FIG 2: XMCD spectra of as grown and annealed layer has been observed. Again, this is samples. The loss of oxygen can be reduced, but at indicated by change of the relative height of 500°C the magnetite is destroyed due to interdiffusion of the Fe2+Oh and Fe3+Oh -peaks. For all aluminum and iron. investigated temperature the reference shape corresponding to magnetite was not found. At 500°C no magnetic signal has been observed. This can be explained by an interdiffusion of the adjacent aluminum and iron atoms. This also has been found in surface sensitive first arrival measurements utilizing Auger electron spectroscopy (AES).

219

Another possibility how to avoid the loss of oxygen is an additional oxidation after the annealing step (post-oxidation). The idea is to increase the amount of oxygen atoms in the surface layer of the magnetite, the aluminum layer can react with. Fig. 3 shows two XMCD measurements of annealed samples with additional oxidation and Al2O3 capping. With increasing ion dose, as well as increasing ion energy the amount of oxygen at the interface is increased. In the case of U=-10V and tOx=200s the characteristic XMCD shape of FIG 3: Post-oxidation method leads to a strong the magnetite was found. But the XMCD amount of additional oxygen at the interface, but the amplitude is by a factor of 10 smaller than magnetic moment is too small. This could be due to a for the uncapped samples. A possible not ordered Fe-O layer after second oxidation step. reason is that the post-oxidation implements oxygen, but also destroys the ordering of the magnetite. Maybe a thermally activated reorganization of the Fe3O4 at the interface is necessary. This could be done by an additional annealing step at a moderate temperature (below 400°C). A third approach to get a stoichiometric ordered Fe3O4/Al2O3 interface is to increase the total amount of oxygen during the deposition/oxidation of the magnetite. This as well as the moderate annealing of post-oxidized samples will be investigated in the near future. The goal of this project was to investigate the influence of alumina layer on the stoichiometry of the magnetite at the alumina interface with the help of the XMCD asymmetry at the L2,3edge of iron. All systematic measurements hint that the deposition of an adjacent alumina layer results in a loss of oxygen in the magnetite. An extensive and more detailed investigation to avoid this reduction will be done during a future beamtime. The authors gratefully acknowledge the valuable assistance of Franz Schäfers and Andreas Gaupp prior to and during the beamtime at the BESSY II. The support of BMBF is also acknowledged (grant number: BMBF 05 ES 3XBA/5).
1 2

M. Julliere, Phys. Lett. 54A, 225 (1975) S. Kaemmerer et.al., J. Appl. Phys. 93, 7945 (2003) 3 S. P. Lewis et al., Phys. Rev. B 55,10253 (1997); M. Penicaud, J. Magn. Magn. Mater. 103, 212 (1992) 4 J. J. Versluijs et al., Phys. Rev. Lett. 87, 026601 (2001) 5 X. W. Li et al., Appl. Phys. Lett. 73, 3282 (1998); P. J.van der Zaag, J. Magn. Magn. Mater. 211, 301 (2000) 6 N. Berdunov, Phys. Rev. Lett. 93, 057201 (2004) 7 F. Schedin et. al., J. Appl. Phys. 96, 1165 (2004); P. Morall et al., Phys. Rev. B 67, 214408 (2003) 8 A. Thomas et. al., J. Vac. Sci. Techn. B 21, 2120-2122 (2003)

220

Investigation of the electronic properties of bio-organic/inorganic semiconductor interfaces
S. Seifert, G. Gavrila, D.R.T. Zahn Institut für Physik, Technische Universität Chemnitz, D-09107 Chemnitz, Germany W. Braun BESSY GmbH, Albert-Einstein-Str. 15, D-12489 Berlin, Germany In recent years the possibility to use DNA bases in electronic devices such as organic field effect transistors[1] has attracted much attention. Also the transport properties of DNA strands (natural as well as artificial, e.g. poly(G)-poly(C) DNA) and whether or not these would be suitable molecular nano wires were subject of scientific discussions and a number of publications [2,3]. There is, however, little experimental work, addressing the electronic structure (i.e. the density of states) of the DNA bases in the condensed phase, which is of utmost importance for transport properties and device performance. Therefore a systematic photoelectron spectroscopy study of thin DNA base films was performed. The experiments were carried out at the Russian German Beam Line at BESSY employing the Multi User Stage for Angular Resolved Photoemission (MUSTANG) experimental station. This system is equipped with a Phoibos150 electron analyzer (SPECS) and consists of two main chambers for analysis and in situ sample preparation. Experimental: The Si(111) substrates (n-type, resistivity 7.5Ω/cm) annealed under ultra high vacuum conditions (base pressure < 3.10-10mbar) by direct current heating at 800°C to desorb possible contaminants. Several DC flushes of 20s duration were applied to heat the samples up to 1100°C1300°C in order to remove the natural oxide. After letting the sample cool down slowly enough to preserve the (7x7) reconstruction, the surface was passivated in situ by exposure to (2.0±0.5) Langmuir atomic hydrogen. This dose should be enough to passivate the dangling bonds without etching the surface. The DNA base layers were deposited onto the H-Si(111)(7x7) surface by organic molecular beam deposition (OMBD). The nominal layer thickness was monitored by a quartz microballance. The Valence band photoemission spectra (VB PES) of layers of the DNA bases adenine, cytosine and guanine were recorded for different thicknesses (1nm - 10nm) and with excitation energies of 55eV and 150eV. Exemplary valence band photoemission spectra (VB PES) are presented in Fig.1. The width of the spectra was determined by linear extrapolation of the onset of the VB PES at low binding energies and the
Fig.1: Valence band photoemisssion spectra of 10 nm layers of adenine (green), cytosine (brown) and guanine (blue) taken at Eexcite=150eV(dots) and Eexcite=55eV(circles)

cut-off at the vacuum level (i.e. low kinetic

221

energies). The ionization potential (IP) of the DNA base film can be derived by subtraction of the spectral width from the excitation energy. The IPs, determined by this procedure for bulk-like, thick DNA base layers are summarized in Tab.1. The experimentally determined IPs are significantly smaller than the values calculated by Preuss et al. for single molecules in [5], using a ∆SFC ansatz. This can be understood qualitatively as the result of the interaction between the molecules in the condensed phase which distributes the effect of the ionization over several molecules so that the individual molecular orbitals are less effected than in the single molecule case. The fact, that the experimental values are much closer to the calculated ground state eigen energies of the highest occupied molecular orbitals (HOMO's) (DFT/B3LYP functionals, 6-311G+(d,p) basis set [6]), suggests a strong interaction of the molecules. This is supported by the DFT/B3LYP calculation of the ionization potential of an infinite stack of guanine molecules published by Prat, Houk and Foote in [7], where they observe a similar shift in calculations including the neighbor interactions with respect to the calculation for single molecules. DNA base Cytosine Adenine Guanine Experimentally determined IP (6.89±0.10)eV (6.70±0.10)eV (6.41±0.10)eV IP(∆SCF s.mol.) [2] 8.66eV 8.06eV 7.63eV HOMO position IP(DFT/B3LYP;inf. (DFT/B3LYP; s. mol.) stack)[3] 6.67eV 6.34eV 6.12eV

6.64eV

Tab 1: The ionization potentials determined from the valence band spectra follow the same trend as the values calculated for single molecules (s.mol.) with DFT/B3LYP(6-311+G(d,p)) and by a DSCF formalism. A better match in the absolute value can be achieved, if neighbor interactions are included, as performed by F. Prat and coworkers [3] for an infinite stack of guanine molecules.

By broadening the eigen energies calculated for single molecules with Gaussian functions, the density of occupied states (DOOS) displayed in Fig.3 is derived. The calculated curves compare very well to the measured VB PES after a shift towards higher binding energies is introduced. The full width at half maximum (FWHM) of the broadening functions is chosen to match the VB PES. It varies from 0.9eV in the case of the adenine layers to 1.3eV for cytosine and is considerably larger than the estimated overall experimental resolution of 0.07eV. The experimentally observed FWHM is considerably larger than that observed in the case of molecules in the gas phase [8]. The broadening due to inelastic electron scattering should be very similar for all three investigated DNA bases and

Fig.3: Comparison of the measured VB PES of 10nm thick layers of the DNA base on H-Si(111)(7x7) with the calculated DOOS (DFT/B3LYP; basis set: 6-311+G(d,p)). The molecular ground state eigen energies were broadened by Gaussian functions and shifted towards higher binding energies to fit the experimental data.

222

therefore one can conclude, that there is an additional, significant contribution to the broadening at least in the cases of cytosine and guanine. A possible broadening mechanism is the overlap of the molecular valence orbitals which could lead to delocalized, Bloch-type orbitals along the stacking direction and band-like behavior or at least to the lifting of the degeneration of the eigen energies of the molecular orbitals of neighboring molecules as proposed by Calzolari, Felice, and Molinari [9]. Since both possibilities would lead to the observed broadening and the spacing between the multiplet lines (0.02eV) described by Calzolari, Felice, and Molinari is smaller than kBT and well below our experimental resolution, we can not distinguish between the two. Yet an overlap in the valence orbitals that causes either of them, is in very good agreement with the above conclusion of strong molecular interaction. Acknowledgments: The authors would like to use this opportunity to thank all BESSY staff members for their assistance during beam times (especially Mike Sperling for the technical support). We also acknowledge the financial support granted by the BMBF(FK MUSTANG 05KS40C1/3, FK 05KS1OCA1). References: [1] G. Maruccio, Field effect transistor based on a modified dna base. Nano Letters 3, 479 (2003). [2] D. Porath, A. Bezryadin, S. de Vries, and C. Dekker, Direct measurement of electrical transport through dna molecules. Nature 403, 635 (2000) [3] K.-H. Yoo, D.H. Ha, J.-O. Lee, J.W. Park, Jinhee Kim, J. J. Kim, H.-Y. Lee, T. Kawai, and Han Yong Choi Electrical Conduction through Poly(dA)-Poly(dT) and Poly(dG)-Poly(dC) DNA Molecules, Phys. Rev Lett., Vol. 87, No. 19, 198102 (2001) [4] K. Takayanagi, Y. Tanishiro, M. Takahashi and S.Takahashi, Structural analysis of Si(111)-7×7 by UHV-transmission electron diffraction and microscopy, J. Vac. Sci. Tech. A3, 1502 (1985) [5] M. Preuss, W. G. Schmidt, K. Seino, J. Furthmüller, and F. Bechstedt, Ground and excited-state properties of dna base molecules from plane-wave calculations using ultrasoft pseudopotentials. Journal of Computational Chemistry 25, 112 (2003) [6] Gaussian 03, Revision C02, M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, J. A. Montgomery, Jr., T. Vreven, K. N. Kudin, J. C. Burant, J. M. Millam, S. S. Iyengar, J. Tomasi, V. Barone, B. Mennucci, M. Cossi, G. Scalmani, N. Rega, G. A. Petersson, H. Nakatsuji, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O. Kitao, H. Nakai, M. Klene, X. Li, J. E. Knox, H. P. Hratchian, J. B. Cross, V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R. E. Stratmann, O. Yazyev, A. J. Austin, R. Cammi, C. Pomelli, J. W. Ochterski, P. Y. Ayala, K. Morokuma, G. A. Voth, P. Salvador, J. J. Dannenberg, V. G. Zakrzewski, S. Dapprich, A. D. Daniels, M. C. Strain, O. Farkas, D. K. Malick, A. D. Rabuck, K. Raghavachari, J. B. Foresman, J. V. Ortiz, Q. Cui, A. G. Baboul, S. Clifford, J. Cioslowski, B. B. Stefanov, G. Liu, A. Liashenko, P. Piskorz, I. Komaromi, R. L. Martin, D. J. Fox, T. Keith, M. A. AlLaham, C. Y. Peng, A. Nanayakkara, M. Challacombe, P. M. W. Gill, B. Johnson, W. Chen, M. W. Wong, C. Gonzalez, and J. A. Pople, Gaussian, Inc., Wallingford CT, 2004 [7] F. Prat, K. N. Houk, and C. S. Foote, Effect of guanine stacking on the oxidation of 8-oxoguanine in b-dna. J. Am. Chem. Soc. 120, 845 (1998) [8] D. Dougherty, E.S. Younathan, R. Voll, S. Abdulnur, S.P.McGlynn, Photoelectron Spectroscopy of some Biological Molecules, Journal of Electron Spectroscopy related Phenomena, 13, 379 (1978) [9] A. Calzolari, R. Di Felice, and E. Molinari, G-quartet biomolecular nanowires, Appl. Phys. Lett., Vol. 80, No. 18, 3331 (2002)

223

Conformational changes in adsorbed molecules
P. M. Schmidt, Th.U. Kampen, J. Hugo Dil, and K. Horn Department of Molecular Physics, Fritz-Haber-Institut der Max-Planck-Gesellschaft Stilbene is the monomer buildingblock of phenylenevinylene-type oligomers and polymers. It undergoes an interconversion around the central C=C double bond upon irradiation with UV-light of 250-320nm wavelength1. This cis-trans-isomerization of the free molecule follows an in-plane hula-twist mechanism2 whose pathway can be assumed to work as well in constraint systems as on surfaces – thus assessing a ‘molecular switch’. To investigate the isomerization of stilbene-molecules on surfaces, planar trans-stilbene has been studied on Si(100) surfaces at ~90 K by means of photoemission spectroscopy (PES) and near-edge x-ray adsorption fine structure spectroscopy (NEXAFS).

C1s core level spectra of different coverages of cSB on Si(100). The left image shows the data for 1L, 2L, 3L, 5L and 15L coverage while the right image shows the fits of exemplary spectra. The signals can be fitted by two Voigt functions. These two signals can be attributed to the ring atoms (dark red) and the bridge atoms (light-red) of the stilbene-molecules. Development of the intensities of the signals for different coverages show that cSB is bond via the bridge atoms to the Si(100)-surface.

The first layer of trans-stilbene adsorbs in an ordered fashion, with the conjugated πsystem of the molecule being parallel to the substrate surface. No dissociation or isomerization takes place during gas phase deposition. This can be shown by comparing valence band spectra with calculated binding energies of molecular orbitals of stilbene3 and valence band spectra from possible dissociation products (benzene)4,5. No energy shifts can be observed in the Si2p core level emission indicating that the interaction of stilbene with the Si surface is only a weak physisorption. The intensity of the

224

contribution from 2nd layer atoms relatively to the bulk contribution is at a maximum for a coverage of about 3L which corresponds to 1ML coverage. In trans-stilbene multilayers the molecules are randomly oriented. C1s core level emission signals with contributions from the different C atoms in stilbene reveal no shift with increasing coverage indicating that the interaction is strongest between the subsrate and the first monolayer. Annealing results in a desorption of the multilayer leaving a monolayer on the surface.

photon energy

NEXAFS spectra of 3L (~ 1ML) cSB on Si(100) for different sample azimuths. The left image shows exemplarily the spectra for normal incidence (0° inc.) and normal emission (54° inc.), the right image shows the development of the signal intensity over increasing incidence angle. The maximum intensity at grazing incidence angle shows the molecule to be lying flat on the surface.

First comparison of valence band spectra of trans- with cis-stilbene reveal significant differences which will allow identification of both isomers upon laser-2-photonphotoemission studies to be performed for further investigation of the isomerizationprocess.
1

Sension, R. J. et al., J. Chem. Phys. 98, 6291 (1993) 2 Fuß, W. et al., Angew. Chem. Int. Ed. 43, 4178 (2004) 3 Molina, V. et al., J. Phys. Chem. A 101, 3478 (1997) 4 Kim, Y. K. et al., Phys. Rev. B 71, 115311 (2005) 5 Gokhale, S. et al., J. Chem. Phys. 108, 5554 (1998)

225

Electronic structure of ultrathin NiO(111) films studied by angleresolved photoemission
A. Gottberg, E. Weschke, A. Helmke, and G. Kaindl
Institut für Experimentalphysik, Freie Universität Berlin, D-14195 Berlin, Germany

Fig. 1: LEED pattern of a 7 ML-thick Ag(111) film on W(110). The double structure of the hexagonal spots indicates Kurdjumov-Sachs growth geometry.
7.5 7.0 6.5 6.0 5.5

11 ML

5.0 4.5 4.0 3.5 3.0 2.5

10 ML

9 ML

8 ML 5 ML

2.0 1.5 1.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0.0 EB [eV]

Fig. 2: Angle-resolved photoemission spectra of Ag(111) film on W(110). The spectra were recorded at hν = 41 eV in normal emission.

NiO is a prototype antiferromagnetic transition-metal oxide, which has regained interest due to its relevance for exchange-bias thin-film systems. While bulk NiO was studied extensively, the properties of thin films, in particular the electronic structure, are not very well explored. Of special importance in this systems is the (111) surface, which is the polar surface of the rocksalt structure and hence attracts substantial scientific interest. From an experimental point of view, essentially nothing is known about its electronic structure due to the difficulties in the preparation of a clean surface. In this experiment, the valence electronic structure of NiO(111) ultrathin films was studied by angle-resolved photoelectron spectroscopy at beamline U125/2-SGM, using a SES-100 spectrometer. The samples were prepared in the following way: As a first step, a clean Ag(111) film was prepared by epitaxial growth on W(110). Fig. 1 displays a LEED pattern which demonstrates wellordered growth in Kurdjumov-Sachs geometry [1]. The quality of the Ag films is also reflected in the normal-emission photoemission spectra shown in Fig. 2, which are characterized by pronounced quantum-well states. These Ag(111) films are well suited as substrates for the growth of NiO(111). Thin films of NiO (111) were prepared by evaporation of Ni in an oxygen atmosphere of 5*10-6 mbar, with the substrate held at 280°C. The films exhibit hexagonal (1x1) LEED patterns and show pronounced dispersion of d-derived states. Fig. 3 shows corresponding spectra, recorded with a photon energy of hν=100 eV. The figure displays a colour intensity plot as a function of both binding energy and electron emission angle, which directly shows the dispersion of NiO states with the parallel momentum k along the ΓM direction of the surface Brillouin zone. A particularly interesting feature of this electronic structure is the band at the lowest binding energy, which crosses the Fermi energy. This band is originating from Ni 3d states, as verified by photoemission spectra recorded at various photon
 

intensity [arb. units]

226

Fig. 3: k dispersion of electronic states of a thin NiO(111) film on Ag(111) on W(110). The photoemission spectra were recorded with hν=100 eV.

energies in normal emission (not shown here). The intensity of the features as a function of hν is shown in Fig. 4. The behavior resembles that of Ni-3d states, with a typical resonance maximum around hν = 63eV, corresponding to the Ni 3p-3d excitation energy [3]. Furthermore, we do not observe any dispersion of the features with perpendicular momentum as expected for the two-dimensional character of electronic states in an ultrathin 50 film. The interesting finding of this study is 49 the existence of a Ni-3d-derived band, that 48 crosses the Fermi energy. This is not expected from the bulk properties of NiO, 47 which is well-known to be a chargetransfer insulator with a large band gap [2]. 46 Obviously, the ultrathin film prepared here 45 exhibits metallic properties, which may be due to the interaction with the Ag substrate 50 55 60 65 70 photon energy [eV] or a property of thin NiO itself. At present, there are still open questions concerning the sample preparation to obtain these Fig. 4: Photoemission intensity of the electronic states close to the Fermi energy particular films and concerning the origin of the metallic character. Experiments to from Fig. 3 as a function of photon energy. clarify the present results as well as on the polar (111) surfaces of other transition metal oxides are in progress.
integrated intensity [arb. units]

References: [1] R. Clauberg et al., Phys. Rev. B 31, 1754 (1985). [2] M. L. Hildner et al., Surf. Sci. 388, 110 (1997). [3] M. Imada et al., Rev. of Mod. Phys. 70, 1039 (1998) .

 

227

Coupling between spin and orbital order in La0.5Sr1.5MnO4 ?
C. F. Chang, M. Buchholz, C. Schüßler-Langeheine, J. Schlappa, M. Benomar, and L. H. Tjeng

II. Physikalisches Institut der Universität zu Köln, Zülpicher Str. 77, 50937 Köln
E. Schierle, G. Kaindl, and E. Weschke

Institut für Experimentalphysik. Freie Universität Berlin, Arnimallee 14, 14195 Berlin
Funded by the DFG through SFB 608, TP C4. La1-xSr1+xMnO4 is the prototypical material for order phenomena in correlated transition-metal oxides because for x = 0.5 charge, spin, and orbital degrees of freedom form an ordered phase. This compound is also the material for which the particular power of resonant soft x-ray diffraction has been predicted first by Castleton and Altarelli calculating the Mn-L2,3 spectral shape of the (¼,¼,0) orbital order superstructure peak and showing the high sensitivity of the resonance to the character of this order [1]. The analysis of the experimental resonant diffraction data is however more involved: below the Néel temperature of 119 K the shape of the resonance changes and the intensity of the superstructure peak shows a steep increase for certain photon energies, which was originally explained as a stabilization of the orbital-order by the onset of spin order [2], which means a coupling between spin and orbital de grees of freedom. The interpretation is a present matter of debate: a polarization dependent diffraction experiments indicates rather a cross talk between orbital-order and spin-order intensity due to finite in tensity from magnetic ordering at the same momentum transfer as the orbital order peak [3], while a combined theoretical and experimental work indeed suggests a coupling between spin- and orbital degrees of freedom [4]. We used resonant soft x-ray diffraction at the Mn L2,3 resonance to study the momentum space around the orbital order peak. The experiments were carried out at the U49/2-PGM2 beamline with the UHV diffractometer built at the Freie Universität Berlin. For the experiment a newly-designed sensitive photon detector was installed to this instrument. A single crystal of La1.5Sr0.5MnO4 was grown cut and polished with a (115) surface orientation and the 6 (110) and (001) direction in the diffraction plane. 5 a) Sample position 1 The propagation vectors of the orbital- and 4 79 K spin order point into perpendicular directions with 3 120 K the obital-order superstructure peak at (¼,¼,0) and 2 the spin-order peak at (¼,-¼,½). Because of the 1 tetragonal symmetry of the crystal a twinning of the 0 order pattern should be expected, which would lead 5 b) Sample position 2 to about the same amount of domains with the spin4 order peak and domains with the orbital order peak 3 in the diffraction plane. 2 While we found an intense superstructure peak 1 at (¼,¼,0) for all positions of the sample, we did 0 not find any peak at (¼,¼,½) in contrast to our ex-0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 pectations. What we found at this momentum transL (r.l.u.) fer for some parts of the sample, was a background, Fig. 1: Scans along the (001) (L) direction which was well defined along the (110) direction but through the orbital order peak for two different diffuse along the (001) (L) direction. Interestingly temperatures below and above the Neél temperathis background is almost flat along L and has con- ture recorded from two different spots on the sample surface.
Intensity (arb. units)

228

siderable intensity also at the position of the a) background at (¼,¼,½) (¼,¼,0) orbital-order peak [Fig 1a), blue curve]. T < TN For other parts of the sample, this background does not exist as can be seen in the blue curve in Fig. 0 1b), which was recorded from a different spot, and (¼,¼,0) with background b) which only shows the slope towards the specular reT < TN flectivity at small L-values as all scans do. By probing the intensity of the diffracted signal 0 (¼,¼,0) with background as a function of photon energy, we obtain spectroc) T > TN scopic information about its origin [5]. Fig. 2a) contains the spectrum of the diffuse background re0 corded at L = ½. This spectrum is the fingerprint of (¼,¼,0) without background d) the spin-order peak [4] and the same spectrum is T < TN found for the whole background up to the onset of the orbital order peak at L = 0.1, which means that 0 despite the missing peak at (¼,¼,½) the background (¼,¼,0) without background e) is definitely from magnetic origin. This assignment T > TN is further confirmed by the temperature dependence with the background intensity vanishing at the Néel 0 630 640 650 660 670 temperature [red curve in Fig. 1a)]. Photon Energy (eV) From this finding it is not surprising that the orbital-order peak appears to gain intensity below Fig. 2: Spectra recorded at different positions in q-space and for different spots on the sample bethe Néel temperature as found in Ref. 2. Because of low and above the Néel temperature. the different spectral shape of the magnetic and the orbital order spectrum the relative intensity of both contributions is strongly energy dependent as demonstrated in Fig. 2b) and c), where we show spectra taken at the position of the orbital-order peak at a sample spot with particularly high magnetic background. The low temperature spectrum [Fig. 2b)] is a sum of magnetic and orbital-order contributions and the spectral shape hence changes strongly when the sample is heated above the Néel temperature [Fig. 2c)]. This finding explains the peculiar different temperature dependences found for different photon energies [2]. For spectra recorded from sample parts without magnetic background as shown in Fig. 2d) and e) the spectral shape does not change across the Néel temperature. It is therefore plausible that the observed apparent coupling between spin and orbital order is only due to this diffuse magnetic background. Further insight can be expected from experiments using samples with different doping levels, which will affect the orbital orientation [6] and should give additional contrast between spin and orbital contributions. We gratefully acknowledge the excellent working conditions BESSY, the experimental support from the BESSY staff and from the U49/2-PGM2 team. We further acknowledge helpful discussions with M. Braden and D. Senff.
[1] [2] [3] [4] [5] [6] C. W. M. Castleton and M. Altarelli, Phys. Rev. B 62, 1033 (2003). S. B. Wilkins et al., Phys. Rev. Lett. 91, 167205 (2003). U. Staub et al., Phys. Rev. B 71, 214421 (2005). S. B. Wilkins et al., Phys. Rev. B 71, 245102 (2005). C. Schüßler-Langeheine et al., Phys. Rev. Lett. 95, 156402 (2005). D. Senff et al., Phys. Rev. B 71, 24425 (2005).
Intensity (arb. units)

229

Electronic structure of Si(100) at high laser intensities
T. Gießel1 , H. Prima-Garcia1 , R. Schmidt1 , R. Weber1 , M. Weinelt1,2 , W. Widdra3
1

Max-Born-Institut, Berlin, Germany, 1,2 Freie Universit¨t Berlin, Berlin, Germany a 3 Martin-Luther-Universit¨t Halle-Wittenberg, Halle (Saale), Germany a

The interest in optically excited dense e-h plasmas in semiconductors, mainly in connection with the investigation of laser-induced phase transitions, initiated several experimental and theoretical studies in recent years [1]. Laser-induced changes of the optical and structural properties were probed applying pump-probe techniques. Conclusions regarding changes of the electronic structure could be drawn from laser-induced changes of the optical and structural properties. Laser-induced changes of the electronic structure can be probed directly using photoelectron spectroscopy. Moreover, bulk and surface electronic states are in many cases clearly distinguishable allowing for the specific assignment of the laser-induced changes to the surface and bulk, respectively. In this respect Si(100) is a particularly interesting candidate. Due to the technological importance of this system detailed knowledge about its ground state electronic and geometric structure is available [2]. According to both theoretical and experimental studies the surface atoms of the Si(100) surface form asymmetric dimers, which are alternately tilted with respect to the surface plane thereby forming a c(4x2) super structure. The corresponding electronic surface states with an energetic splitting between occupied and unoccupied surface states of ∼ 0.8 eV [3] are localized mainly at the upper dimer atoms for the occupied and at the lower dimer atom for the unoccupied states and are therefore referred to as Dup and Ddown , respectively. Using the laser fundamental of 800 nm (1.56 eV) electrons at the surface can be excited into unoccupied states by single photon processes, while electron excitation in the bulk is governed by the absorption of at least two photons, because of the larger direct bulk band gap of around 3.2 eV. Therefore, a higher excitation rate at the Si(100) surface is expected under these conditions. We have detected laser-induced reversible changes of the electronic structure of Si(100) using time-resolved photoelectron spectroscopy with combined laser and synchrotron radiation (SR). By combining valence band and core level photoelectron spectroscopy we were able to separate surface photovoltage and space-charge effects from transient changes of the electronic structure. Changes of the electronic structure could be assigned specifically to surface and bulk. For the optical excitation of the sample we used a Ti:sapphire regenerative amplifier system (Coherent, RegA 9050) producing pulses of 70 fs duration at 208,3 kHz repetition rate (pulse energy: 6 µJ at the fundamental wavelength of 800 nm). The laser beam was focused to a diameter of <100 µm in order to access the full intensity range up to the damage threshold (0.18 J/cm2 ) of the sample. The excited state of the sample was probed as a function of the delay time between pump and probe by valence band and core level photoelectron spectroscopy using VUV and soft X-ray synchrotron radiation pulses with
230

Center of Mass (eV)

80x10

-3

40 0

Si 2p valence band

47.5 47.0

Si 2p

EKin (eV)

46.5 46.0 45.5 45.0 44.5 44.0 valence band

EKin (eV)

43.5 43.0 42.5 42.0 -200 -100 0 100 200 300 400 Time Delay (ps)

Fig. 1: 2D plots of series of Si(100) valence band spectra (bottom) and of Si 2p spectra (middle) as a function of the time delay between laser and SR. The top panel shows the corresponding center of mass for Si 2p and valence band spectra.

a pulse length of 10 ps. The laser pulses are synchronized (I. Will, MBI) to the 500 MHz BESSY master clock with a jitter < 10 ps, thereby matching the SR pulse length in the low-α mode. Photoelectrons were energy- and angle-selected using a hemispherical electron analyzer with a modified electronics for time-resolved photoelectron detection. Valence band spectra were recorded at an emission angle of 15◦ and core level spectra at an emission angle of 60◦ , both with an angular acceptance of ±8◦ . When interpreting photoemission spectra from a semiconductor excited by intense laser pulses surface photovoltage and space-charge effects have to be addressed properly. Laserinduced changes in the spectra due to these effects have to be separated from those caused by changes of the electronic structure itself. Surface photovoltage and space-charge effects at a certain excitation state are characterized by a rigid shift and possibly broadening of the whole spectrum. Hence, these effects can be identified by comparing laser-induced changes in different regions of the photoemission spectrum. Regions with distinct features, which can be used for that purpose, are in the case of Si(100) the Si 2p core level and the valence band region. Figure 1 shows 2D plots of series of Si(100) valence band spectra (bottom) and of Si 2p spectra (middle) as a function of time delay between laser and SR (SR pulse length < 50 ps). The top panel shows the corresponding center of mass for Si 2p and valence band spectra. The change of the center of mass as a function of the time delay for both Si 2p and valence band spectra is mainly induced by a shift of the whole spectrum and is very similar for the two spectral regions. These laser-induced changes are therefore assigned to SPV and/or space-charge effects and can be used within certain limits as independent detectors for the determination of the coincidence time of laser pump and VUV (or soft X-ray) probe pulses. In order to extract information about changes of the electronic structure from the photoemission spectra a more detailed quantitative analysis is necessary. Figure 2 (left) shows valence band spectra recorded in the BESSY low-α mode (SR pulse length ∼10ps) at a laser fluence well below the damage threshold (0.07 J/cm2 ) for two different time delays. In a pragmatic approach the data are fitted to a function, which consists of a broadened step function for the bulk states, two gaussian functions for the split occupied Dup surface bands and an additional gaussian for the lower branch of the transiently populated unoccupied Ddown band. The panel on the right shows the result for a whole series of spectra recorded as a function of the time delay between laser and SR. Changes in the edge position of the bulk region are governed by SPV and space-charge effects with an initial
231

Ddown amp. /counts

60 40 20 0 240 200 160 τ = 24 ± 3 ps τ = 26 ± 3 ps

Si(100) valence band hν = 49 eV

photoemission signal /a. u.

∆t = -75 ps

∆t = 10 ps

rel. edge pos./eV

Dup amp. /counts

S U R F A C E

0.10 0.00 -0.10 0

τ = 51 ± 2 ps

0 43.5 44.0 44.5 45.0 45.5 46.0 Kinetic energy /eV

B U L K

100 Time delay /ps

200

Fig. 2: Left: Valence band spectra recorded in the BESSY low-α mode (SR pulse length ∼ 10 ps) at a laser fluence well below the damage threshold (0.07 J/cm2 ) for two different time delays. Right: Temporal evolution of the transient population and depopulation of the Ddown and Dup surface state, respectively and dynamics of SPV and space-charge relaxation characterized by the change of the valence band edge position.

relaxation time of 51 ps. Depopulation of bulk states is expected to be in the low percentage range and cannot be resolved. The Dup surface bands are depopulated by around 40%. Note, that in the bulk already 10% depopulation leads to a collapse of the band structure and irreversible changes of the sample. The reversibility of the laser-induced changes at the surface at the observed high excitation densities can be understood in terms of a stabilizing effect of the underlying substrate. Relaxation processes at excitation densities above 1020 cm−3 in the bulk are governed by auger recombination. However, population and depopulation of the Ddown and Dup surface state, respectively, show a simple exponential decay. The matching initial decay times of approximately 25 ps for both population and depopulation confirm the assignment of the transient shape changes at the foot of the valence band edge in the spectra to the Ddown surface band. The position of the transiently populated Ddown band was found 0.65 eV above the Dup surface band in its ground state. This change in the surface band gap by at least 0.15 eV is interpreted in terms of a transient surface band gap renormalization.

References
[1] K. Sokolowski and D. von der Linde, Phys. Rev. B 61, 2643 (2000). [2] S.B. Healy, C. Filippi,1 P. Kratzer, E. Penev, and M. Scheffler, Phys. Rev. Lett. 87, 016105 (2001). [3] M. Weinelt, M. Kutschera, T. Fauster and M. Rohlfing, Phys. Rev. Lett. 92, 126801 (2004).

232

Unravelling the Verwey transition in magnetite via soft X-ray resonant scattering
Y. Su1, H.F. Li1, J. Persson1, W. Schweika1, D. Schrupp2, M. Sing2, R. Claessen2, A. Nefedov3, H. Zabel3, V.A.M. Brabers4 and Th. Brueckel1
Institut fuer Festkoerperforschung, Forschungszentrum Juelich, D-52425 Juelich, Germany Experimentelle Physik 4, Universitaet Wuerzburg, Am Hubland, D-97074 Wuerzburg, Germany 3 Institut fuer Experimentalphysik 4, Ruhr-Universitaet Bochum,D-44780 Bochum, Germany 4 Department of Physics, Eindhoven University of Technology, 5600 MB Eindhoven, The Netherlands
2

1

1.

Introduction

The nature of the Verwey transition in magnetite (Fe3O4) is one of the longest standing puzzles in modern condensed matter physics. In particular, whether and how charge and orbital ordering (CO/OO) take place below the Verwey transition temperature (TV) at ~ 123 K is still among the central focus of current investigations. Significant (~ 20%) charge ordering was found in a latest high-resolution X-ray and neutron powder diffraction refinement [1]. But such a picture was discarded in a later single-crystal resonant X-ray scattering experiment undertaken at the Fe K-edge [2], where no resonance was observed at the corresponding superstructure wavevector (0, 0, ½) and (0, 0, 1), respectively. The situation was further complicated by two recent LDA+U calculations, where possible orbital ordering of the B-site minority t2g electrons was predicted [3-4]. In fact, a prominent role of multi-orbital electronic correlations with associated Jahn-Teller distortions on the B-sublattice was further demonstrated in a more sophisticated calculation via LDA+DMFT method [5]. Therefore, it seems that the orbital degree of freedom is crucial to the occurrence of the Verwey transition. It should be mentioned that in an alternative approach within a more itinerant-electron picture, Khomskii proposed a new model of CO/OO by using the idea of an interplay of site- and bond-centered ordering, which may explain both the structural data and even the presence of multiferroics in magnetite [6-7]. In view of such a resurgence of the exciting new ideas, it is not surprising to notice that new waves of pursuits are being carried out worldwide to unravel the nature of the Verwey transition, among which soft X-ray resonant scattering stands out due to its tremendous sensitivities to charge and orbital degrees of freedom. In the following part, we will report the first observation of significant resonances of the superstructure reflection (0, 0, ½) at both the O K- and Fe LIII-edges on high-quality and stoichiometric single crystals of magnetite. Together with our latest observation of the similar resonance at the Fe K-edge, i.e. in the hard X-ray regime [8], it would become possible now to establish a complete “band topology” of the Verwey transition.

(b) (a) Fig. 1 Resonant scattering from the superstructure refection (0, 0, 1/2)) observed at the oxygen K-edge, (a) energy dependence, (b) temperature dependence of the resonance at 530.5 eV.

2.

Results and Discussions
233

Experiments were undertaken with the two-circle ALICE diffractometer at the beamline of UE56/1-PGM-b. The first major obstacle to work with very high-quality single crystals of magnetite is that even a slight misorientation of the sample would easily ruin your experiment due to the extreme sharpness of both the rocking curve (< 0.1°) and longitudinal width ( ~ 0.001-0.002 Å-1). To overcome this difficulty, a new portable UHV/Low-T goniometer was constructed and successfully tested with the ALICE chamber. This new option finally enabled us to perfectly align all single crystals measured so far. Another obstacle to overcome is the formation of the multiple twins below TV due to the lowering of the symmetry from cubic. A unique way has been figured out to allow us to as much as possibly have access to the desired twins. As shown in Fig. 1 (a), a tremendous resonance of (0, 0, ½) has been observed in the pre-edge regime of the oxygen K-edge. The width of the energy profile is extremely small at only ~ 0.5 eV, suggesting a rather long lifetime of the core holes and a very narrow bandwidth of the corresponding unoccupied intermediate bands, i.e. possibly a localized band. Away from the resonant energies, (0, 0, ½) is completely undetectable, which is consistent with the hard x-ray observations [2, 8], suggesting that (0, 0, ½) is not resulted from lattice displacements due to the low-T structure phase transition. The temperature dependence of (0, 0, ½), as shown in Fig. 1 (b), clearly indicates its association to the Verwey transition. In a similar way, a giant resonance was also observed near the Fe LIII-edge. The temperature dependence of the resonance is shown in Fig. 2. The inverse correlation length along the longitudinal direction of the resonant (0, 0, ½) is estimated at ~ 0.002 Å-1, suggesting a longrange correlation. To summarize, we have finally established an unprecedented picture on the spatial modulation of all relevant bands at the Verwey transition, which would represent a giant step toward a final understanding on the nature of the Verwey transition.

Fig. 2 Temperature dependence of the resonant (0, 0, ½) superstructure reflection near the Fe LIII-edge ( longitudinal scans were shown)

Acknowledgements
We would like to thank BESSY technical and scientific staffs for providing excellent conditions of the storage ring and the corresponding beamline. This project was also partly funded by the BMBF grant O3ZA6BC2.

References
[1] J.P. Wright, et al., Phys. Rev. Lett. 87, 266401 (2001) [2] G. Subías, et al., Phys. Rev. Lett. 93, 156408 (2004) [3] H.-T. Jeng, et al., Phys. Rev. Lett. 93, 156403 (2004) [4] I. Leonov, et al., Phys. Rev. Lett. 93, 146404 (2004) [5] L. Craco, M.S. Ladd and E. Mueller-Hartmann, cond-mat/0511390 (2005) [6] D.I. Khomskii, et al., Phys. Rev. Lett. 94, 156402 (2005) [7] D.I. Khomskii, et al., cond-mat/0601096 [8] Y. Su, et al. unpublished

234

Soft X-ray resonant scattering study of magnetic ordering in La2-2xSr1+2xMn2O7 (x = 0.5)
H.F. Li1, Y. Su1, T. Chatterji2, A. Nefedov3, H. Zabel3, and Th. Brueckel1
1

Institut fuer Festkoerperforschung, Forschungszentrum Juelich, 52425 Juelich, Germany 2 Institute Laue-Langevin, BP 156, 38042 Grenoble Cedex 9, France 3 Institut für Experimentalphysik 4, Ruhr Universität Bochum, 44780 Bochum, Germany

1. Introduction
The physics of highly correlated transition-metal oxides (TMO), including high-Tc cuprates and CMR manganites, is currently attracting enormous attention. In particular, observations of complex ordering phenomena are extremely important for understanding the natures of different intriguing phases, e.g., ferromagnetic insulating (FMI) phase and charge-ordered antiferromagnetic (AF) phase. Like neutron scattering, element-specific resonant soft X-ray scattering also appears to be a powerful probe to magnetic ordering in 3d TMO since a very large resonant enhancement can be expected at the corresponding L3/2 edges. The existence of the charge and orbital ordering (TCO/OO

La2-2xSr1+2xMn2O7 with doping level of x = 0.5 was first suggested by the observation of the resistivity anomaly [1] and the associated superstructures were then revealed by the electron diffraction measurements [2]. However, the magnetic phase diagram seems to be complicated, in which the CE-type AF phase coexists with the layered AF phase. Wilkins et al [3] for the first time performed resonant X-ray scattering measurements for the La1.05Sr1.95Mn2O7 and observed extremely large resonant enhancements of the antiferromagnetic (AF) reflection at the L3 and L2 edges of Mn. Here we report the results of magnetic resonant exchange scattering experiments on La1Sr2Mn2O7. The experiments were carried out at UE56/1-PGM-b beamline of BESSY-II using the UHV-diffractometer ALICE [4].

100 (a) 80 Intensity (a.u.) 60 40 20 0 0.90

(0, 0, 1) @ 70 K

100 Intensity (a.u.) 80 60 40 20

π σ

0.95

1.00 L (r.l.u.)

1.05

1.10

0 630 636 642 648 654 660 666 Photon Energy (eV)

Fig. 1 Resonant magnetic soft X-ray scattering of (001) supperlattice reflection of the La1Sr2Mn2O7 at 70 K. (a) Longitudinal scans at 641.7 eV, (b) the energy dependence of intensity at Mn L2/ 3-edges at the fixed wave vector with σ and π polarizations.

235

~

210 K) in the bilayered manganite

(b)

(0, 0, 1) @ 70 K

π σ

2.

Results and Discussions

At 70 K, i.e. well below the magnetic ordering temperature (

enhancement of the AF superstructure reflection (001) was found at the Mn L3-edge, as shown in Fig. 1(a). Such a resonance is due to an electric dipole transition from the 2p3/2 to 3d levels directly probing the magnetic-exchange split 3d bands. The magnetic correlation length along the L direction can be estimated to be less than 100 Å. The energy dependence of the AF (001), as shown in Fig. 1(b), indicates two noticeable peaks showing an extremely large resonant enhancement in the broad region of the L3-edge. In addition, the magnetic resonant enhancement was also observed in a broad region at the L2-edge. However, there is no obvious polarization dependence. The temperature dependence of the (001) peak suggests TN ~ 210 K. Surprisingly, as shown in Fig. 2, a strong resonance of the AF reflection (001) was also observed near the O K-edge with the σ-polarized incident photons. That may indicate the existence of spin-polarized oxygen holes. The origin of the observed O K-edge resonance will clearly be the subject of further investigations.

12
(0, 0, 1) @ 70 K

10 Intensity (a.u.) 8 6 4 0.94 0.96 0.98 1.00 1.02 L (r.l.u.) 1.04 1.06

Fig. 2 Longitudinal scan of the superstructure reflection (001) near O K-edge with σ incident polarization

Acknowledgements
We would like to thank BESSY technical and scientific staffs for providing excellent conditions for the storage ring and the corresponding beamline. The project was also partly funded by the BMBF grant O3ZA6BC2.

References
[1] R. Seshadri, C. Martin, M. Hervieu, et al., Chem. Mater. 9 270 (1997) [2] J.Q. Li, Y. Matsui, T. Kimura and Y. Tokura, Phys. Rev. B 57, 3205 (1998) [3] S.B. Wilkins, P.D. Hatton, et al., Phys. Rev. Lett. 90, 187201 (2003) [4] J. Grabis, A. Nefedov and H. Zabel, Rev. Sci. Instr. 75, 4048 (2003)

236

~

210 K), a pronounced resonant

Self-assembled monolayers of molecular switches azobenzene alkanethiols
Roland Schmidt1 , Helena Prima Garcia1 , Tanja Gießel1 , Ramona Weber1 , Wolfgang Freyer1 , and Martin Weinelt1,2
1. Max-Born-Institut, Max-Born-Straße 2 A, 12489 Berlin 2. Freie Universit¨t Berlin, Fachbereich Physik, Arnimallee 14, 14195 Berlin a

Alkanethiols are known to form self-assembled monolayers (SAMs), e.g., on gold films. The self-assembly takes about 24 hours in a 10−4 molar solution at room temperature. Due to the specific chemical bond between the thiol head-group and the gold surface as well as lateral interactions densely packed self-assembled layers of alkanethiols form ordered layers (molecular carpet). The average tilt angle between alkane chain and surface normal amounts to about 35◦ [1]. This result is rather independent from the alkane chain length (CH2 )n , n = 16, 20, 22. Therefore, the alkane may be used as a spacer to decouple a functional molecule from the thiol head-group which is in turn used as a linker to the surface. Thus SAMs may be promising candidates for organizing molecular switches at surfaces and may be used as a building block for molecular devices [2]. For this purpose the thiols need to be fuctionalized by an appropriate end-group. We choose azobenzene, i.e. two phenyl rings connected via two double-bonded nitrogen atoms. The azobenzene molecule itself has two isomers, the thermally stable trans- and the metastable cis-form. They can be interchanged by optical excitation with a time constant of the isomerization reaction in the (sub)-ps range [3]. In solution the molecule is successfully used as a photoswitch, e.g., in triggering structural changes of peptides [4]. In our case azobenzene is attached to the alkane chain via a single oxygen bond CF3azobenzene-O-(CH2)12-SH (Az-12, cf. inset in Fig. 1b). To characterize the molecules prior to adsorption we have measured steady-state absorption in solution. We observe high switching efficiency upon irradiation at around 350 nm (trans → cis) and 440 nm (cis → trans). Furthermore, the trans-form of azobenzene is the stable conformation at 300 K. Adsorbed on the gold surface the potential of these molecules as molecular switches has first been pointed out by Stiller and coworkers [5]. They observed work-function changes upon irradiation which were interpreted as a change of the molecular dipole moment associated with a switch of the molecular conformation. A first prerequisite for this model to hold is a well oriented azobenzene entity. Therefore, we studied the orientation of the molecules on a gold substrate (200-300 nm Au / 5 nm Ti / Si(100)) by near-edge X-ray absorption spectroscopy (NEXAFS). The data in Fig. 2a and b show the Auger yield at the C1s - and N1s - absorption edges, respectively. To avoid radiation damage samples have been cooled to 100 K and spectra are recorded by scanning the sample through the de-focussed beam [6]. Typically, one sample with a size of 1 cm2 is used for two to four scans. A load-lock system and sample garage allows us to quickly exchange samples. The spectra have been normalized to the photon flux measuring the transfer characteristics of the beamline by a photodiode. Overall, the spectra in Fig. 1 show significant variations of the intensity when changing the incidence angle of the incoming X-ray beam. For 90◦ the polarization of the exciting radiation is parallel to the surface, while the component of the field vector normal to the surface dominates for 20◦ incidence angle. An analysis of the C1s → C-H∗ resonance at 288 eV (Fig. 1a bottom) gives an average tilt angle of 30◦ of the pure alkane chain with respect to the surface normal. The spectrum

237

a) C1s - NEXAFS
p* s* s*

90o 20o T = 90 K

b) N1s - NEXAFS
Auger yield (arb. units)
90o 20o T = 90 K p*

Auger yield (arb. units)

30 % Az12 / Au

s* 15 % Az12 / Au 400 410 Photon energy (eV) 420

CH3(CH2)12 - S / Au 280 290 300 310 Photon energy (eV) 320

Figure 1: Near edge X-ray absorption spectra for azobenzene - alkanethiols on a 300 nm gold / 5 nm Ti / Si(100) surface. a) C1s edge for 30 % mixture (top) and for a pure alkanethiol layer (0 % mixture, bottom); b) N1s edge for 15 % mixture. The inset shows the average adsorption geometry deduced from the NEXAFS spectra.

recorded for a mixture of 30 % of alkanethiols with azobenzene vs. pure alkanethiols (Fig. 1a top) shows a comparable contrast at around 288 eV. We therefore conclude, that for the mixed layer the average tilt angle is little altered. The latter spectrum is dominated by the C1s → π ∗ transition at a photon energy of 285 eV. The absorption line shows some structure which unfortunately coincides with a dip in the beamline transmission and needs further experimental clarification. Nevertheless, the C1s → π ∗ transition exhibits a polarization dependence similar to that of the C1s → C-H∗ resonance. Hence, both the alkane C-H bond and the pz -orbitals of the phenyl ring are similarly oriented. The analysis gives an tilt angle of about 10◦ between the phenyl-ring plane and the alkane chain (cf. Fig. 1). This result is corroborated by absorption measurement at the nitrogen edge depicted in Fig. 1b. The N1s → π ∗ resonance at 399 eV shows a polarization ∗ dependence similar to the C1s → π transition. As expected, the N1s → σ ∗ transition at 409 eV shows the opposite behavior. Thus both phenyl ring-planes are comparably oriented and the trans conformation is favored at 100 K. SAMs are well-ordered at 100 K and the molecular orientation of the switch follows the tilt angle of the spacer thiol (see inset of Fig. 1b). A second prerequisite for a photoswitch to work is that the optical excitation is long lived and not quenched by charge transfer to the substrate. Charge transfer-times can be studied by Resonant Raman Auger [7] applying the so-called core-hole clock method [8]. After, e.g., N1s → π ∗ excitation the N1s core-hole is mainly filled via Auger decay. When the electron is still present in the π ∗ LUMO (lowest unoccupied molecular orbital) it may either participate in the Auger process or may screen the core-hole and thus lead to a spectator shift of the decay spectrum. In both cases the kinetic energy of the outgoing electron shifts as the photon energy of the exciting radiation (Resonant Raman Auger). If the electron is in contrast transferred to the substrate or a neighboring molecule before decay of the core-hole takes place we observe a Auger-like spectrum, i.e. with the

238

hn (eV)
399.5 399.2

N1s

p*

Intensity (arb. units)

398.9

398.6 398.3

397.0

360

380 Kinetic energy (eV)

400

Figure 2: Resonant photoemission spectra for photon energies spanning the N1s → π ∗ transition. The emission lines shift with the photon energy and are resonantly enhanced.

energetics identical to continuum excitation. Figure 2 shows Auger Resonant Raman spectra recorded at photon energies spanning the N1s → π ∗ resonance. The spectrum is clearly dominated by emission lines, which shift linearly with photon energy but resonate reflecting the change of the π ∗ transition intensity. To estimate the lifetime of the excited electron in the π ∗ LUMO we compare the intensity of Resonant Raman vs. pure Auger channels. Note that this is of course a rather simplifying approach considering the complex wave-packet dynamics initiated by the resonant transition. We obtain a lifetime of the excited electron of ≥ 30 fs at resonance. We conclude, that the azobenzene switch is significantly decoupled from the gold surface by the alkane chain and optically induced switching should be feasible. [1] G. H¨hner et al., J. Vac. Sci. Technol. A 10, 2758 (1992). a [2] M. Zharnikov and M. Grunze, J. Vac. Sci. Technol. B 20, 1793 (2002). [3] T. N¨gele, R. Hoche, W. Zinth, and J. Wachtveitl, Chem. Phys. Lett. 272, 489 a (1997); T. Fujino, S. Y. Arzhantsev, and T. Tahara, J. Phys. Chem. A 105, 8123 (2001). [4] H. Satzger et al., Chem. Phys. Lett. 396, 191 (2004). [5] B. Stiller et al., Materials Science and Engineering C 8, 385 (1999). [6] P. Feulner et al., Phys. Rev. Lett. 93, 178302 (2004). [7] O. Karis et al., Phys. Rev. Lett. 76, 1380 (1996). [8] W. Wurth, P. Feulner and D. Menzel, Phys. Scr. T41, 213 (1992); O. Bj¨rneholm o et al., Phys. Rev. Lett. 68, 1892 (1992).

239

Electrochemical formation of ultrathin CdS-films on Cu(111): An SXPS study using the SoLiAS Experimental Station at BESSY II
Sascha Hümann, Peter Broekmann and Klaus Wandelt, Department of Surfaces and Interfaces, Institute of Physical and Theoretical Chemistry, University of Bonn Ralf Hunger, Thomas Mayer and Wolfram Jaegermann Surface Science Division, Institute of Materials Science, Darmstadt University of Technology Cadmium sulphide (CdS) thin films are widely used in a large number of solid-state device applications such as photoconductive detection, xerography, photovoltaic solar energy conversion and thin-film transistor electronics. In order to prepare these thin films various ways have been used, e.g. spray pyrolysis, dip techniques [1,2] and molecular beam epitaxy [3]. Our work deals with the deposition of ultrathin CdS films using the so-called electrochemical atomic layer epitaxy (ECALE). This method was established by Stickney and co-workers [4] as a low-cost procedure for the production of ultrathin compound semiconductor films of e.g. CdS, CdSe, CdTe and ZnS. This ECALE method is based on the alternating underpotential deposition (upd) of both constituents of the respective film. The deposition of submonolayer or monolayer amounts of a metal on an unlike metal substrate at electrode potentials more anodic than its reversible Nernst potential is well known under the term underpotential deposition (upd). The shift in the deposition potential is caused by an interaction of the adatoms with the substrate which is stronger than the adatom-adatom interaction that is relevant for the bulk deposition.

Fig. 1:

a) EC-STM image of the b) LEED pattern of the

7 × 7 R19.1o 7 × 7 R19.1
o

sulphide layer, 17 nm x 17 nm, E = -375 mV sulphide layer at 116 eV

c) EC-STM images of Cd upd dislocation network, 17 nm x 17nm, E = -375 mV

Fig. 1a) shows a representative electrochemical scanning tunneling (EC-STM) image of the 7 × 7 R19.1o sulphide phase on Cu(111) in 5 mM H2SO4 at -375 mV vs. RHE (reversible hydrogen electrode). This phase is highly ordered and nearly defect free with lattice constants of a = b = 0.667 nm [6,7]. After emersion and air-free transfer of this sample into UHV as described below the LEED-pattern shown in Fig. 1b could be detected. It is consistent with the existence of 7 × 7 R19.1o -S domains on the surface which are rotated by 21.8° with respect to each other, in accordance with the in-situ EC-STM findings. This agreement

240

proves the successful transfer of the electrochemically prepared sulphide layer and its stability in UHV. Underpotential deposition of one monolayer of Cd from a 0.1 mM CdSO4 / 5 mM H2SO4 electrolyte at a potential of -375 mV onto the S-precovered Cu(111)-surface leads to an EC-STM image as shown in Fig. 1c. Both, the observed dislocation network being dominated by angles of 60° and the interatomic distance of 0.667 nm imaged within the domains, point to a structural “template” effect from the S-precovered Cu-substrate. However, it is unclear which atoms are actually imaged within the domains of Fig. 1c. A previous study on the underpotential deposition of Cd onto a chlorine precovered Cu(111) electrode clearly showed an exchange between the preadsorbed Cl- and the postadsorbed Cdlayer leading to a final stacking of Cl/Cd/Cu(111) [8,9,10].

a)

b)

Cu crystal on sample plate

electrolyte compartment

Fig. 2: Electrochemical setup

Thus, in order to answer the question how the domains of the present Cd-S bilayer are terminated we have preformed photoelectron spectroscopy measurements using the SoLiAS experimental station at the U49/2-PGM2 beamline at BESSY II [11]. Photoelectron spectra were recorded with a Phoibos 150 MCD 9 analyzer after each preparation step, using photon energies of 125, 245, 650, and 1040 eV, respectively, thereby varying the information depth of the data. All spectra were recorded in normal emission and are referenced to the Fermi level of a clean, sputtered Cu foil. The electrochemical processing was performed in an Ar purged ambient pressure electrochemistry cell from which a direct air-free transfer of the sample into the UHV chamber was possible (Fig. 2). The electrical contact between the working electrode, i.e. the Cu(111) crystal, and the Pt counter electrode through the electrolyte was established by a “hanging meniscus”, while the reference electrode (RHE) was connected via a luggin capillary. After each electrochemical treatment and emersion of the electrode remainders of electrolyte clinging to the surface were blown off with argon before transferring the sample into UHV. In the following we concentrate on the S2p- and Cd3d core level spectra only. First the clean Cu(111) surface was exposed at -375 mV to pure 5 mM H2SO4 solution, the supporting electrolyte in all further experiments. The corresponding S2p spectrum is displayed in Fig. 3a (1). The three observed doublets correspond to the spin-orbit split 2p states of sulphur in the form of some minor sulphide contamination (S2p3/2 at 161.7 eV), sulphate (S2p3/2 at 168.2 eV) and sulphite (S2p3/2 at 166.15 eV), the latter being a reduction product due to irritation damage. The Cu(111) surface was then brought in contact at -375 mV with an electrolyte consiting of a 1:1 mixture of 5 mM H2SO4 and 10 mM Na2S which leads to a stable 7 × 7 R19.1o -sulphide layer (see Fig. 1a and 1b). The corresponding S2p spectrum is shown in Fig. 3a (2) with the S2p3/2 component at 161.9 eV. Only very weak

241

sulphate induced intensity can be seen in this spectrum indicating that S2- is bound more strongly than SO42- and that only traces of electrolyte may have been transferred. In a third step the sulphide precovered surface was exposed at -375 mV to a mixture of 5 mM H2SO4 and 0.1 mM CdSO4 in order to deposit an additional layer of Cd. The corresponding S2p- and Cd3d-spectra are shown in Fig. 3a (3) and Fig. 3b (3), respectively. Note that the S2pspectrum shows both sulphide (S2p3/2 at 161.75 eV) and sulphate (S2p3/2 at 168.2 eV) with a minor sulphite component (see above). The binding energies of the spin-orbit split 3dcomponents of Cd are 405.5 (d5/2) and 412.5 (d3/2), respectively. Finally, in order to explain the coexistence of sulphide- and sulphate-species in the presence of Cd on Cu(111) we have deposited a further layer of sulphide (from 5 mM H2SO4 and 10 mM Na2S) onto the surface characterized by the spectra a(3) and b(3) in Fig. 3. After this only the signals of sulphide (161.6 eV) and Cd (405.35 eV) can be seen (Fig. 3a(4) and 3b(4)), the sulphate component has completely vanished. A careful analysis of both binding energies and intensities of the series of SXPS spectra shown in Fig. 3 leads to the conclusions that sulphide displaces sulphate from the Cu(111) surface, that the first Cd layer resides on the underlying 7 -sulphide layer, and that it is terminated by sulphate. This sulphate is again displaced by sulphide after exposure to the sulphide containing electrolyte in step 4. These conclusion are visualized by the ball models in Fig. 3c.

Fig. 3: SXPS spectra of a) S2p (hν = 650 eV) b) Cd3d (hν = 650 eV); c) Ball models for the different layers

References: [1] M. K. Karanjai and D. Dasgupta, Mater. Lett. 4 (1986) 368 [2] M. K. Karanjai and D. Dasgupta, Thin Solid Films 155 (1987) 309 [3] K. L. Chopra and I. Kaur Thin Film Device Applications 1983 (New York: Plenum) [4] T.E. Lister, J. L. Stickney, Applied Surface Science 107 (1996) 153-160 [5] S. Hümann, P. Broekmann and K. Wandelt to be published [6] D. Wang, Q. M. Xu, L. J. Wan, C. Wang and C. L. Bai, Surf. Sci. 499 (2002)159-163 [7] A. Spänig, P. Broekmann and K. Wandelt, Electrochimca Acta 50 (2005) 4289-4296 [8] C. Stuhlmann, Z. Park, C. Bach and K. Wandelt, Electrochimca Acta 44 (1998) 993 [9] J. Hommrich, S. Hümann and K. Wandelt. Faraday Discuss. 121 (2002) 12 9-138 [10] S. Hümann, J. Hommrich and K. Wandelt, Thin Solid Films 428 (2003) 76-82 [11] T. Mayer, M. V. Lebedev, R. Hunger, and W. Jaegermann, Applied Surface Science 252 (2005) 31-42

242

Probing the Ground State Electronic Structure of a Correlated Electron System by QuantumWell States: Ag/Ni(111) A. Varykhalov1, A. M Shikin2, W. Gudat1, P. Moras3, C. Grazioli3, C. Carbone3, O. Rader1 BESSY, Albert-Einstein-Str. 15, D-12489, Berlin, Germany St. Petersburg State University, 198904, St. Petersburg, Russia 3 Istituto di Struttura della Materia, Consiglio Nazionale delle Ricerche, I-34012 Trieste, Italy
2 1

The observation of quantized electron states in ultrathin metal films using photoelectron spectroscopy has meanwhile become an experimental routine [1]. Likewise, the theoretical description of the observed discrete electronic states on the base of simple models like the phase accumulation model [2] has considerably matured in recent years. We have recently focused our attention on changes of the quantization conditions given by the electronic structure of the metal substrate comparing the cases where a band gap of appropriate symmetry is present and where quantum well states are degenerate with bulk electronic states of the substrate [3]. The behaviour was described by a phase accumulation treatment extended by the different phase shifts in the gap and the degenerate energy regions. The quantum-well states appear so sensitive to the parameters of the substrate electronic structure that a determination of these parameters through the measurements of quantum well states of the overlayer appears possible. Recently, this possibility has been demonstrated in an impressive way for Al films on Si(111) where the band dispersion of quantum well states in Al parallel to the film plane is strongly modified by the Si band gap [4]. It was also pointed out in Ref. 4 that in this way properties of the gound state of the substrate are probed. Ni serves as the test case for metal systems in the present work since its band structure deviates strongly between the ground state as calculated by local density methods and band dispersions derived from angle resolved photoemission experiments on Ni [5]. One of these band parameters is accessible in normal emission from the (111) surface: The bottom of the Λ1-band is predicted by local density theory at 2.7 eV binding energy whereas photoemission experiments give values between 1.7 and 1.8 eV [5].

Fig. 1. Photoelectron spectra in normal emission of quantum well states in Ag film on Ni(111). (a) Energy range of Ag d and sp states. (b) Overview spectra of clean Ni(111) and 14 ML Ag.

243

Fig. 1. shows the behavior of quantum well states with the thickness of the Ag layer. The measured energy positions were introduced into Fig. 2 as circles. Within the phase accumulation model, we have varied the band gap parameters of the Ni substrate, the upper border of which corresponds to the bottom of the Ni Λ1-band. A fit routine gave the gap parameters in Ni for which the Ag quantum well energies agree best with the experimental data. Fig. 2 shows for example the cases of the gap determined by direct photoemission from Ni (red), the local density gap (green, gap in gray), and the hypothetical case of an absent band gap (magenta). It is seen that the green curve for quantum number 1 ("sp1") predicts a jump in energy in the region of 2.6 eV which is observed also in experiment. The close agreement is confirmed by the behaviour of states with higher quantum numbers. The behaviour of the sp1 branch is highlighted in Fig. 3 where the energy difference of quantum well states for consecutive thicknesses (n monolayers and n+1 monolayers) is plotted. It is concluded that the band dispersion of Ni in the ground state is (at a scale of ±150 meV) sufficiently described by standard local density theory. This means that the strong deviations observed in direct valence band photoemission of Ni are mainly an effect of the photoemission process [6].
[1] T.-C. Chiang, Surf. Sci. Rep. 39, 181 (2000). [2] N.V. Smith, N. B. Brookes, Y. Chang, P. D. Johnson, Phys. Rev. B 49, 332 (1994). [3] A. M. Shikin, O. Rader, G. V. Prudnikova, V. K. Adamchuk, W. Gudat, Phys. Rev. B 65, 075403 (2002). [4] L. Aballe, C. Rogero, P. Kratzer, S. Gokhale, K. Horn, Phys. Rev. Lett. 87, 156801 (2001). [5] See, e. g., O. Rader, W. Gudat in Landolt-Börnstein, New Series, Vol. 23C2 (edited by A. Goldmann, Springer, Berlin 1999). [6] A. Varykhalov, A. M Shikin, W. Gudat, P. Moras, C. Grazioli, C. Carbone, O. Rader, Phys. Rev. Lett. 95, 247601 (2005).

Fig. 2. Thickness dependence of measured quantum well states (black circles) and phase-accumulation-model values for different parameters of the Ni band gap (color).

Fig. 3. Plot of the energy difference of consecutive quantum well states for the experiment and three different models. The parameters from local density theory (green) give best agreement.

244

Valence Band Structure of Intercalated Layered Transition Metal Dichalcogenides
M. Marczynski 1 , M. Helle 1 , K. Rossnagel 1 , and L. Kipp 1 P. Stojanov 2 , A. Tadich 2 , E. Huwald 2 , N. Janke-Gilman 2 , R. Leckey 2 , and J. Riley 2
1 Institut

¨ fur Experimentelle und Angewandte Physik, Universitat Kiel, D-24098 Kiel, Germany. ¨ 2 Department of Physics, La Trobe University, Victoria 3086, Australia.

An interesting and useful property of layered transition metal dichalcogenides (TMDCs) is their ability to intercalate electron donating atoms and molecules between the layers. This can be exploited, for instance, for storing alkali ions as in battery cells as well as for the controlled modi cation of the host electronic structure [1, 2, 3]. We have investigated the effect of Fe and Rb intercalation on the valence band structure of several TMDCs taking advantage of a second generation toroidal electron spectrometer for fast data acquisition [4] and present here angle-resolved photoemission measurements (ARPES) for VSe 2 / Rbx VSe2 taken at the wiggler beamline TGM4 (hν = 30 to 120 eV) and TiTe 2 / Fe0.25 TiTe2 taken at beamline U125/2-SGM (hν = 50 to 100 eV). The expectation is that iron/rubidium intercalation leads on the one hand to an electron transfer to the host material and on the other hand to a decoupling of the layers because of their increased separation. The TMDCs, as well as their iron intercalates, are grown by chemical vapor transport. All of the crystals grow in the 1T structure with octahedral coordination of the transition metal. Cleavage in ultrahigh vacuum leads to clean and at surfaces. Rubidium deposition on VSe 2 is achieved with a carefully outgased SAES Getter source in two steps. The distance to the sample was 3 cm and the evaporation time 10 minutes in both steps at a current of 6 A.

Figure 1: ARPES spectra of VSe2 taken along the ΓM direction before (a) and after (b) the rst adsorption of rubidium at a photon energy of 60 eV. Dashed lines indicate high-symmetry points in the Brillouin zone. In gure 1(a) we present ARPES spectra of pristine VSe2 . The V 3dz2 band crosses the Fermi energy at almost halfway between Γ and M and is clearly separated from the highest Se 4p bands
245

that touch the Fermi energy at the Γ point. After rubidium adsorption ( g. 1(b)) this separation is hardly resolvable. In terms of the rigid band model we can explain this with an electron transfer that occurs from the adsorbed alkali metal rubidium to the originally half lled V 3d z2 band. This leads to a movement of the Fermi energy crossing towards the Γ point and closer to the Se 3p bands.

Figure 2: CIS measurements of VSe2 taken along the ΓM direction before (a+c) and after (b+d) the rst adsorption of rubidium. (c) and (d) show line pro les extracted from (a) and (b), respectively. Dashed lines indicate high-symmetry points in the Brillouin zone. The parameters of the measurement are: hν = 30 eV ... 120 eV, V0 = 14.7 eV, Einitial = EF ermi . To investigate if rubidium atoms intercalate into the van der Waals gaps of VSe 2 and thereby decouple the layers we have taken constant initial state (CIS) measurements at the Fermi energy along high-symmetry directions. A layer decoupling would result in a reduction of the band dispersion perpendicular to the layers. Figure 2 shows our results for the ΓM direction. Before rubidium evaporation ( g. 2(a)+(c)) there is a clearly visible dispersion of the V 3d z2 band. (The inner potential V0 of VSe2 could be experimantally determined to be 14.7 eV.) After evaporation ( g. 2(b)+(d)) the band dispersion is reduced, re ecting weaker interlayer interactions. Furthermore, we present the effects of iron intercalation on the electronic structure of TiTe 2 ( g. 3). Again an electron transfer occurs from the iron atoms to the Ti 3d z2 band which leads to larger electron pockets around the M points (see Fermi surface maps in gure 3(a)+(b)). In contrast to rubidium intercalation, no apparent reduction of the band dispersion perpendicular to the layers is observed in our CIS measurements ( g. 3(c)+(d)). This could be due to a stronger interaction between the Fe 3d and Ti 3d states which acts against the decoupling of the layers. This work was supported by the DFG Forschergruppe 353.

246

Figure 3: Fermi surface maps of TiTe2 (a) and Fe0.25 TiTe2 (b) taken at a photon energy of 100 eV. Yellow lines represent the Brilloin zone boundaries. The CIS measurements of TiTe 2 (c) and Fe0.25 TiTe2 (d) are taken at photon energies between 50 eV and 100 eV with E initial = EF ermi . To calculate the k⊥ -values, an inner potential of 12 eV was assumed.

References
[1] H. I. Starnberg, H. E. Brauer, Phys. Rev. B 58, 10031 (1998). [2] R. Adelung, J. Brandt, K. Rossnagel, O. Seifarth, L. Kipp, M. Skibowski, C. Ramirez, T. Strasser, W. Schattke, Phys. Rev. Lett. 86, 1303 (2001). [3] K. Rossnagel, E. Rotenberg, H. Koh, N. V. Smith, L. Kipp, Phys. Rev. Lett. 95, 126403 (2005). [4] A. Tadich, L. Broekman, E. Huwald, R, Leckey, J. Riley, T. Seyller, L. Ley: BESSY Annual Report 2003, p. 515.

247

The Fermi surface of the single CuO2 -layer of Pb-Bi2201
L. Dudy, B. Müller, B. Ziegler, L. Lasogga, A. Krapf, H. Dwelk, C. Janowitz, and R. Manzke Humboldt-Universität zu Berlin, Institut für Physik Among the essential features determining the macroscopic electronic properties of high temperature- superconductors the Fermi surface (FS) topology plays a key role. Over the last years important progress has been made in highly resolved photoemission on various HTcmaterials, from which Bi2 − y Pby Sr2CaCu2O8+δ (Pb-Bi2212) is the most commonly studied. For the investigation of the intrinsic features it was found more appropriate to study crystals with partial substitution of Bi by Pb to suppress the ( ≈ 1× 5 ) superstructure in the BiO- planes, which lead otherwise to unwanted diffraction replicas [1]. The situation in the one- CuO2 layer material Bi2− y Pby Sr2 − x Lax CuO6+δ (Pb-Bi2201) is not very different from the two-layer material. Here it is also possible to suppress the ( ≈ 1× 5 ) superstructure by partial substitution of Bi by Pb. For the study presented here single crystals of Pb-Bi2201 were grown out of the stoichiometric melt, similar as in [2]. The samples were characterized by energy dispersive x-ray analysis (EDX), susceptibility measurements and LEED. The crystals studied had a typical Pb-content of y = 0.4 and a typical La-content of x = 0.4 giving optimally doped samples with a hole-concentration per CuO2 - layer of nh = 0.15 ± 0.02 [3] and a transition temperature Tc=32K. Regarding the LEED-patterns of fig. 1, the high crystal quality is obvious from the sharp spots, and no sign of a superstructure is visible over a sufficient large energy range. In addition, the homogeneity of the samples was controlled by moving the electron beam over the sample surface.

The high-resolution photoemission measurements have been carried out with a Scienta SES100 analyzer at the U125/2 10m-NIM undulator beamline at BESSY [4]. All data were taken at 22eV photon energy and a sample temperature of 25 K. The overall energy resolution was 18 meV. The angular resolution was below 0.2°. The Fermi level has been determined by measuring the Fermi edge of an evaporated Au film. The normalized photoemission spectra are shown on Fig. 2. The upper inset shows the first Brillouin zone and its orientation relative

Fig. 1: LEED-patterns of optimally doped Bi2-yPbySr2-xLaxCuO6+δ with x=0.4 and y=0.4 for electron energies of 60, 90 and 120 eV. The peaks are quite sharp at all energies. No sign of any superstructure is visible

to the Cu-O bonds of the CuO2 – layer. The electrical field vector of the synchrotron light was orientated parallel to the line ΓM as indicated by the blue arrow in the Brillouin zone of fig. 2.

248

The spectra in the lower part of Fig. 2 at different Brillouin zone locations clearly show the typical parabolic-like narrow band near the Fermi level. We computed the Fermi map by integrating the spectra over an energy window of ± 10 meV about the Fermi-level. It shall be remarked that this window has the size of the resolution. The resulting Fermi map is shown in fig. 2. An anisotropy in the intensity at different Brillouin zone locations with respect to the symmetry direction ΓM is notable. The momentum distribution curves (MDC) near Γ are less symmetric than near M. Between the main FS other features are visible. We simulated the FS assuming the main signal of the photoelectrons is disturbed by a modulated charge distribution. According to the information gained by LEED, the modulation length in real space should be larger than the typical coherence length of LEED of ~10-30 nm. The intensity of the Fermi surface is then computed by using a intensity function I 0 (k , ω = 0, T ) . This intensity function is weighted by a factor (Ni) and shifted by a reciprocal wave vector kS. For example for a (π/|kS|)x1 like superstructure one gets:
I (k , ω = 0, T ) = N 0 I 0 (k , ω = 0, T )
Fig.2: Upper panel: First Brillouin zone and its orientation relative to the Cu-O bonds of the CuO2 – layer. The direction of the electrical field vector used in the photoemission experiments is indicated by an arrow. Middle Panel: The resulting Fermi map computed by integrating the spectra in an energy window of ±10meV around the Fermi-level. For better visualization the density of points along the horizontal line in the Fermi-map were doubled by interpolation with a cubic spline. Typical spectra are shown in the lower part of the figure. The colored lines near the spectra point to the position of the integrated spectra in the Fermi map.

+ N1 I 0 (k + kS , ω = 0, T ) + N 2 I 0 (k − kS , ω = 0, T )

A(k , ω , T ) =

1 Σ λ (ω − ε (k )) 2 + Σ 2

Please note that only the first order shift of kS is taken into account. Higher orders would result in a ribbon-like structure along the kS direction [5]. To simulate the I 0 (k , ω = 0, T ) we use a simplified approach of [6]. We started with the intensity of the photoemission signal given as the Fermi function times the spectral function (the matrix element was set constant): I 0 (k , ω , T ) ≈ f (ω , T ) ⋅ A(k , ω , T ) . The spectral function is due to [6] chosen as ,

249

where we set λ −1 = 1 eV . The imaginary part of the self energy Σ is not only energy dependent but also weakly temperature dependent:
Σ = (αω ) 2 + ( β T ) 2

.

The energy and temperature dependence is controlled by the constants α = 1 and β = 1 meV K −1 , respectively. The constants α,β and λ are only correct within an order of magnitude. The dispersion relation is the bare temperature independent tight-binding-like dispersion ε (k ) = ∆E − 2t (cos(k x ) + cos(k y )) + 4t ' cos(k x ) cos(k y ) − 2t ' ' (cos(2k x ) + cos(2k y )) . From a fitting routine the values are ∆E = 0.18 eV ; t = 0.17 eV ; t ′ = 0.015 eV ; t ′′ = 0.039 eV . In Fig. 3 two computed FS’s are compared with the measured one. Obviously the experimental FS is asymmetric with respect to ΓM, showing that some superstructure must be superposed. Therefore the FS in the middle is calculated with a 7x1 superstructure with a weight of 40% between the refracted and the main band. For the FS on the right a 7x1 superstructure with weight of 40% and a 1x7 with weight of 20% are included. Note that possible matrix-element effects due to the polarization of the incident light are not taken into account.

-3.4

-3.2

M
-3.0 -2.8 -2.8 -3.0

-3

-2.6

-2.6

-3.2

-3.4

-0.8

-0.6

-0.4

-0.2

0.0

0.2

0.4

0.6

0.8

0.8

0.6

0.4

0.2

0.0

-0.2

-0.4

-0.6

-0.8

0.8

0.6

0.4

0.2

0.0

-0.2

-0.4

-0.6

-0.8

high
-2.4 -2.4

Γ

kS
-1.8 -1.8

Intensity

-2.2

-2.0

-2.0

-2

-2.2

0

Fig. 3: Measured FS (left) compared with (middle) the simulated FS with a 7x1 modulation and (right) with a 40% weighted 7x1 and a 20% weighted 1x7 modulation.

To conclude: We investigated single crystals of optimally doped Pb-Bi2201. The LEED patterns show no sign of superstructure while the measured Fermi surface is clearly of asymmetric shape. It can be theoretically reproduced by a modulation of about 7x1 or 7x7. Such a modulation can have its origin in the electronic structure or the crystal structure. This has to be organized in domains larger than the typical coherence length of LEED. We would like to thank the staff of BESSY and in particular Dr. G. Reichardt for excellent support during our measurements. [1] H. Ding et al. , Phys. Rev. Lett. 76, 1533 (1996) [2] I. Matsubara et al. , Appl. Phys. Lett. 58, 409 (1991) [3] L. Lasogga diploma thesis, Humboldt Univ. (2005) and M. Schneider doctor thesis, Humboldt Univ. (2005) [4] G. Reichardt et al. , NIM A 467,462 (2001) [5] M.S. Golden et al., Physica C 341-348, 2099 (2000) [6] A.A. Kordyuk et al., Phys. Rev. B 67, 064504 (2003) and cond-mat/0208418

250

Cobalt and manganese valence states in complex oxides La0.75 Ca0.25 Co0.5 Mn0.5 O3
V. R. Galakhov1∗ A. S. Shkvarin1 , A. F. Tak´cs2 , M. Raekers2 , M. Prinz2 , , a M. Neumann2 , A. V. Korolyev1 , G. V. Bazuev3 , O. I. Gyrdasova3 , T. I. Chupakhina3 , D. V. Vyalykh4 , Yu. S. Dedkov4 , and S. L. Molodtsov4 Institute of Metal Physics, Russian Academy of Sciences — Ural Division, 620041 Yekaterinburg GSP-170, Russia 2 Universit¨t Osnabr¨ck — Fachbereich Physik, D-49069 Osnabr¨ck, Germany a u u 3 Institute of Solid State Chemistry, Russian Academy of Sciences — Ural Division, 620041 Yekaterinburg GSP-145, Russia 4 Institut f¨r Festk¨rperphysik,Technische Universit¨t Dresden, D-01062 Dresden, Germany u o a
Transition-metal perovskites have been known to exhibit varieties of electrical properties, from a large-gap insulator to a metal, and of magnetic properties such as paramagnetism, ferromagnetism, antiferromagnetism, and diamagnetism. LaMnO3 is an antiferromagnetic insulator with high-spin state Mn3+ ions (3d4 , S = 2). When doped with divalent cations, the valence of Mn becomes mixed 3 + /4+. 30 The mixed-valence La1−x Ax MnO3 compounds undergo 20 P1 a phase transition to a ferromagnetic metallic state and ex10 P2 hibit the colossal magnetoresistance effect near room tem0 perature. Ferromagnetism is also induced in LaMnO3 by the substitution of other transition metal such as Co, Ni, -10 etc., for Mn [1]. -20 Goodenough et al. [1] have reported that Mn3+ –O– -30 Mn3+ superexchange interaction are responsible for ferro-60 -40 -20 0 20 40 60 magnetism in the series LaMn1−x Cox O3 and that the trivaH (kOe) lent Co ions are present in their low-spin state. On the other hand, X-ray absorption and photoelectron spectral Figure 1: Magnetization hysteresis studies [2, 3] have showed that dopant Co in manganese loops of the phases P1 and P2 of perovskites is divalent, the same as in CoO and the Mn La0.75 Ca0.25 Co0.5 Mn0.5 O3 at 2 K. ions are Mn3+ –Mn4+ mixed valence. It has been shown that two different phases of 3 La(CoMn)0.5 O3 can be synthesized in single-phase form P1 P2, x 5 and that the spin states of Mn and Co are different in 187 K the two different phases [4]. The phase obtained after 2 annealing at 700 ◦ C (P1) undergoes to the ferromagnetic state at the Curie temperature of Tc ≈ 230 K. 1 156 K The phase synthesized at 1300 ◦ C (P2) has the Curie temperature Tc ≈ 150 K. Core-level Mn and Co 2p X0 ray photoelectron spectroscopic studied at room tem0 50 100 150 200 250 300 perature and magnetical measurements indicate that T (K) the spin states of Mn and Co are different in these two phases of La(MnCo)0.5 MnO3 . Mn and Co ions are as in Figure 2: The temperature detheir trivalent states with Co in the low-spin configura- pendence of real part of the action in the P1 phase and as Mn4+ and high-spin Co2+ susceptibility χ of the two phases of in the P2 phase [5, 6]. La0.75 Ca0.25 Co0.5 Mn0.5 O3 .
∗

1

e-mail: galakhov@ifmlrs.uran.ru

251

χ’ x 103 (cm3/g)

M (emu/g)

Here we report the first magnetical and X-ray absorption spectroscopy studies of the two phases of La0.75 Ca0.25 Co0.5 Mn0.5 O3 . The phase P 1 was prepared by a citrate method at 700 ◦ C. The sample P2 of La0.75 Ca0.25 Co0.5 Mn0.5 O3 was synthesized by a ceramic method at 1300 ◦ C. The oxygen content of the samples was determined by means of a thermogravimetric analysis during reduction in hydrogen flow at 900 ◦ C. Both phases, P1 and P2 show ferromagnetism (see magnetization hysMn 2p3/2 teresis loops M (H) in Fig. 1). The samples P1 and P2 show magnetic Mn 2p1/2 transitions at TC = 187 K and 156 K, La0.07Ca0.93MnO3 as one can see in Fig. 2 where the temperature dependence of real part of P1 the ac-susceptibility χ is presented. At temperatures of 230–300 K, P2 magnetical susceptibility χ follows the Curie-Weiss low χ = C/(T − Θ). La0.87Ca0.13MnO3 From the measurements of χ(T ), the following values of effective magnetic LaMnO3 moments, peff have been estimated: 4.67 µB and 4.32 µB for the phases 630 635 640 645 650 655 660 P1 and P2, respectively. It is not possible to estimate separately valence Co 2p3/2 states of Mn and Co ions in these two phases using these effective magnetic Co 2p1/2 moments only. In order to determine CoO valence states of Mn and Co ions, we have used the method of X-ray abP1 sorption spectroscopy. The Co 2p and Mn 2p x-ray P2 absorption spectra (XAS) of the LaCoO3 La0.75 Ca0.25 Mn0.5 Co0.5 MnO3 oxides were carried out at BESSY at the 770 775 780 785 790 795 800 Russian-German Beam Line. The Photon energy (eV) spectra were normalized to the incident current as measured from a Figure 3: Mn 2p and Co 2p X-ray absorption spectra of gold grid located at the entrance the phases P1 (prepared at 700 ◦ C) and P2 (prepared chamber. While P1 and P2 exhibit at 1300 ◦ C) of La0.75 Ca0.25 Mn0.5 Co0.5 MnO3 . For comparamagnetic-to-ferromagnetic tran- parison, Mn 2p and Co 2p spectra of some manganites, sition at TC , the spectra have been CoO and LaCoO3 are shown. The spectrum of a single measured at about 100 K (at the tem- crystal of LaCoO3 is reproduced from Ref. [2]. perature lower than TC ) and at about 500 K also (much more than TC ). Figure 3 (a) shows the Mn 2p X-ray absorption spectra of two phases, P1 and P2, of La0.75 Ca0.25 Mn0.5 Co0.5 MnO3 in comparison with spectra of manganites. The maxima of the spectra of the P1 and P2 phases are shifted toward the higher photon energy by ∼ 1 eV. This shift arises due to the higher absorption energy of Mn4+ ions. Therefore, valence state of manganese ions in the P1 and P2 phases of La0.75 Ca0.25 Co0.5 Mn0.5 O3 is closer to 4+ than to 3+. In figure 3 (b), the Co 2p X-ray absorption spectra of the samples P1, P2, CoO, and LaCoO3 are presented. The spectrum of LaCoO3 is reproduced from Ref. [2] and could be served as a reference compound of Co3+ ions in the low-spin state. From the comparison of the Co 2p spectra of P1 and P2 with those of CoO (Co2+ ions) and LaCoO3 (Co3+ ions), we have estimated the Co2+ /Co3+ concentrations as (0.82±0.4)/(0.18±0.4) and (0.64±0.4)/(0.36±0.4) for the phases P1 and P2, respectively. While the magnetic moment of Co3+ in the low-spin state is equal to
Intensity (arb. units) Intensity (arb. units)

252

zero, an increase of the relative Co3+ amounts should lead to lowering the effective magnetic moments peff , as it has been found in our measurements of the temperature dependence of the magnetic susceptibility χ(T ). Figure 4 shows Co 2p3/2 X-ray absorption spectra of the phases P1 and P2 in ferromagnetic states Co 2p3/2 XAS 500 K (measurements at ∼ 100 K) and paramagnetic 300 K states (measurements at room temperature and at 100 K ∼ 500 K). One can see that the ferromagnetic-toparamagnetic transition is accompanied by an increase of the Co2+ -ion concentration. On the other hand, an increase of the temperature from 300 K to about 500 K is accompanied by the increase P1 of amount of Co2+ ions. For the sample P1, the temperature changes in the Co 2p spectra are less pronounced in comparison with those for the samP2 ple P2. It means that concentration of Co2+ ions in P1 has achieved saturation. Heating does not change Mn 2p spectra of the both phases. The ob775 780 785 served behavior of the temperature dependence of Photon energy (eV) the Co2+ /Co3+ concentrations found in the X-ray absorption spectra may be due to the polaron for- Figure 4: Co 2p3/2 X-ray absorpmation in La0.75 Ca0.25 Co0.5 Mn0.5 O3 and this needs tion spectra of the phases P1 and P2 further detailed investigations. of La0.75 Ca0.25 Co0.5 Mn0.5 O3 measured at room temperatures (green lines), heated at This work was supported by the Russian Foun- about 500 K (red lines) and cooled down to dation for Basic Research (Grants Nos 04-03- about 100 K (blue symbols). 96092-Ural and 05-03-32355), by the grant “Cooperation between the Ural and Sibirian Divisions of the Russian Academy of Sciences”, the Research Council of President of the Russian Federation (Project NSH-4192.2006.2), and by the bilateral Program “Russian-German Laboratory at BESSY”. V. R. G. acknowledges financial support by BESSY. A. F. T. acknowledges financial support by the Ph.D. program of Lower Saxony, Germany.
Intensity (arb. units)

[1] J. B. Goodenough, A. Wold, R. J. Arnott, and N. Menyuk, Phys. Rev. 124, 373 (1961). [2] J.-H. Park, S.-W. Gheon, and C. T. Chen, Phys. Rev. B 55, 11072 (1997). [3] M. C. Falub, V. Tsurkan, M. Neumann, I. O. Troyanchuk, V. R. Galakhov, E. Z. Kurmaev, H. H. Weitering, Surface Science 53, 488 (2003). [4] P. A. Joy, Y. B. Khollam, and S. K. Date, Phys. Rev. B 62, 8608 (2000). [5] V. L. J. Joly, P. A. Joy, S. K. Date, and C. S. Gopinath, J. Phys.: Condens. Matter 13, 649 (2001). [6] V. L. J. Joly, Y. B. Khollam, P. A. Joy, C. S. Gopinath, and S. K. Date, J. Phys.: Condens. Matter 13, 11001 (2001).

253

Preferential CO oxidation in H2 (PROX) on Pt/CeO2 catalyst, high-pressure XPS and in-situ DRIFTS study D. Teschner , E. Vass , S. Zafeiratos1, P. Schnörch1, M. Hävecker1, A. Knop-Gericke1, H. Sauer1, J. Kröhnert1, F. Jentoft1, R. Schlögl1, O. Pozdnyakova2, A.Wootsch2 1 Fritz-Haber-Institut der Max-Planck-Gesellschaft, Faradayweg 4-6, D-14195 Berlin 2 Institute of Isotopes, CRC, HAS, POB 77, Budapest, H-1525, Hungary Aim of the work and scientific background: The CO content of hydrogen feed to proton exchange membrane fuel cells (PEMFC) must be kept under 1-100 ppm for their proper operation [1]. This can be achieved by using catalysts able to selectively oxidize CO in the presence of excess hydrogen (PROX). Ceria supported Pt catalysts show remarkable activity in the PROX reaction [2]. In order to gain further insight into the mechanism of the PROX reaction on Pt/CeO2 catalyst, catalytic tests, high-pressure XPS, in-situ DRIFTS and HRTEM techniques were utilized. In this report we summarize the main results obtained using Pt/CeO2 catalysts [3]. Some preliminary experiments were also conducted with PtSn and Au/FeO2 samples (they are also promising candidates for good CO removal performance); however they will be investigated in details during our forthcoming beamtime in March. Results Activity pattern of Pt/CeO2 measured by different techniques (flow reactor, in-situ DRIFT, both at p=atm, and in high-pressure XPS at p=1 mbar) showed similar trends: a maximum selectivity towards CO oxidation at T=360-370 K decreasing with increasing temperature. Bulk metallic, pronounced adsorbateinduced surface Pt, interface Pt and a small amount of oxidized Pt sites were shown by high-pressure XPS under PROX conditions (Figure 1). The pre-oxidized ceria surface was strongly reduced in pure H2 but significantly re-oxidized under PROX conditions (i.e. O2+CO in high excess of hydrogen) at T=358K (Figure 2). The remaining small amount of Ce3+ decreased with increasing temperature.
Fig. 1. Pt 4f region of 5% Pt/CeO2 at different conditions: 1, in 0.48 mbar H2 at RT (after O2 activation at 573 K); 2, in ~0.5 mbar PROX mixture at 358 K
1 1

HRTEM found well-crystallized CeO2 particles (8-10 nm) in the case of activated

254

(pre-oxidized) sample that transformed to an oxygen deficient ceria super-cell structure (CeO1.695) after PROX reaction. Metallic Pt particles (2-3 nm) and small (0.5-0.6 nm) Pt clusters were indicated by HRTEM. These findings were in accordance with the variations in relative intensity of the corresponding Pt-CO bands (in-situ DRIFTS). Different types of carbonate and formate species were detected (XPS and DRIFTS). The negative correlation between formate species and CO2 yield clearly indicates that formates are not intermediates in the PROX reaction. Moreover their accumulation near the metallic particles is suggested at high
Fig. 8. Part of the Ce 3d region of 5% Pt/CeO2 at different conditions: 1, in 0.5 mbar O2 at 573 K; 2, in 0.48 mbar H2 at RT; 3, in ~0.5 mbar PROX mixture at 358 K

temperature (T=523 K). Broad, however not resolvable structure in the OH stretching region could be found by DRIFTS in the PROX

reaction mixture indicating significant amount of adsorbed water (identified also in XPS) in a hydrogen-bonded structure. Its amount decreased with increasing temperature, with parallel decrease in the selectivity towards CO oxidation. Thus, the presence of adsorbed surface water seems to suppress hydrogen oxidation while CO oxidation still takes place, as the metallic particles are covered by CO (DRIFTS). The direct contribution of surface water in a low-temperature water-gas-shift (LTWGS) type reaction in the PROX mixture is proposed [4] as follows (Figure 1):

Fig. 1: Proposed model describing the reactions happening on Pt/ceria in the PROX reaction mixture at low and high temperature.

At low temperature, a significant amount of water accumulates on the ceria, via spillover of adsorbed hydrogen atoms from the platinum. This water reacts in LTWGS reaction with linearly bonded CO at the Pt/ceria interface, forming CO2 and hydrogen. The by-product

255

hydrogen before desorption (or another hydrogen molecule from the gas-phase) regenerates the water structure on the support in the close vicinity of Pt. This adsorbed water, which ensures that CO is linearly bonded to the interface Pt site, might be stabilized by the oxygendeficient character of ceria (CeO1.695). At higher temperature the hydrogen-bonded structure decomposes and water desorbs allowing the liberation of coordinatevely unsaturated (cus) Cen+ sites at the interface. Thus CO can be adsorbed at the interface in a bridged-like manner (both found in DRIFTS and XPS), oxygen coordinating to this cus site. The adsorbed CO can then dissociate or hop to the ceria and react with the first OH group forming formates. The latter species are stable on the ceria at this condition and might block the way of still existing surface water to the Pt particles. The negative correlation of surface formates and CO2 yield strengthens the proposed model. To sum up our results point to the beneficial effect of surface water, which suppresses further hydrogen oxidation and can directly participate in a water–gas-shift-like reaction under PROX condition. Whether this mechanism holds for other PROX catalysts will be the topic of our next investigation. Acknowledgement The authors thank the BESSY staff for their continual support during the measurements. The work was supported by the Athena Consortium. References
1 A. J. Appleby, F. R. Foulkes, Fuel Cell Handbook, Van Nostrand Reinhold, New York, 1989. 2 A. Wootsch, C. Descorme, D. Duprez, J. Catal. 225 (2004) 259. 3 O. Pozdnyakova, D. Teschner, A.Wootsch, J. Kröhnert, B. Steinhauer, H. Sauer, L. Toth, F. C. Jentoft, A. Knop-Gericke, Z. Paál, R. Schlögl, J. Catal. 237 (2006) 1. 4 O. Pozdnyakova, D. Teschner, A.Wootsch, J. Kröhnert, B. Steinhauer, H. Sauer, L. Toth, F. C. Jentoft, A. Knop-Gericke, Z. Paál, R. Schlögl, J. Catal. 237 (2006) 17.

256

Selective gas-phase hydrogenation of aliphatic triple and double bond using palladium based catalysts D. Teschner, E. Vass, S. Zafeiratos, P. Schnörch, E. Kleimenov, M. Hävecker, A. KnopGericke, R. Schlögl Fritz-Haber-Institut der Max-Planck-Gesellschaft, Faradayweg 4-6, D-14195 Berlin Aim of the work and scientific background: Heterogeneous catalysis plays an essential role in the field of chemistry and in chemical manufacturing. The complexity of a catalytic process usually requires that chemistry and engineering intimately mix to deliver the desired effect. The research project Athena (Advanced technology in catalytic chemistry and engineering for novel application) is a joint collaboration of Fritz-Haber-Institut and several British and American institutions presenting a multi-disciplinary approach to investigate heterogeneous catalytic reactions. To produce polymer-grade alkene stream the removal of multiple unsaturated hydrocarbons has a crucial importance. This can be done by using catalysts showing high selectivity towards hydrogenating C≡C triple bond instead of C=C double bond. In this study we aimed to understand the governing factors of selective triple C≡C hydrogenation (i.e. only hydrogenating to alkene) on palladium by using in-situ X-ray photoelectron spectroscopy. For these purpose single crystals, foils and supported catalysts are used and studied by surface sensitive XPS, in-situ, i.e. in the desired manner during catalytic experiment. Results All catalytic samples (5% Pd/carbon-nanotubes, 3% Pd/Al2O3, Pd(111) and Pd foil) showed activity in the hydrogenation of 1-pentyne (~1 mbar). Both, single and total hydrogenation

Fig. 1: Pd 3d region of 5%Pd/carbon-nanotube (a), Pd foil (b) in the reaction mixture of 0.85 mbar H2 + ~0.05 mbar 1-pentyne at 358 K. As a comparison Pd 3d of Pd foil in the hydrogenation of t-2-pentene (c) is also shown. Incident photon energy, hν = 720 eV.

products were formed, however mainly selective hydrogenation to 1-pentene occurred at steady 358 K. This single hydrogenation is related with carbon retention as a special “Pd-C surface phase” builds up in the reaction (Fig.1). The 335.6 eV Pd 3d5/2 component is surface related (shown by non-destructive depth profiling; Fig. 2); however calculation revealed its
257

thickness as 2-3 atomic layers. A direct link between the “Pd-C” component and the 1pentene yield could be established; therefore the active surface in selective triple bond hydrogenation is a non-metallic Pd phase. Valence band spectra point also to a massive charge redistribution. It is important to mention that in the hydrogenation of trans-2-pentene (Figure 1c) this Pd-C component does not form, only adsorbate induced surface-core-level appeared [1]. This spectroscopic evidence correlates perfectly with recent catalytic work [2] showing that propene formation from propyne occurs only when
Fig. 2: Pd3d5/2 as a function of photon energy (information depth)

the catalyst retained a significant amount of carbon on its surface. For propane formation carbon retention was

not a prerequisite. HRTEM experiments on used catalysts indicate lattice expansion (i.e. carbon incorporation) which is more pronounced in the surface-near area. Depth-profiling XPS experiment during catalytic run on both palladium and carbon core level reveals maximum carbon content with intermediate information depth (Fig. 3, red line). This clearly indicates that a significant amount of carbon is situated in the near-surface region i.e. in subsurface positions. The remarkable increase of palladium at the most
Carbon Composition (%)
60 55 50 45 40 35 30 25 20 15 780 380 260 140
Deeper layers

surface sensitive energy suggests that the surface is not fully covered by any type of adsorbates and that the subsurface carbon is located below the 2-3 palladium-atom-thick “Pd-C” layer. By using “switching off” (H2/C5) experiments we conclude that the “Pd-C phase” is heterogeneous: a Pd rich layer builds the surface, below of which a

Carbon Composition (%)

Reaction C5 off H2 off
Top surface

Photoelectron Kinetic Energy (eV)

Pd-to-C 2-to-1 layer is found. This latter

Fig. 3: Carbon distribution as a function of information depth

decomposes and forms as pentyne is switched off respectively introduced (Figure 4), emphasizing the dynamics of the system. The switching experiments were performed in depth profiling manner hence carbon depth distribution could be calculated (dashed lines in Figure 3). Although the absence of pentyne in the gas-phase reduces the carbon content, however not on the surface. Carbon was depleted from the subsurface region (note the much less pronounced maximum curve), therefore desorption of pentyne is accompanied not only by a

258

partial decomposition of “Pd-C” but also by the migration of carbon from subsurface to surface position. Pure pentyne in the feed gas induced desorption of hydrogen and increased the surface coverage of C5. Some subsurface carbon should have moved into deeper layers, as well. The different carbon depth-profiles underline the importance of in-situ surface analysis. At high temperature (523 K) in the reaction mixture both hydrogen the and pentyne is carbon desorb and/or decomposes, double-layer destroyed, and a blocking surface/subsurface layer builds up inhibiting
Fig. 4: Pd3d5/2 and C1s core levels at three stages. 1: during reaction; 2: pentyne switched off (only H2); (Spectrum after switching back 1-pentyne is identical to curve 1, not shown.) 3: H2 switched off (only C5).

any further reaction. (Note HRTEM indicated also graphitic carbon after high-T

reaction.) The active double-layer can be restored only after regeneration, however by the reaction itself. Pd-C surface phase was found also when acetylene was hydrogenated, but not during ethylene hydrogenation. The rate of the latter reaction was suppressed roughly by a factor of three if Pd-C like phase was built up from acetylene previously, before switching to ethylene feed. This result give us a new insight why palladium based catalysts are able to handle the job removing multiple unsaturation of hydrocarbons from mainly alkene streams. Acknowledgement The authors thank the BESSY staff for their continual support during the measurements. The work was supported by the Athena Consortium. References
1 D. Teschner, A. Pestryakov, E. Kleimenov, M. Hävecker, H. Bluhm, H. Sauer, A. Knop-Gericke, R. Schlögl: J. Catal. 230 (2005) 195. 2 D.R. Kennedy, G. Webb, S.D. Jackson and D. Lennon: Appl. Cat. A 259 (2004) 109.

259

BESSY Annual Report 2005 : HE-SGM, MPI-FKF Stuttgart / U Bochum

Single atoms are the most active: Cu-catalyzed carboxylic deprotonation
D. Payer,1 T. Strunskus,2 A. Dmitriev,1 N. Lin,1 J. V. Barth,3,4 Ch. Wöll,2 and K. Kern1,3 1 Max-Planck-Institut für Festkörperforschung, Stuttgart, 2 Lehrstuhl für Physikalische Chemie I , Ruhr-Universität Bochum 3 Institut de Physique des Nanostructures, Ecole Polytechnique Fédérale de Lausanne, 4 Departments of Chemistry and Physics & Astronomy, University of British Columbia Almost 80 years ago Taylor coined the concept of ‘active sites’ in heterogeneous catalysis suggesting that adsorbate bond cleavage or formation occurs preferentially at specific arrangements with low-coordinated surface atoms. The identification of such active sites is decisive for the understanding of surfa ce reaction mechanisms, the corresponding rate- limiting steps, and the design of advanced catalysts with improved efficiency or selectivity. Here we demonstrate a new paradigm in this field: the function of highly mobile adsorbed atoms (adatoms) as dynamic active sites in surface chemical reactions. In view of the appreciable twodimensional vapour pressure of many metal catalysts at typical reaction temperatures (>400 K), these findings indicate that mobile adatom deserve general consideration in surface chemical reactions and can bestow dynamic heterogeneity to materials. Combined scanning tunnelling microscopy and X-ray photoelectron spectroscopy studies reveal that the deprotonation reaction of carboxylic groups of 1,3,5-benzenetricarboxylic acid molecules (trimesic acid, TMA, cf. figure 1A) adsorbed at the Ag(111) surface readily occurs in the presence of a diluted 2-D Cu adatom gas at the surface, while negligible reaction rates occur under similar conditions with Cu in the form of 2-D condensed islands. We demonstrate the function of highly mobile adsorbed atoms (adatoms) as dynamic active sites in surface chemical reactions, as illustrated by scheme 1.

Scheme 1 The hydrogen-bonded open networks can be fabricated by deposition of TMA on a cold substrate (T = 120 K) followed by warming to room temperature, as shown by the STM image in figure 1B. The corresponding model in figure 1C shows how the dimerization of the selfcomplementary carboxylic groups accounts for the dominating planar honeycomb domains. The XPS data shown in figure 1D prove that the carboxylic groups are protonated. In order to address the reactivity of coadsorbed Cu, TMA molecules and small concentrations of Cu atoms (0.05 ML) were sequentially deposited on the cold Ag(111) surface (120 K). Under these conditions, the molecules remain protonated at 120K, as evidenced by the corresponding XPS measurements. The chemical activity of the Cu atoms becomes apparent upon increasing the substrate temperature. The spectroscopic data show dramatic changes in the TMA carboxylic groups above 200K. The analysis of the XPS chemical shifts clearly reveals the formation of a tricarboxylate species at 300K, i.e., there is definitely a complete deprotonation of the carboxylic groups which is associated with the presence of Cu adatoms (cf. figure 2A). These findings are substantiated by STM topographic data (reproduced in figure 2B) showing complete inhibition of honeycomb network formation since the underlying H-bond motif is absent. Rather, TMA molecules aggregate in disordered agglomerates containing bright protrusions, which are Cu islands formed in the annealing process. Since carboxylic groups are still present after Cu deposition before the sample warm- up, it is concluded that the deprotonation reaction is not

260

BESSY Annual Report 2005 : HE-SGM, MPI-FKF Stuttgart / U Bochum

triggered by the impact of Cu in the deposition process, rather it must be mediated by thermal activation and Cu adatoms during warm- up. In order to clarify whether the boost of chemical reactivity correlates with the highly dispersed Cu adatoms, control experiments were performed, where Cu was offered in a 2-D condensed form by predepositing the same amount of Cu at room-temperature on the clean Ag surface. Subsequently the substrate was cooled down to 120 K and TMA was added. Following warm- up to room temperature, strikingly the formation of perfect honeycomb structures coexisting with the Cu islands was observed, as shown by the STM image in figure 2C. The underlying hydrogen bonding implies that TMA deprotonation does not occur in the presence of pre-grown Cu islands. The sharp distinction between the two cases demonstrates that the Cu condensation must be associated with a drastically decreased chemical reactivity. Consequently the deprotonation reaction rate depends on the Cu adatom density, and Cu adatoms are the decisive element mediating carboxylic deprotonation, i.e., this mobile species represents the true active site in this surface chemical reaction. Since the two-dimensional vapour pressure of metals is appreciable at typical reaction temperatures (>400 K), the findings indicate that activation by a mobile adatom gas is of general relevance in catalysis and bestows a dynamic heterogeneity to materials.

Figure 1 A Structural model of the 1,3,5-benzenetricarboxylic acid C6 H3 (COOH)3 . (TMA) molecule. B Assembly of extended hydrogen-bonded TMA honeycomb networks on Ag(111) following sub-monolayer deposition on the 120 K-cooled substrate and warm- up to room temperature. C Model of the hydrogen-bonded nanoporous supramolecular TMA layer with hydrogen-bond mediated dimerization of self-complementary carboxylic groups. D XPS data testify the integrity of the organic molecules (photon energy 400 eV for carbon, 670 eV for oxygen spectrum). Indicated are the C 1s position of aromatic ring and carboxylic group, and the convoluted O 1s signal with contributions from carbonyl and hydroxyl groups, respectively.

261

BESSY Annual Report 2005 : HE-SGM, MPI-FKF Stuttgart / U Bochum

Figure 2. A XPS data (solid curves) monitoring the chemical changes occurring in the 300 K warm-up of an intermixed TMA/Cu layer grown at low temperature on Ag(111) (photon energy 400 eV for carbon, 670 eV for oxygen spectrum). For comparison, spectra of the protonated species are shown as dashed curves. The formation of a TMA tricarboxylate species is reflected by the distinct chemical shift of the higher-energy C 1s peak and the characteristic narrowing of the O 1s peak. B STM image of irregular TMA agglomerates coexisting with Cu islands when 120 K co-deposited TMA and Cu are annealed to room temperature. C In the presence of pre-deposited condensed 2-D Cu islands deprotonation in warm- up is negligible, and regular hydrogen-bonded TMA honeycomb networks evolve.

262

Local structure determination of a chiral adsorbate: Alanine on Cu(110)

D.I. Sayago1, M. Polcik1, G. Nisbet2 , C.L.A. Lamont2 and D.P. Woodruff3 1 Fritz-Haber-Institut der MPG, Faradayweg 4-6, D 14195, Berlin, Germany 2 Centre for Applied Catalysis, Department of Chemical and Biological Sciences, University of Huddersfield, Queensgate, Huddersfield, HD1 3DH, UK 3 Physics Department, University of Warwick, Coventry CV4 7AL, UK

In an earlier study of the adsorption of the simplest (deprotonated) amino acid, glycine(ate), NH2CH2COO-, on both Cu(110) and Cu(100) using scanned-energy mode photoelectron diffraction (PhD) [1], we showed that the molecule bonds to the surface through the two carboxylate O atoms and the amino N atom, each of these bonding sites being off-atop. One interesting feature of the (2x3) ordered phase formed on (110) surface is the presence of glide symmetry lines. This may be understood in terms of two different rotational orientations of the amino groups when bonded to the surface. In effect, although glycine in the gas phase is not chiral, when bonded to the surface it becomes chiral, and the (2x3) ordered phase contains equal numbers of the two enantiomers which are mirror images of one another (see fig. 1). Fig. 1 Plan view of the Cu(110)(2x3) glycinate adsorption phase Alanine, NH2CH3CHCOOH, is the simplest truly chiral amino acid and also deprotonates to form alaninate when adsorbed on Cu(110). On this surface it is also found to also adopt a (2x3) phase, even when a single enantiomer is deposited. In this case, it is not possible that the structural phase is heterochiral, as in fig. 1 for glycine, and indeed we have shown through measurements of CDAD [2] (circular dichroism in the angular dependence of photoemission) that the adsorbed species does display the anticipated chirality. The adsorption structure must therefore be homochiral with each (2x3) unit mesh containing two alaninate species which have the same chirality but must have different local adsorption geometries. Using O 1s and N 1s PhD we have therefore conducted a detailed structure determination of the adsorption geometry of alaninate on Cu(110) in this (2x3) adsorption phase. Analysis of the data clearly shows that the local adsorption geometry is the same as for glycine in that the amino N and carboxylate O atoms are in off-atop sites, with the O atoms significantly further from atop than the N atoms, and with Cu-N and Cu-O bonding distances essentially equivalent to those for glycinate. Obtaining a more precise description of the off-atop offset values without

263

prior constraints is, however, extremely difficult, because there must be at least two different offset values for the N atoms and four different offset values for the O atoms, with two inequivalent alaninate species per surface unit mesh. To gain further information on this aspect we have therefore tested the complete set of structural parameters obtained from a recent density functional theory calculation of the minimum energy structure by Rankin and Sholl [3]. Fig. 2 shows a schematic view of their structure, which also includes small relaxations of the outermost Cu layer atoms both perpendicular and parallel to the surface. Fig. 2 Plan view of the Cu(110)(2x3) alaninate adsorption phase as determined by the DFT calculations of Rankin and Sholl [3] We find that this theoretical structure does give a good fit to the experimental PhD spectra, with a slightly lower R-factor than the best-fit structure obtained without this final stage of optimisations, but only if the whole molecular adsorbate layer is displaced towards the surface by approximately 0.10 Å. Without this shortening of the Cu-O and Cu-N bondlengths, the fit to the data is very poor. Specifically, we find N-Cu and O-Cu bondlength values, averaged over the cooccupied sites, of 2.02 Å and 1.98 Å respectively, with an estimated precision of ±0.03 Å. These bondlengths are shorter than those obtained from DFT by 0.08 Å and 0.10 Å respectively. Such a discrepancy in the bondlengths is surprisingly large; DFT calculations commonly give bondlengths accurate to within ~0.02-0.03 Å.

1 J.-H. Kang, R. L. Toomes, M. Polcik, M. Kittel, J.-T Hoeft, V. Efstathiou, D. P. Woodruff and A. M. Bradshaw, J.Chem.Phys. 118 (2003) 6059 2 M. Polcik, F. Allegretti, D.I. Sayago, G. Nisbet , C.L.A. Lamont and D.P. Woodruff, Phys. Rev. Lett. 92 (2004) 236103 3 R.B. Rankin and D.S. Sholl, Surf. Sci. 574 (2005) L1

264

The chemisorption bondlength of molecular water on TiO2(110); a key parameter for theoretical understanding
F. Allegretti1,S. O’Brien1, M. Polcik2, D.I. Sayago2, D.P. Woodruff1
2

Physics Department, University of Warwick, Coventry CV4 7AL, UK Fritz-Haber-Institut der MPG, Faradayweg 4-6, D 14195 Berlin, Germany

1

The (110) face of rutile phase TiO2 is perhaps the most studied of all oxide surfaces as a model system to investigate the range of catalytic applications of this material, of which one of the most interesting is the photochemical production of hydrogen from water, first discovered more than 30 years ago. Most of the very extensive work on the interaction of water with this surface has recently been reviewed [1, 2]. The extent to which H2O does, and should, dissociate to produce surface hydroxyl species on clean and well-ordered TiO2(110), remains a subject of controversy, at least between theory and experiment. However, experimentally it is well-established that molecular water can be adsorbed on this surface intact at low temperatures. Prior to this investigations there has been no quantitative structural information regarding this adsorption, although STM studies have been interpreted in terms of molecular adsorption on the five-fold coordinated (i.e. under-coordinated) Ti atoms at the TiO2(110)(1x1) surface. This is the site for molecular water adsorption which seems to be implicit in theoretical total energy calculations, although most of these are primarily concerned with whether or not dissociation, to produce surface hydroxyl species, occurs on the perfect surface; most of the earlier studies and some very recent ones predict facile dissociation on the perfect stoichiometric surface, a conclusion inconsistent with experimental results. Using scanned-energy mode photoelectron diffraction (PhD) [3] in the O 1s emission we have conducted a quantitative determination of the adsorption geometry of molecular water on TiO2(110) at low temperature [4, 5]. The O 1s signal from the water is clearly resolved from that of the underlying oxide by a large chemical shift, and we found no evidence of dissociation to produce surface hydroxyl species for which there is an intermediate value of the O 1s chemical shift. A defocussed incident synchrotron radiation beam ensured that no radiation damage was seen during the course of the measurements. The structural analysis provides confirmation that the adsorption site is atop the five-fold coordinated surface Ti atoms. In addition, however, the results provide quantitative data on the adsorption bondlength and the surface relaxation. A key finding is that the Ti-Owater bondlength is 2.21±0.02 Å. This is significantly longer that the strong Ti-O chemisorption bondlengths found to formate (COO-) and hydroxyl (OH) coadsorbates on this surface of 2.08 Å and 2.02 Å respectively [6], reflecting a weaker bond of rather different character for the intact water molecule. This bondlength is also much longer that the Ti-O bonds in bulk TiO2 (1.94-1.99 Å). However, the measured Ti-Owater bondlength is rather significantly shorter than the value provided by theoretical calculations of this quantity for water adsorbed in this site on TiO2(110), with

265

published values in the range 2.25-2.41 Å. Unfortunately, despite the many theoretical calculations of this adsorption system, rather few reports of these studies quote values of this adsorption bondlength. Bearing in mind that many of these calculations fail to correctly predict the stability of molecular water to dissociation on this surface it seems likely that the fact that the Ti-Owater bondlength is longer than the experimental value is significant. If the calculations fail to correctly describe the interaction of the intact molecule with the surface it may not be too surprising that they also fail to correctly determine the activation barrier to dissociation. Interestingly, of the four published theoretical values for the bondlength, the one closest to experiment (2.25 Å) corresponds to a calculation which does predict that the molecular species is stable on the surface [7]; it would be of interest to know whether similar near-agreement exists for more recent calculations predicting non-dissociative behaviour. The PhD study also provides some information on the near-surface relaxations of the TiO2 surface in the presence of the adsorbed water. Most notable is the relaxation of the bondlength the Ti surface atom, bonded to the adsorbed water, makes with the O atom directed below in the surface. This relaxation is found to be essentially identical to that found for the clean TiO2(110) surface by both experimental and theoretical methods. The absence of a change in this relaxation provides further evidence of the relatively weak Ti-water bonding.

Schematic diagram of the local adsorption site of molecular water on TiO2(110) as found in this study. The blue spheres represent O atoms, the smaller green sphere represent Ti atoms.

1 U. Diebold, Surf. Sci. Rep. 48 (2003) 53 2 M. A. Henderson, Surf. Sci. Rep. 46 (2002) 1 3 D. P. Woodruff, A. M. Bradshaw, Rep. Prog. Phys. 57 (1994) 1029 4 F. Allegretti, S. O’Brien, M. Polcik, D. I. Sayago, D. P. Woodruff, Phys. Rev. Lett, 95 (2005) 226104 5 F. Allegretti, S. O’Brien, M. Polcik, D. I. Sayago, D. P. Woodruff, Surf. Sci. in press 6 D. I. Sayago, M. Polcik, R. Lindsay, J. T. Hoeft, M. Kittel, R. L.Toomes, D. P. Woodruff, J. Phys. Chem. B 108 (2004) 14316 7 E. V. Stefanovich, T. T. Truong, Chem. Phys. Lett. 299 (1999) 623

266

Dehydrogenation of C4 hydrocarbons using vanadium based catalysts E.M. Vass, D. Teschner, A. Knop-Gericke, M. Hävecker, S. Zafeiratos, P. Schnörch, E. Kleimenov, R. Schlögl Fritz-Haber-Institut der Max-Planck-Gesellschaft, Faradayweg 4-6, D-14195 Berlin Scientific Background and Aims Dehydrogenation of n-butane to butene and butadiene is a highly interesting commercial process due to increasing demand for unsaturated hydrocarbons for industrial applications. Many commercial processes for the dehydrogenation of light alkanes use catalysts containing chromia or platinum supported on alumina. The dehydrogenation of alkanes is an endothermic process, which requires high reaction temperatures and low pressures. Unfortunately at higher temperatures light alkanes are produced and coke formation is encouraged causing catalyst deactivation. Oxygen treatment can be used to regenerate the catalyst, hence prolonging the life-time of the catalyst. This study forms part of the ATHENA project one of the goals of which is to investigate selective gas-phase alkane dehydrogenation.
Figure 1. XPS of V2p3/2 region during n-butane dehydrogenation (2 mbar).

As part of an international

collaboration our aim is to utilise the techniques of high pressure in-situ XPS and NEXAFS to investigate the electronic structure of the vanadium species under reaction conditions. The techniques are used to gain a better understanding of aluminasupported investigation relationships. VxOy of based their catalysts through structure-reactivity

In addition, ex-situ experiments

(typically at 100 mbar pressure) were used to examine the differences in the reaction products and final state catalyst at increased pressure. By performing the experiments in an adjoining reaction chamber, the sample could be transferred under reaction gas atmosphere. Results and Discussion The electronic structure of vanadia/alumina catalysts was examined during the dehydrogenation of n-butane. High-pressure measurements were made possible due to a specially designed, differentially pumped, electrostatic lens system. Hence spectra could be

267

measured while the cell pressure was maintained at 2 mbar. The reaction products were detected using a Proton Transfer Reaction Mass Spectrometer (PTRMS). An 8% V/alumina catalyst was examined both in- and ex-situ for the n-butane dehydrogenation reaction at partial pressures of 0.4 and 100 mbar of butane respectively. After an initial treatment in oxygen, the 8% V/alumina catalyst contained predominantly V2O5 crystallites. This is in agreement with literature, as at vanadium loadings above one monolayer of V on alumina crystalline V2O5 is expected to be the major species
1,2,3

. However, immediately after introduction of butane gas NEXAFS (not shown) reveals This effect is confirmed by the XP spectra, which show a

that the V2O5 structure is lost.

reduction of the vanadium species coupled with a decrease in intensity during the reaction (figure 1). The initial reduction may be due to removal of oxygen from the catalyst in the form of oxygenated products. Furan and dihydrofuran/crotanaldehyde) were observed as the catalyst was heated to reaction temperature. After 87 mins there is a further shift in binding energy and a reduction in intensity. The shift to lower binding energy suggests a decrease in the vanadium oxidation state. This correlates with the formation of a surface carbon species, which is the main source of deactivation for dehydrogenation catalysts.
Figure 2. PTRMS reaction profile for 8% V/alumina during n-butane dehydrogenation and area of surface carbon peak from XPS.
2500 PTRMS Intensity/a.u. 2000 1500 1000 500 500 0 0 50 100 Time/min
Butadiene Furan DHF/Crotanaldehyde Benzene Area of Surface C1s Peak
Temp Ramp 110-450 oC Constant Temp of 450 C
o

2000 Surface C1s Peak Area/a.u.

1500

1000

0 150

Figure 2 shows the trends of the

reaction

products:

butadiene,

furan,

DHF/crotonaldehyde and benzene (butene not shown). There is a steep increase in the surface carbon content, which correlates with a maximum in benzene formation and is proportional to the decrease in butadiene formation. The carbon was identified as chain/graphitic carbon.

268

Therefore investigation of the surface species under reaction conditions has identified two stages. During reaction the vanadium species is reduced to contain a mixture of V5+ and V4+. At this point the catalyst is active and shows only slow deactivation. The second step is the deposition of carbon on the surface, which leads to further deactivation. This deactivation is mainly due to carbon laydown, which blocks the active vanadium sites.
Figure 3. NEXAFS VL3 edge before and after reaction with n-butane (100 mbar).
1.4

As with the in-situ experiment, the ex-situ reaction showed formation of oxygenated products during heating in n-butane. Due to the higher partial pressure of n-butane, the catalyst rapidly deactivated over a period of only 30 minutes. maximum. A steep decline in activity was Hence it is likely that this point Once detected where benzene production reached a coincided with the formation of surface carbon, similarly to the in-situ reaction. transferred to the measurement cell, NEXAFS

V2O5, 350 C, O2 8% V/Al2O3, 350 C, O2
o

o

1.2

0.4 mbar n-butane, 450 C o Gas off, 450 C

o

1.0

516.8 518.5

TEY/a.u.

0.8

0.6

0.4

0.2

(figure 3) showed a great reduction in the vanadium species and XPS indicated high
514 516 518 520 522

0.0 512

Photon Energy/eV

surface carbon content. The total percentage of elemental carbon detected on the catalyst

surface was 46 %. The ratio of vanadium to aluminium decreased on comparison of pre- and post-reaction catalysts, suggesting that carbon laydown occurs preferentially on the vanadium. Acknowledgements The authors would like to thank the staff of BESSY for their support during the measurements. The funding for this work was provided by the ATHENA Project, which is jointly funded by the EPSRC (UK) and Johnson Matthey Catalysts plc. References
1 2

L.J. Burcham, G.T. Deo, X.T.Gao, I.E. Wachs. Topics in Catalysis 11, 85 (2000). N.R. Shiju, M. Anilkumar, S.P. Mirajkar, C.S. Gopinath, B.S. Rao, C.V. Satyanarayana. J. Catal. 230, 484 (2005). 3 Z. Wu, H.-S. Kim, P.C. Stair, S.R. Rugmini and S.D. Jackson. J. Phys. Chem. B 109, 2793 (2005).

269

Catalytically Active States of Ru(0001) Catalyst in CO Oxidation Reaction
R. Blume1, M. Hävecker1, S. Zafeiratos1, D. Teschner1, E. Kleimenov1, A. Knop-Gericke1, R. Schlögl1, A. Barinov2, P. Dudin2, and M. Kiskinova2*
† §

Fritz-Haber-Institut der Max-Planck-Gesellschaft, Faradayweg 4-6, 14195 Berlin, Germany Sincrotrone Trieste, AREA Science Park-Basovizza, Trieste-34012, Italy

1. Introdiction The recent XPS microscopy and TDS studies clearly showed that the formation of a rutile RuO2 phase, starting from an atomically clean Ru(0001) surface, is kinetically hindered at temperatures lower than 500 K and readily occurs at temperatures higher than 550 K 1 . An important finding is that ‘surface oxide’, which forms by incorporation of O atoms below the top Ru layer and the RuO2 can coexist in a wide T-pressure range, even when formed in pure O2 ambient 2 . Undoubtedly, when CO oxidation reaction takes place, the CO will drive the oxidation state away from the equilibrium achieved in O2 ambient, which implies that the T-p space of coexistence of the two structures may be expanded. The temperature dependence of the actual ‘oxidation’ state and the complex morphology of the Ru surface evidenced by XPS microscopy have reopen the disputable issue about the active state of Ru catalysts during CO oxidation 3,4,5 . We verified the catalytic activity of the different oxidation states of Ru(0001) catalyst, starting from a metallic Ru surface and following in-situ the temperature evolution of the catalyst surface composition and yield of CO2 during CO oxidation reaction carried out close to the realistic reaction conditions. 2. Experimental The experiments were performed in the high pressure XPS station designed and constructed in FHI-MPG, attached to the beamline U49/2-PGM2 at BESSY 6 . The XPS spectra, were measured in-situ using a set-up combining differential pumping and electrostatic focusing of the emitted photoelectrons and simultaneously the CO2 yield was monitored by a mass spectrometer. The Ru(0001) sample was cleaned before each reaction cycle using the well established procedures of alternating Ar ion bombardment and oxidation-annealing cycles. Photon energies 450 eV and 650 eV were used for monitoring the Ru 3d and O 1s spectra, respectively. 3. Results The dynamic response of the O 1s and Ru 3d5/2 core level spectra was used for precise assignment of the catalyst oxidation state in the course of the reaction correlated to the corresponding CO2 yield. The already available Ru 3d5/2 and O 1s core level spectroscopy data provided the necessary basis for identification of the adsorption, ‘surface oxide’ with incorporated oxygen, and stoichiometric RuO2 states and verifying their actual role in CO oxidation reaction 1,7 . We started from a clean Ru(0001) surface and followed the changes after introducing 0.1 mbar CO+O2 (O2:CO partial pressure ratio 1) and slowly increasing the temperature. The excess of oxygen with respect to the reaction stoichiometry provided slightly oxidizing conditions ensuring the formation of the different Ru oxidation states. Fig. 1 (a) shows the CO2 yield as a function of the reaction temperature. There is a clear sharp onset of the reaction at ~ 420 K, the reaction rate increasing continuously in the temperature range 420-500 K. The selected set of Ru 3d5/2 and O 1s spectra in Fig. 2, measured at different reaction temperatures, represent the milestones in the evolution of the catalyst
*

M. Kiskinova thanks AvH foundation for the Award to pursue research in FHI-Berlin in 2004-2005.

270

oxidation state. The first Ru 3d5/2 and O 1s spectra represent the status of the Ru catalyst after the onset of CO2 production. The Ru 3d5/2 spectra of this ‘low temperature’ state require the characteristic components of the ‘surface oxide’, Ru(II)-Osub and Ru(I)-2OadOsub, where Ru(I) and Ru(II) account for emission from the Ru atoms in the 1st and 2nd layer, bonded to oxygen residing on the surface, Oad, and below the surface, Osub1. The Ru 3d5/2 and O 1s spectra undergo negligible lineshape changes in the temperature range 420-480 K despite the gradual increase of the CO2 yield. They resemble those measured for the ‘surface oxide’ with ~ 2 ML oxygen, but with less oxygen on the surface (the Ru(II)-Osub component at 0.5 eV is dominant). Apparently the adsorbed oxygen is kept low, since it is very effectively consumed by the ongoing reaction. The very fast dynamics at the surface is confirmed by the absence of a COrelated feature in the O 1s spectra at binding energy > 531. 0 eV (Fig.2), indicating that the lifetime of the CO on the surface before being reacted is shorter than that of the O species.

Fig. 1 (a) CO2 yield as a function of reaction temperature. (b) Plot of the CO2 yield versus O content at the surface and near surface region. The dashed line in (a) - (c) indicates the onset of the RuO2 growth.

Natural consequence of the progressive incorporation of oxygen with further increasing the reaction temperature is the nucleation and growth of stoichiometric RuO2. This is manifested by the significant changes of the Ru 3d5/2 spectra in Fig. 2 undergone above 500 K, until a ‘steady-state’ composition is reached and maintained in the 550-600 K range, when the ‘oxide’ component is dominant in the Ru 3d5/2 spectra. Rubulk component can still be distinguished in the Ru 3d5/2 spectrum of this ‘steady-state’ as well as the O1s component corresponding to the ‘surface oxide’. This indicates a patchy structure consisting of RuO2 islands and ‘surface oxide’ areas, as reported in ref. 2, reflecting the kinetic limitations imposed by the presence of CO. The most striking result is that the growth of the RuO2 phase above 500 K does not affect the monotonous increase of the CO2 yield (Fig. 1(a)), suggesting that the nucleation and growth of the oxide phase barely affects the reaction barrier. The plot of the CO2 yield vs O content in Fig. 1(b) is the best illustration that the high catalytic activity of the Ru catalyst is not exceptionally correlated to the formation of RuO2 with a well defined surface structure. It clearly shows that the ‘surface oxide’ formed via progressive incorporation of oxygen already exhibits high catalytic activity and there is no significant increase with the formation of stoichiometric RuO2. Here, it should be noted that since the formation of RuO2 occurs above 500 K the temperature effect on the reaction rate should be also taken into account when comparing the catalytic activity of the ‘surface oxide’ and RuO2. 4. Concluding remarks The present results demonstrate that one cannot draw a clear line between the catalytical activity of the stoichiometric RuO2 phase and a few layers thick not well-ordered ‘surface oxide’. Our findings are in qualitative agreement with the theoretical predictions 8 that the

271

catalytically active region under realistic dynamic reaction conditions can often lie at the boundary between two structures. Present results also apply well to real Ru catalyst systems which are nano-particles. The ‘oxidized’ states of these Ru nanoparticles, often described as RuxOy or ultra-thin Ru oxide films covering the metallic core, are comparable to the ‘surface oxide’ with subsurface oxygen rather than with the well-structured RuO2(110) surface 9,10 .

Fig. 2. Ru 3d5/2 and O 1s spectra illustrating the catalyst composition developed during CO oxidation with increasing of the reaction temperature from 370 to 600 K. dT/dt = 2 K/min. Reaction conditions: PCO = 0.5x10-1 mbar, PO2 = 0.5x10-1 mbar. The Ru(II)-Osub and Ru(I)-2OadOsub components are labelled as ‘ad’ and ‘sub’ in the Ru 3d5/2, respectively. The component labelled ‘RuxOy’ correspond to amorphous film which is precursor to growth of the RuO2 islands, characterised by the component ‘ox’1. The component used as the zero-energy reference corresponds to metallic bulk Ru at binding energy at 280.1 eV. In the O 1s panel the component at 530.0 corresponds to adsorbed and ‘surface oxide’ phase, whereas the one at 529.5 eV to oxygen in the RuO21,3. The position of the O1s component corresponding to CO is indicated as well. It is observed only at temperatures below the real onset of CO2 production.

References
R. Blume, H. Niehus, H. Conrad, A. Böttcher, L. Aballe, L. Gregoriatti, A. Barinov, M. Kiskinova, J. Phys. Chem. B 109 (2005) 14052. 2 A. Böttcher, U. Starke, H. Conrad, R. Blume, L. Gregoriatti, B. Kaulich, A. Barinov, M. Kiskinova, J. Chem. Phys. 117 (2002) 8104. 3 H. Over, Y.D. Kim, A.P. Seitsonen, E. Lundgren, M. Schmid, P. Varga, A. Morgante, G. Ertl, Science 287 (2000) 1474. 4 J. Wang, V.Y. Fan, K. Jacobi, G. Ertl, J. Phys. Chem. B 106 (2002) 3422. 5 R. Blume, H. Niehus, H. Conrad, A. Böttcher, J. Phys. Chem. 108 (2004) 14332. 6 H. Bluhm, M. Hävecker, A. Knop-Gericke, E. Kleimenov, R. Schlögl, D. Teschner, V.I. Bukhtiyarov, D.F. Ogletree, M. Salmeron, J. Phys. Chem. B. 108 (2004) 14340. 7 H. Over, A.P. Seitsonen, E. Lundgren, M. Wiklund, J.N. Andersen, Chem. Phys. Lett. 342 (2001) 467. 8 K. Reuter, M. Scheffler, Phys. Rev. Lett. 90 (2003) 46103; Phys. Rev. B 60 (2003) 45407. 9 V. Narkhede, J. Assmann, M. Muhler, Z. Phys. Chem. 219 (2005) 979. 10 W. Vogel, N. Alonso-Vante, J. Cat., 232 (2005) 395.
1

272

Steps make the difference – CO oxidation on Pt(355) and Pt(322) investigated by in-situ high resolution XPS B. Tränkenschuh, C. Papp, T. Fuhrmann, R. Denecke and H.-P. Steinrück Lehrstuhl für Physikalische Chemie II, Universität Erlangen-Nürnberg, Egerlandstr. 3, 91058 Erlangen The CO oxidation on differently stepped platinum surfaces was studied in-situ using a combination of a supersonic molecular beam and high-resolution x-ray photoelectron spectroscopy (HRXPS). The Pt(355) and Pt(322) surfaces are vicinal surfaces with five atom rows wide (111) terraces. The structural difference of the two surfaces is the orientation of the monatomic steps, being (111) and (100), respectively. The experiments were performed at beamline U49/2-PGM1, using a transportable apparatus described elsewhere [1]. It combines HRXPS with a supersonic molecular beam for gas dosing at well-defined pressures. In our study, first a layer of atomic oxygen was prepared, and subsequently its reaction with CO was followed by time-dependent XPS at distinct temperatures. O 1s and C 1s spectra were acquired in-situ to distinguish the different species and adsorption sites (O and CO on step and terrace sites) and their surface coverages. The product (CO2) desorbs at reaction temperature and was detected by a quadrupol mass spectrometer. Fig. 1 (a) shows O 1s XP spectra of atomic oxygen layers (~0.28 ML) before reaction on both surfaces. One can clearly see that on Pt(322) two different oxygen species are present, whereas on Pt(355) only one peak is detected, although part of the O is adsorbed close to the step edge on both surfaces. In Fig. 1 (b) the C 1s XP spectra of CO saturation coverages on both surfaces are shown, depicting the situation after complete reaction at 260 K. The obvious difference between both surfaces is the amount of CO observed in adsorption sites on the platinum steps. On Pt(322), a smaller CO coverage occupies two step sites, while on Pt(355) only one step site with a higher occupation is observed [2]. By continuously recording such spectra during reaction and merging all quantitative information it is possible to follow the occupation of all these sites on the time scale of seconds. The reactions are carried out at high CO pressures (1*10-6 mbar), so that the reaction rate becomes independent of the impinging flux of CO molecules.
O/Pt(355)
O 1s 260 K oxygen

CO/Pt(355)
C 1s 260 K

terrace bridge

terrace on-top

step on-top terrace hollow

Intensity [a.u.]

Intensity [a.u.]

O/Pt(322)
oxygen 2

CO/Pt(322)
terrace on-top

terrace bridge step on-top step bridge

oxygen 1 (a)
(b)

536

534 532 530 Binding Energy [eV]

528

289

288 287 286 285 Binding Energy [eV]

284

Fig. 1: (a) O 1s XP spectra of the atomic oxygen layers on Pt(355) (upper part) and Pt(322) (lower part). (b) C 1s XP spectra for CO saturation (after reaction at 260 K) on Pt(355) (upper part) and Pt(322) (lower part). Sites labelled “terrace hollow” on Pt(355) are only occupied at high CO coverages (obtained here at CO pressures of 1*10-6 mbar).

273

Low temperature (~130 K) co-adsorption experiments of O and CO (data not shown) reveal that on both surfaces oxygen covers the platinum steps, blocking these sites for CO. Therefore, if the occupation of step sites by CO is observed in C 1s spectra during reaction, the oxygen has to be reacted away from the steps. This reaction, indeed, occurs very quickly on both surfaces, as signalled by early CO saturation of step sites. The further reaction path is not that clear because there is a fast exchange of CO adsorbed on step and terrace sites above 200 K, shown by experiments with isotopically marked CO [3]. For both surfaces the CO oxidation was studied at temperatures between 220 and 300 K. Since the reaction rate is independent of CO pressure at the experimental conditions (see above), we are able to evaluate the rate constant only from the oxygen coverage decrease as a function of reaction time. In Fig. 2 the oxygen coverage is plotted versus reaction time for both surfaces. The decrease of oxygen is faster, i.e., the reaction rate is higher for higher temperatures. Comparing both figures, the reaction occurs faster on Pt(355) (Fig. 2a) than on Pt(322) (Fig. 2b) for all temperatures. Fitting a suitable rate law (assuming surplus CO) to the oxygen decrease, it is possible to determine the activation energy for the reaction of oxygen with CO by an Arrhenius evaluation. In this way, a value of 530 meV was determined by a similar experiment for CO oxidation on Pt(111) [4]. The situation on the stepped surfaces is more complex, with at least two different reaction channels, as suggested by the change of slopes observed in the curves of Fig. 2. While we have hints that the fast channel at the beginning of the experiment is dominated by the reaction at steps, the lower reaction rate observed in the further course of the experiments might not only be attributed to CO oxidation on the (111) terraces. However, the overall reaction rate, just given by the time needed to reduce the O coverage by a certain amount, is on both stepped surfaces significantly higher than on Pt(111). This clearly demonstrates the influence of steps on surface reactivity. A detailed analysis to determine the reaction pathways is in progress. Work was supported by DFG (Ste620/4-2). [1] [2] [3] [4] R. Denecke, M. Kinne, C.M. Whelan, H.-P. Steinrück, Surf. Rev. Lett. 9 (2002) 797. B. Tränkenschuh, N. Fritsche, T. Fuhrmann, C. Papp, J.F. Zhu, R. Denecke, H.-P. Steinrück, J. Chem. Phys. 124 (2006) (in print). A. Szabo, M.A. Henderson, J.T. Yates, Jr., J. Chem. Phys. 96 (1992) 6191. M. Kinne, T. Fuhrmann, J.F. Zhu, C.M. Whelan, R. Denecke, H.-P. Steinrück, J.Chem. Phys. 120 (2004) 7113.
(a)

Pt(355)

Oxygen Coverage [ML]

0.3
300 K 275 K 260 K 240 K 220 K

0.2

0.1

0.0 0 50 100 150 200 Reaction Time [s] 250 300

(b)

Pt(322)
300 K 275 K 260 K 250 K 240 K 225 K

Oxygen Coverage [ML]

0.3

0.2

0.1

0.0 0 50 100 150 200 Reaction Time [s] 250 300

Fig. 2:.Oxygen coverages vs reaction time plotted for different reaction temperatures (a) for Pt(355) and (b) for Pt(322).

274

A detailed analysis of the adsorption of benzene on Ni(111) with HR-XPS C. Papp, B. Tränkenschuh, T. Fuhrmann, R. Denecke and H.-P. Steinrück Lehrstuhl für Physikalische Chemie II, Universität Erlangen-Nürnberg Benzene is the prototype of an organic molecule with an aromatic system and, despite its size, it can be described rather simply because of its high symmetry. Aside from fundamental interests, benzene may occur as an intermediate in the nickelbased heterogeneously catalysed petroleum reforming process. Nickel is a particularly relevant substrate, as it plays an important role in various heterogeneously catalysed reactions (1). We recorded C 1s C6H6 (Θ=0.143 ML) spectra with an excitac adiab. peak 1 320 1st exc. state 1 β= 0.5 K/s tion energy of 380 eV at b 280 beamline U49/2-PGM1 in a time-resolved man240 ner, while adsorbing benzene with a pres286.0 285.0 284.0 283.0 200 1.0 sure of 2*10-9 mbar on Binding Energy [eV] a 0.8 100 a Ni(111) crystal. This C6H6 (Θ=0.08 ML) 0.6 d adiab. peak 1 type of experiment is 1st exc. state 1 50 0.4 adiab. peak 2 called “uptake” and is 1st exc. state 2 0.2 0 shown in Fig. 1a as a 0.0 colour-coded density 286.0 285.0 284.0 283.0 plot. We also performed Binding Energy [eV] experiments, 286.0 285.0 284.0 283.0 TP-XPS Binding Energy [eV] shown in Fig. 1b, by heating the sample with Fig.1: a) Benzene (C6H6) adsorption experiment at 200 K; a fixed rate while inset: intensity scale. b) TP-XPS experiment, both as colour- measuring every 10 K. coded density plots. c) C 1s spectrum and fit for a saturated The apparatus used is described in detail elselayer at 200 K; d) same for a dilute layer. where (2).
Exposure [L] T [K]

On the right side of Fig. 1 representative C 1s spectra and their corresponding fits are shown. In Fig. 1c two peaks at 284.3 and at 284.7 eV are needed to model the spectrum for saturation coverage. These are assigned to an adiabatic transition at 284.3 eV and a final state with excitation of the C-H stretching mode at 284.7 eV. This assignment was proven by an isotope exchange experiment with C6D6 (data not shown). These two peaks have a constant intensity ratio, called S-factor (3), and a constant binding energy difference over the whole coverage range. The C 1s spectra at lower coverages show an additional shoulder at 283.9 eV (Fig. 1d). This shoulder originates also from benzene molecules and is, thus, again fitted with two peaks, with the same parameters as the saturated layer, including binding energy difference, line width and S-factor. The peak areas of the corresponding adiabatic and the first vibrationally excited states are added, as they belong to different carbon components, C1 and C2, within the benzene molecules. In Fig. 2a, we see these carbon components increasing up to a benzene coverage of ~0.09 ML; corrected for photoelectron diffraction effects, a value of 0.10±0.01 ML is obtained from a larger data set. For higher coverages up to saturation of 0.143 ML, the C1-component looses its intensity completely to component C2. The dotted lines show the data for the TP-XPS experiment of Fig. 1b, demonstrating the reversibility within this coverage or temperature range.

Intensity %

275

The red lines and symbols show a similar experiment with deuterated benzene (C6D6), leading to the same result. In Fig. 2c we show schematic drawings of benzene adsorbed at the sites discussed in this context (4). A benzene molecule adsorbed on the hollow site, with C-C bonds along the [1 1 0] direction of the substrate, is depicted on the right hand side. In this geometry, which is reported for saturation coverage (4), all six C atoms are equivalent and they are pair wise attached to one Ni atom. Therefore, we only detect one signal in the C 1s spectrum (C2). For benzene adsorbed on bridge sites (see Fig. 2c) with the CC bonds aligned along the [2 1 1] direction of the substrate, we find again pair wise coordinated carbon atoms but also singly coordinated carbon atoms (C1) (4). The ratio of the two different carbon components, C1/C2, in the bridge adsorption site is 0.5, which is well matched by the ratio of 0.45 found at coverages up to ~0.09 ML (see Fig. 2b). The ratio C1/C2 for the hollow adsorbed benzene should be zero, as only component C2 is present, and, indeed, a value of zero is found at saturation coverage. At coverages between 0.09 and 0.143 ML the rapidly changing ratio is interpreted as a coverage dependent site change of benzene from bridge to hollow in case of adsorption, and from hollow to bridge in case of a TP-XPS experiment.
Partial Coverage [ML] 0.12 0.08

C6H6 C6D6

a

C2

C1
0.04 0.00

Ratio (C1 / C2)

0.6 0.4 0.2 0.0

b

C6H6 C6D6
bridge hollow

Partial Coverage [ML]

0.12 0.08 0.04 0.00

c

0.04 0.08 0.12 Total Coverage [ML]

Fig. 2: a) Comparison of adsorption and thermal desorption experiments of C6H6 and C6D6. b) Ratio of the two components for C6H6 and C6D6. c) Occupation of adsorption sites during adsorption at 200 K. (Dotted lines mark results of TP-XPS experiments).

In summary, we performed a detailed analysis of the coverage-dependent site preference of benzene on a Ni(111) single crystal surface. Our quasi-continuous measurement of the adsorption of benzene (C6H6 and C6D6), summarized in Fig. 2c, shows up to a coverage of 0.10±0.01 ML only bridge adsorbed benzene. For higher coverages we find the beginning of a coverage-dependent site change, accompanied by reorientation of the molecules (5), resulting in a mixture of hollow and bridge adsorbed benzene. At saturation coverage only hollow adsorbed benzene is found on the surface. Our desorption experiment showed, that this site change is reversible. This work was supported by the DFG through grant Ste620/4-2. 1. G. A. Somorjai, Introduction to Surface Chemistry & Catalysis (Wiley, New York, 1994). 2. R. Denecke, M. Kinne, C. M. Whelan, H.-P. Steinrück, Surf. Rev. Lett. 9, 797-801 (2002). 3. S. J. Osborne et al., J. Chem. Phys. 106, 1661-1668 (1997). 4. O. Schaff et al., Surf. Sci. 348, 89-99 (1996). 5. H.-P. Steinrück, W. Huber, T. Pache, D. Menzel, Surf. Sci. 218, 293-316 (1989).

276

Combined application of XPS, XANES and mass-spectrometry to in-situ study of methanol oxidation over vanadium based catalysts
1

V.V. Kaichev1, V.I. Bukhtiyarov1 Boreskov Institute of Catalysis, Lavrentieva prosp., 5, 630090, Novosibirsk, Russia

2

D.Yu. Zemlyanov2,3, S. Belochapkine2, B.K. Hodnett2 Materials and Surface Science Institute and Physics Department, University of Limerick, Limerick, Ireland 3 Birck Nanotechnology Center, Purdue University, West Lafayette, IN 47907-2057, USA E. Kleimenov4, D. Teschner4, S. Zafeiratos4, M. Hävecker4, A. Knop-Gericke4, R. Schlögl4 4 Abteilung Anorganische Chemie, Fritz-Haber-Institut der Max-Planck-Gesellschaft, Faradayweg 4-6, D-14195 Berlin, Germany

Vanadium-based systems are widely applied as catalysts for selective oxidation of hydrocarbons, alcohols, etc. For example, V-P-O oxide catalysts convert n-butane to maleic anhydride [1], V-Ti-O mixed oxides – beta-picoline to nicotinic acid [3]; V-P-O/TiO2 catalysts are active in methylpyrazine ammoxidation [2]. Oxidation of methanol to formaldehyde (or to methyl formate) and formaldehyde to formic acid over V2O5/TiO2 catalysts attracts a special attention of researchers due to its practical importance [4-6]. Despite numerous reports, the exact mechanisms of these reactions are not clear yet. In this report we present the results of in situ study of methanol and propane oxidation over V2O5/TiO2 catalysts performed with X-ray photoelectron spectroscopy (XPS) and near edge X-ray absorption fine structure (NEXAFS). Catalytic properties of the catalysts studied were tested simultaneously using mass-spectrometry. All experiments were performed at beam line U49/2-PGM1 at BESSY. The spectrometer was equipped with a special gas cell which allowed increase in pressure for in situ XPS and NEXAFS measurements up to 2 mbar. The gas flows into the experimental cell were regulated using calibrated mass-flow controllers. Before experiments the catalyst powders were pressed to pellets and then mounted on a temperature-controlled heating stage. The sample temperature was measured using a chromel-alumel thermocouple pressed directly at the back of the sample. The overall spectral resolution was 0.1 eV at the Oxygen K-edge. All spectra were normalized by the incident photon flux, which was measured using a photodiode with known quantum efficiency. XPS spectra were calibrated against C1s line from adventitious carbon (284.8 eV). To extract the information about chemical states of the elements, the narrow regions of their core level spectra have been measured, original XPS spectra being decomposed on separate components. The latter procedure involved Shirley background subtraction and a curve fitting using Doniach-Sunjic functions. Vanadia-titania catalysts 20V2O5 – 80TiO2 (wt.%) were prepared by spraying titanium dioxide (anatase) suspension in an aqueous vanadyl oxalate solution followed by calcination of the obtained powder in air at 400°C [6]. The specific surface area of the catalyst was equal to 140 m2/g. In spite of the high content of V2O5, X-ray diffraction (XRD) analysis indicates that the catalysts contains only TiO2 anatase phase. At the same time, transmission electron microscopy (TEM) shows the presence of

277

the V2O5 nanocrystals (d = 1-2 nm) located on the surface and inserted between small (3-8 nm) anatase particles, which are joined into aggregates with irregular shape. This sample exhibits high activity in methanol oxidation to formaldehyde and in formaldehyde oxidation to formic acid [6]. This sample was used as an object for our in situ experiments which include the step-wise heating of the sample from 50°C to 150°C under near-equimolar CH3OH/O2 mixtures at total pressure of ∼ 0.1 mbar. MS, XPS and NEXAFS spectra were measured simultaneously at three different temperatures: 50, 90, 110 and 150oC. The corresponding XPS spectra are shown in Fig.1. Before in situ experiments were started, the catalyst was activated in 1 mbar of oxygen at 300°C for 30 min directly in the gas cell. This led to full oxidation of vanadia and removal of any carbon-contained impurities. Only sharp single feature at 517.6 eV, which corresponds to V5+ ions, is observed in the V2p3/2 spectrum (fig.1). Under influence of the CH3OH/O2 mixtures, vanadium(V) ions are reduced to V4+ state that is identified on

V

3+

V

4+

V

5+

7 Intensity [arb. un.] 6 5 4 3 2 1
512 514 516 518 520 522

B in d in g E n e r g y [e V ]
Fig.1. V2p3/2 core-level spectra from the V2O5/TiO2 catalyst obtained in situ: 1 – under 1 mbar oxygen at 300°C; 2-5 – under near-equimolar CH3OH/O2 mixtures at pressure of ∼0.1 mbar at temperature 50, 90, 110 and 150°C, respectively; 6 – after switch off methanol flow at 150°C; 7 - similarly after switch off oxygen flow at 150°C. appearance of wide V2p3/2 peak at 516.4 eV. Increasing the temperature led to partial oxidation of vanadium(IV) ions and two features at 516.4 ± 0.1 eV (V4+) and 517.6 ± 0.1 eV (V5+) are observed in the V2p3/2 spectra (fig.1). It should be noted, that the fraction of V5+ ions are increased constantly with temperature. Significant part of V4+ ions are remained on the catalyst surfaces even after removal of

278

methanol form the gas phase. On the other hand, removal of oxygen at 150°C results in further reduction of vanadium and two features at 515.6 and 516.6 eV, which can be attributed with V3+ and V4+, respectively, are observed in the V2p3/2 spectrum (fig.1). Thus, our data unambiguously show that lattice oxygen of vanadium oxide takes a part in the methanol oxidation via Mars-van Krevelen mechanism, which consists of reduction of the oxide catalyst surface by methanol and subsequent reoxidation by gas phase oxygen. The reduced V4+ ions are believed to be the active sites involved in the Mars-van Krevelen redox cycles. It has been also found that surface composition of the V2O5/TiO2 catalysts is changed as a function of the reaction atmosphere and temperature. At 150oC titanium signal disappears from the surfacesensitive spectra, but it is still detectable in the bulk-sensitive spectra. This result can be explained by the accumulation of carbonaceous species selectively on the low-active titania surface or by the redistribution of the elements so that the titania surface is covered with vanadium. The former concept seems to be more preferable, because recently some author observed agglomeration of the dispersed vanadia during methanol oxidation over V2O5/SiO2 catalysts [7]. Acknowledgement. This work was supported by the European Community - Research Infrastructure Action under the FP6 "Structuring the European Research Area" Programme (through the Integrated Infrastructure Initiative" Integrating Activity on Synchrotron and Free Electron Laser Science Contract R II 3-CT-2004-506008).V.I.B. and V.V.K. gratefully acknowledge the Max-PlanckGesselschaft for financial support of the visit to FHI and work at BESSY. We gratefully acknowledge the BESSY staff for the support during beamtime. References: 1. B.K. Hodnett, Heterogeneous Catalytic Oxidation, Wiley, New York, 2000, Chapter 5. 2. V.M. Bondareva, T.V. Andrushkevich, O.B. Lapina, A.A. Vlasov, L.S. Dovlitova, Reac. Kinet. Catal. Lett. 79 (2003) 165. 3. E.M. Al'kaeva, T.V. Andrushkevich, G.A. Zenkovets, G.N. Kryukova, S.V. Tsybulya, Catal. Today 61 (2000) 249. 4. P. Forzatti, E. Tronconi, A.S. Elmi, G. Busca, Appl. Catal. A 157 (1997) 387. 5. E. Santacesaria, A. Sorrentino, R. Tesser, M. Di Serio, A. Ruggiero, J. Mol. Catal. A 204-205 (2003) 617. 6. G. Popova, T.V. Andrushkevich, I. Zakharov, Yu. Chesalov, Kinet. Catal. 46 (2005) 217.

7. T. Feng, J.M. Vohs, J. Phys. Chem. B 109 (2005) 2120.

279

The Local Adsorption Geometry of Alanine on the Chiral Cu{531} Surface studied by NEXAFS
M. J. Gladys1, A. Stevens1, N. Scott1, D. Batchelor2 and G. Held1
1 2

University of Cambridge, Department of Chemistry, Lensfield Rd, Cambridge CB2 1EW, UK. BESSY and Universität Würzburg, Experimental-Physik II, Am Hubland, Würzburg, Germany.

Chiral (enantiomeric) molecules often exhibit drastically different reactivity in living matter. The ‘wrong’ enantiomer of a chiral molecule may have such damaging side effects that any beneficial properties of the ‘right’ enantiomer are completely overshadowed. The importance of producing single enantiomer products cannot be understated and beyond the pharmaceutical sector, chiral modified surfaces have considerable appeal in the production of flavours and fragrances, microelectronics, and magnetic components [Barl03,Coll95]. This report details experiments on the adsorption of the chiral amino acid alanine (OOC– C*HNH2–CH3 ) on chiral copper surfaces that may exhibit enantiospecific selectivity, which can be used in molecular sensors or heterogeneous catalysis. Specifically, synchrotron techniques such as High Resolution XPS and NEXAFS, using variable polarised X-rays provide accurate determination of bonding geometry and the chemical nature of metalorganic interfaces. It was established that certain single crystal fcc surfaces with high Miller indices are chiral because they contain two inequivalent types of steps constituting a kink site which has no mirror symmetry [Atta01]. The goal of our studies is to find the correct combination of substrate unit cell and organic molecule that will combine to produce stereo-selectivity. We have chosen the Cu{531} surface which contains the smallest chiral unit cell, with only three atoms dividing the step kinks on the surface. This surface incorporates {110} and {311} step kinks and directly correlates with the knowledge gained from the investigations on Cu{110} in which the alanine adsorbs in a three point bonding geometry [Jone06]. The experiments were performed at beamline UE52-PGM (CRG) of BESSY II, which has a spot size of about 100 x 200 at the sample. The base pressure of the UHV endstation was 3x10-10 mbar; all data were recorded at room temperature. XPS data were recorded using a Scienta 200 mm electron energy analyser with pass energies of 20 and 40 eV at a photon energy of 630 eV. The binding energies (BE) were calibrated with corresponding measurements at the Fermi energy for the same photon energy and pass energy. For the NEXAFS experiments the synchrotron beam hit the surface either at normal incidence or at 70˚ off normal incidence, the orientation of the electrical field vector, E, within the surface plane was controlled by the undulator settings to be parallel or perpendicular to the [1-21] direction (see Figure 2) as well as angles of 15o, 45o and 65o from parallel (which is now a permanent feature at UE52-PGM). A partial yield detector (PYD) was used with the retarding voltage set to accept electrons in the kinetic energy range up to 50 eV below the lowest photon energy in order to avoid detection of Cu d-band photoelectrons. The raw C, N, and O K-edge NEXAFS data were normalised with respect to the ring current and spectra of the clean sample. The Cu sample was prepared using standard procedures including electro-polishing, Ar-ion sputtering and oxygen treatment in UHV followed by a final annealing step to 1000 K. R and S-alanine was adsorbed by evaporating a coverage in excess of the amount actually needed at a sample temperature of 300K. After the adsorption was complete the sample was annealed to about 400K to produce the (1x4) ordered structure as determined by LEED. The

280

LEED pattern of the R-alanine overlayer indicates a lesser degree of order than that observed with S-alanine, for which the data are shown here. XPS spectra of C1s, O1s and N1s core levels (not shown) were identical for R and Salanine layers prepared under the same conditions. Narrow O1s and N1s XPS peaks were found at BE 531.3 and 399.5eV, respectively. The absence of a second O 1s peak indicates that both oxygen atoms in the molecule are involved in the bond formation with the substrate and the molecule is in its deprotonated alaninate form. The C 1s spectra show two peaks at BE 288.0 and 285.5eV. According to [Hass98,Barl04,Jone06] we assign the first peak to the OOC– carbon atom of the carboxylate group and the second, more intense, peak to the two – CHNH2–CH3 carbon atoms, which only take part in C-C single bonds. The XP spectra are very similar to data for alanine on Cu{110} [Jone06]. Decomposition of the alanine layer occurs around 470K where an abrupt change in all core levels is observed. The raw NEXAFS spectra for the Carbon K-edge are shown in Figure 1, for the inplane E orientations listed above. There is a clear polarisation dependence of the π* resonance at 289eV for in-plane angles. The polarisation dependence of the * resonance for the 70o NEXAFS data plainly show that the O-C-O triangle is tilted with respect to the surface normal [Stoe92]. The finding that the molecules form bonds with the Cu substrate through three atoms, two O and one N, explains the tilt of the O-C-O group and is in good agreement with results from recent DFT calculations [JoneXX]. Figure 2(a) shows the variation of the π* resonance as the linear polarization is rotated within the surface plane. If only a single adsorption site was occupied the intensity of the resonance should follow a cos2θ relation, θ being the angle between the polarisation vector and the normal of the O-C-O triangle [Stoe92], and the intensity at the minima should be zero. The fact that the intensity does not decrease to zero indicates the existence of multiple molecule orientations on the surface. DFT calculations by [JoneXX] show little difference in the cross sections of the two molecules. It can therefore be assumed that each of the differently oriented molecules will follow a cos2θ relation with a similar pre-factor. Using circular polarised synchrotron radiation as an average over all angles, we are able to normalise the spectra and determine the maximum of the cos2θ dependencies. Assuming two orientations of the molecules, rotated by about 75° with respect to each other, provides an excellent fit to the data, as shown in Figure 2(a). Combining HRXPS, NEXAFS and LEED results, suggests that there is only room for two molecules within the (1x4) unit cell, one binding to a {110} and the other to a {311} step as shown in Figure 2(b). It should be noted that without the newly acquired ability to measure the carbon K-edge at angles between the horizontal and vertical, determination of the exact angles would be impossible. A comparison of in-plane polarisation dependences of the π* resonance for R and Salanine show a sizeable deviation. The minimum is at the same angular position, however the intensity variation is larger. This indicates that the orientations of the two types of molecules are different to that of the S-alanine case, which may be the result of slightly different stabilities of the two systems due to different degrees of inter-molecular hydrogen bonding. This study was supported by the European Commission under the Contract No R II 3-CT-2004-506008 / project BESSY-ID.05.2.090 and the EPSRC. We would like to thank the BESSY staff for their support during the beamtime.

281

vert Horz 67.5 45 15

275

280

285

290

295

300

305

310

315

320

Beam Energy (eV)

Figure 1: Carbon K-edge spectra for S-Alanine on Cu{531} for in-plane polarisation at different azimuthal angles between and including the horizontal and vertical angles.

0.8

π * intensity (arbitrary)

0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 -30 -10 10 30 50 70 90 110 130 150 170 E-vector angle from Horizontal

[1-21] (b

(a)

Figure 2: (a) The intensity of the carbon π* resonance for the ordered (1x4) structure of S-Alaninate on Cu{531}. The experimental results (squares) can be fitted by including two molecular orientations (bottom curves) of Alaninate on the surface. (b) The two orientations of the molecule adsorb onto the {110} and {311} steps.

References:
[Atta01] [Barl03] [Barl05] [Coll95] [Hass98] [Jone06] [JoneXX] [Stoe92] G. Attard, J.Phys. Chem. B, 105, (2001) 3158. S.M. Barlow and R. Raval. Surface Science Reports 50, (2003) 201. S. M. Barlow et al. Surf. Sci. 590 (2005) 243. A. N. Collins, G. N. Sheldrake and J. Crosby (Eds), Chirality in Industry: The Commercial Manufacture and Applications of Optically Active Compounds (John Wiley, 1995). J. Hasselström et al. Surf. Sci. 407 (1998) 221. G. Jones, L.B. Jones, F. Thibault-Starzyk, E.A. Seddon, R. Raval, S. J. Jenkins, G. Held, Surf. Sci. (2006) in press. G. Jones et al. to be published. J. Stöhr ‘NEXAFS spectroscopy’, Springer, 1992.

282

A comparative in situ XPS study of PtRuCo catalyst for CH3OH and CO oxidation using water. S. Zafeiratos1, G. Papakonstantinou2, F. Paloukis2, A. Knop-Gericke1, S.G. Neophytides2, R. Schlögl1, 1 Fritz-Haber-Institut der MPG, Faradayweg 4-6, D-14195 Berlin (Dahlem), Germany 2 ICEHT-FORTH, GR-26504 Rion, Achaias, Greece
Grant Number :RII 3 CT-2004-506008

Introduction Fuel cells using methanol as a fuel are promising future energy technology. Methanol is electrooxidized in the presence of water at the anode generating CO2, hydrogen ions and the electrons that travel through the external circuit as the electric output of the fuel cell. The excellent catalytic activity of Pt for methanol oxidation makes this metal electrocatalysts ideal for use as an anode in Direct Methanol fuel cells (DMFCs). However, Pt metal surface is easily poisoned at low temperatures by trace amounts of CO, which exists as a byproduct of methanol electrooxidation. Theoretical and experimental studies have shown that ternary PtRuCo catalyst significantly promotes the methanol oxidation reaction [1]. In this study we investigate the PtRuCo surface for the CH3OH and CO oxidation in the presence of H2O.

Experimental The Pt37.5Ru37.5Co25 catalyst was prepared by combustion synthesis and pre-treated in-situ under oxidation (0.5 mbar O2 at 400°) and reduction (0.5 mbar H2 at 300°) cycles. XP spectra were recorded at 300° under steady state conditions using depth profiling by variation of the incident photon energy. The overall pressure was 0.5 mbar and the CH3OH:H2O and CO:H2O mixing ratios were 1:2 and 1:10 respectively. Spectra under CO, H2O, O2 and H2 atmospheres were also recorded as a reference. Synchrotron radiation delivered by the undulator U49/2 was used. Changes in the gas phase composition were monitored by on-line mass spectrometry simultaneously to the spectroscopic characterization of the catalyst surface. Results On line Mass Spectrometry data for CH3OH and CO oxidation are given in fig. 1a and b respectively. The activity of the catalyst is demonstrated by the detection of the reaction products (H2 and CO2) in the gas phase and the consumption of CH3OH when the catalyst was heated in the reaction mixture. Blank experiments were performed in order to support the above picture.

283

9.0x10

-10

(a) H2O/MeOH : 2/1
H2

1.0x10

-9

(b) H2O/CO : 10/1
CO2 CO H2
300

QMS Intensity / a.u.

8.0x10 7.0x10 6.0x10 5.0x10 4.0x10 3.0x10 2.0x10 1.0x10

QMS INtensity / a.u.

-10

300

9.0x10 8.0x10 7.0x10 6.0x10 5.0x10 4.0x10 3.0x10 2.0x10 1.0x10

-10 -10 -10 -10 -10 -10 -10 -10 -10

Temperature / °C

Temperature/ °C

-10

CO2

-10

CO

250 200 150

250 200 150 100 50

-10

-10

-10

CH3OH
100 50

-10

-10

0

18

Relative time / min

36

54

72

5

15 20 Relative time / min

25

30

Figure 1. On line Mass Spectrometry data a) Methanol oxidation b) CO oxidation
70 60
O2 H2 CO/H2O CH3OH /H2O

Surface segregation, i.e. the enrichment of one element at the surface relative to the bulk, is a ubiquitous phenomenon in metal alloys [2]. In situ XPS gives a unique opportunity to determine the surface composition under various gas atmospheres. In figure 2 the calculated C, Ru, Pt, Co and O atomic concentration at the first 4 atomic layers (electron KE = 180 eV) are presented under various atmospheres. From the results is concluded that Pt segregates on the surface during methanol

% Atomic Ratio

50 40 30 20 10 0 Nominal Pt&Ru

Nominal Co

C

Ru

Pt

Co

O

Figure 2. The surface atomic concentration calculated from XPS intensities recorded under various gas atmospheres

oxidation, while for CO oxidation the picture changes and cobalt enrichment occurs accompanied with higher O amounts and the decrease of Pt. The chemical state of surface elements is very much depended on the type of the reaction. In figure 3 characteristic C1s&Ru3d5/2, Pt 4f, Co 2p3/2 and O 1s spectra are presented. In Fig 3a graphitic carbon at 284.4eV dominates C1s region, while a surface located component at 288.1 eV corresponds most probable to adsorbed CO. Ru 3d5/2 peak is found at 279.8 eV, very close to the binding energy of metallic Ru. A comparison of Pt 4f peaks for methanol and CO oxidation (fig. 3b) reveals that in the later case a new Pt doublet at 72.5 eV is needed in order to fit the overall spectrum. This peak is primarily surface located and can be attributed to hydroxyl-Pt complexes [3] as supported also from O1s spectra (see below). In fig. 3c, Co 2p3/2 spectra under reaction as well as under H2 and O2 atmospheres are presented. For methanol oxidation cobalt found completely reduced (778.1 eV), contrary to CO oxidation where it is partially oxidized. O 1s spectra in fig. 3d showed a complex structure, also due to gas phase peaks appear above 534 eV.

284

a
hv = s ; 4 7 5 e V b; 865 eV

Pt 4f
hv = 225 eV 72.5

71.2

b

C 1s
284.4 eV

Ru 3d3/2
279.8 eV

Assignment of the O 1s peak is speculative and should be done taking into account

288.1 eV

XPS Intensity / a.u.

C O /H 2 O ( b ) C O /H 2 O (s ) CH3OH /H 2 O ( b ) CH3O H /H 2 O (s )

XPS Intensity / a.u.

CH 3OH/H2O

characteristics of the Pt and Co peaks. During CO oxidation non-gas phase peaks are

CO/H2O

80

78

76

74

72

70

68

292

290

288

286

284

282

280

278

Binding Energy / eV

centered at 529.5 and 531.1 eV. Depth analysis showed that the peak at 531.1 eV is mainly surface located testifying for the presence of hydroxyl

B in d in g E n e r g y / e V

O 1s

Co 2p3/2
hv = 965 eV

780.4

778.1

c
O2

hv=720 eV gas phase O

532.2 530.4 531.1 529.5

d

XPS Intensity / a.u.

XPS Intensity / a.u.

CH3OH/H2O

CO/H 2 O

species on the surface. The bulk component at 529.5 eV is probably related to the oxidized
CO/H2O

C H 3 OH /H 2 O

H2

790

785

780

775

Binding Energy / eV

538

536

534

532

530

528

526

cobalt as showed in fig. 3c. In case of methanol oxidation two broad O1s compounds were found at 530.4 and 532.2 eV.

Binding Energy / eV

Figure 3. XPS spectra recorded at 300°.

Although definitely assignment of these components is unfeasible, it comes out that different oxygen species are involved in each reaction. In summary, it was showed that the composition and the chemical state of PtRuCo surface depend on the performed reaction, indicating the dynamic character of the catalyst. It was found that the oxidation state of Co varies, probably acting as a carrier or reservoir of oxidation agents participating in the reaction. This nicely correlates with observations on realistic fuel cells, where PtRuCo found to facilitate CO oxidation at lower potentials compared to Pt and PtRu [1]. The excess of oxidized species on the surface, related with the presence of cobalt, promotes CO oxidation most probable providing OH species on Pt where CO oxidation actually takes place. References
[1] Strasser P. Fan Q., Devenny M. Weinberg W.H. , Liu P. Norskov J.K., J Phys. Chem B 2003, 107 11013 [2] Markovic, N.M. Ross, P.N. Surf. Sci. Rep. 2002, 45 117 [3] J.E. Drawdy, G.B. Hoflund, S.D. Gardner, E. Yngvadottir, D.R. Schryer, Surf. Interface Anal. 16 (1990) 369.

285

In-situ XPS study on (MoV)5O14 selective oxidation catalysts
P.Schnörch, E. M. Vass, S. Zaferiatos , D.Teschner, M. Hävecker, A. Knop-Gericke, R. Schlögl Department of Inorganic Chemistry, Fritz-Haber Institute of the MPG, Faradayweg 4-6, 14195 Berlin, Germany

Introduction Catalytic reactions involving partial oxidation belong to the most important processes in the chemical industry [1]. Therefore, Mo based selective oxidation catalysis (propane and propylene to acrylic acid) is one of the main research area in our department. In-situ experiments are necessary to identify the active phase under working conditions. In-situ spectroscopy techniques in the soft X-ray range like X-ray absorption spectroscopy and photoelectron spectroscopy were applied to investigate the electronic structure of the working catalyst surface. In this work we measured two different oxide containing catalysts, (MoVW)Ox and (MoV)Ox which are very good candidates for these reactions. Results The experiments were performed at beamline U49/2-PGM1 at the synchrotron source. The insitu XPS system is a modified standard XPS spectrometer. Three differential pumping stages keep the hemispherical analyzer at high vacuum while the pressure in the sample cell is in the mbar range. The reaction cell is separated from the synchrotron beam line by a 100 nm thick SiNx X-ray window. The (Mo0. 91V0.09)5O14) catalyst was prepared by spray-drying technique of a mixed solution of ammonium heptamolybdate (AHM) and vanadyl oxalate. The precursor was treated at 623K in air and 773K in helium. The (Mo0.68V0.23 W0.09)5O14 catalyst was prepared also by spray-drying technique of a mixed solution of ammonium heptamolybdate (AHM), ammonium metatungstate (ATM) and vanadyl oxalate. The obtained product was calcined at 623 K for 2h in static air and in flowing He at 12 h at 713 K. [2] The catalysts were investigated under ~0.5 mbar of propylene-oxygen (1:2) mixtures. We have recorded the Mo3d, O1s/V2p, C1s and valence band regions under reaction conditions. In addition we have performed depth-profiling by varying the excitation photonenergy applied to the same core-level, which leads to a change of the photoelectron kinetic energy, and as a consequence to a change in the information depth. The catalytic reactivity was measured using PTRMS. Other than CO2, catalysts (Mo0. 91V0.09)5O14) and (Mo0.68V0.23 W0.09)5O14 produced only aldehydes and no acids under these low-P conditions at higher pressure (atmospheric) the catalysts are produced acrylic acid as well. The reason might be a pressure barrier existing between atmosphere and mbar condi-

286

tions. However further investigations are required to examine the effects of pressure dependence on the reaction. The mass spectra were recorded simultaneously with the XPS spectra, which allowed us to correlate the XPS results with the catalytic activity of the material.

40 35 30
623K

Intensity

25 20 15 10 5 0 60 160
298

473K

473K

Acetaldehyde Acrolein Propanal/Aceton Formaldehyde

The samples were measured using two different temperature profiles. At 298K – 473K – 623K – 473K, and 298K – 623K – 473K, the heating ramp was in both cases 10K/min. (Fig1.) The catalyst showed different catalytic behavior depending on reaction conditions.

260

360

Time (min)

Fig. 1. Catalytic activity under reaction atmosphere

From the XP spectra vanadium enrichment was observed on the catalyst subsurface in both catalysts. (Fig2.)
Mo and V ratio #1862

V Mo

100.00 98.00 96.00 94.00 92.00 90.00 88.00
T, bu lk ,s ur f 0b ul k 0s ur f lk Fe ed RT Fe ed 35 Fe ed 35 0, bu Fe ed 20 Fe ed 20 Fe ed R 0, su rf
97.02 93.58 92.02 92.89 92.52 2.98 6.42 7.98 7.11 7.48

7.92

92.08

In the case of W containing catalyst the Mo3d oxidation state it was not really changed but the only vanadium and molybdenum containing sample showed different oxidation states in room temperature ,during the reaction the Mo5+ state is disappeared and the remaining state is Mo6+. It means the molybdenum completely oxidized under the reaction mixture. (Fig 3. ,Fig 4.)

Fig.2. Molybdenum and Vanadium ratio in “bulk” (~18Ǻ) and surface (~8Ǻ) sensitive mode

287

232,5

.

298K 231,4 623K 473K
Intensity (a.u.)

Mo3d

RT- Mo3d 350 - Mo3d 200 - Mo3d

Intensity (a.u.)

232,2

. .

Intensity (a.u.)

232,2
. . 240 238

236 234 232 230 Binding Energy (eV)

228

238 237 236 235 234 233 232 231 230 229 Binding Energy (eV)

Fig 3. Mo5+statet disappear during the reaction (MoV)Ox

Fig4. Small changes on the oxidation state of the Mo3d core level under the reaction (MoVW)Ox

The V2p core level was more oxidized under the reaction condition. The V4+ state increased during catalytic reaction.(Fig5) The initial tungsten oxidation state was 6+ and there was no changing under the reaction.

V2p3/2
Intensity (a.u.)

RT - V2p 350 - V2p 200- V2p

In the future we would like to investigate more precisely the role of the Mo, V oxidations state during the reaction atmosphere because it seems there is no significant changing in the W oxidation states. According to some preliminary investigations the role of the tungsten in the catalysts is to stabilize the structures.

518

517 516 515 Binding Energy (eV)

514

513

Fig5. V2p spectra at different temperatures

Acknowledgement The authors thank the Bessy staff for their continual support during the measurements.

References
[1] J. Holmberg, R.K. Grasselli and A. Andersson, Appl. Catal. A: General 270 (2004), p. 121 [2] S. Knobl, G. A. Zenkovets, G. N. Kryukova, O. Ovsitser, D. Niemeyer, R. Schlögl, G. Mestl, Journal of Catalysis 215 (2003) 177 .

288

Remnant XMCD investigation of bimetallic oxalate based magnets: local magnetic contributions to the remnant magnetization C. Cartier dit Moulin, C. Train LCIM2, Université Paris 6, 4 place Jussieu, 75252 Paris Cedex 05, FRANCE We are engaged in the enantioselective synthesis of oxalate-based magnets of general formula A[MIIMIII(ox)3] where A is a monocation, MII and MIII two transition metal ions and ox=C2O42-. Apart from being good candidates to observe magnetochiral dichroism, a new physical effect arising from the synergetic breaking of space and time symmetries, these materials exhibit rich magnetic properties. In particular, depending on the choice of the metal ions, one can obtain coercive magnets 1,2 and magnets exhibiting negative magnetization for TBA[FeIIFeIII(ox)3] (TBA=tetra(butyl)ammonium).3 Up to now, quite all XMCD signals are recorded while applying a high magnetic field in order to saturate the sample and therefore increase the intensity of the XMCD signals. Rare are the XMCD experiments performed with a null or weak magnetic field. 1. On one hand, recording remnant XMCD signals shall allow to precise the respective impact of single ion anisotropy, magneto-structural anisotropy and exchange interaction on the remnant state of these magnets. 2. On another hand, through the XMCD measurements of the respective contribution of FeII and FeIII in TBA[FeIIFeIII(ox)3] in a low applied magnetic field (100 G), we want to precise the origin of the negative magnetization observed in this compound at low temperature. 2.Aims of the experiment 2.1 Study of the remnant state in TBA[FeIIFeIII0.77CrIII0.23(ox)3] We have recently performed XMCD measurements on a two dimensional oxalate based magnet, [N(C4H9)4][FeIIFeIII0.77CrIII0.23(ox)3], at the Fe and Cr K-edges (Figure 1).4

Figure 1: (a) Isotropic Fe K-edge XAS spectrum for {[N(C4 H9)4][FeIIFeIII0.77CrIII0.23(ox)3]}; (b) XMCD signals (x1000) measured in H = ±2 T; (c) XMCD signals (x1000) measured in zero field after applying H = ±2 T(Ref 4).

This material is a ferrimagnet with a Curie temperature of 36 K. The coercive force at 2K is 1.03 T and the remnant magnetization Mr is 0.25 µB. We have evidenced a remnant XMCD signal at H = 0 T at the two edges. The remnant to saturation XMCD signals ratios follow the macroscopic measurements for both ions. We therefore show that the contribution of the local magnetic moments to the total remnant magnetization is the same for Cr(III) and Fe(II) cations

289

despite the higher single-ion anisotropy of the Fe(II) ion. This result is in line with other XMCD investigations on the origin of the hysteretic behavior of single molecule magnets5 and of extended networks.6 Measurements at the K-edges do not allow to determine the spin and orbital contributions to the XMCD signal. 2.2 Study of the low field XMCD in TBA[FeIIFeIII(ox)3] versus temperature According to macroscopic magnetic measurements,3 the magnetization in 100 G goes from positive just below Tc (44 K) to negative below 30 K when the temperature is lowered. This result has been somehow anticipated by the classic theory of ferrimagnets developed by Néel. In this approach, this behavior is strongly related to the higher single-ion anisotropy of the Fe(II) ion compared to Fe(III). Recording remnant XMCD signal at the Cr and Fe L2,3 edges for two compounds, we want to precise the respective impact of single ion anisotropy, magneto-structural anisotropy and exchange interaction on the remnant state of these magnets. 3.Results Out of two different compounds, we could fully measure one sample at low temperature (1.5K) in a variable magnetic field between 0 Tesla (remnant state) and 5 Tesla. We appreciate the very high quality beam of the beamline necessary to measure the small signal expected in the remnant state. We had not enough beamtime to measure XMCD signals on the second compound and on two reference compounds to calibrate our measurements. Given the well known fragility of such compounds built with transition metal ions linked with organic ligands, we encountered strong difficulties. Due to the high flux delivered by the beamline, we observed a very rapid radiation damage of the samples. We used quite all the beamtime to try to optimize the recording conditions to reduce as well as possible the radiation damage: - we check the different ways to reduce the flux with conservation of the circular polarization (the difficulty is the limit of the signal/noise ratio); - during the series of scans, we changed regularly the position of the sample in the beam, using(and destroying progressively) all the surface of the pellet. For other molecular compounds as Mn12-acetate (XMCD measurements at BESSY with the same beamline), we used successfully this way to avoid radiation damage and we obtained high quality and reliable XMCD signals. Despite our efforts, for these iron compounds, the sample radiation damage remains very rapid. After careful analysis of our data, it seems that the results obtained are not reliable and the measured signals traduce only the evolution of the sample. 4. References (1) Bhattacharjee, A.; Iijima, S.; Mizutani, F. J. Magn. Magn. Mater. 1996, 153, 235. (2) Coronado, E.; Galan-Mascaros, J. R.; Gomez-Garcia, C. J.; Martinez-Agudo, J. M. Adv. Mater. (Weinheim, Fed. Repub. Ger.) 1999, 11, 558. (3) Mathonière, C.; Nuttall, C. J.; Carling, S. G.; Day, P. Inorg. Chem. 1996, 35, 1201. (4) Train, C.; Baudelet, F.; Cartier Dit Moulin, C. J. Phys. Chem. B. 2004, 108, 12413. (5) Moroni, R.; Cartier dit Moulin, C.; Champion, G.; Arrio, M. A.; Sainctavit, P.; Verdaguer, M.; Gatteschi, D. Phys. Rev. B 2003, 68, 064407/1. (6) Train, C.; Giorgetti, C.; Baudelet, F.; Champion, G.; Cartier dit Moulin, C. C.R. Chimie 2003, 6, 337.

290

Effect of molecular conformation on packing density and orientational order of aromatic self-assembled monolayers
A. Shaporenko1, M. Elbing2, A. Błaszczyk2,3, C. von Hänisch2, M. Mayor 2,4, and M. Zharnikov1 1 Angewandte Physikalische Chemie, Universität Heidelberg, Im Neuenheimer Feld 253, 69120 Heidelberg, Germany 2 Forschungszentrum Karlsruhe GmbH, Institute for Nanotechnology, P. O. Box 3640, 76021 Karlsruhe, Germany 3 Faculty of Commodity Science, Al. Niepodleglości 10, 60-967 Poznań, Poland 4 University of Basel, Department of Chemistry, St. Johannsring 19, CH-4056 Basel, Switzerland The on-going miniaturization of the silicon-based electronic technology will reach its physical limit in the foreseeable future and hence, alternative concepts that allow further reducing the size of electronic active components are highly desirable. In particular, the concept of molecular electronics that envisages the use of molecular structures to build electronic devices has received considerable scientific as well as popular interest recently. Prototypes of future devices are metal-molecule-metal junctions, in which a single molecule or a molecular assembly is placed between two metal electrodes. In the best case, the molecules should be chemically coupled to the electrodes, e.g., via reaction between organic thiol moieties, attached at both ends of the “device” or “wire” molecule and a metal electrode surface.

BPT

1

2

3

4

5

Fig. 1. Biphenyl-derived dithiol molecules with different conformations used for the SAM fabrication. It is generally assumed that the conformation of BPT and 1 changes to planar one in the solid state or upon the formation of densely packed SAMs. The conformations of 3 and 5 remain unchanged (torsion angles of 20° and 80°, respectively). 2 and 4 have planar conformation both in the molecular and solid state.

Keeping this general approach in mind, a series of biphenyl-derived dithiol (BDDT) compounds with terminal acetyl-protected sulfur groups and different structural arrangements of both phenyl rings have been synthesized and fully characterized. The different arrangements were achieved by introducing hydrocarbon substituents in the 2 and 2´ positions of the biphenyl backbone as shown in Fig. 1, where the respective molecules marked by 1-5 are shown (without the protection groups), along with the reference system biphenylthiol (BPT). The presented model compounds enable the investigation of the correlation between the intramolecular conformation and other physical properties of interest, like, e.g., molecular assembly or electronic transport properties. As a first step, we studied the ability of these model compounds to form self-assembled monolayers (SAMs) on Au(111) and Ag(111). The deprotection of the target molecules was performed in situ, using either NH4OH or triethylamine (TEA) deprotection agent. The fabricated films were characterized by

291

Fig. 2. C 1s (left panel) and S 2p (right panel) HRXPS spectra of 3/Au prepared using a deprotection by TEA (top curves) and NH4OH (bottom curves). The spectra are decomposed into the individual contributions.

synchrotron-based high-resolution X-ray photoelectron spectroscopy (XPS) and near-edge absorption fine structure (NEXAFS) spectroscopy. Whereas the deprotection by NH4OH resulted in the formation of multilayer films, the deprotection by TEA allowed the preparation of densely packed BDDT SAMs. This is illustrated by Fig. 2, where the C1s and S2p XPS spectra of the differently deprotected 3/Au (as an example for 1-5) are presented. Note that in the TEA-deprotected SAMs, not all acetyl protection groups were removed, but 10-20% of the thiol groups at the SAM-ambient interface remained protected (see Fig. 2). The signature of the acetyl group was also seen in the NEXAFS spectra (Fig. 3). BPDT SAMs are characterized by a noticeably higher orientational order and smaller molecular inclination on Ag than on Au, as shown in Fig. 4, where NEXAFS difference spectra are depicted. Most important, the introduction of the alkyl bridge between the individual rings of the biphenyl backbone does not lead to noticeable change in the structure and packing density of the BDDT SAMs as far as the molecule had a planar conformation in the respective films. However, the deviation from this conformation, as it e.g. happens in the case of 3 and 5, results in the deterioration of the film quality and decrease in the packing density and orientational order. Note that the SAM 1 has the similar properties as SAMs 2 and 4, which is a proof that 1 has a planar conformation in the SAM.

Fig. 3. C K-edge NEXAFS spectra of the TEA-deprotected SAMs 1-5 on Au (left panel) and Ag (right panel) acquired at an X-ray incidence angle of 55°.

Figure 4. The C K-edge NEXAFS difference (90°-20°) spectra of the TEA-deprotected SAMs 1-5 on Au (left panel) and Ag (right panel).

This work has been supported by BMBF (05KS4VHA/4 and 05 ES3XBA/5).

292

Magnetic domain structure in bilayers of antiferromagnetic NiMn and ferromagnetic Co C. Tieg1, R. M. Abrudan1,2, M. Bernien2, W. Kuch2, J. Kirschner1 1 Max-Planck-Institut für Mikrostrukturphysik, Weinberg 2, D-06120 Halle, Germany 2 Freie Universität Berlin, Institut für Experimentalphysik, Arnimalle 14, D-14195 Berlin, Germany

Many modern devices in the field of magnetic data storage and sensing consist of mulilayered magnetic films in which the exchange bias effect between an antiferromagnetic (AFM) and a ferromagnetic (FM) film is exploited in order to tune the switching behaviour of the latter. Despite the enormous technological importance of AFM/FM systems, the fundamentals of the magnetic interaction in such systems are not well understood. Promising candidates to gain a deeper insight are model systems composed of single-crystalline films. In comparison to the widely studied polycrystalline samples, singlecrystalline bilayer systems have the advantages of a higher structural and magnetic homogeneity (no grains), in addition to atomically flat interfaces, which greatly reduces the complexity and also simplifies the theoretical description. We have employed a photoelectron emission microscope (PEEM) in order to study the magnetic domain structure with sub-micrometer resolution in single-crystalline bilayers composed of AFM NiMn and FM Co on Cu(001). The x-ray magnetic circular dichroism (XMCD) at elemental L3 absorption edges was utilised as a contrast mechanism for magnetic domain imaging. The experiments were performed at the beamline UE56/2-PGM2. Bulk NiMn exhibits an L10 phase in the equiatomic concentration range, in which the magnetic moments order in a collinear AFM structure with a Néel temperature of 1070 K. In our experiments, the alloy films were obtained by co-evaporation of Ni and Mn from separate sources with the same deposition rates on Cu(001) held at 300 K. The bilayer structures were prepared as crossed wedges of NiMn and Co, which allows a convenient investigation of thickness dependences. In Fig. 1 we show the domain structure of the Co layer of an as-grown Co/NiMn/Cu(001) sample. The inset illustrates the sample geometry and the position of the field of view. The thickness of the NiMn layer increases from zero at the bottom to 16 monolayer (ML) at the top of the image. The thickness of the Co wedge with a plateau (7.3 ML) on the right side is indicated at the abscissa. Crystallographic directions and the projection of the direction of the photon beam onto the sample surface (hn) are indicated at the bottom right. The thick arrows show the magnetisation directions of the domains. The grey level reflects the Co L3 XMCD asymmetry. Magnetic order in the Co film is absent in the left part of the sample (region with a homogeneous grey level labelled by i)), as concluded from the vanishing XMCD asymmetry. This is due to the thickness dependence of the Curie temperature, which is lower than 300 K in this sample region. The domain structure in the region where the Co film is ferromagnetic exhibits a clear dependence on the thickness of the NiMn layer. Only one single domain with a magnetisation along the [-110] direction can be seen in the sample region where the Co film is in direct contact to the substrate (lower part of the image). Large domains, several mm in size, are present in the region where Co is on top of NiMn thinner than about 8 ML (region ii)). The four different grey levels (black, dark grey, light grey, white) in this part of the image correspond to magnetisation along the four in-plane <110> directions. Above this NiMn thickness, the Co layer is broken up into much smaller domains of irregular shape and size (region iii)). The domain sizes are close to the chosen instrumental resolution, which impedes us to determine exact magnetisation directions. However, as in region ii), four different grey levels are readily recognisable in this region. We ascribe the

293

change of the Co domain structure upon exceeding a NiMn thickness of about 8 ML to the magnetic phase transition in the NiMn layer from para- to antiferromagnetic, similar to the observations in Co/FeMn systems [1]. The small Co domain structure indicates a laterally fluctuating coupling of the FM Co film to the AFM NiMn film, which may stem from a locally varying AFM spin structure due to the presence of terraces and magnetic domains in the AFM.

Fig. 1: Magnetic domain structure in an as-grown Co(wedge)/NiMn(wedge)/Cu(001) sample imaged by XMCDPEEM at the Co L 3 absorption edge at 300 K. The Co is paramagnetic in the left part of the image (region i)). Ferromagnetic order is established above a certain critical film thickness. The magnetic transition line is indicated by arrows. The domain structure in the Co depends on the NiMn film thickness, as can be concluded from the different domain structures (domain sizes) in the lower (region ii)) and upper right (region iii)) part. The change of the Co domain structure at a NiMn thickness of about 8 ML is attributed to the magnetic phase transition in the NiMn film from para- to antiferromagnetic, which results in a locally fluctuating interaction at the Co/NiMn interface. The inset illustrates the sample geometry.

References: 1. W. Kuch, F. Offi, L. I. Chelaru, M. Kotsugi, K. Fukumoto, and J. Kirschner, Phys. Rev. B, 65, 140408(R) (2002).

294

Commensurate-to-incommensurate phase transition of sodium ordering in single crystal Na0.48CoO2
E. Dudzik, R. Feyerherm, C. Milne, D. Argyriou, D. A. Tennant Hahn-Meitner-Institute, 14109 Berlin, Germany The system NaxCoO2 currently arises considerable interest after it has been demonstrated that intercalation of water into samples with x = 0.35 leads to superconductivity with a TC of up to 5K [1]. In order to explain this extraordinary phenomenon, many researchers aim at understanding first the water-free system. In NaxCoO2, variation of the Na concentration x allows for a control of the magnetic and electronic degrees of freedom of the quasi twodimensional triangular CoO2 sheets, because compositions with x < 1 nominally are mixed valence systems NaxCo3+xCo4+1-xO2 where Co4+ has a LS S = 1/2 configuration. Variation of the Na content has also been shown to result in various types of ordered Na superstructures resulting in the interesting question what constraints the Na ordering may impose on the CoO2 layer. There are indications, e.g., that Na0.5CoO2 exhibits Co3+/Co4+ charge ordering. The aim of the present x-ray diffraction study was to investigate in detail the temperature dependence of the Na ordering in a single crystal NaxCoO2 with an x value close to 0.5 and to look for indications of possible charge-ordering coinciding with the magnetic ordering transitions at 52 and 87 K reported previously [2]. A single crystal NaxCoO2 with average composition x = 0.48 was grown at HMI, and was studied by x-ray diffraction on the beamline MAGS at the 7 Tesla wiggler in the temperature range 10-550K. A number of weak superstructure reflections were identified and studied as a function of temperature. In order to enhance possible Co3+/Co4+ charge ordering effects, resonant scattering at the Co K-edge was involved. Our sample is a piece of a batch used simultaneously for neutron diffraction studies of the magnetic ordering. The first central result is that there are no indications in our data for any significant structural changes around the magnetic ordering temperatures. In contrast, we unexpectedly found that Na0.48CoO2 exhibits a reversible transition around 225 K from a commensurate superstructure, stable below that temperature, to an incommensurate superstructure existing between 225 and 430 K (see Figure 1). The commensurate phase is consistent with the orthorhombic superstructure reported previously for polycrystalline Na0.5CoO2 [3], related to the original hexagonal cell by a’ = √3a, b’ = 2a, c’ = c, i.e., a doubling of the unit cell along both basal plane axes. The incommensurate phase is a modulation of this orthorhombic cell with a modulation wave vector q = (0, δ, 0)’. The value of δ exhibits a temperature dependent Figure 1: Longitudinal scans along (0k0)’ at various variation between 0.055 and 0.11, temperatures. The (020)’ splits off in two satellites showing a broad plateau at the latter around 220 K marking the commensurate-tovalue between 260 and 360 K (see incommensurate transition. The broad bump at k = 1.97 Figure 2). Above 430 K, the sodium stems from the (220) reflection of a highly textured Co3O4 impurity. ordering breaks down. 1
295

An incommensurate phase has been observed recently in electron diffraction studies after long irradiation of a Na0.5CoO2 sample and was discussed in terms of a regular arrangement of extra sheets of voids introduced to the NaxCoO2 system for x < 0.5 [4] (see Figure 3). In this model, the observed plateau value of δ is consistent with x = 0.47, in close agreement with x = 0.48 measured on our sample by neutron activation analysis. We argue that there are two possible explanations for the incommensurate-to-commensurate phase Figure 2: Temperature dependence of the position of transition observed in this phase. Below the incommensurate satellites (0 2±δ 0)’. The grey 225 K, either a phase separation into line plotted in the center indicates the average two volume fractions with x = 0.5 and between the two satellite positions, reflecting the x<<0.5 takes place, or the extra sheets lattice expansion in that temperature range. of voids loose their regular arrangement and distribute in a random fashion in the commensurate phase. The former model would point to a surprisingly high long-range mobility of Na ions at 225K. The latter model would imply an even more interesting transition from a disordered low-T to an ordered high-T phase, hence a structural analogue to re-entrant spin glass behavior. We have also studied several other crystals NaxCoO2 with various stoichiometries x = 0.65, 0.75, 0.85. Also in these samples, we identified a number of weak superstructure reflections and studied their temperature dependence. So far, however, the results are inconclusive and therefore not reported in more detail here. For a continuation of our studies of these samples we resubmit or previous proposal.

Figure 3: Model for the formation of an incommensurate superstructure by insertion of extra sheets of voids (after [4]). Only a regular arrangement of these extra sheets along the b axis would produce incommensurate Bragg reflections. Red and yellow circles denote Na atoms.

References [1] K. Takada et al., Nature 422 (2003) 53 [2] P. Mendels et al., PRL 94 (2005) 136403 [3] Q. Huang, et al., J. Phys.: Condens. Matter 16 (2004) 5803 [4] H. W. Zandbergen, et al., PRB 70 (2004) 024101 2
296

Incommensurate structural distortion induced by magnetic ordering in TbMnO3 and DyMnO3
E. Dudzik, R. Feyerherm, N. Aliouane, D. Argyriou Hahn-Meitner-Institute, 14109 Berlin, Germany The interplay of ferroelectricity and magnetism in rare earth manganites currently arises much interest [1,2]. In orthorhombic TbMnO3 the Mn spins order in an incommensurate magnetic structure below TMn = 42 K and the corresponding incommensurate magnetic propagation vector qMn = (0,0.28,0) locks below Tlock = 27 K [2]. This lock-in transition is associated with a jump of the electric polarization. The Tb spins also order antiferromagnetically below 7 K with a different qTb = (0,0.42,0). Similar behaviour is observed for DyMnO3 for which, however, the Dy ordering is unknown yet. Here, TMn ~ 40 K and Tlock ~ 20 K. The goal of the present experiments was to study the lattice distortions induced by the incommensurate ordering of the Mn magnetic moments through the magnetic and the successive lock-in transition. Since the propagation vectors of the magnetic ordering and the lattice distortion are closely related, this experiment aimed at yielding additional information on the magnetic ordering and the lock-in transition. For TbMnO3, the magnetic transitions have been investigated previously by neutron diffraction [3]. For DyMnO3, no neutron diffraction results are available yet because Dy is a strong neutron absorber, making neutron diffraction experiments difficult. Here, x-ray diffraction provides the only way to determine the propagation vector for the magnetic ordering, both of Mn and Dy. We have performed high-resolution diffraction experiments on single crystals of TbMnO3 and DyMnO3 using the beamline MAGS at x-ray energies between 8 and 14 keV. These crystals were produced and characterized at HMI. The TbMnO3 crystal was a small piece (0.2×0.3×0.5 mm³) of a sample previously used for neutron diffraction experiments at HMI. The DyMnO3 single crystal was plate-shaped (3×3×0.5 mm³). For both crystals we looked for the incommensurate lattice distortion induced by the Mn ordering below TN = 42 K down to the lowest temperatures achievable with the cryostat at the MAGS beamline (7 K). We observed crystallo40 35 Intensity (arb. units) 30 25 20 15 10 5 0 2.53 2.54 2.55 2.56 k 2.57 2.58 2.59

(0 k 3) superstructure peak TbMnO3 MAGS data

6.2 K 13.1 K 19.2 K 23.8 K 26.1 K 28.2 K 30.4 K 32.6 K 34.7 K 36.8 K 39.0 K

Figure 1: (0, 2+q’, 3) superstructure reflection measured at various temperatures.
2.580 2.575 2.570

X-rays (2+q') Neutrons (2+2q)

2+q'

2.565 2.560 2.555 2.550 2.545 0 5 10 15 20 25 30 35 40 45

T(K)

Figure 2: Temperature dependence of the position of the crystallographic and the magnetic superstructure reflections measured by x-ray and neutron diffraction, respectively.

297

graphic superstructure reflections of type (0 2+q’ L) with integer L. The intensity of these reflections were less than 10-6 of that of the neighbouring (0 2 3) standard reflection, showing that the corresponding superstructure can be hardly studied with a laboratory x-ray instrument or neutron diffraction. Figure 1 shows the (0 2 + q’ 3) Bragg reflection observed in TbMnO3 at different temperatures. Our data confirm previously published data by a Japanese group [4] but are of better resolution. Figure 2 shows the temperature dependence of the value of 2+ q’ compared to the corresponding data for the magnetic superstructure measured by neutron diffraction at HMI. We observe the relation q’ = 2 q, which is expected if the crystallographic distortions at the magnetic ordering are due to exchange striction effects. Our experiments are the first to verify this relation in TbMnO3 by two experiments on the same crystal. In addition, we observed that below Tlock the T-dependence of the intensity of the crystallographic superstructure reflection varies from that of the magnetic order parameter in showing a decrease below about 20 K. We speculate that this decrease is produced by some disorder in the crystallographic distortion induced by the deviation from the commensurate value q‘ = 0.5. The superstructure peaks were not enhanced when measuring at the Mn K-edge, suggesting the distortion is due to the displacement of O rather than Mn atoms. Similar measurements were carried out on DyMnO3. The crystal was mounted in a suitable way to be able to access large parts of the reciprocal space (0, k, l) plane in Bragg geometry (reflection). Compared to the Laue geometry used for TbMnO3 the scattered intensity was by more than an order of magnitude larger. For DyMnO3, no neutron diffraction results are available yet. Dy is a strong neutron absorber, making neutron diffraction experiment difficult. Therefore, x-ray diffraction provides the only way to determine the DyMnO3 propagation vector related to the Mn (0 k 3) superstructure peak 2.77 ordering.
2.76

Peak area (arb. units)

Figure 3 shows the temperature dependence of the position and intensity of the crystallographic superstructure reflection in DyMnO3. We observe an incommensurate value of q’ = 0.76 below the lock-in transition temperature, consistent with previously reported results [4]. For the first time, however, we observed a pronounced hysteresis of the lock-in transition. Comparison with the results on TbMnO3 suggest that the unusual non-monotonous temperature dependence of the intensity does not directly reflect the behaviour of the magnetic ordering parameter. References
[1] M. Fiebig et al., Nature 419 (2002) 818. [2] T. Kimura et al., Nature 426 (2003) 55. [3] R. Kajimoto et al., Phys. Rev. B 70 (2004) 012401. [4] T. Kimura et al., Phys. Rev. B 68 (2003) 060403.

k (r.l.u)

2.75 2.74 2.73 2.72 2.71 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 5 10 15 20

Peak position , heating , cooling

correction for temperature variation of lattice parameter b

Peak area , heating , cooling

25

30

35

40

Temperature (K)

Figure 3: Temperature dependence of the position (upper panel) and intensity (lower panel) of the superstructure reflection in DyMnO3.

298

Determination of the structural role of Fe as a function of the waste content in a series of stabilized industrial waste glasses. F. Pinakidou, M. Katsikini, E. C. Paloura∗, P. Kavouras, Th. Kehagias, Ph. Komninou, Th. Karakostas Aristotle University of Thessaloniki, School of Physics, GR54124 Thessaloniki, Greece. A. Erko BESSY GmbH, Albert Einstein Str. 15, 12489 Berlin, Germany. The solidification and stabilization of toxic industrial waste (lead-oxide-contaminated ash) is a process that allows for its safe disposal. The main advantage of the vitrification process is that it produces a chemically stable material, which can homogeneously incorporate into its matrix numerous toxic elements1. The studied samples were produced with an incineration process of Pb- and Fe-rich sludges that produces a highly toxic fly ash, which consists of 47 wt% Fe2O3, 42 wt% PbO·PbBr2 oxides and 11 wt% other oxides. The Pb-contaminated ash is vitrified via comelting with the appropriate quantities of vitrifying (SiO2) and flux (Na2O) agents at 1400°C, followed by quenching. The studied vitrified products contain various concentrations of fly ash, ranging from 10 % to 60 wt% with a step of 10%, and a SiO2/Na2O concentration ratio equal to 2.33. Previously reported XRF maps demonstrated that the samples are homogeneous2. The Fe-K-EXAFS measurements were conducted at room temperature, using the KMC2 monochromator at BESSY. The fluorescence spectra were recorded with a Si-PIN photodiode, at an angle of incidence θ=85o. The Fourier transforms (FT) of the Fe-K edge EXAFS spectra of the vitrified samples are shown in Fig 1(a). As shown in the figure, only the 1st nearest neighbor (nn) shell is resolved, indicating that the samples are amorphous. The Fe-K EXAFS spectra of studied glasses were fitted using one shell that consists of oxygen atoms.
Table I. EXAFS analysis results for the waste contaminated glasses; R, N and DW correspond to the interatomic distances, coordination numbers and Debye-Waller factors. The asterisk indicates the parameters that were kept fixed during the analysis. DW Sample N (O) RFe-O (Å) (x10-3 Name (±10%) Å) 5.2 1.92 4.3 10 ash% 4.7 1.91 3.2 4.6 1.90 3.3 20 ash% 4.2 1.89 2.6 3.8 1.87 2.3 30 ash% 3.7 1.88 2.3 5.2 1.92 4.3 40% ash 4.7 1.91 3.2 4.6 1.90 3.3 50% ash 4.2 1.89 2.6 60% ash

20

16

FT k *χ(k) (arb. units)

16

60% ash 50% ash

FT k *χ(k) (arb. units)

12

40% ash
8

12

40% ash 30% ash 20% ash 10% ash

30% ash 20% ash

3

8

3

4

4

10% ash
5
0 0 1 2 3 4 5

0

0

1

2

3

4

R (Å)

R (Å)

Figure 1: (a) The FTs of the k3×χ(k) spectra of the studied glasses. (b) The FTs of the k3×χ(k) Fe-K edge spectra of the glasses containing 10 to 40wt% ash. The fitting was performed using the mixed model. The raw data and the fitting are shown in thin and thick solid lines, respectively.

The EXAFS analysis results, shown in Table I, disclose that the coordination environment of Fe changes with increasing ash content. More specifically, as the ash content increases from 10 to 60 wt%, the Fe-O bond length (R1) decreases from 1.92Å to 1.87Å while the coordination number in
∗

E. C. Paloura, paloura@auth.gr, tel.:+302310998036, fax: +302310998036

299

the 1st nn shell (N1) decreases from 5.2 to 3.7. The simultaneous decrease of R1 and N1 indicates that, the ash content increases, the coordination environment of Fe changes gradually from octahedral to tetrahedral. Therefore, it can be proposed that in the samples with low ash content, the majority of the Fe atoms are octahedrally coordinated and thus the dominant role of Fe is that of a modifier3. In the intermediate ash concentrations (30-40 wt%) the Fe-O bond length decreases and takes the value 1.89-1.90Å, which is intermediate to the distances corresponding to octahedral and tetrahedral coordination of the Fe atom. Finally, in the high ash concentration limit (50-60 wt%) the measured Fe-O bond length is 1.87Å, a value that corresponds to tetrahedral coordination of Fe. In order to determine the percentage of FeO4 and FeO6 polyhedra in the studied samples we fitted the Fe-K EXAFS spectra using an alternative model, the mixed model, according to which X% of the Fe atoms belong to octahedral sites while the rest (100-X)% occupy tetrahedral sites. The fitting of the Fourier transforms (FT) of the Table II. EXAFS analysis results using the mixed EXAFS spectra (k-range 2.8-9.8Å-1), using the model; RFe-O and DW correspond to the interatomic mixed model, are shown in Fig 1(b). The distances and Debye-Waller factors. The asterisk spectra were fitted in the 1st nn shell and the indicates the parameters that were kept fixed during fitting parameters were the percentages of the the analysis. Fe tetrahedra and octahedra and the value of the RFe-O DW Sample Percentage of Debye-Waller factor. The fitting was performed (Å) (x10-3 Å) Name FeOx (%) simultaneously for all the samples and the 1.95 3.1 FeO6 = 55 ± 5 10 ash% distance in the tetrahedral coordination was FeO4 = 45 1.88* 2.3* 1.95* 3.1* kept fixed to the value derived from the FeO6 = 35 ± 4 20 ash% FeO4 = 65 1.88* 2.3* previous EXAFS analysis (1.88 Å). The 1.95* 3.1* FeO6 = 38 ± 6 iterated distance for the octahedral arrangement 30 ash% FeO4 = 62 1.88* 2.3* was kept the same among all the samples 1.95* 3.1* FeO6 = 13 ± 4 (1.95Å). Analysis using the mixed model 40% ash FeO4 = 87 1.88* 2.3* (Table II) reveals that in the glass with the lowest ash concentration (10 wt%), 55% of the Fe atoms are octahedrally coordinated in the vitreous matrix, while the rest constitute tetrahedra. As the ash content increases, the percentage of the FeO4 polyhedra increases, i.e. in the glasses with 20 and 30 wt% ash, the percentage of the octahedrally coordinated Fe is approximately equal to 35%, while it is drastically reduced to 13% when the ash content reaches the value of 40 wt%. In the glasses with ash concentration higher that 50 wt%, the fitting using the mixed model was not possible because the majority of the Fe atoms occupy tetrahedral sites in the glass matrix (i.e. the number of the FeO6 octahedra is lower than 11 wt%). Given that Fe+2 forms only octahedra, the observed changes in the Fe-O distance can only be attributed to a change in the coordination environment of Fe3+, from octahedral to tetrahedral. This change in the coordination environment of Fe+3 causes a change in the structural role of Fe3+, which is a glass modifier in the low ash limit (10-20 wt%) and becomes a glass former (along with Si) when the ash content increases to 50-60 wt%, as it is affirmed by the decrease of the number of FeO6 octahedra. It should be pointed out that while the SiO2/Na2O ratio did not vary in the glasses, the Si/O ratio is different. In samples with 10-30 wt% of fly ash the Si/O ratio is larger than the critical value of 0.33 that renders Si the dominant glass former. Therefore, Fe+3 is an intermediate. On the other hand, in the samples containing 40, 50 and 60 wt% ash, the Si/O ratio is equal to 0.324, 0.294 and 0.255 respectively and the formation of the vitreous matrix is made possible only due to the glass forming ability of tetrahedrally coordinated Fe+3.
1 2

P. Kavouras et al, J. Eur. Ceram. Soc. 23 (2003) 1305. F. Pinakidou et al, Journal of Non-Cryst. Solids 351 (2005) 2474. 3 W. Vogel, Glass chemistry 2nd ed., Springer-Verlag, Berlin, 1994, p. 251.

300

Micro-XRF mapping and micro-EXAFS study of glasses containing Electric Arc Furnace Dust F. Pinakidou, M. Katsikini, E. C. Paloura∗, P. Kavouras, Th. Kehagias, Ph. Komninou, Th. Karakostas Aristotle University of Thessaloniki, School of Physics, 54124 Thessaloniki, Greece. A. Erko BESSY GmbH, Albert Einstein Str. 15, 12489 Berlin, Germany. Electric arc furnace dust (EAFD) is one of the largest solid waste streams produced by steel mills. It contains mainly heavy metals and thus is considered as a toxic waste. Recycling of the valuable metals (Fe, Zn and Pb) reduces the disposal problems and results in resource conservation but can recover only a part of the heavy metals from the EAFD. Vitrification, leading to the formation of vitreous or glass-ceramic materials1, is a promising process to stabilize metallic Zn and Fe and hence permit the safe disposal of the EAF dust2. Therefore it is of great importance to study the structural role of both Fe and Zn in vitrified EAFD-rich industrial wastes, since the structural integrity of the glass matrix depends strongly on the type of polyhedra that Fe and Zn form. The under study samples are vitrified products of EAF dust (which mainly consists of ZnO and ferric oxides (ZnFe2O4)) that is co-melted with SiO2, Na2CO3 and CaCO3 at 1400°C for 2h, followed by quenching. The under study sample consists of 20% EAFD, 55% SiO2, 15% CaO and 10% Na2O and was heat-treated at 20°C above the glass transition temperature (Tg) for residual stress relaxation. Previous work reported on the studied sample disclosed that upon annealing, surface crystallization of wollastonite (CaSiO3) initiates from the edges towards the center. The XRF and µ-XAFS measurements were conducted at the KMC2 beamline using capillary optics that reduce the beam diameter to 5µm. The XRF maps were recorded using excitation photons of 7200 eV. The µ-XAFS spectra were recorded at the Fe-K and Zn-K edges, using an energy dispersive fluorescence detector, at two spots of the sample surface, one at the edge (position E) and one Zn 2.0 at the centre (position C). Fe The XRF spectra shown in Fig.1 were recorded 1.5 Annealed with excitation energy of 9400 eV, i.e. high enough to Mn Fe Edge excite both Fe and Zn. As shown in the spectra, the Annealed 1.0 studied samples contain Fe, Zn, Ca and Mn. The Center concentration of either Fe or Zn is the same in both Annealed 0.5 the as-casted and heat-treated sample and does not As casted Ca vary across the sample surface. 0.0 The XRF map recorded from the annealed sample, 2 3 4 5 6 7 8 9 10 shown in Fig.2, reveals that the concentration of E (keV) either Fe or Zn varies less than 15% across the sample and the formation of Fe and Zn clusters has been Figure 1: Fluorescence spectra normalized at avoided. This observation, along with the EXAFS the Zn Kα peak recorded at the edge and the centre of the annealed sample and at random results discussed in the following, indicates that both positions of the as-casted and annealed sample. Fe and Zn have been successfully incorporated into the vitreous matrix. In order to investigate possible changes in the local coordination of Fe and Zn, due to local compositional variations or due to differences in the cooling rate between the center and at the edge of the samples, we recorded µ-EXAFS at the Fe-K and Zn-K edges, from the two spots of
∗

E. C. Paloura, paloura@auth.gr, tel.:+302310998036, fax: +302310998036

301

Intensity (arb. Units)

the sample surface. The Fourier transforms (FTs) of the k3 - weighted µ-EXAFS spectra recorded at the Fe- and Zn-K edges are shown in Fig. 3(a) and (b), respectively. As shown in the figure, the FTs of the studied samples have well-resolved structure up to a distance of about 3.5Å. Therefore, mid-range order exists, possibly in the form of nanocrystallites, around both the Fe and Zn atoms. The EXAFS spectra recorded at the Fe-K-edge were fitted using a mixed model according to which X% of the Fe atoms occupy octahedral sites that belong to the ZnFe2O4 phase, while (1X)% form tetrahedra, and participate in the formation of the glass. In both the as-casted and annealed samples, the fitting procedure was performed using four shells: the first nearest neighbor (nn) shell consists of oxygen atoms that are either octahedrally or tetrahedrally coordinated to the Fe atom. The second, third and forth nn shells comprise of Fe, Zn and O atoms, according to the ZnFe2O4 model. The Zn-K EXAFS spectra were fitted in the four nn shells, using the ZnFe2O4 model: the Zn atom in the 1st nn shell is tetrahedrally coordinated with four oxygen atoms at a distance 2.02Å, while the 2nd, 3rd and 4th shells comprise of Fe, O and Zn atoms, respectively

Figure 2: 7500x15µm XRF maps of the annealed sample. The regions depicted as E and C, refer to the positions at the edge and at the centre, respectively.

Figure 3: Fourier Transforms of the (a) Fe-K and b) ZnK edge µ-EXAFS spectra recorded at the edge and the center of the annealed samples. The raw data and the fitting are shown in thin and thick line, respectively.

The analysis of the µ-EXAFS spectra from the annealed sample reveals that the percentage of the Fe atoms that constitute octahedra is, within the error bar, equal at the center and at the edge of the sample. More specifically, 55% of the Fe atoms form octahedra at the edges of the sample, while 65% form octahedra at the center of the sample. The Fe-O bond length in the FeO6 polyhedron is equal to 1.93Å whilst the respective value in the FeO4 tetrahedron is 1.86Å. Therefore, the intermediate role of Fe is disclosed: Fe participates in the formation of the vitreous matrix by forming FeO4 tetrahedra and acts as a glass modifier when bonded in FeO6 octahedra. Thus, Fe partially occupies octahedral sites that belong to ZnFe2O4 nanocrystallites, which are not resolved in the XRF maps, due to the limited spatial resolution of the experimental setup. On the other hand, the µ-EXAFS analysis at the Zn-K edge reveals no alteration in the bonding environment of the Zn atom. In both regions, the Zn atom participates in the formation of ZnO4 tetrahedra, i.e. Zn is coordinated with 4 oxygen atoms at a distance 1.94Å and thus acts as a glass former.
1 2

Sheng, J., 2002, Fuel 81, 253 Pellino M., A. Karamanov, P. Pisciella, S. Criscussi and D. Zonetti, 2002, Waste Management 22, 945.

302

Resonant x-ray Bragg scattering on a magnetic grating patterned by ion bombardment in a magnetic field
V. H¨ink, M. D. Sacher, J. Schmalhorst, and G. Reiss o
Thin Films and Nanostructures,Department of Physics, University of Bielefeld, P.O. Box 100131, 33501 Bielefeld, Germany

D. Engel, T. Weis, and A. Ehresmann
Institute of Physics and Center for Interdisciplinary Nanostructure Science and Technology (CINSaT), Kassel University, Heinrich-Plett-Str.40, D-34132 Kassel, Germany

The exchange bias effect (EB) in antiferromagnet (AFM) / ferromagnet (FM) systems results in a shift of the hysteresis loop of the FM with respect to zero magnetic field [1], caused by the exchange interaction at the interface between both materials (pinning). The EB is usually initiated by field cooling (FC) resulting in a homogenous distribution of the EB orientation within the sample. An alternative method for initializing the EB is the ion bombardment (IB) of the sample in the presence of an external magnetic field HIB [2–5]. Besides the initialization of the EB it is also possible to change of the size and orientation of an existing EB-coupling by IB. Furthermore it is possible to define the area in which the unidirectional anisotropy is manipulated by a local restriction of the area exposed to IB. For small external fields below HEB this local manipulation of the EB-coupling results in a corresponding pattern of the magnetization of the pinned FM. Our measurements show no unintentional change of the topographical microstructure due to the magnetic patterning. This technique opens the opportunity to build a magnetic grating showing a soft x-ray interference pattern based on the x-ray magnetic circular dichroism (XMCD) effect [6]. It will be demonstrated that it is possible to switch the magnetic scattering on and off by an external magnetic field. The sample used in the present experiment consists of a layer stack with 3 nm Co70 Fe30 pinned to 15 nm Mn83 Ir17 with a Cu seed layer and 1.4 nm Al oxide as an oxidation preventing capping layer. A homogenous EB was initialized by FC in an external magnetic field HF C . Ion bombardment induced magnetic patterning (IBMP) with He-ions (10keV, 1 × 1014 ions/cm2 ) in an external magnetic field HIB was performed through a resist mask with 1.6 µm wide lines parallel to HF C and a periodicity of 5 µm (HIB ↑↓ HF C ). The test of the magnetic grating was performed with the ALICE experimental setup [7] at beamline UE56/1-PGM-b at BESSY. A circular polarized x-ray beam was directed at the rotatable sample with a variable angle while the detector was fixed at an angle of Θ=20◦ relative to the incident beam (Fig. 1). The grating was tilted by about 45◦ relative to the plane of the incident/reflected beam. This made it possible to measure the interference pattern which results from the XMCD effect with a detector in the plane of the incident / reflected beam. The direction of the external magnetic field was oriented parallel to the direction of the incident photon beam. All measurements were done at room temperature.

T o p v ie w :

R e fle c tiv ity (a rb . u n its )

X -ra y
d e te c to r

φ
θ

S id e v ie w :

X -ra y

d e t.

R e fle c tiv ity (a rb . u n its )

θ

FIG. 1: ExperT o p v ie w : imental setup 1 .8 1 .8 with θ b o m b a r Θ d angle d e b o m b a rd e d 1 .6 1 .6 o n g ra tin g o n g ra tin g X -r φ betweena y sample (F C & IB ) (F C & IB ) 1 .4 1 .4 surface d e t and e c to r n o t n o t incident v b i photon S b i d o e m e. w : 1 .2 1 .2 b o m b . beam and θ ann e x t to g ra tin g n e x t to g ra tin g 1gle φ between .0 1 .0 (F C , n o IB ) (F C , n o IB ) detector and 0 .8 0 .8 X -ra y incident photone t . d 0 .6 0 .6 beam, rotation axis perpendicu-2 .0 -1 .5 -1 .0 -0 .5 0 . 0 - 2 0 . 0 . 5 - 1 1. 5 . 0 - 1 1 . 0 . 5 - 0 2 . 5 . 0 0 . 0 0 .5 1 .0 1 .5 lar to scatteringa l m a g n e t i c f i e l d E [ k x O t e e r ] n a l m a g n e t i c f i e l d [ k O e ] E x te rn plane;

2 .0

FIG. 2: Hysteresis loops measured on the magnetic grating (black diamonds) and next to the magnetically patterned grating (red circles) on the same sample (Co L3 -edge, Θ= 10◦ , φ=20◦ , loops shifted on y-axis for better visibility)

303

5

Reflectivity [arb. units]

4 3 2

+2700 Oe (saturation)

remanence

-2700 Oe (saturation)

1 10,2 10,4

Sample rotation Θ [°]

10,6

10,8

11,0

FIG. 3: Sample rotated by angle Θ with the detector fixed to φ=20◦ in saturation in a magnetic field of +2700 Oe (green line) and -2700 Oe (red line) and without an external magnetic field after saturation at 2700 Oe (black line). All measurements were done at the Co L3 -resonance.

As it can be seen in Fig. 2, bombarded areas as well as not bombarded areas contribute to the hysteresis loop measured on the magnetic grating. The change in reflectivity due to the not bombarded area between the lines is approximately twice as high as for the bombarded lines. This corresponds to the fact, that the average distance between the lines is by a factor of about two larger than the width of the bombarded area. The hysteresis loop of the same layer system measured on the only field cooled area aside the magnetically patterned grating can be seen in the lower part of Fig. 2. The shape of the hysteresis loops is rounded in both cases because the measurement is not performed parallel to the easy axis of the pinned Co70 Fe30 -Layer. Magneto optical Kerr effect measurements carried out parallel to the easy axis (not shown) show that the shift of the hysteresis loop due to EB is for all areas much larger than the the coercivity field. Therefore, one can assume an approximately antiparallel orientation of the FM-magnetization in the bombarded and not bombarded areas in remanence. A small deviation of the antiparallel orientation of up to 5◦ might occur because of a possible small inaccuracy in the alignment of the directions of HF C , HIB and the lines. Apart from the charge scattering of the x-rays which is independent of the IB-induced magnetic changes, a resonant magnetic scattering occurs at the L2,3 -white lines of Co and Fe. When investigated in remanence, the part of the scattered x-rays which is due to resonant magnetic scattering will have a different intensity on areas with a magnetization direction oriented predominantly parallel to the incident beam than on areas with the opposite orientation of the magnetization. Therefore a superposition of an interference pattern like that of a topographic reflective grating which is due to resonant magnetic scattering and the signal obtained by the field independent charge scattering can be expected. Figure 3 shows a Θ-scan measured with a rotating sample and the detector fixed at Θ=20◦ (Co L3 -resonance). The maximum possible magnetic signal is defined by the difference between the two measurements for the sample saturated in the X-ray propagation direction and opposite to it. The expected oscillation of the reflectivity clearly can be seen at the measurement without an external magnetic field. This oscillation vanishes when a magnetic field of ±2700 Oe, which is sufficient to saturate the ferromagnetic layer (see Fig.2), is applied. This shows that the observed oscillation at the measurement in remanence is due to interference at the magnetic pattern and not related to any topographic structures. Θ-scans have been measured with various magnetic fields in the range between saturation at -2700 Oe and saturation at +2700 Oe (not shown). The angles of the individual maxima are not changed considerably by the magnetic field. The magnitude of the interference pattern of the magnetic grating decreases with increasing magnetic field and vanishes at a magnetic field larger than 675 Oe which is in the range of the external field necessary to saturate the sample (see Fig. 2). This shows that it is possible to build a gradually tuneable magnetic grating by IBMP. The grating period d can be calculated by applying the grating equation [8] mλ = d × (sin(α) + sin(β)) (spectral order m, wavelength λ=1.59 nm, angle α between the incident beam and the surface normal, angle β between the scattered x-rays and the surface normal) to the experimental results. The first visible maxima next to the specular peak at 10.02◦ are located at 9.81◦ and at 10.32◦ . The measured distance between adjacent maxima is 0.12◦ to 0.13◦ . Therefore, the maximum at 10.32◦ (9.81◦ ) is of order -3 (2). The maxima of lower orders are hidden by the specular peak. The position of the maxima on the flank of the specular peak can be shifted in the direction of the peak by the superposition of the maxima due to

304

-1,5 -2,0 -2,5 -3,0 -3,5 -4,0 10,3 10,4 10,5 10,6 10,7 10,8 10,9

FIG. 4: Θ-scan measured in remanence subtracted by the Θ-scan measured in saturation at +2700 Oe; vertical lines: angle at which a maximum is predicted by the grating equation for a grating with d=1.6 µm (big black lines), d=3.4 µm (medium sized red lines) and d=5 µm (small blue lines)

XMCD and the specular peak itself (compare Fig.4). By taking into account that the bombarded lines were tilted by 45◦ relative to the horizontal plane defined by the incident beam and the detector and that the measured angles represent the projection of the interference pattern on this plane, values in the range of d=1.58 µm to d=1.92 µm can be calculated. This result does not match with the periodicity of the magnetic grating of d = 5µm, but it is in the range of the width of the bombarded lines. This is a hint that the N´el wall like boundaries between e the bombarded and the not bombarded areas are of a great importance for the interference process. The maxima one would expect from the interference of x-rays scattered at N´el wall like e boundaries with a distance of 3.4 µm corresponding to the width of not bombarded areas have a smaller distance. Therefore more maxima of this kind are hidden by the specular peak and only weaker maxima of higher order can contribute to the observed interference pattern. Figure 4 shows the angles at which maxima can be expected according to the grating equation for the used experimental setup and a grating with d=1.6 µm, d=3.4 µm and d=5 µm. To compare this values with the experimental results, the difference between the reflectivity measured at a Θ-scan in remanence and that resulting from a measurement done in saturation at +2700 Oe is printed in the same graph. This emphasizes the interference signal due to magnetic scattering. It can be seen, that interference at a grating with a periodicity of d=5 µm can not explain the observed interference pattern and therefore the observed angular distribution of the reflected xrays might be connected to the resonant scattering of x-rays at the magnetization of the N´el walls. e We have shown, that it is possible to measure an x-ray interference pattern with a purely magnetic grating patterned by ion bombardment induced magnetic patterning. Regular topographical patterns as an origin of the interference pattern can be ruled out. Scattering at the edges between bombarded and not bombarded areas as an important contribution for the observed interference pattern has been suggested. The authors thank H. Zabel, A. Nefedov and A. Remhof (Ruhr-Universit¨t Bochum) for the a possibility to use the ALICE experimental setup und the valuable assistance during the measurements at the BESSY. The support of the DFG and the BMBF is acknowledged (grant number: BMBF 05 ES 3XBA/5).

[1] J.Nogues and I. K.Schuller, Journal of Magnetism and Magnetic Materials 192, 203 (1999). [2] C. Chappert, H. Bernas, J. Ferre, V. Kottler, J.-P. Jamet, Y. Chen, E. Cambril, T. Devolder, F. Rousseaux, V. Mathet, et al., Science 280, 1919 (1998). [3] J. Fassbender, D. Ravelosona, and Y. Samson, Journal of Physics D: Applied Physics 37, R179 (2004). [4] A. Ehresmann, Recent Research Developments in Applied Physics 7 part 2, 401 (2004). [5] D. Engel, A. Ehresmann, J. Schmalhorst, M. Sacher, V. H¨ink, and G. Reiss, Journal of Magnetism o and Magnetic Materials 293, 849 (2005). [6] J. St¨hr, Journal of Electron Spectroscopy and Related Phenomena 75, 253 (1995). o [7] J. Grabis, A. Nefedov, and H. Zabel, Review of Scientific Instruments 74, 4048 (2003). [8] X-Ray Data Booklet, second edition (Center for X-Ray Optics , Advanced Light Source, Berkeley, 2001).

305

Long-range order in thin epitaxial Fe3 Si films on GaAs(001)
B. Jenichen, V. M. Kaganer, J. Herfort, D. K. Satapathy, H.-P. Sch¨ nherr, W. Braun, and K. H. Ploog o Paul-Drude-Institut f¨ r Festk¨ rperelektronik, Berlin u o

Fe3 Si on GaAs is a promising candidate for spintronic applications. It can be grown by molecularbeam epitaxy at GaAs substrate temperatures near 200 ◦ C. The Curie temperature of Fe3 Si is as high as 840 ◦ C. Fe3 Si has the face-centered cubic D03 structure. This structure can be re1 1 garded as four interpenetrating fcc sublattices A, B, C, and D with origins at (0, 0, 0), ( 4 , 4 , 1 ), 4

( 1 , 1 , 1 ), and ( 3 , 3 , 3 ), respectively. In the ordered Fe3 Si crystal, Fe atoms occupy the three 2 2 2 4 4 4 sublattices A, B, and C, while Si atoms fill the sublattice D. Si(D) and Fe(B) are located on opposite corners of the same cube. In as-grown thin epitaxial films of Fe3 Si on GaAs, the lattice mismatch and the long-range order depend on the stoichiometry. Significant changes of the resistivity with film stoichiometry have been observed [J. Herfort et al., J. Vac. Sci. Technol. 22, 2073 (2004)]. A clear minimum of the resistivity near stoichiometric conditions is found, which had previously been ascribed to reduced alloy scattering. The scattering is smallest for an ordered crystal structure, but the degree of order has not been determined. The aim of the present work is the determination of long-range order in thin Fe3 Si films by grazing incidence x-ray diffraction. In addition, we use the same measurements for a precise determination of the position of the Fe3 Si unit cell with respect to the GaAs substrate. Since the structures of the substrate and the epitaxial layer are different, the relative positions of the unit cells are not known in advance, and an additional phase factor in the polarizability of one layer with respect to the other may appear. A common origin for all structures is chobut it is not taken into account in conventional dynamical diffraction calculations. Figure 1 shows the influence of the relative shift of the crystal lattices on the calculated diffraction curves. We compare a crystal truncation rod (CTR) scan at the 111 reflection calculated for relative positions of the Fe3 Si and GaAs lattices without a lateral shift with the same scan for the case that the Fe3 Si lattice is shifted by one half of the unit cell along the [100] direction in the interfacial plane. The corresponding shift of the layer thickness oscillations in the
Intensity (arb. units)

sen. This is standard in surface diffraction,

11l

0.9

1.0 l

1.1

Fig. 1. Calculated CTRs at the 111 reflection from a stoichiometric fully-ordered Fe3 Si layer (30 nm thick) on GaAs(001) for the lattice registry without lateral shift (full line) and for the case of the Fe3 Si structure shifted by 1/2 of the unit cell in the [100] direction (dashed line).

diffraction curve is obvious. Thus, we have included the relative positions of the layer and the substrate crystal lattices in our calculations to obtain the registry of the layer on the substrate.

306

The z-component (normal to the interface) of the shift corresponds to a relaxation at the interface. Three types of reflections are expected for the ordered stoichiometric compound Fe3 Si. Fundamental reflections, which are not influenced by disorder, are given by h + k + l = 4n, where n is an integer and (hkl) are the Miller indices of the diffracting net planes. There are two distinct types of superlattice reflections sensitive to the disorder, which is described by the two order parameters α and β , defined as the fractions of Si atoms occupying the Fe(B) and Fe(A,C) sites, respectively.
GaAs Fe3Si hkl 002

Reflections with odd h, k, and l are sensitive to both types of disorder. Reflections given by the condition h + k + l = 2n (where n is an integer), however, are sensitive to disorder in the Fe(A,C) sublattice alone. In order to obtain α and β , we measure diffraction curves of all three types of reflections and fit them using dynamical diffraction calculations. Figure 2 combines
111 113 311 022 004

Normalized intensity

222

CTR scans performed at different bulk reflections. The width of the GaAs substrate peak is limited by the resolution of the experimental setup. The Fe3 Si layer peak is broadened due to the small layer thickness of 33 nm. Periodic side maxima reveal the high quality of its top and bottom interfaces. The layer peak is fairly close to the substrate peak, indicating that the sample is nearly stoichiometric. Well-pronounced Fe3 Si layer reflections in all these scans with their different sensitivity to disorder indicate a high degree of long-range structural order in the layer. The long-range order in Fe3 Si epitaxial layers strongly depends on their stoichiometry. In our example the sample contains 25.5% Si,

-0.05

0.00

∆l

0.05

Fig. 2. Measured (gray lines) and fitted (black lines) CTRs on a logarithmic scale near different reciprocal lattice points. The fits of all curves are performed simultaneously with the same set of parameters. The sample is nearly stoichiometric with a slight surplus of Si and exhibits an ordered structure with some disorder due to migration of Si atoms into the Fe(A,C) sublattice. The coordinate L along the CTRs has its origin at the respective GaAs peak.

i.e., it is nearly stoichiometric. The resulting order parameters are α = 0 and β = 0.3 with an accuracy of ±0.05, i.e., 30% of the Si has left its sublattice exchanging solely with Fe(A,C) atoms. For a clearly nonstoichiometric sample with 16.5% Si, we detected complete disorder (α = 0.25 and β = 0.5, vanishing Fe3 Si peaks in the superlattice reflections), which resulted in a higher resistivity of the layer. Note that the almost complete long-range order in nearly stoichiometric samples is observed without additional thermal treatment, which is usually required for the preparation of bulk samples.

307

NEXAFS spectra of polymer-fullerene (C60) composites A.O. Pozdnyakov1, M.M. Brzhezinskaya2, D.A. Zverev2, E.M. Baitinger3, A.S. Vinogradov2, K. Friedrich4
Institute of Problems of Mechanical Engineering, St. Petersburg,199178, Russia V.A. Fock Institute of Physics, St. Petersburg State University, St. Petersburg, 198504 Russia 3 Preussiches Privatinstitut für Technologie zu Berlin, D-13187 Berlin, Germany 4 Institut für Verbundwerkstoffe GmbH, Technisches Universität Kaiserslautern, Kaiserslautern, D67663, Germany
2 1

Polymer-fullerene composites (PFC) are studied in depth because they exhibit interesting optical, mechanical, biological and other properties [1]. The properties of the composites are strongly dependent on the dispersion state of fullerene molecules inside the matrices. The degree of dispersion can vary over a wide range from molecules to clusters and crystallites. Donor-acceptor properties of polymer matrices and the solvent used for the formation of the composite may determine the degree of dispersion and, as a consequence, the character of thermal desorption of fullerene C60 molecules [2]. Below we report on the study of the electronic subsystem of PFCs using the NEXAFS technique. Polymer matrices with different physical properties have been chosen. One of the polymers, poly(pyromellitic dianhydride-co-4,4'-oxydianiline) (PMDA-ODA), is a typical high temperature rigid chain polymer which has the glass transition temperature (Tg) above 600 K. Polydimethylsiloxane is a flexible chain polymer with low Tg~126 K. Poly(methyl methacrylate) (PMMA) and polystyrene (PS) are amorphous polymers with intermediate Tg~400 K and 378 K, respectively1. Polymer-C60 suspensions were prepared by co-dissolution of the C60 solutions either in toluene (Tol) or in 1-methyl2-pyrrolidinone (NMP). PS and PMMA were dissolved in Tol. PDMA-ODA is a 15 wt. % NMP solution. The solution of C60 in NMP was added to this solution. The C60 content in NMP and Tol before mixing with a polymer solution amounted to the equilibrium solubility of C60 in these solvents at room temperature (2.9 and 0.89 mg/ml for Tol and NMP, respectively [3]). Details of handling the suspensions and coatings formation can be found elsewhere [2]. Common solutions were pipetted onto the steel foil (12X18H10T) with a roughness of about 1 µm. The mean rated coating thickness of PFC coatings was ~ 5 µm. PMDA and PMDA-ODA-C60 coatings were 286.1 286.6 C60 powder heated up to ~ 320oC in air for 284.7 C60 from NMP imidization. C60 from Tol 288.4 All measurements were performed during the single-bunch beam time at the Russian-German beam line at BESSY-Π. The NEXAFS spectra of these composites were obtained at room temperature (RT) in the total electron yield mode by detecting a sample current. No noticeable charging effects were observed in the experiments. The photon282 284 286 288 290 292 294 energy resolution was set to 0.15 eV Photon Energy (eV) at the C1s edge (~285 eV). The Fig. 1 C1s NEXAFS spectra of C60 powder and C60 layers absorption spectra were normalized deposited from toluene and 1-methyl-2-pyrrolidinone to the incident photon flux, which was monitored by measuring the at the surface of stainless steel substrate total electron yield from a clean
The following materials have been used: PMMA Fluka, standard Mw=500000; PS Fluka, 81414, standard Mw=500000; poly(pyromellitic dianhydride-co-4,4'-oxydianiline) amic acid (Aldrich, 575801, ca. 15 wt.% sol. in 1-methyl-2-pyrrolidinone (NMP); fullerene C60 (522500, Aldrich, sublimed, 99.9%); toluene (650579, SigmaAldrich, Chromasolv plus, for HPLC >99.9%); 1-methyl-2-pyrrolidinone (270458, Aldrich, 99+% HPLC grade).
1

Total Electron Yield (arb. un.)

308

gold substrate with the use of a channeltron. The photon energy in the region of the C1s absorption spectra was calibrated using the Ti2p absorption spectrum of K2TiF6 (459.0 eV [4]). Lateral resolution of the method is about 300 µm. All spectra were normalized to the intensity of the main absorption band in the spectrum. In this report NEXAFS spectra at the C1s threshold obtained on PMMA-C60, PS-C60 and PMDA-ODA-C60 composites are briefly compared with the spectrum of neat C60 layers. The N1s and O1s absorption spectra (not reported here) were investigated additionally. Fig 1 shows C 1s NEXAFS spectra for the C60 fullerite powder covered C60 powder 284.7 286.6 over the steel substrate as well as C60 films PS + RT o formed at the steel substrate from NMP PS-C60 + 180 C and Tol solutions. The general similarity in PS-C60 + RT the shape of the NEXAFS spectra shown in Fig 1 is obvious. This is due to the rigidity of the structure of the C60 molecule. The energy position of LUMO (284.7 eV), LUMO+1 (286.1 eV), LUMO+2 (286.6 eV), LUMO+3 (288.4 eV) corresponding to the first, second, third and forth unoccupied orbitals (vertical dotted lines) are in a good agreement with literature [5]. Note the differences in the shape of both 282 284 286 288 290 292 294 the LUMO (284.7 eV) and LUMO+1 Photon Energy (eV) (286.1 eV) for C60 layer deposited from Fig. 2 C1s NEXAFS spectra of PS-C60 composite NMP compared with the spectrum of C60 subjected to different heat treatment regimes layer deposited from Tol and the spectrum taken from C60 powder. Additionally, the LUMO+3 is evidently more intense for C60 powder 284.7 286.6 C60 deposited from Tol. These effects PMMA + RT 286.1 may be tentatively explained to result PMMA-C60 + RT from the donor-acceptor (polar) o PMMA-C60 + 180 C short interaction of NMP and Tol molecules o PMMA-C60 + 180 C long with C60 [6]. Similar effects are probably characteristic also of the solutions containing carbon nanotubes since NMP is a good solvent for these objects as well [7]. Fig 2 presents spectra of the PSC60 composite after heating at different temperatures in air along with the spectra of neat PS and C60 powder. The 282 284 286 288 290 292 294 vertical dashed lines indicate the Photon Energy (eV) positions of LUMO and LUMO+2 in Fig. 3 C1s NEXAFS spectra of PMMA-C60 composite the spectrum of C60 powder. Note that additional contribution of C60 in the subjected to different heat treatment regimes spectrum of PS-C60 composite is seen as clearly discernable shoulders on both sides of the main absorption band of PS spectrum. The decrease and yet disappearance of these shoulders has been observed for the PS-C60 composite heated at 180oC. Fig. 3 shows the spectra of PMMA-C60 composite formed at RT and after treatment at about 180oC for different time periods in air. The spectrum of a composite formed at RT indicates the presence of the LUMO, LUMO+1 and LUMO+2 of C60. Notable is the decrease in relative intensity of LUMO+1 upon heating the PFC and the loss of intensity of the peaks characteristic of PMMA matrix on prolonged heating at 180oC. More detailed quantitative data analysis is under way to reveal if this is an effect of thermal degradation of the matrices and/or the indication of its interaction with C60. Fig. 4 shows the spectra of PMDA-ODA-C60 composite along with the spectra of neat PMDAODA and C60 powder for comparison. No discernable effect of C60 on the shape of C1s NEXAFS spectrum is observed. One may speculate that the sedimentation process results in the drop of C60
Total Electron Yield (arb. un.)

Total Electron Yield (arb. un.)

309

concentration below the sensitivity threshold of the experimental technique (~2 wt. %). In order to elucidate this effect the experiments on the samples formed from the sediment are needed. Finally, the results obtained reveal changes in the shape and position of electronic states of C60 in different PFCs and solvents. The effect of temperature in C60 powder redistribution the intensity of these 284.7 PMDA-ODA 286.6 states is observed pointing to the PMDA-ODA-C60 probable changes in the interaction mechanism between polymer matrices and C60 upon heating. The results provide good grounds for further studies of the interrelated problems of the nature of C60 dispersion states in the polymer matrices and the polymer-fullerene interaction mechanisms.
Total Electron Yield (arb. un.)

282

284

This work was supported by the bilateral Program "Russian-German Laboratory at BESSY". We thank Dr. Yu.S. Dedkov for valuable technical assistance. A.O. Pozdnyakov expresses gratitude to the Alexander von Humboldt Foundation for support of his research stay at IVW, Technical University of Kaiserslautern.
Fig. 4 C1s NEXAFS spectra of PMDA-ODA-C60 composite
References 1 2 3 4 5 6 7 The physics of fullerene-based and fullerene-related materials, Ed. W. Andreoni, Kluwer Academic Publishers, 2000, 445p. A.O. Pozdnyakov, B.M. Ginzburg, T.A. Maricheva, et. al., Fiz. Tverd. Tela, 2005, 12, 2239-2245 [Phys. Solid State, 2005, 47(12), 2333-2340] V.N. Bezmelnizin, A.V. Eleckii, M.V. Okun, Usp. Phys. Nauk, 1998, 168(11), 1195 A.S. Vinogradov, A.Yu. Dukhnyakov, V.M. Ipatov, et al., Fiz. Tverd. Tela, 1982, 24, 1417 [Sov. Phys. Solid State 24, 803 (1982)] J.-H. Guo, P. Glans, P. Skytt, N. Wassdahl, et. al., Phys. Rev., B, 1995, v. 52, № 15, 10681-10684 I. Baltog, M. Baibarac, L. Mihut, N. Preda, T. Velula, S. Lefrant, Romanian Reports in Physics, 2005, 57(4), 807-814 K. D. Ausman, R. Piner, O. Lourie, and R. S. Ruoff, The Journal of Physical Chemistry B, 2000, 104(38), 8911-8915

286 288 290 292 Photon Energy (eV)

294

Acknowledgements

310

X-ray emission and photoelectron spectroscopy at Cu(In,Ga)(S,Se)2 thin film solar cells in the CISSY endstation 2. Cd2+/NH3 treatment -induced formation of a CdSe surface compound on CuGaSe 2 thin film solar cell absorbers M. Bär1,2, S. Lehmann1 , L. Weinhardt2,3, M. Rusu1 , A. Grimm1 , I. Kötschau1 , I. Lauermann1 , P. Pistor1 , S. Sokoll1 , Th. Schedel-Niedrig1 , M.C. Lux-Steiner1,4, C. Heske2 , Ch. Jung5 , and Ch.-H. Fischer1,4, (1) Solarenergieforschung (SE 2), Hahn-Meitner-Institut, 14109 Berlin, Germany (2) Department of Chemistry, University of Nevada, Las Vegas, NV89154, USA (3) Experimentelle Physik II, Universität Würzburg, 97074 Würzburg, Germany (4) Freie Universität Berlin, 14195 Berlin, Germany (5) BESSY GmbH, 12489 Berlin, Germany
CuGaSe 2 (“CGSe”)-based high-gap thin film solar cells have up to date not reached their potential performance level. To elucidate possible shortcomings of the electronic interface structure, we have studied the initial stage of the interface formation between the CGSe absorber and the CdS buffer layer by use of a simple Cd2+/NH3 -treatment [1]. For our investigations, we used CGSe/Mo/soda-lime glass structures. Mo was evaporated onto the glass substrate, followed by the deposition of the CGSe absorber using chemical close-spaced vapor transport (CCSVT) [2] with a [Ga]/[Cu] ratio of approx. 1.16. Each sample was cut into two pieces, one of which was treated in an aqueous solution of 1.5 mM CdSO 4 and 1.5 M NH3 for 10 min inside the glovebox of the CISSY apparatus. The other half was used as a reference. Se 3d Ga 3d Cd 4d Subsequently, both types of CGSe surfaces EPhot c) a) b) were characterized by photoelectron as174 depos. spectroscopy (PES) using synchrotron [eV] radiation (BESSY II, beamline UE 41-PGM). 174 For normalization purposes, the excitation intensity was also recorded. A CLAM4 254 electron spectrometer (Thermo VG 354 Scientific) was used. The electron spectrometer was calibrated using XPS and 454 Auger line positions of different metals (Cu 654 3p, Au 4f 7/2 , Cu L3 VV, and Cu 2p3/2 ). For the 854 synchrotron measurements, the zero point of the energy scale was adjusted for each 1254 excitation energy used, such that the Au 4f 7/2 56 55 54 53 21 20 19 13 12 11 10 Binding Energy [eV] reference line appears at a binding energy of 84.00 eV. From the Cd M45 N45 N45 Auger peak at Fig. 1 Se 3d (a), Ga 3d (b), and Cd 4d (c)
Norm. Intensity

E Cd M 4 N 45N45 = (381.6 ± 0.1) eV, CdSe is Kinetic
identified on the surface of the Cd /NH3 treated CGSe absorber (literature
Cd M N N
2+

4 values E Kinetic 45 45 (CdSe) between 381.4 and 381.7 eV). In order to investigate whether Cd atoms diffuse into the absorber and thereby occupy vacant cation sites or whether a separate CdSe surface compound is formed, we have varied the spectroscopic information depth by applying different excitation energies (EPhot). Since the valence states of Se 3d, Ga 3d, and Cd 4d are energetically close together (binding energy between 11eV and 54 eV) and the escape depth λ?cosϕ?and the spectrometer characteristic are almost identical, these signal intensities can be directly compared. Variation of the excitation energy between 1254 eV and 174 eV resulted in a decrease of kinetic energies of the considered photoelectrons from

photoemission lines measured at different excitation energies (EPhot ). All spectra are presented on the same intensity scale. After removal of a linear background they were normalized by the excitation intensity and the respective photoionization cross sections.

311

Cd /NH3-treated

2+

1221 eV to 141 eV and, consequently, also a decrease of their escape depth from approx. 1.2 nm to 0.4 nm (for ϕ ≈ 45°) [3], i.e., the surface sensitivity is strongly enhanced. Fig.1 shows the Se 3d, Ga 3d, and Cd 4d signals of the treated CGSe sample for such an energy variation, as well as that of the untreated reference for 174 eV. The intensity of the Cd 4d peak increases with decreasing excitation energy. The Se 3d signal of the treated sample stems from both, CuGaSe 2 (higher binding energy) and CdSe (lower binding energy). Peak-fitting based on a Voigt function doublet reveals that the CdSe contribution also increases for higher surface sensitivity. Both observations clearly indicate a very thin CdSe layer on the very top of the treated surface. In Fig. 2a the calculated Cd/Ga- and Escape Depth, λ∗cos ϕ [nm] ϕ [nm] Photon Energy, EPhot [eV] Escape Depth, λ ∗cos Photon Energy, E [eV] Cd/Se CGSe-ratios are plotted as a 1.13 1.02 0.91 1.02 0.91 0.78 0.62 0.40 200 200 380633633 959 380 1.13 0.78 0.62 0.40 959 1358 1358 8 2 8 2 function of excitation energy and Cd/Ga SeCdSe /Se CGSe Cd/Ga SeCdSe/SeCGSe escape depth. For the latter, the area Cd/Se CGSe Cd/SeCGSe of the Cd 4d line was compared with the intensity of the Se 3d 4 1 contribution of CGSe. Both ratios 4 1 increase with decreasing photon a) b) energy. In Fig. 2b, the ratio of the ( ) a) b) ∼c ∗[exp(d/λ ∗cosφ)-1] two contributions to the Se 3d signal 0 0 1200 1000 800 600 400 200 c∗ [exp(d/λ∗cos φ)-1] ∼ 0.4 0.6 0.8 1.0 ( ) 1.2 (Se CdSe/Se CGSe) are plotted in the Photon Energy, E [eV] Escape Depth, λ∗cosϕ [nm] 0 0 same way. Assuming an abrupt 1200 1000 800 600 400 200 0.4 0.6 0.8 1.0 1.2 Photon Energy, EPhot [eV] Escape Depth, λ∗cos ϕ [nm] CdSe/CGSe interface, the area FIG. 3 intensity ratio of Se CdSe/Se CGSe can Fig. 2a. Cd/Ga - and Cd/SeCGSe-ratios calculated from the spectra FIG. 3 be calculated to be proportional to shown in Fig. 1 versus excitation energy (corresponding escape the corresponding concentration depth at the top). ratio cCdSe/c CGSe = c, multiplied by b. Ratio of the two contributions to the Se 3d line (exp(d/(λ⋅cosϕ)) - 1). As shown by (SeCdSe/SeCGSe) versus escape depth of the respective the fit in Fig. 2b, the escape depth photoelectrons and fit of the data to determine the thickness of dependence of the Se CdSe/Se CGSe the CdSe surface layer (solid line). At the top, the corresponding ratio can be described well by this excitation energy of the respective spectra in Fig. 1 is given. approach. The thickness of the CdSe surface layer is determined to d = (0.7 ± 0.1) nm, corresponding to about one monolayer of CdSe. This is in good agreement with the value determined based on the attenuation of the absorber peaks induced by the CdSe layer, as judged from the Ga and Cu signals [4]. In summary, we do not find any experimental evidence for a significant Cd diffusion into the CGSe absorber. We conclude that Cd–Se bonds are formed at the CGSe absorber surface after a Cd2+/NH3 treatment, which is in close analogy to the previously observed formation of a CdS monolayer at the S-rich CIGSSe absorber surface [5]. This demonstrates the possibility to modify the electronic surface and interface structure by deliberate surface modification treatments and describes the early stages of the CdS/CGSe interface formation by a conventional CdS chemical bath deposition.
Phot

Intensity Ratio [a.u.]

Intensity Ratio [a.u.]

Phot

References [1] M. Bär, S. Lehmann, M. Rusu, A. Grimm, I. Kötschau, I. Lauermann, P. Pistor, S. Sokoll, Th. SchedelNiedrig, M. Ch. Lux-Steiner, Ch.-H. Fischer, L. Weinhardt, C. Heske, Ch. Jung, Appl. Phys. Lett. 86 (2005) 222107 [2] M. Rusu, S. Wiesner, D. Fuertes Marrón, A. Meeder, S. Doka, W. Bohne, S. Lindner, Th. Schedel-Niedrig, CH. Giesen, M. Heuken and M.Ch. Lux-Steiner, Thin Solid Films 451-452 (2004) 556 [3] M.P. Seah, W.A. Dench, Surf. Interf. Anal. 1, (1979) 2 [4] M. Bär, M. Rusu, S. Lehmann, S. Sokoll, A. Grimm, I.M. Kötschau, I. Lauermann, P. Pistor, L. Weinhardt, O. Fuchs, C. Heske, Ch. Jung, W. Gudat, Th. Schedel-Niedrig, M.Ch. Lux-Steiner, and Ch.-H. Fischer, Proc. 31st IEEE PVSC, Lake Buena Vista, 2005, p. 307-310 [5] L. Weinhardt, Th. Gleim, O. Fuchs, C. Heske, E. Umbach, M. Bär, H.-J., Muffler, Ch.-H. Fischer, M. Ch. Lux-Steiner, Y. Zubavichus, T. P. Niesen, F. Karg, Appl. Phys. Lett. 82, (2003) 571

312

X-ray emission and photoelectron spectroscopy at Cu(In,Ga)(S,Se)2 thin film solar cells in the CISSY endstation 1. XES of a liquid/solid interface through a 100 nm Si3 N4 -window I. Lauermann(1), A. Grimm(1) , M. Bär(1,2), S. Lehmann(1), Ch. Loreck (1), H. Mönig(1), S. Sokoll(1) , M. Ch. Lux-Steiner(1,3), C. Heske(2), Ch. Jung(4) , and Ch.-H. Fischer(1,3) (1) Solarenergieforschung (SE 2), Hahn-Meitner-Institut, 14109 Berlin, Germany (2) Department of Chemistry, University of Nevada, Las Vegas, NV89154, USA (3) Freie Universität Berlin, 14195 Berlin, Germany (4) BESSY GmbH, 12489 Berlin, Germany

Liquid flow Sample movement

Chalcopyrite (“CIGSSe”) thin film solar cells with a typ ical structure n+-ZnO/iZnO/CdS/Cu(In,Ga)(S,Se)2 /Mo/glass
Sample Window Lever for sample movement

are

promising candidates for future low-cost, high efficiency power conversion devices. Some functional materials in chalcopyrite

Figure 1: Cross-section of the flow-through cell, top view (drawing by PINK® GmbH).

thin film solar cells, e.g., buffer layers like CdS, are usually deposited wet-chemically.

However, the exact mechanisms of material formation, inter-diffusion, and possible chemical changes of the absorber are often unknown. In-situ analysis is necessary to answer some of these questions. Therefore, we developed a UHV-compatible flow-through cell for in-situ x-ray emission spectroscopy at liquid/solid or gaseous/solid interfaces. The set- up of the flowthrough cell is shown in Figure 1. The sample is mounted on a moveable stub inside the flowthrough cell. The variable distance sample-window between 0 and 5 mm allows either XESmeasurements (without prohibitive absorption by the liquid) or liquid flow (for continuous chemical reactions), respectively. A crucial part of the cell is the Si3 N4 window, which is only 100 nm thick (supplied by Silson Ltd, England). Yet is has to withstand a pressure difference of around 1000 mbar between the inside of the flow-cell (which is under ambient pressure) and the UHV in the analysis chamber. For that reason it was chosen to be as small as possible, i.e. the diSi3N4 window

mensions are 0.5 x 1 mm2 . The front view of the window mounted in the front flange of the flow-through cell is shown in Figure 2.

Figure 2: View of the Si3 N4 window of the flow-through cell

313

The CISSY flow-through cell was tested for UHV-compatibility, and first O Kα XES spectra of liquid water flowing through the cell were acquired in 2004. During the beam time in June 2005, we overcame window rupture problems and monitored Cu L2,3, Se L2,3, and S L2,3 XES spectra from a Cu(In,Ga)(S,Se)2 sample mounted inside the flow-through cell filled with water. To prove that the signal actually originated from the sample and not from contamination of the wet cell window or the solution, we measured spectra with the sample close (distance of > 1 µm) to the window and about 3 mm away from the window, respectively. While the attenuation of x-rays in water in the utilized energy range is sufficiently low to allow the detection of signals from a sample covered by a few µm of water, it is impossible to detect a signal through 3 mm of water. For example, at 1380 eV, the emission energy of the Se L3 peak, the attenuation length for an incidence angle of 45° is around 4 µm. Furthermore, when the sample is moved away from the window by more than 1 mm, it will be out of the line-of-sight of the spectrometer, which is mounted at an angle of 45° with respect to the wet cell window. Figure 3 shows Cu L2,3 and Se L2,3 XES spectra, acquired at a small and at a large samplewindow distance, respectively. Clearly no signal is visible at a large distance in the case of Cu. Since only the sample-window distance was changed between the two spectra, this shows that the detected signal actually originated from the sample. In the Se L spectrum recorded at 2,3 large sample-window distance, a small signal is visible at the Se L3 emission energy (vertical line). This could be due to traces of a dissolved Se species in the circula ting liquid which might have been produced in a photo-(electro)chemical reaction at the illuminated CIGSSe-surface. With this setup it is now possible to examine wet-chemical processes at surfaces in-situ.
1000

800

Cu L2,3 XES hν = 1150 eV Flow-through cell
800

Se L2,3 XES hν = 1500 eV Flow-through cell

600
Intensity [counts]

Intensity [counts]

>1µm samplewindow distance 3 mm samplewindow distance

>1 µm samplewindow distance 600 3 mm samplewindow distance

400

400

200

200

0 850 900 950 1000 Emission Energy [eV]

0 1300

1350

1400

1450

1500

Emission Energy [eV]

Figure 3: Cu L2,3 XES (left) and Se L2,3 XES spectrum (right) of CIGSSe “under water”, mounted in the CISSY wet cell. Red: sample is close to the window, black: sample is 3 mm away from the window

314

Structural and magnetic properties of [Co2MnGe/Al2O3]n multilayers
´ M.Vadala, K. Westerholt and H. Zabel
¨ Experimental Physik IV, Ruhr-Universitat Bochum, ¨ Universitatstrasse 150, 44780 Bochum Germany

From electronic energy band structure calculations the Co2 MnGe Heusler alloy has been predicted to be a half metallic ferromagnet with a 100% spin polarization at the Fermi level. This property is very attractive for application in spin dependent electron transport devices such as giant magnetoresistance (GMR), spintronics devices or for the injection of a spin polarized current into semiconductors. We have grown multilayers of the Heusler phase Co2 MnGe and normal metals like V and Au [1] and found that the formation of non ferromagnetic interlayers at the interfaces is a common feature. It was found before [2] that x-ray resonant magnetic scattering (XRMS) is well suited for the determination of the magnetization profile inside the Heusler layer. Following these lines, in our present experimental investigations we show the results obtained on [Co2 MnGe/Al2 O3 ]n multilayers. This system is very interesting because magnetic tunnelling junction using the fully spin polarized Heusler phases as electrodes typically possess Al2 O3 as tunnelling barrier. Since the TMR reacts sensitively on the spin polarization of the first few monolayers of the ferromagnetic electrodes, detailed knowledge of the surface magnetism of very thin Heusler layers at the interfaces with Al2 O3 is very important. Recent magnetization measurements on [Co2 MnGe/Al2 O3 ]n multilayers indicated that the interface magnetism in this system is similar to that observed for the Heusler/normal metal interfaces before. Thus in this contribution we present results of the determination of the magnetization profile using XRMS technique. [Co2 MnGe/Al2 O3 ]n Multilayers ˚ [Co2 MnGe] [thickness in A] ˚ [Al2 O3 ] [thickness in A] P14 P34

27.3 20.7 41.7 41.7

Table 1: Characteristics of the Heusler multilayers, investigated at the beamline UE56/1-PGM-b BESSY. In Table 1 the parameters of the samples used for the present study were summarized. The multilayers were deposited on single crystalline Al2 O3 substrates at room temperature; both samples have 25 bilayers. The experiments have been carried out at the beamline UE56/1-PGM-b, using the diffractometer ALICE [3]. The measurements have been performed with a fixed photon polarization (the right hand circularly polarized radiation) and a reswitching of the magnetization direction by applying a magnetic field. All measurements have been done at room temperature. In Fig.1 -the first row- we show the energy dependence of the scattering intensities measured on the sample P14 (taken at the first three multilayer Bragg peaks positions) after applying the magnetic field of 2000 Oe (blue curves) and the negative field of -2000 Oe (green curves). The corresponding asymmetry (the ratio between their difference and their sum) around the Co L3,2 absorption edges is shown in the second row. One clearly observes a non vanishing asymmetry, which gives evidence that the Co atoms in the multilayer possess a magnetic moment. The precise determination of the magnetization profile

315

Figure 1: Scattering intensity (the first row) and asymmetry ratio (the second row) taken for the sample P14 at the positions of the first three Bragg peaks around the Co L3,2 edges.

needs sophisticated model calculations and requires accurate fitting procedures which are now in progress. The authors acknowledge the financial support provided through the European Community’s Marie Curie actions (Research Training Networks) under the contract MRTN-CT-2003-504462, ULTRASMOOTH.

References
[1] U. Geiersbach, A. Bergmann and K. Westerholt, Thin Solid Films 425, 225 (2003) [2] J. Grabis, A. Bergmann, A. Nefedov, K. Westerholt and H. Zabel, Phys. Rev. B 72, 024438 (2005) [3] J. Grabis, A. Nefedov and H. Zabel, Rev. Sci. Instr. 74, 4048 (2003)

316

e ‰ h ‘ Ž v hp • Œ vp h xp x ‘ ‡ d „ q v x u ” ‘d x ‘ ‡ d qd w ‘ j h ‰p q ” ‘p ‘ s u ‘ 
p ˆ † w ‘ wp q † ‘ ¨–l—†ngSr4$‰‚‹˜@I‹‚d ˜ft$”™˜‹fT††‚ kxuY—nXÁœ†rTƒn†ƒnp j'œG­Yfƒf)©q gh 'j$jgt'”XV˜It©r‡‘$”9ƒVr‰¨fd kxj ~'j4‘©mqgtkˆ©o†nÁƒhgtX˜ƒX`7¡ j ˆ ep qd † ‘d ‘ w ” x d x d x †u ‘ ‡ d „ q d ” qp„ x w ‘u ˆ ‘p ‘ s u u j h ‰ u ” x d j q u ‘p j x d h x u ‘ ‡ d ‘ † ” h ‡ ” ‘ q e0'”gt©gri‘ oft©˜d ˜f†Bm¥l†f–'”uf‡t–V”†©mqgtX4™IIYa˜@IY'j¢bfœ–q 'jd 4”†oˆ € x d ” ˆ q ‰ • ‘p j x ‰ h u d ˆ q ‡ x s u ‘ ˆ x ” ‡ † ‘d € xp ‘dd h †u w ‘ ” x d ” ‘ ‰ ‘ 
 q Ž h ‘ ‡ d „ q j h p q„ u s j h x u h w ‘ u u h s p ‘d ‘ ‰ qd † hp x w j † x † u x x d hup ‘ 
 ‘ ‡ ¡ e ™ ‘ w » Ñ Æ º „ q v €p ‘ ” ‘ ” qd q ‡ x h ‡ x s ‘ j ‰ h ‘ Ž d ‘ ” € h i‰f4Af†H@ ©‹‘ Xn'pX‡ x• –‘ $jgton˜˜†gA'qANJ{IiInXi©fIt–˜d {$”’x–IS†©‰ € x w ” ‘ Ž · Ð — Ï d h w ‘ ‰p q„p ‘ x ‘p ‘ s u d ” ‘ xp ‘ x ‘ ‘ ‡ d ” x d x w hp ” qp d qp ‡ † ” v u „ q ” x d h x d ˆ ‘ ‡ d u w ” h ‰ ‘ $”©Xl–7–œfiƒ¦b&ln‰¶ftX4'‰uIPŒ ˜©mqgtƒh©H—©r˜InXV7IS©mqgt†m‚kx'jgt˜iY4w g€ „ „ q ‘ u h † x v u ” ‘ d x w ‘p ‘ d d h † u w ” h € x ‰ q † x ” ‘ ‘ sd ‘ Ž x d hp u ˆ q ‘ d ” h 
 w h x w ‘ ‡ d u ˆ e ‘ † h„p u ‘ ‡ Ev˜ITVXƒt$”Šu xd Xft$”©ƒrf‡t–V‘)$”S©†$”`†Áb†l®mqgtH‹©†ftYXlomu©i˜a‰”Iƒ‚rˆ –˜d q d ‘ x x u ” ‘ u ‘p h ” x d † ‘ ± ‘p } q · · z ‘ ‡ d d h } g€ „ z ” x d † hp x w v hp • Œ ‘ † ” ‘ x † x € x ‚ hp € w ” h u d ” ‘ ‰ ‘p u h ‘ f
 gd ˜V¨nXi©mqgƒA†n4Il­i˜ft{¶evV”©mqgXn¢Â ©™†ngS†w X$”t$”tH©”ftX4Xnˆ Xƒ4‰ } ¸ ¸ ªz vp d ‘ ‰ qd † ‘ ± ‘p v hp • Œ u h ‘p ‘ ‡ s y ‘ x dd j j hdu vp † ‘ ‡ d „ q u x dp ‘ x qp x j ˆ Ž ” h ‘ ‰ ‘ ‡ d d ˆ q Ž h ” x d h ‰p q„ 7¢{u¨r˜4©fƒA†r†H(¢Xƒn˜Iƒm†g‡tmhIYVfrX˜£‰ƒ‘ grI©nXkwa–˜t©IY©q g4¦bt$”x u j x v ” x d † ‘ ± ‘p € € hp h } · q qz ‘ ‡ d d h } g ¸ ªz ” x d † hp x w v hp • Œ j h xp d ‘ ‰ ‰ v 
 e w ‘ x Ž ‰ q † ‘p ‘ s u ‘ ˆ x ” ‡ † ‘ w Xm‘l©mqgƒA†n‹lH“vll¦a˜@ti¶@{u©q g†H@ ©†H(t–† u˜4aA‰¤†$”la©ƒ˜nXÁ‰ƒB¥ lƒfd € xp ‘ dd h † u v hp • Œ ‘ x d † ˆp d u ‘ w• ” q ” d ” ‘p ‘ x w u u ‘ ” ‡ € ˆ qp ‘ † h„p ‘ d x w ” h ‘ † h„p ˆ u u h j ‘ s u h ‘p ˆ d † ˆp d u ‘ x dd '”uXf‡ƒo4†H(˜
 g¦˜fƒ‡–A˜tXnXF ©vVƒ©An¨†X‚&ft$”iY†X‚&¶'jXÁÁXSr˜†u˜Vƒm†g‡tmhj ‘ † h„p u p h ‘ ” ‘ ‡ d w ” h j ˆ Ž ‘ ‡ d v w ˆ du q ¡ e ‘ † ” ‘p ‘„ ‘p h u h w ‘ 
p ‘ u w ” h ” x d x w hp x € xp ˆ w ‰ h ‘ Ž ” ƒ‚rˆ 4X–¯˜Y‘'w®˜)l˜f¯F´IƒXnbn©—9ƒrXo°Y)©mqgtƒh©Hu$p³'”uBΖl¯©mqx ‘ ‡ d ‰ qp„ w ‘ s q w h u u h s ” ‘ x † ‘ x u ‘ ‡ d „ q dp h x ˆ e j d ” h † x ” x u u u j v €p ‘ ” ‘ x ” qp d † j ‘ ‘ ‡ d € x € ” h ‡ ˜Í©n‡`ƒ‰–XX‡ itX4'‰XƒI–˜fIS&XÇt–v ˜tYƒ©¢ l©m€˜aolq `InX”† l©n˜ƒ‘ T˜•$”AX† d ˆ q ‡ x s u x d † ‘Ì qp x ‘ ‡ d „ q ‘d hdu ‘ €p h ‡ † ‘ ‡ d ‘ € ” h ‡ † qd x q„ p ‘ w hp € ‘ w j q € w x ‡ d ‰ Ë · h v Ž w ‘p ‘ 
 q † u h tA˜d {¶†‘ $jgƒ‡˜ra˜I4ftVV4lnXX4˜˜lYXX™fX$jAb“H©†™w ©I† l˜´“‰†tƒrI†vis u j x ‰ h u „ q u xp ‘ u ‘ ” Ê e jp ‘ h ‘ d ˆ x d u •p ‘ ” x ‘ • ” ‡ h ’ y É € ‘ ‡ d „ q ” qp d j † v † ‘ ‡ d d h À ‰ † È ˆ ˆ ¾½ q p ¼ ƒ‘ o“I{†m‘uXoS¶P$”kxu7ift˜d gf"”€ &X˜d ™— –‰7P7k
i˜@I©n˜Iq a˜vœut¯i¹ » » À q p ¼ p Æ d Å u ‘ † ” ‘ ˆ ± d ” ‘p ‘ x w qd x ˆ u ” x ˆ ˆ ™ ‘ — q Ä ¹ ‡ x s w ‘d x w hp x ‘p ‘ s u j x ‰ h u } p q qz | ” –ƒ» ¦½ t©aÇvv†ƒ£ftXn“ ©if¨©q Tœ3†™•I—˜d {iƒftƒh©Hu$p—n‰fƒ‘ a(œAl­@Ye€ „ q u ‘ † x x u ” x x w ” q † ” x d x w hp x ‘ ‡ d ” q € x w ” ‘ x ‘ w ” x d † ˆ w qp x ‘ € h ‰ h w ‘ ‡ d ‘d h x du ‘ 
 x qd p ‘ wp q ” I–†ƒ†m‘lt©mqgd ©©ƒ‘©mqgtƒh©Hu$p`˜‘©¯$”©Xl)©mqgBAr`lY•˜•ftm€gfƒt'”³f˜rY‘e€ e w † hp d j € x u h x j u ‘p j ˆ q ‡ u j x ‰ h u ‘ ‡ d qd € x ” x x x ” x ˆ ˆ ‡ † h º e u w † hp IH˜‘ A$”˜™™'”4d ˆ ƒrŠw AS‘ a(S˜‹f‹$”l©€ l'‰–©q –œB{ƒ†H˜d u ˆ q ‡ xp q ‰ h „ q ” x d hp ‘ ” ‘ € ‘ ‡ d p q„ vp h uu ‘ † ‘ ” ‘ Ž j ˆ q ‡ u € x € h ‰ h w• ‘p x q d e ” x d x w hp x ˆ ˆ „ q ‘u h † ©uYItAmqgtHXli˜vbtrX(Vƒ†ƒl°w ©•$”PX4YY­r“‘P©q gmh©Xn¦'p‘œf–VXƒt$”x w ‘d † ‘ x ‘ x ” x d h ‰p q„ w † hp d w ‘ † ” h ‡ ” ‘ u ” x ‘ ª ™ ‘ — q s ¹ qd w ‘p h x ‰ q † p ‘ €p j x u ” x ˆ ˆ ™ ‘ — q Ä ¹ „ †f††lPŒ {u T©mqgt4¦bSH˜”ƒ†YXX©mq—ÁŸ™ŠIlTf–ƒnXX©†“XlnXmhmu©q Tœy™ŠlIÁIq | € x u u j v €p ‘ ” ‘ p h j † ˆ ” w ” h x ” qp d † j ‘ ‘ ‡ d u ˆ e u ‘p ˆ d † ˆp du d † ‘„ ‘ w d ” ‘p ‘ x w qd w h j ‘d hdu ‘ €p h ‡ Y"”$”—olq ®InXƒ‘ X©Y† l©n˜ƒ‘ XT˜¨“®†ƒr˜†u˜Va††b†atr“ ©`f)–‘ ‘fVf”lrX† w ” h v €p ‘ ” ‘ d ” ‘p ‘ x w ‡ x s u x † ‘ x u ” x d ” ‘p ‘ x w e ex y † x † ‘ u ” x x j xp ‘ d h ‰ h x u u j v €p ‘ ” ‘ ‘ ‡ Y)InX‹tXn“ ©‘˜d {™ƒ‘ XƒI‘©mqitr“ ©9lƒ‘e˜©¢ X†lx )©q Tmu‰ƒhuXf4°‘$””oIq °nXt˜¡ e¬ « ” x d x w hpp x ” x ‘p ˆ d hp ‘ x ‰ ‘d s j „ q ‘u h † x ” x d hp ‘ ” ‘ € d † ‘„ ‘ w qd v x x x u ” ‘u w ‘u h ‘p † ‘ w h w ” ­I'ÄT©mqgtƒh©Hu¦$p—©q —r˜nlafytq vaV–™'”–©q gnXlfƒb†if¨‚d l
 gd ˜VtƒVXƒnXƒ¨–Yh ¬ « p ‘ v j ‘ † h„p ˆ u ‘ j h d u vp † h y ‡ d q Ž w ‘ ‰p ¢ ” q † u ” x ‘ ª ™ ‘ — q s ¹ ‡ x s w ‘ d x w hp x u ” ‘ x † ‘ u | € ” I'r˜Xƒ†h •†X‚&—'”kx'jVfr”@Y˜II°ƒ¦£t©†©mqtÁƙ®ll˜d {•ƒftƒh©Hu$p‹$‰ƒlx ©Y"”`©q ” x d † hp x w v hp • ª e w ‘ 
p ‘ u Ž q u h s ” x d h x ‡ xp q ‰ h ‘d j x ‰ q † ‰ ” q r ” h ‡ d p ‘ €p j u ‡ d x ‘ w d h x ‡ s ‘ ‘p„•d † ‘„ ‘ ©mqgXn¢Â ©f†ng“~¤ƒrXoAXÁv©q g†m‚luYœf‘ a©ƒ9vYX˜PInh ‰˜t7t¶‘ 'jlÁƒr‡­­ƒb†w du q j h x ‡ x ‡ s p ‘ v j ‘ † h„p u w x ‡ d ‰ ” q r h qd d x q x À ‰ † ¿ ‘ ª ¾½ q p ¼ » p qd x ˆ w ‘d x w hp x | € „ q ¬ fl'‰4u ”† l@††mh–ƒ‚rˆ ™† l˜¤—tfft$”Pl‹4Xt‰gutG–krl‘f‹)ƒftƒh©H¦'p9Y"”{I“$·« u j h ” h } — º ¡z v x q † u qp x ‰ ” qp d † j ‘ ” x u x u ” hp d ” x d † ‘ u• u u qp † w ” h ¬ « vp d ‘ ‰ qp d † ‘ x u € xp ‘ dd h † u w † h xu v YXv—FuT©ƒVlrm†Si©n˜ƒ‘ X¨Amq˜ou a‰ Yn˜¨©q g††o­fVlnX—Y~X$¹”&˜†©n˜ƒI—$”‚f‡t–VƒXXXŽ wp q„p ‘ ‡ d ˆ ¸ y j € x du ‘p ‘d € e¬ « d h ‘ ‡ „ q ” x d h u x w du h„ h p q„ u s j h ‡ x ‡ s v x x d † ˆ w ” q † j h ‰p ‘ ‡ d ‡ x rbr˜t~¶ƒv ©$”gfƒrft"”•­$·¨ƒ”©q gX'xx Vmu©4V‚–vb4vtq 'j† l˜–‚d l
 gB©ƒ£u˜i©m€l‡ ‘ ‡ d v Ž w ‘ uu ‘p x x u j x u u q x x u w † hp d ‘ ‡ d „ q € x j ‘ ‰ u ‘p ˆ d hp ‘ x ‰ ‘d s j d ˆ e u ˆ q ˆ x d ” q † x w ‘p h u w † hp ˜–°†oVƒrˆ Tv $Ž˜oIl¨u ¨ƒXXn˜—˜¶I$”gd ‹ƒn˜tHXlafµtq ¨™†©$”gt©†ou ©‹nS†H˜d ‘ ‡ d ‘p ˆ d hp ‘ x ‰ ‘d ” ‘ € qp x ” x ˆ j d h x ‡ s w ‘d hp ‘ ” ‘ € ‘p h u w † hp d u ˆ q ˆ x d ” q † ‘p ˆ d hp ‘ x ‰ ‘d ‰ q qp d ˜`n˜tHIXf¯Xllr˜d l´mwlBm¥’x®‘ $jl˜³†fnI`n†H˜TA$”gt©†r˜nIXf²©Pn•~ˆ e u ‘ † ” ‘ ˆ ± ‡ x ‡ j d ” x † § ˆ u d h ” x d h x ‡ xp q ‰ h ‘d j x ‰ q † qd u w h j x jp ‘ 
 q w † hp d w ” h u ˆ q ‡ xp q ‰ h ‘p ƒ†ƒX“©m€l3v ˜tXm‘X™t–°©q g†m‚luY¯f‘ a©ƒ9fT–‘ 3mhuI®XXn˜GY©u4®nh u w † hp d ‘ ‡ d „ q u ‘p q † ‘ ‡ e¬ p« u ” x ‘ ª ‡ x s ” x d x w hp x p ‘d„ h w ‘ 
p ‘ u Ž q ‘p h u w † hp d | € w ‘ € h ‰ h w• ‘p x ” ƒXXn˜i˜vIaƒn†–{’¡­ƒr‹Amq•‰”˜d {–Amqgtƒh©H¦'p˜XfH)ƒrXoA–r¨†H˜©Y"”G†lX4Y‡rŠe€ e w † hp d d ” ‘d j h „ lXXn˜œtftmhi‰Iq ” x d h ‰p q„ qd € x w h j ‡ d h x ” x d ‡ x hp du ‘ ‡ d € ” j h w ‘ † ˆ w x ‘ Ž ” h † u ‘ x x u j h ‰p ‘ ‡ d ‘ €p j j d ” x † § ˆ u ©mqgtubTfT$”©–‘ t˜XtAmqt©m€lH˜f¨˜–©q ƒ†B$”–ItYƒƒ˜'wl©Y4¦X˜lrmh•v ˜tXm‘‹t˜mux p ‘ s q x € x x x qdu x ” qp d † j ‘ ‘ ‡ d € e ” x d † hp ‘d x • ” q ” q ‡ x • ” qp d † j ‘ e e ‘ y u ‰ x ” h ‡ † ‘ ‰ u ˆ xp h 
 x 
 ‰ ‘du v ‰tI)$”l©ff‹m†lAn˜††‘ X‹˜‰ƒ„—P©q g†HXft$”(–A©g–©r˜††‘ B’€I†u lYXXƒaAmquXX—mhl¦ffVu m†S©ft˜—ft$”•ƒnubfYn˜™v ˜tXBƒV–Y¤XfVf† l©n˜ƒ‘ a˜ifiƒfd ˜lIfn£fmuiInXS˜¡ x ‰ qd h ‘ ‡ d qd x w ‘pp ‘„u ” hp d j d ” ‘ ˆ ¥ ‘u Ž ˆ u w ” h ‰ ‘du vu x ” qp d † j ‘ ‘ ‡ d qd w ‘ x u q x ‘ w dup ¢ x v €p ‘ ” ‘ ‘ ‡ e†AVlq •nX—m†l©r˜†ƒ‘ ™˜—ƒhl—˜–vtq iv $”lSd @0mwkxAoiƒftH˜l©q ttƒaH„ {fiTXŸž   
 uu j v €p ‘ ” ‘ x ” qp d † j ‘ ‘ ‡ d x 
 ” s q w u s j u j x h ‰ x y j q u h u ‘d hp d ‘ ” ‘ x ” x v 
 h ‘ ‡ d x su h ” ‘ ‡ e w ‘ x d ˆ ‘ Ž qd u h ‡ ” x d hd ” j x x ” x v €p ‘ ” ‘• ‡ x ‡ up ‘ v j w xp ˆ Ž p q w x ‡ d „ q ” x d h xp Ž h ¤ƒm‚’x$jgt™lfœXT©q gVtmha$‰—©q iIrX‡–©€ ln††h Šƒ‘ uiam†l˜@I©mqgt–† uXY‚„ ‘ ‡d p q › e u ‰ h ‘ Ž ” x ™ ‘ — ‡ x s ” xd hd ” j x x ” x x j xp ‘d h ‰ ‘ ‡d ‘d j q x j h xpd † j ‘ p q ‘ x q w qd v h ˜œX‘ƒ–lŠ©mqš†™˜d ˜i©mqgtVtmha$‰ŠAmq™mu~YmhuXf4Š˜—fh ©Vmu‘v 'j–m†‚˜††‘ v–I©•f”†is d ” ‘ ‰ qp x du q ‰ ‘ ‡ d | € „ q v x Ž hd j h ‰p ‘ ‡ d s j ‘ ‡ d „ q ‘u ˆ h † ‘ h e ” x d † ˆ w qp x ‘ x 
 ‘ w p q„ w ‘ w ‘ ‘ ” j xp ‘d h tX$”xS©nX“flf˜Y"”“If‚d ’x$jlVVu u˜‘tq f˜YIio–†7vP©q g†BPnXƒm†lb†ƒƒtƒh‚ft‰ €$”gtmhj©Vmuxft–nXSfƒ†BAr˜t$”l‹fftXfƒ‘ gd ua$‰if††b†X©mqgd ©X{w Y–H˜tX‹fXXŠYe{‰ˆ x d q ‘ d h ‘p † q d w ‘ † ˆ w qp d x ‘ Ž q d ‘ 
 h ‡ u x xp ˆ x x p q u d † ‘„ ‘ w j h ” x x w w h ‘ x Œ q j hp ˆ d h ” q ” u h ‡ | ”€ u eu xd †”©mqgth–† kxjxv† l©n˜ƒ‘ ‡Xft©vm†l©r˜†ƒ‘ 7AmquXXPb˜ƒh‚ftS$”˜muS©nX~u @7Y"”g{mwlx ly'ˆ©e€ x h x ” qp d † j ‘• qd x q w ” h x ” qp d † j ‘ u ˆ xp h 
 p q„ j xp ‘d h ‰ € x x ‰ qp x h x } | € z ‘ x ‡ u q ‡ x ‰ x w ” vt4¦@FY'”kxu7tlt©–˜“7AltioVXn˜A“˜Xm†l‘ kxiFY'”kx¦X7ift˜d gfe˜&˜d X™˜–Y“’ ” h ‰p ‘ „ y jp ‘ h s q p r p • g y q q p ‘ u u hp d 
 p ‘ w x ” j „ y jp ‘ h ‘ d ˆ x d u ”€ •p ‘ ” x ‘ — • ” ‡ h p ‘ ‡ † h ‰ ˆ ‡ † 
 e „ y w h x  e€ x u s qp h g e 4©BvFƒ‚ @0&@$yXw vtri7fd c %C ¥ D % #  § # 6 U £ D R Q 1  ¨ ¡ 1 %C % DC  5£  % 8 6 # 2 1 % §  # % !   £   ¨ §¥£ ¡ FbaB`F$DYX$WB¤V¤TSPI©FBHGFE¤¦BA@97'543©0)('©&$#"!©§ ¦©¦¦¤¢ 

317

‘ x Y# ™# # @ & B 2 2 B ™ & 5H  d H 5 & 5 2 0 (H ( x ™ ! X# Y 8 Q B& ( D Y – 5& X B # 2 8 0 B D 6 5 ’ ‘  2 h #&  & B ! (& ˆ B ! 8 u Q „  h # e ! Ga #2 U9UiD 9g9dP39f$e7c˜”$di1ƒƒ3yG Y#B U42 ƒ7)˜GB —9)9
c@ 4c1c•C§”“3c§ˆ d‰91€§E7"‡†
ƒi2 g‚16 (  2H Y  Y 2 0 B&# ! D 6 # 8 u t B r  q p h # e ! B 2 # ! (& &# D Y X ( D V 6S Q 5& H ! B !& 5 0 B& D B @& 8 0 6 # 2 0 (& # !    m ¨ ¦ ¤ ¡ €3B dyx iicAUbwcR5 I"v37s7i2 gfGdc( b1)aU`GB FQ 4W$UTRAPIG)$EG)FECA97§5 431)'% $" ©§¥£ ¢ 

e ‘p ˆ d † ˆp d In˜†u˜Vu j h d u vp † p h ‘ ” ‘ † h„p u ‘ ‡ d „ q ‘ € h ‰ h w ‘p ‘ 
 ‘ u h € x d h x w x y À ‰ † ¿ ˆ ˆ j½ q p ¼ p Æ d Å p q„ w ‘ 
p ‘ u Ž q ‘ Ž j ˆ q YVfrXƒ™ƒ‚rˆ ˜Il4YYroa—$”gt–m†©'”£tvg4t`¤iv@b‹ƒ&V©Iw ©ƒ† j'˜t™–I)“e À Xœˆ ¦½ t9i¹ lvÁb†nXƒP‹o–rƒi‚d ˜ft'””Y–˜d {“X˜†lIfT$”lXXrŽ j h dh w h‘x q d ‰ † ¿ ˆ À q p ¼ k d Å p q„ w ‘p ˆ † † q ‘ u h ‘p † ‘ w v x u ” ‘d x ” h ‡ x s p ‘ ‡ d ‘ € qd € x ” ‘ w h qp € ” qp du h }d ‡ xpz p ‘ w hp € ‘ w ‡ x s ” x d x w hp x p q › ©n˜f–ot©€ uV{H©†”˜d {‹©q gmh©Xn¦$pX–e À œˆ 4½ t3™p vfI—ƒX“—•f–ƒn†ƒnp ‰†¿ˆ j qp ¼ Æ d Å „ q ‘ † ” ‘ ˆ ± h qd x ˆ w ‘ wp q † ‘ u h s j ‘ 
 j ‘ € h ‰ h w s j h e x y ‡ d x s w ‘ € ” h ‡ † ” ˆ jp h ‘ ” „ q w h ‘ x € € hp h xp h u h }d„ jz ” x d x w hp x d † ‘ x iA'y†‘ l4YYŸ–q iÁI’‘"e“Y˜lmw{™ƒIYXXiv ¦ƒiI–ITlH“–¦X‡ voH†‘ V—©mqgtƒh©Hu$p{††n$p©w p q › e · ‘p ˆ € ¢ x ” s q u vp d ‘ ‰ q ‘ g€ „ ‘ 
 x u ” ‘ u ‘ † h„p ˆ u „ q ‘ u h † x d ” ‘ x ‰ qp x ‘p q ‰ ” ‘ 
 ‘ ‘p h p ‘ w hp •BTr©AT$”Š–‡ tr˜4Aƒl€ Ev‡d ˜V–†X‚&œI–V–t'”¨t'”a©rn)–rvH©€ • ‘ w ‡ x s w ” h j d † ‘ x w w ‘ d x w hp x u j x ‰ h u d h w ‘ wp q † ‘p u ‘ 
p ˆ † ” x d † hp x w ‘ ‡ d ” ‘ ‘ sd ‘ Ž u ‘ † ” ‘p ‘ x w ‘ ‡ —˜d {–Yiv ˜††n$p©t†fmh©Xn¦'pS†‘ aYotŠ†nƒ†nSƒ&—©q g†H@ ©a˜–XƒÁb†I†ƒr“ ©a˜¡ e À ‰ † ¿ ˆ ˆ j½ q p ¼ p Æ d Å q d x ˆ w ‘ 
p ‘ u Ž q u h s ‘ u h ‘p † ‘ Xœ(4t°a¤Áv”fa‘ƒ&V©io–rƒw v x u ” ‘d x d ” h † x ” x u d ˆ q ‡ x s w h ‘ x € € hp h xp h ‡ u h }d ‡ xpz p ‘ w hp € ‘ w j q € ‡ x s ” x d x w hp x „ q ‘u h † ” ‚d ˜ft'”‹tYƒ©¢ l©m€˜¨tA˜d {i–I°lH“¯¦XX¨ot©€ uoH©†`w ©l•˜d ˜”Amqgtƒh©H¦'pI”VXƒe€ el†£¢mul˜0bSƒV‰mu©q g†m‚lu4‰lƒ‘e©'w˜{'”I‘ 4l4YY˜©€ laÁ˜ItXœ˜4t‹‰p ‘ † ” ‘ ˆ ± x ‡ d p q„ w ‘ ‰ ˆ u u h x ” x d h x ‡ xp q ‰ h e ex y j ˆ Ž ‘ ‡ d x j ‘ 
 j ‘ € h ‰ h w ‡ x ‡ h u ˆ ‡ d y À ‰ † ¿ ˆ ˆ j½ q p ¼ ‡ x s ” x d x w hp x p q„ w ‘ 
p ‘ u Ž q ‘ Ž j ˆ q † w h ‘ x € € hp h q d e ‘ † ” ‘ ˆ ± € u h ‘p † x ‡ x s € x ” ‘ w h qp Ž w ” h vp ˜d {—Amqgtƒh©H¦'pœb•†rXo©¨l`w ©†‹–lŠlXn“®“ilƒX“t'”x –r$”˜d {T'”lnX`Y–r˜d •aV)©r˜V”‘$”{–@ƒrXt©q g†H@ ©i˜T$”ƒVmuuXiYƒl®lXn“‰H˜tXƒi˜v– xw ¥š– jq ‘ ‰ ‰ vu h € ” qp du h € x s q ‡ u y u ‘ 
p ˆ † ” x d † hp x w ‘ ‡ d x w ‘ xp h w h ‘ x € € hp h j hp d ” ‘ † ‘ ‡ d „ q ‘ u i · s ‘ ‡ d d h ” x d ˆ xp d ” q † € xp ‘dd h † u h j h ” x x w w ˆ e w ‘ 
p ‘ u Ž q x u ‘ j h 
• i · p ‘ ‡ x ‡ qd d x u w h ‘ x j h u w ” ˜{T©mqgt'Žu˜t©†$”uXf‡ƒoi–v 'jX©q gd ©‰a¤ƒrXoA4muSƒkˆYX¦¥iX©m€l`f{H„ l‡ ‹–I7'j‰ Yh € x ” ‘ w h qp Ž w h ‘ x y ‘u h ‘p † ‘ w v x u ” ‘d x w h ‘ x € € hp h € x w ” ‘ x ‘ w ‘ † ” ‘ ˆ ± h } d„ j z u ” x ˆ ˆ ™ ‘ — q Ä ¹ ‡ x $”lr–YƒlÁIVXƒnXƒ”‚d ˜ft'”––I)lXn“$”©I‹†£ioH†‘ o™Amqœ´†™`”˜d ˜s ” x d x w hp x d † ‘ x w „ q ‘u h † € e j x ‰ h u w ‘d x w hp x ” ˆ ‘ ‡ d „ q j € ” h € € hp h ‘ ‡ d qd d † ‘ u ‘p ‡ x s j € ” h € xp ‘ ©mqgtƒh©Hu$pf††n'p©I4V–"”l‘ o™ƒftƒh©Hu$pl˜FI‘ ©¨lH“˜¨fÁƒlx ƒr‹˜d {‘ ©™'”ufd •d h † u ‘ ‡ d u ‘d q ” ‘ w i · h e ” x d † ‘ ± ‘p € € hp h } · q qz ‘ ‡ d d h w ‘ wp q † ‘p u ‘ 
p ˆ † ” x d † hp x w ‘ ‡ d u s q u p ‘p ˆ x ­ƒo˜v†fIa¥—ƒPAmqg††©n—lH“{ll¦™˜‰tT†nƒ†n{ƒrXS©mqgXn¢Â ©4˜¶œ–‡ ¨r©m€©›
‘ x Y# ™# # @ & B 2 2 B ™ & 5H  d H 5 & 5 2 0 (H ( x ™ ! X# Y 8 Q B& ( D Y – 5& X B # 2 8 0 B D 6 5 ’ ‘  2 h #&  & B ! (& ˆ B ! 8 u Q „  h # e ! Ga #2 U9UiD 9g9dP39f$e7c˜”$di1ƒƒ3yG Y#B U42 ƒ7)˜GB —9)9
c@ 4c1c•C§”“3c§ˆ d‰91€§E7"‡†
ƒi2 g‚16 (  2H Y  Y 2 0 B&# ! D 6 # 8 u t B r  q p h # e ! B 2 # ! (& &# D Y X ( D V 6S Q 5& H ! B !& 5 0 B& D B @& 8 0 6 # 2 0 (& # !     ¨ ¦ ¤ ¡ €3B dyx iicAUbwcR5 I"v37s7i2 gfGdc( b1)aU`GB FQ 4W$UTRAPIG)$EG)FECA97§5 431)'% $" ©§¥£ ¢ 

318

¢ ˆ

&  5 g g 9 ‚ 5 4 3 1 0 5 y w w v '
„h†!8)%8‡"bW'

u !

u $

r & q p 9 g g Q ''6ihfe%dc%bb(YWUS6£I)%22 © a 9 4 ` 9 a ` X V T R Q P H 4 3 1 0

¤ ¥

¤ ¥

¢ ¢ )(

&   #   '!%$"!

e r Ä p } rs spz 0lI 4YIlto y t‹ƒˆ“YYnP“7¤m†’x“¤©v « d ˆ u h — e e ¸ y h wp q q ¸ e 
 y ‘ ‡ j Ð e É ¬ e ¹p p } ssspz B–© 4lIt( y ” q q e º y ” q q — e y ‘ ‘ É e ¡ y ” h q › e ›e „ y ˆ e —e Ï y v h s € x ¸ e e — yp ‘ 
 j „ e e Ð ¬ Y©P“ 77YAA™˜ ƒX7e ž 0YYX7¤evFY¢ t&v7††iw ~‰"Ðt~¦Iq i£Y ‰Á « e p p Ä q ·z BÑtI4}YIlql¦ yp ‘ ‡ † h ‰ ˆ ‡ † 
 e „ y w h x  e€ x w u s qp h g e d ¬ ­XX©0v“Y†m‚@0E@"yv–nXi7f¶I'Ä« e p ¹ p } ¹ q q ·z ©–4lll¦ lll¦³Î } ¹ q q ·z h yp j w ” ‘ ž e º yp ‘ ‚ ” ‰ ˆ j Ï e 
 yp ‘ ‡ † h ‰ ˆ ‡ † 
 e „ y w h x  e€ x w u s qp h g e d ¬ ­X‘ BX®A77¦ˆ mhi77£­XX©0v“Y†m‚@0E@"yv–nXi7f¶I'r« e ·¹Ñr } Ñsspz Blllvll–(®  h y x — e y ¦pm‘‘™70|¦p‘‚”ˆ mhiF7¤YIrA¢7¤e¤IBw ¢¤&E0­X‘ BX®A7~YVƒ`Pe Bn¦X“FS“$¹« ‰ ˆ j Ï e 
 y 
 qp h ‰ q Ï e ›e › y w ˆ x h „ e€e | y p j w ” ‘ ž e º y ‡ † u ‘ ž ž y ‘pp ‘ ’ e Ê ¬ e ¹ } q q q ·z B fllI­` h Î YXo†®PyAYIrA¢¤e¤7IBw ¢¤&E0¶$·« y ‡ † u ‘ ž e ž y 
 qp h ‰ q Ï e ›e › y w ˆ x h „ e€e | ¬ e p r¹ } Ñsspz ©IœlIt(® h Î yp x ‘ — e | yp ‘ ‚ ” ‰ ˆ j Ï e 
 y p j w ” ‘ ž e º y w ˆ x h „ e€e | y ‘pp ‘ ’ e Ê y ‡ † u ‘ ž e ž ¬ p ¦‘ X™70¦ˆ mhi77£‡‘ B®Y7Bmwl¢£&E0~Ir¦X“7S£YVƒ®Pyƒr«
¢ s ¢ †  r 5 g g 9 ‚ "
„hƒ!8)%2€s"bxv 5 4 3 1 0 5 y w w t £ u    s G F D ''E © § ¨¦ C & 'B 5 @ 9 7 6A286)%20 5 4 3 1 © § ¦      © § ¨¦ ¤ ¥

e ” x d h d ‘p xp ‘ d x x ‡ d v xp ‘ 
 q d ‘p ˆ d ˆ„ p h ‘ ” ‘ ‡ d p q„ w ‘ ” ” j x x xp d h ‰ ‘ j h d u vp † P©mqgtV†ruft'”4mul˜aH„ uXifr˜t‡@X–‹˜~Ybiƒmh4u ”Œ u˜t‹$”’x'jVV&¨h $”ƒXXn˜œ©uY‰I–w X©©mqg&Yffmu©i$”©XlXa†£˜¢I©q gh † –ft‘ XSfd l£©–ƒVXXŽ x u w † hp d u ˆ q ‡ xp q ‰ h „ q j ‘ ¢ ” x dp q d x w € x w ” ‘ x ‘ w ‘ † ” ‘ ˆ ± ‘ ‡ d „ q ” x d j ˆ j h † d ” ‘ ‰ j ‘ ‘ x ” ¢ ” q w ‘ u h u j ” h ” ˆ e u j ¢ qp x v x u ” ‘ d x v hp • Œ ‘ ‡ d x u ” x d ˆ xp d ” q † w ‘ 
p ‘ u Ž q ‘ ‡ d q d w h j j ˆ q s u ” x € ‘p w ‘ dp q d x xu v Y™œÁƒ†‘ £InXT‚d ˜ft'”—Hg4˜a$”4©q gt$Ž‚˜t©ƒ–ƒrXoAa˜‹f—Xƒ‘ w ©‰œ©mqPƒn•ƒf&Yffmu©w ‘ u ‘ ‡ ¡ e u w † hp d u ˆ q ‡ xp q ‰ h ‘ ‡ d ” ‘ ‘ sd ‘ Ž xp d h ‰ ‘ j h d u vp † ‘ ‡ d x u ” x € ‘p w ‘ dp q d x w x u ” ‘ d u h j ‘ s u o†˜aƒ†H˜©¦Y—˜–ƒ‰‚l‘Œ ‚˜4”$”’x'jVV&‹˜—$”AmqA†n¯†frffmu©–v $j˜f˜X“kjÁ{Xh w ‘uu ‘p x ‰ q † j h ‘ 
 ‘p ” x d x ‰ uu h x ‡ d ” q w ‘u h Ž j ‘ w q ‰ x hp du j h x € j q ” ‘ ‰ q ” ‘ ‡ x j x u ˆ e } ” x d x w hp ƒVVƒnXa©ƒu Yƒ†n–©mqgtaˆ o˜mul˜a©T†oX¤P4–$”lH˜fA–† Alq ©©‹‘ '‰x 3‰†©q gmh©Xn¦$px d † ‘ x wz u u ‘ ” w x ‡ d x hdp ‘ † h „ q p ‘ v j ‘ j hdu vp † h ‡ x s w ‘p ‘ 
 q † x ‡ x ‡ s wp q sd ‘ ” w † hp d h w ” h }p ‘ w hp € ‘ ƒr$p©¦4o†Xm†l˜$”lVrXƒ~IF††mh$”kxkjYVfrXa˜d {¨ƒrI†vu ™m†l˜˜¦Áb†H˜‹vbXXn©ƒw ‡ x s ” x d x w hp x z ‘ † h„p u j x ‰ h u ‘ ‡ d d h u dp h d u w † hp d u ˆ q ‡ xp q ‰ h ‡ † h ‘ ‘p ‘ ‡ s wp q sd ‘ ” w † hp d h „ q j ‘ ˜d {¨©mqgtƒh©Hu$pfTƒ‚rˆ ¨‘ o¨˜ftf&VVH˜{©uYi–™r˜uY‰‚–XXn˜™iItw X©¢ ” x dp qd x w ‘ ‡ d ” ‘ ‘ sd ‘ Ž u ‘ † ” ‘p ‘ x w ‘ €p j ‘p h ‘p ‘ ‡ d u ˆ x 
 Ž Ê e ” x d h Œ j ‘p ‘ † h„p u qd ‘ ˆ w d ” ‘ w ” ‘ x ‘ Amqgrffmu©–˜•Xƒ‰‚l—ƒƒXnXF ©–lrmh•nTnX˜T jv ©q lttPAmqgtI–h n•ƒ‚rˆ ‘f—BStXlw ‡ d x ‘ w x j ‘ ¢ ” x dp q d x w w ” h x hp d u € x j u ‘p ‘ ‡ ¡ e xp d h ‰ ‘ j h d u vp † ‘ ‡ d q d x u u ‘p d u u ‘ † ˆ w qp d x u w † hp ˜tXœu tw AAmqgrffmu©‹$”lH˜fi'”gd ˆ ƒr{‰©Œ u˜t4¨$”’x'jVV&˜™ft'”4o†n˜fœƒ†BAr˜t$”†H˜d u ˆ q ‡ xp q ‰ h ‘ ‡ d x ” x d h † u ” ‘ w ‘ ‡ d ¬ y « | ”€ ‘ j h d u vp † j € u q d w ‘p h x ‰ q † | € u ˆ q ‡ xp q ‰ h p q„ p ‘ €p j ©uYœ˜v$”4©q gƒ© x¢ Xv˜Y © tYe$”’x'jVV&¶‘ ©'”x ˜f4ƒrXa©ƒ‹Y"”¶©¦YAbAXlnXmh‰mux v x u ” ‘ w uu h ‰ ‘ ‡ d u ˆ e u w † hp d j € x u € ” j h ” x d h x ‡ xp q ‰ h v Ž w ‘u ˆ h † ” x d ˆ xp d x w x hp du € x w ” ‘ x ‘ w ‡ d x ‘ ‚d ˜@VX4˜@‰“†ƒXXn˜¶‘ ©$”˜4©q Y˜Amqgtƒ‚ l¦4{™ƒVYƒ{©mqgt'Žu˜fmu©{'”n˜V'”©I{˜tw w ‘d h j x ‰ q † h v Ž w ‘ x j x Œ ‘ ‘ Ž ” h † u ‘ † ” ‘ ˆ ± d ” ‘p ‘ x w p q„ w ‘ wp q † ‘p u ‘ 
p ˆ † ” x d † hp x w ‘ ‡ d x u ‘ † ” ‘p ‘ x †fƒm†’x©†i†$”lmhP˜˜I¨–¶ƒ†£“tr“ ©£b™ƒrY†ƒrfƒrX©q g†H@ ©˜˜'”{†ƒXn“ ©w ‘ ‡ ¡ e ” x d hp ‘ ” ‘ € d † ‘„ ‘ w ‘ x d † ‘  ‘ ‘p q ‰ qd w h j qd u ‰ ‘ ‘u ‡ x ‡ s ‘d hdu ‘ €p h ‡ † ‰ xp x ˆ ¥ ‘ u x ‡ † h ‘p q {{PAmqgtHXl{††b†™
 g††FrY‘fT–‘ Šf˜4XƒV—m†l˜SfVfInXXÇ'ˆuX$Žkx'jlBƒ˜fd ”XXƒnŠfd mwkxAoa˜'”—†YVfmu©™$”lVr†¨AtH˜‹ffXi‘ $jg††‡˜nX‹˜œtX˜fmu—©mqgtXYh PŒ S‘ 'Žx VlIx ‰P©q gmh©Xn¦$px j q u ‘ ‡ d x ‘ † ” hd x w x hdp ‘ † h j ‘ 
 hp d qd u h ‡ x d † ‘Ì qp x ‘ ‡ d d h ‡ d x ” x d h ” j x ‘ j uu q ˆ e ” x d x w hp p ‘d„ h j ‘ 
 j ‘ € h ‰ h w ‘ ‡ d ” q ” x ‘ ‡ d „ q ‘d hdu ‘ €p h ‡ † ‘ ‡ d „ q ‘ † ” ‘ ˆ ± x w ‘ u q x x ˆ u ‘ ‡ d v xp ‘ 
 u j u ‘p ‘ u ‘ ‡ fHtP‘ ¨l4YYS˜S©©mq¨˜@IftVVSlrX˜@ISƒX“t'”†olIS˜SH„ ‚fd ˆ ƒr™o†˜¡
¢ £   ¡

319

NEXAFS - spectroscopy on freshly prepared and aged plasma-polymerised films Sufal Swaraj, Umut Oran, Andreas Lippitz, Wolfgang Unger Bundesanstalt für Materialforschung und –prüfung (BAM), D-12203 Berlin, Germany NEXAFS spectroscopy has proved to be a powerful tool in the analysis of certain aspects of plasma-polymerised films. Along with XPS (photoelectron spectroscopy) and ToF-SIMS (Time of flight static ion mass spectroscopy) we have employed NEXAFS spectroscopy for the characterisation of plasma polymerised films to obtain information which either XPS and ToF-SIMS are incapable of obtaining or less sensitive to. A dedicated plasma chamber was designed and added to the respective spectrometers. This enabled analysis of freshly prepared samples without exposure to air as well as after exposure to air (aged). In our investigations we have used the organic monomers such as ethyene, allyl alcohol, styrene, allyl amine and acrylic acid to prepare different homopolymers and copolymer films. The aim was to study the correlation of deposition parameters like duty cycle, power, pressure, and monomer (or comonomer) flow rate on the chemical character of the deposited film. In addition to this, the important effect of 'aging' in air of these films was also studied by our unique multi-method approach [1-5]. Here we present some NEXAFS results of the selected case of allyl amine homopolymer films as well as allyl amine/styrene copolymer films. Plasma deposited allyl amine films: Figure 1a presents the N K-edge spectra of a single plasma polymerised allyl amine film measured at different intervals of aging time. The main features observed are the π*(C=N) resonance at 398.2 eV, π*(C≡N) resonance at 399.2 eV, σ*(C-N) resonance above 405 eV and a feature A prominent in the aged sample probably due to an amide π* resonance at 400.6 eV. The extent of unsaturation of these films in terms of the presence of C=N and C≡N bonds was observed in terms of the sum of maximum intensity of the π*(C=N) and π*(C≡N) resonances for samples prepared at different deposition parameters. From figure 1b it is clear that the samples prepared at high duty cycles or power show a high degree of unsaturation when measured in-situ. After exposure to air there is a decrease in this unsaturation, and the decrease is high for samples prepared at hard deposition conditions.
(a)
Normalized TEY [a.u.] 3
σ* Α π*(C=N)

(b)
π∗(C=N) + π∗(C=N)max. intensity

20W, 3.3mbar Duty cycle
"in-situ" 120 days

1.0, 3.3mbar Power

"In-situ" 120 days expo. 330 days expo.

7 6 5 4 3 2 1

2

1

π*(C=N)
(10W, 0.1 duty cycle, 3.3mbar)

Photon energy [eV]

Figure 1: (a) NEXAFS N-K edge spectra of plasma polymerized allyl amine film. (b) Sum of maximum intensity of the π*(C=N) and π*(C≡N) resonances for different plasma polymerised allyl amine films (in-situ and aged) (boxes on top indicate constrained deposition parameters). From XPS studies an increase in the O/C and a decrease of the N/C atomic ratios was

320

1 10 W 20 W 40 W 50 W

0. 1 0. 5 0. 7

0 390

400

410

420

430

440

observed due to aging. A loss of nitrogen after absorption of humidity from the ambient air has been discussed in ref. [6] to be due to the reaction R-CH=NH + H2O → R-CH=O +NH3. The concentration of C=N species was high in the case of samples prepared at hard plasma conditions. Consequently N/C decrease was found to be stronger in these cases. The oxygen uptake due to aging originated from an oxidation preocess starting at embedded C● radical sites adjacent to N atoms. In ref. [7] the authors proposed such kind of a reaction:
O O HC N + O2 HC N

This reaction leads to the formation of amides and similar groups during the aging process. The amides are clearly identifies in the case of the NEXAFS N K-edge spectrum as shown in figure 1a. Plasma deposited allyl amine/styrene copolymer films: Figures 2a and 2b present the C K-edge and N K-edge of plasma deposited allyl amine/styrene copolymer measured in-situ, respectively. The area under the resonance C1s→π*ring (C K-edge), is found to decrease with the decrease in the partial flow rate (PFR)
(a)
20 Normalized TEY [a.u.] 15 10 5
π*(C=N) π*ring

C-H*
π*ring2

Sty%:Alam% 100%:0% 90%:10% 70%:30% 30%:70% 10%:90% 0%:100%

(b)
15 Normalized TEY [a.u.] 10 5 0 397

π*(C=N) π* (C=N)

Sty%:Alam% 90%:10% 70%:30% 30%:70% 10%:90% 0%:100%

0 282 284 286 288 290 292 294

398

399

400

401

402

Photon energy [eV]
(d)
3.0 2.5 2.0 1.5 1.0 0.5 0.0 0% 10% 30% 70% 90% 100% Relative PFR [%] of allyl amine
π∗(C=N)+π∗(C=N) max. intensity

Photon energy [eV]

(c)
3.5
π∗(C=C) or π*(ring) area

"in-situ" 60 days

5 4 3 2 1 0

"in-situ" 60 days

0% 10% 30% 70% 90% 100% Relative PFR [%] of allyl amine

Figure2: (a) NEXAFS C K-edge (b) NEXAFS N K-edge (c) C1s→π*ring/π*C=C area (d) sum of N1s→π*(C=N) and N1s→π*(C≡N) resonance maximum intensities for freshly prepared and 60 days aged allyl amine/styrene copolymer films [20W, 0.1, 3.3mbar, 20sccm].

321

of styrene. This indicates a decrease in the overall unsaturation of the deposited films (Figure 2c). However, in the case of N K-edge the unsaturation (sum of N1s→π*(C=N) and N1s→π*(C≡N) resonance maximum intensities) first increased with the increase in the PFR of allyl amine and then decreased (Figure 2d). The increase was observed as long as there was a sufficient styrene concentration in the feed gas (up to 30% PFR) and then found to decrease. Styrene along with a sufficient concentration of allyl amine favors the formation of unsaturated CN bonds such as nitriles and imines. This leads to the increase in unsaturation till the percentage of partial flow rate of allyl amine was 70% as shown in Figure 2d. This increase in the extent of unsaturation does not continue further because there is insufficient styrene to favor the formation of unsaturated CN bonds. This result implies a higher level of unsaturation in the copolymer films as compared to the respective homopolymers. The films when measured after exposure to air were found to lose unsaturation in terms of π*ring/π*C=C bonds, π*(C=N) and π*(C≡N) bonds. There is no particular trend found in the loss of unsaturation of the copolymer films. Summary: NEXAFS can be successfully used along with XPS and ToF-SIMS for the analysis of freshly prepared and aged plasma polymerised organic films. In some cases it has been found that aging is influenced by the deposition parameters. Key intermediate species and end products of auto-oxidation reactions during aging can be successfully identified by NEXAFS. Acknowledgement: Thanks are due to BESSY staff and other CRG members for the excellent collaboration during commissioning of the HESGM beamline. Special thanks to Matthias Mast (BESSY) and Olaf Schwarzkopf (BESSY) for support during experiments. References: 1. U. Oran, S. Swaraj, J. F. Friedrich and W. E. S. Unger; Plasma Processes and Polymers, 1, 2004, 123. 2. Swaraj S, Oran U, Lippitz A, Schulze RD, Friedrich JF, Unger WES; Plasma Processes and Polymers, 2 (4): 310-318 MAY 12 2005. 3. Swaraj S, Oran U, Lippitz A, Friedrich JF, Unger WES; Plasma Processes and Polymers, 2: 572-580 2005. 4. Swaraj S, Oran U, Lippitz A, Schulze RD, Friedrich JF, Unger WES; Plasma Processes and Polymers, 1 (2): 134-140 SEP 24 2004. 5. U. Oran, S. Swaraj, J. F. Friedrich and W. E. S. Unger; Plasma Processes and Polymers, 1, 2004, 141. 6. L. J. Gerenser, J. Adhesion Science Technol. Vol. 1 (1987), p. 303. 7. T. R. Gengenbach and H. J. Griesser, J. Polym. Sci.: Part A: Polymer Chemistry, Vol. 37 (1999), p. 2191.

322

Photoelectron Diffraction on Transition Metal Oxide Ultrathin Films
K.-M. Schindler, M. Huth, A. Chassé, Ch. Langheinrich, Ch. Hagendorf, Stefan Großer, Steffen Sachert, W. Widdra FB Physik, Martin-Luther-Universität Halle-Wittenberg, D-06099 Halle, Germany

D. Sayago, E. Kröger Fritz-Haber-Institut der MPG, Faradayweg 4-6, D-14195 Berlin, Germany

The particular physical properties of transition metal oxides make them promising candidates in applications such as nonvolatile computer memory. Therefore, there is much interest in the growth mode, structure and properties of ultrathin films. This investigation aims at the determination of their structure with photoelectron diffraction (PED) in energy scanned mode. Furthermore, the feasability of PED to study their magnetic structure is investigated. Their elemental composition, reactions and electronic structure is probed with XPS and NEXAFS as necessary. Ultrathin films of manganese oxides were studied on a Ag(001) substrate. The films were prepared by reactive metal evaporation in an oxygen atmosphere with the substrate heated to 400 K. Former studies have shown that these conditions lead to a mostly layer-by-layer mode with the smallest roughness of the film. The resulting manganese oxide strongly depends on the deposition parameters Previous investigations have shown that the splitting of the Mn 3s lines is very sensitive to the occupation of the Mn 3d levels. Fig. 1 shows spectra of two preparations. Whereas the splitting of the annealed film is 5.6 eV, it is 6.2 eV for the RT deposited film. The splitting of 6.2 eV is in a g reement with published data for MnO, whereas the reduced splitting indicates that annealing to 350 °C results in a different occupation of the Mn 3d levels, i.e. a change of
Fig. 1: XPS Mn 3s spectra of Ag(001)/MnOx

the oxidation state of Mn. Most likely, Mn diffuses into the Ag bulk resulting in a decreased Mn/O ratio. Contrary to claims in the literature, the details of the preparation clearly show that the oxidation state of the film is not controlled by thermodynamics, but kinetics.

323

Further difference of the two systems become evident in their NEXAFS O K-edge spectra (fig. 2). The spectra of the RT film (bottom) have far less peaks and only a minimal difference between normal and
MnO deposited at RT MnOX after annealing to 350 °C

gracing incidence. Both findings indicate higher symmetry in the RT film than in the annealed one. Another significant difference is the shift of 2.6 eV between the low eFig. 2: NEXAFS O K-edge of Ag(100)/MnOx

nergy resonances of the two films. This difference will be the topic of theoretical multiple scattering calculations. The precise knowledge of the geometrical structure, however, is a major help in such calculations. In order to obtain such information, photoelectron diffraction data were recorded and compared to theoretical calculations. The high symmetry in the RT film is confirmed by strong and simply structured PED O 1s intensity modulations in normal emission (fig. 3). The first step towards a complete structure determination are theoretical cluster calculations with one emitter in the top plane of the film. Figure 3 shows the dependence of the intensity modulations on the unit cell lengths. Good
Fig. 3: Experimental and theoretical PED O 1s intensity modulations of Ag(001)/MnO in dependence on the unit cell length.
unit cell length

agreement is achieved for unit cell lengths ranging from 4.25 Å to 4.45 Å, i.e. larger than the Ag unit cell (4.09 Å), but probably slightly smaller than the MnO unit cell (4.45 Å). However, the magnitude of the experimental modulations were adjusted by a factor of 3 and the experimental modulations vanish for energies above 250 eV contrary to the calculated ones. Both findings could be due to averaging over similar scattering processes, such as modulations from the emitters in deeper layers of the film with different geometric environments. The success of this preliminary analysis of the PED data gives good reasons to record additional data in off-normal emission directions and perform calculations in order to address further structural questions. This is the route to an increase in precision and reliability of the structure determination.

324

Determination of the illumination function for coherent x-ray reflectivity experiments at the EDR-beamline at BESSY II
T. Panzner ,G. Gleber and U. Pietsch Institut für Physik, Universität Potsdam, 14415 Potsdam, Germany Now: FB7- Physik University of Siegen, 57068 Siegen, Germany funding source : BMBF 05KS41PC/7

For coherent reflectivity experiments the knowledge about the incident beam intensity distribution at the sample site is one of the important parameters for the correct reconstruction of an unknown height function [1]. Up to now the incident beam function B(x) is approximated by a Gaussian und used in formular (1) for calculating the scattered amplitude:
A(q x ) ⋅ eiα ( qx ) ≡ 1 2π
∞

−∞

∫ B( x) ⋅ e

iq z h ( x )

⋅ e −iq x x dx

(1)

This approximation is not correct if a circular pinhole is used in front of the sample to select the coherent part of the incident beam. In this case one has to consider Fresnel diffraction also within the incident beam. In order to quantify the effect we have calculated the intensity distribution at sample position. We used the Lommel-solution-algorithm [2] for a numerical calculation of the Fresnel integrals. The results of the calculations have been compared with measurement taken at the sample position using the set-up shown in figure 1. Here the sample was replaced by the energy-dispersive detector which was scanned through all sample positions illuminated by the incident beam.

Scan direction simulating trace at sample

αi sample

absorber box

source pinhole

Figure 1: set up for measuring the intensity distribution of the incoming coherent radiation at sample position

The set-up shown in fig.1 is a small modification of our coherent reflectivity set-up reported in 2005 [3]. The results for the measurements and calculation for two different pinhole diameters are shown in figure 2 (part a)). The good agreement between the measurements and the calculations of the intensity distributions are clearly visible. This confirms our approach for considering the Fresnel diffraction from incident pinhole and that the Lommel algorithm works well. The main advantage of this formalism consists in the fact that it provides amplitudes and phases of the incident beam at various sample positions (figure 2 part b). Therefore one is able to replace the Gaussian illumination function B(x) in equation (1) by the correct illumination function B(x, φ) which includes amplitudes and phases.

325

Using this improved formalism we hope for better results in the reconstruction of surface height profile of samples from coherent x-ray reflectivity measurements.

energy [keV]

8.00 6.00 4.00 15 µm pinhole , α 0.102 degree =i

energy [keV]

8.00 6.00 4.00 -5.00 0.00 5.00

8.00 6.00 4.00 -5.00 0.00 5.00

p-s-distance: 250 mm
-5.00 0.00 5.00

sample position [mm]

sample position [mm]

sample position [mm]
Phaseshift [degree]
50.00 100.00 150.00 200.00 250.00 300.00

Measurement
12.00 10.00

Calculation
12.00 10.00 8.00 6.00 4.00 -5.00 0.00 5.00

12.00 10.00 8.00 6.00 4.00 -5.00 0.00 5.00

8.00 6.00 4.00 35 µm pinhole , α i 0.102 degree =

p-s-distance: 250 mm
-5.00 0.00 5.00

sample position [mm]

sample position [mm]

sample position [mm]

Figure 2: a) measured and calculated intensity distribution at the sample after diffraction at a 15µm and a 35µm pinhole, b) calculated phase distribution at the sample position.

References:

[1] I. Vartanyants et al, PRB Vol 55, 19, 13193 [2] K.D. Mielenz, J. Res. Natl. Inst. Stand. Technol. 103, 497 (1998) [3] U. Pietsch et all, PHYSICA B 357 (1-2): 45-52 FEB 28 2005

326

energy [keV]

energy [keV]

energy [keV]

energy [keV]

a)

12.00 10.00

12.00 10.00

b)

12.00 10.00

EXAFS study of the effect of aging on the microstructure of SmCo3Cu2 magnets. M. Katsikini(a), E. C. Paloura(a), F. Pinakidou(a), A. Gabay(b), G. Hadjipanayis(b).
(b) (a)

School of Physics, Aristotle University of Thessaloniki, 54124 Thessaloniki, GREECE University of Delaware, Department of Physics & Astronomy, 223 Sharp Lab, Newark, DE19716, USA

Samarium - cobalt permanent magnets are important industrial materials used for the construction of travelling wave tubes and DC brushless motors.1 They are characterized by high coercivity that originates from the Sm-sublattice anisotropy while the Co sublattice yields high Curie temperature. Addition of Cu causes an increase of the coercive field which is further increased after prolonged annealing/aging.2,3 The scope of this work, is to study the effect of aging on the microstructure of SmCo3Cu2. Although aging alters the magnetic properties (coercivity and Curie temperature) no differences in the xray diffraction spectra (XRD) occur. Therefore, we resorted to EXAFS spectroscopy which can identify aginginduced local distortions. The SmCo3Cu2 sample was synthesized from pure components by arc melting on a water-cooled copper hearth under an argon atmosphere.4 The sample was studied after solution treatment at 1323K and after subsequent aging at 623K for 200 h. The EXAFS measurements were recorded at the KMC-II beamline at the Co and Cu K edges in the fluorescence yield mode. The use of the XFlash energy-dispersive detector permits the electronic isolation of the Co Kα and Cu Kα and Kβ fluorescence photons that allows elimination of the background due to the preceding absorption edges. The detector was positioned in the horizontal plane at right angle to the beam. The angle of incidence was 80o to the sample surface in order to minimize the self-absorption. The EXAFS spectra were normalized with the I0 current, i.e. the current recorded using an ionization chamber positioned in front of the sample. The Fourier Transforms (FT) of the Co and Cu K edge χ(k) spectra, that result after subtraction of the atomic background and transformation from the energy space to the k-space, are shown in Fig. 1. In order to obtain information on the bonding environment around the Co and Cu atoms, the spectra were fitted using the FEFF8 package.5 The proper model for the fitting was constructed according to the XRD characterization which has shown that the sample before and after aging has the CaCu5 (1:5) structure. The 1:5 structure is shown in Fig.2. Both Co and Cu can occupy the M1 (light pink) or M2 (dark pink) sites. The atoms that occupy the M1 site belong to the plane that contains the Sm (red) atoms while the atoms of M2 are “out-of-Sm-plane”. The coordination numbers, the

5.5 5.0 4.5 4.0 after aging before aging

|FT {k χ(k)}|

3.5 3.0 2.5 Cu K 2.0 1.5 1.0 0.5 Co K 0.0 0 1 2 3 4

2

after aging

before aging 5 6 7 8

R(Å)
Figure 1: FTs of the Co and Cu K edge k2χ(k) spectra, before and after the aging process. The fitting and the experimental curves are shown in color and black lines, respectively.

Debye-Waller factors and the nearest neighbour (nn) distances were iterated. The fitting results for both edges (before and after aging) are listed in Table I.

Figure 2: The CaCu5 (1:5) structure of the SmCoCu magnets. The red sites are occupied by Sm atoms while the pink sites are occupied by Co and Cu atoms.

327

The Co-Co distance is Table I: Fitting results of the spectra recorded at the Co and Cu K found smaller that the Co- edges for the samples before and after aging. R is the nn distance, N is Cu distance which is the coordination number and A is the Debye – Waller factor .M-Sm1 and smaller than the Cu-Cu M- Sm2 correspond to the distance of M=Co,Cu (that belong to sites M1 distance. The coordination and M2 respectively) to the Sm atoms. Before aging After aging numbers in the as-grown Co K edge sample correspond to a N R (Å) A×10-3 Å2 N R (Å) A×10-3 Å2 random distribution of the Co and Cu atoms in the M1 Co-Co 2.43 6.9 4.2 2.43 6.9 3.9 and M2 sites, i.e. the M1 Co-Cu 2.48 7.1 2.8 2.48 7.1 2.2 and M2 sites are occupied Co-Sm1 2.91 14.4 1.5 2.94 14.4 1.4 by 60% Co and 40% Cu Co-Sm2 3.16 7.2 2.0 3.16 7.2 1.0 atoms, respectively. Cu K edge Aging does not cause N R (Å) A×10-3 Å2 N R (Å) A×10-3 Å2 significant alteration in the Cu-Co 2.48 7.1 4.2 2.48 7.4 2.1 bonding environment of Cu-Cu 2.49 4.9 2.8 2.50 4.6 4.2 Co, except from a decrease Cu-Sm1 2.90 6.4 1.5 2.93 6.6 2.0 of the number of Co atoms Cu-Sm2 3.20 8.7 2.0 3.19 8.7 1.6 that occupy the M2 site Cu-Cu 3.62 3.0 1.5 (decrease of the Co-Sm2 coordination number). In the case of the Cu K edge spectra, Cu seems to be preferably coordinated with Cu atoms after aging. Additionally, the number of the Cu atoms that occupy the M1 sites increases at the expense of Cu atoms that occupy the M2 sites. The results suggest that partial ordering takes place after aging, with the Cu atoms preferentially occupying the M2 sites. Analysis using a more sophisticated model that will yield the exact percentage of Cu and Co atoms that occupy the M1 and M2 sites is in progress. In the FT of the Cu K edge spectrum, recorded from the sample subjected to aging, an additional peak at 3.62Å is detected (indicated by an arrow in Fig.1). This distance is not predicted by the model of the 1:5 structure and does not correspond to multiple scattering paths. Using the Fourier Filtering technique, the contribution of that peak was fitted separately. Fitting is almost equally good by using either Cu or Co paths but not Sm. The coordination number, nearest neighbor distance and Debye-Waller factor for this peak are used as fixed values for the fitting of the non-filtered spectra. The Cu-Cu distance of 3.62Å is characteristic of metalic Cu and thus it could be an indication of formation of a “metallic” Cu phase, in the form of very small precipitates, which can not be detected by XRD. The presence of Co atoms in the Cu matrix is not excluded. In conclusion, Co and Cu K edge EXAFS characterization of SmCo3Cu2 shows that there is no preferential occupation of the in-the-Sm-plane and out-of-Sm-plane sites by the Co and Cu atoms. Aging at 623K for 200h results to alterations mainly in the bonding environment of Cu. More specifically, Cu preferentially occupies sites in the out-of-Sm plane. At the FT of the Cu K edge spectra of the sample after aging, an additional Cu-M (M=Co, Cu) peak at the distance of 3.62Å is detected. That distance is not predicted by the model and it could be attributed to the formation of Cu rich precipitates.
M. F. de Campos F. J. G. Landgraf, R. Machado, D. Rodrigues, S. A., Romero, A. C. Neiva, F. P. Missel, Journal of Alloys and Compounds, 267, 257 (1998). 2 J. C. Tellez- Blanco, R. Sato Turtelli, R. Grössinger, D. Givord, D. Eckert, A. Handstein, K. –H. Müller, Journal of Magnetism and Magnetic Materials, 238, 6, (2002). 3 E. Estévez-Rams, J. Fidler, A. Pentón, J. C. Téllez-Blanco, R. S. Turtelli, R. Grössinger, Journal of Alloys and Compounds, 283, 327 (1999). 4 A. M. Gabay, P. Larson, I. I. Mazin, G. C. Hadjipanayis, J. Phys. D: Appl. Phys. 38, 337(2005). 5 A. L. Ankudinov, B. Ravel, J. J. Rehr, S. D. Conradson, Phys. Rev. B58, 7565 (1998).
1

328

H induced restauration of the Fe moment in Fe/V superlattices
A. Remhof, G. Nowak, A. Nefedov, and H. Zabel
¨ ¨ Experimentalphysik IV, Ruhr-Universitat Bochum, Universitatstrasse 150, 44780 Bochum Germany

¨ ¨ ¨ M. Bjork, M. Parnaste and B. Hjorvarsson
Department of Physics, Uppsala University, SE-751 21 Uppsala, Sweden

Due to their unique magnetic and superconducting properties Fe/V multilayers are model systems to investigate interface magnetism [1], interlayer exchange coupling [2] and superconducting spin valves [3]. V metal is paramagnetic and becomes superconducting at low temperatures. However, V it is at the edge of ferromagnetism. The instability of V towards forming magnetic moments leads to complex magnetic interfaces in Fe/V multilayers. In contact with ferromagnetic Fe, V gains a magnetic moment, aligned antiparallel to the Fe moments. At the same time a reduction of the Fe magnetic moment near the Fe/V interface has been observed [4]. Both effects lead to a strong reduction of the total magnetic moment with decreasing thickness of the Fe layers. Samples with thin V show the largest induced V moment of up to 1.1µB /atom, while samples with thin Fe layers display the strongest reduction of the Fe moment down to 1.34µB /atom. Interface alloying plays an essential role in the formation of the magnetic structure of Fe/V interfaces. An increasing number of Fe (V) nearest neighbors to the V (Fe) atoms clearly enhances the effect [1].

10 1 0.1 0.01 1E-3 1E-4 1E-5 1E-6

Normalized Intensity

pristine state

pH 2=10mbar

E=513.25eV
40 60 80 100

Detector Angle [deg]

Figure 1: Reflectivity curves of a [Fe(2)/V(16)]×30 superlattice prior (black) and after hydrogen exposure Exposed to H2 , Pd capped Fe/V superlattices are known to dissolve large amounts of hydrogen. In the loaded state hydrogen atoms fill the interstitial octahedral z sites inside the V bcc unit cell and lattice expansion takes place only in the out-of-plane direction. As the hydrogen solubility in Fe is extremlx low, the amount of hydrogen dissolved into the Fe matrix is negligible.

329

The presence of hydrogen within the superlattice has drastic effects on the magnetic properties [2, 5]. In particular, upon hydrogen loading the saturation magnetization was found to increase with increasing hydrogen concentration [6]. This could be either due to an increase of the Fe moment or to a decrease of the antiparallel V polarization. In order to check these two possibilities we employed element specific X-ray magnetic resonant scattering (XRMS) to investigate the response of the Fe and the V moments individually. To maximize the expected effect an epitaxial [Fe(2)/V(16)] superlattice with 30 repetition was sputter deposited onto a Mg0 [110] substrate, ensuring a high number of V neighbors of each Fe atom. The XRMS measurements were carried out using the ALICE diffractometer [7] at the undulator beamline UE56/1-PGM. The sample is fixed to the cold finger of a closed cycle cryostat, placed between two magnetic poles. This set-up covers temperatures between 30 K and 350 K and magnetic fields up to 2.5 kOe. The magnetic field is applied parallel to the scattering plane. Thus using circularly polarized light longitudinal Kerr effect (L-MOKE) geometry can be realized, in which the magnetic moments are both parallel to the scattering plane and to the sample surface. The reflected beam yields magnetic information as it uses the resonant absorption of polarized synchrotron radiation in the vicinity of the L edges of the transition metals. Hydrogen loading was carried out by exposing the sample to a hydrogen pressure of 10 mbar at room temperature for several minutes. As the diffusivity of H in V is very large, we expect the sample to be in a homogeneously loaded state [8]. Subsequently the chamber was evacuated and the sample was cooled down to 30K. Hydrogen gets absorbed rapidly by the sample under these conditions, while hydrogen desorption is rather slow and stops completely at temperatures below 250K.
0.6 0.5 0.4 0.10

T=30K Fe - edge

unloaded pH 2=10mbar

T=30K V - edge

unloaded pH 2=10mbar

Asymmetry Ratio [(I+-I-)/(I++I)] -

0.2 0.1 0.0 -0.1 -0.2 -0.3 -0.4 -0.5

Asmmetry Ratio [(I+-I-)/(I++I)] 710 720 730

0.3

0.05

0.00

-0.05

700

-0.10 505

510

515

520

525

530

Energy [eV]

Energy [eV]

Figure 2: Magnetig asymmetry ratios measured on a [Fe(2)/V(16)]×30 superlattice in the vicinity of the Fe L-edges in reflection at the 2nd superlattice reflection prior (black) and after (red) hydrogen exposure. The spectra on the right panel were recorded on a [Fe(6ML)/V(16ML)]×30 superlattice close to the V L-edges. The hydrogen uptake can be monitored by the shift of the superlattice peak, as seen in figure 1, indicating the hydrogen induced lattice expansion. At low temperatures the superlattice reflections do not shift in time, proving that there is no hydrogen loss. The comparison between the two reflectivity curves also demonstrate that the hydrogen loading does not affect the structural quality of the sample. The magnetic properties, however, change drastically. Figure 2

330

displays asymmetry ratios recorded at the 2nd superlattice reflection in the vincity of the Fe and V L-edges prior and after hydrogen exposure. The data clearly show a strong increase of the Fe moment upon hydrogen loading. We also investigated the response of the magnetic asymmetry of the V in the same sample. No effect could be seen. However, due to the small magnetic moment of the only two ML thick Fe layers, the induced moment in V and consequently the asymmetry ratio is rather small. Therefore we investigated a second sample, containing 6ML of iron. Figure 2 (right) depicts the asymmetry ratio close to the V L-edges at the 2nd superlattice reflection. Again, the presence of hydrogen obviously does not affect the magnetic state of the V. Also this sample shows an increase of the Fe moment upon H loading. Our measurements confirm nicely the theoretical model by Uzdin et al. [9], predicting an increasing Fe moment and a stable V moment in H loaded Fe/V superlattices. In this model the increase of the Fe-V bond at the interface is responsible for the restauration of the Fe moment. The increasing bond-length leads to a narrowing of the 3d band and to an increase of the Fe magnetic moment. The more V neighbors a Fe atom has, the more pronounced is the effect. Therefore, Fe atoms in the close contact with the V atoms, e.g. those with the reduced moment are most sensiively affected by the presence of dissolved hydrogen. The authors are grateful for technical support by T. Kachel. We would like to thank the DFG for financial support under contract RE 2203-1/1. The soft x-ray work was supported by the BMBF under contracts O3ZA6BC2 and 05 ES3xBA/5.

References
[1] A. Scherz, H. Wende, P. Poulopoulos, J. Lindner, K. Baberschke, P. Blomquist, R. W¨ ppling, F. Wilhelm, and N. B. Brookes, Phys. Rev. B 64, 180407 (2001). a [2] B. Hj¨ rvarsson, J. A. Dura, P. Isberg, T. U. Watanabe, T. J. Udovic, G. Andersson, and C. o F. Majkrzak, Phys. Rev. Lett. 79, 901 (1997). [3] K. Westerholt, D. Sprungmann, H. Zabel, R. Brucas, B. Hjrvarsson, D. A. Tikhonov, and I. A. Garifullin Phys. Rev. Lett. 95, 097003 (2005). [4] G. R. Harp, S. S. Parkin, W. L. O Brian, B. P. Tonner, Phys. Rev. B 51, 3293 (1995). [5] V. Leiner, K. Westerholt, A. M. Blixt, H. Zabel, and B. Hj¨ rvarsson, Phys. Rev. Lett. 91, o 037202 (2003) [6] D. Labergerie, K. Westerholt, H. Zabel and B. Hjrvarsson, J. Magn. Magnet. Mat. 225, 373 (2001). [7] J. Grabis, A. Nefedov, H. Zabel, Rev. Sci. Instr. 75, 4048 (2003). [8] A. Remhof, S. J. van der Molen, A. Antosik, A. Dobrowolska, N. J. Koeman, and R. Griessen, Phys. Rev. B 66, (R)020101 (2002). [9] V. Uzdin, K. Westerholt, H. Zabel, and B. Hj¨ rvarsson, Phys. Rev. B 68, 214407 (2003). o

331

X-ray Standing Wave/EXAFS Measurements with Nano-Resolution A. Gupta1, N. Darowski2, I. Zizak2, C. Meneghini3, A. Erko4, G. Schumaher2
1

UGC-DAE Consortium for Scientific Research, University Campus, Khandwa Road, Indore 4520017, India 2 Hahn-Meitner-Institute Berlin, Glinicker Str. 100, D-14109 Berlin, Germany 3 Via della Vasca Navale 84, I-00146 Roma, Italy 4 BESSY GmbH, Albert-Einstein Str. 15, D-12349 Berlin, Germany

X-rays are highly penetrating radiations, and therefore, any information obtained through x-ray based measurements is averaged over a depth of several microns. However, x-ray based techniques can be made depth selective by generating standing waves inside the nanostructure of interest by making use of the phenomenon of total reflection [1, 2]. X-ray intensity is localized in the anti-nodal regions, the position of which inside the nanostructure can be varied by varying the angle of incidence. Use of such x-ray standing waves in elemental depth profiling, XANES or fluorescence measurements with nanometer depth resolution has been demonstrated for organic [3] and inorganic material systems [4]. For the first time angular dependences of the fluorescence yield from a single organic monolayer on a solid substrate modulated by a standing wave in a total external reflection conditions has been measured experimentally. The scanning by standing wave field of a single organic molecule has been done, the depth positions of particular ions in the molecule structure have been determined [4]. In this experiment a depth resolution on the order of 1-5 nm has been achieved. Depth selectivity can further be enhanced by making use of wave-guide structures [5, 6]. For the first time at BESSY, depth selective EXAFS studies have been performed. The absorption spectra of Fe and W nano-layers were recorded with in-depth resolution on the order of 1 nm. This method is combining total external reflection standing waveguide mode and EXAFS measurements. We present the results of depth resolved tungsten XAFS measurements in a Si/W/Si trilayer embedded in a Au waveguide structure. The graded-crystal monochromator beamline KMC-2 was used to set-up of the in-depth nanoprobe EXAFS system [7]. The x-ray beam in the energy range of 10 keV - 14 keV was monochromatized by the double-graded-crystal monochromator and collimated in both directions using two slit systems and collimating mirror shown in figure 1. The beam size on the sample position was 100 µm horizontal and 700 µm vertical.
Source (banding magnet) Double-crystal monochromator Slit 1 Collimating mirror Fluorescence detector Slit 2 Reflection detector

Figure 1. Experimental setup, top view.

The beam divergence was on the order of 20 arc sec obtained by rocking curve measurements of a Si (111) Sample 31 m reference crystal. The total flux in the spot was in the range of 109 phot/sec/100 mA. Beamline monochromator, stabilized by MOSTAB feedback system for EXAFS measurements, provides RMS intensity variations in the order of 1.7 10-3 in entire energy range. PIN photodiode detectors were used for measurements of reflected (diffracted) beam as well as xray fluorescence signal from the investigated sample. Energy-dispersive detector RONTEC X-

332

FLASH was also used to record the fluorescence spectrum of the sample. Control software was used to realize simultaneous energy scan of the monochromator and angular scan of a sample. The objective has been to study the swift heavy ion induced intermixing in this system. A detailed information about various phases formed as a function of depth can be very valuable in understanding the mechanism of mixing induced as a result of the electronic energy loss incurred by the bombarding ions.
a)
0 5 10
Depth (nm)

Au (2 nm) Si (10.2 nm) W (2 nm)

15 20 25 30 35 0.4 Au (70 nm) 0.5 0.6 q (nm )
-1

Figure 2. The multilayer sample structure. The contour plot represents the intensity of x-rays as a function of depth and q. The hatched strip marks the position of W (a), normalized intensity curves of reflected radiation and W Lα fluorescence radiation at 10.3 keV incidence energy (b).

Si (19.5 nm)

0.7

0.8

b)
Relative count rate

1.0

0.8

0.6

0.4
Fluorescence Reflectance

0.2 0.4 0.5 0.6 0.7
-1

0.8

q (nm )

temperature. In figure 2b the corresponding normalized intensity curves of reflected radiation and W Lα fluorescence radiation are shown vs the q values at 10.3 keV incidence energy. XAFS measurements across the L-edge of W were done for various values of q. With varying q, the depth distribution of x-ray intensity inside the multilayer varies, thus providing weighted information from different depths. A typical W L-edge XAFS data and the corresponding radial distribution function is shown in figure 3.
Figure 3. Calculated XAFS function and the corresponding fourier transform of the sample irradiated with 600MeV Au ions to a fluence of 1x1013 ions/cm2

The structure of the multilayer is shown in figure 2a and consists of several layers: Si substrate 400µm / Cr 20nm / Au 70nm / Si 19.5nm /W 2nm / Si 10.2nm/Au 2nm. The two layers of Au form the walls of the waveguide. A Cr buffer layer has been used for improving the adhesion of the film. For the sake of clarity, depth dependence of the x-ray intensity inside the multilayer as a function of q is also shown as a contour plot. The scattering vector q defined as q = 4π sinθ/λ, θ being the angle of incidence and λ being the wavelength of x-rays. The sample multilayer has been irradiated with 600MeV Au ions to various fluences in order to induce intermixing of W layer with Si at room

The XAFS data of the pristine as well as irradiated samples was fitted by taking a threeshell structure into account: i) a W-Si shell

333

which is the main contribution to the whole XAFS signal, ii) a W-W shell around 2.7 Å, similar to the W-W nearest neighbors distance in metallic W, and iii) a W-W shell around 3.2 Å which is similar to the W-W second neighbors shell in metallic W. The XAFS data of pristine sample suggests about 60 % of W in metallic environment and 40% in WSi2. Irradiation provokes the partial dissolution of W in Si giving rise to higher fraction of W-Si correlation. Figure 4 shows the variation of a) fraction of W-Si bonds, XW-Si and b) W-Si bond length, RW-Si, as a function of the scattering vector q. From figure 4, despite the quite large uncertainty, it is clear that the fraction of W-Si bonds increases with irradiation and seem larger in the deeper region below the W layer. The W-Si distance shows a clear trend as a function of depth and irradiation. At lower irradiation fluences (2x1012 ions/cm2 and 5x1012 ions/cm2) the W-Si distance is systematically shorter but it grows as the depth increases. The low RW-Si value is consistent with W-rich W-Si structures such as W5Si3 and W3Si. In the deeper regions a large W-Si distance is suggestive of a Si rich WSi2 phase. At higher irradiation fluences (1x1013 ions/cm2 and 2x1013 ions/cm2) the structure is dominated by W-Si correlations while the metallic W contributions become weak. The Si content increases as a function of depth. The W-Si distance changes weakly as a function of depth.
a)
1.0 0.9 0.8

Figure 4. Results of the analysis of XAFS data taken at different values of q. XW-Si is the fraction of the bonds of type W-Si, while RW-Si is the corresponding bond length.

XW Si

0.7 0.6 0.5 0.4 0.40 0.26 0.45 0.50 0.55 0.60
-1

XSi 2x10 XSi 5x10 XSi 1x10 XSi 2x10 XSi ref 0.65

12 12 13 13

0.70

0.75

q (nm )

b)

0.25 RSi 2x10 RSi 5x10 RSi 1x10 RSi 2x10 RSi ref 0.45 0.50 0.55 0.60 0.65
12 12 13 13

The combination of EXAFS and standing wave methods in investigation of the layers placed into an x-ray waveguide provides sub-nanometer in-depth resolution with an extremely high sensitivity to the layer structure. The method was successfully used to structure measurements of the nanometer-thick layers exposed to heavyion beam. In the future the EXAFS/XRSW combination can be applied for investigations of different kinds of diffusion processes, as well as interfaces of different materials in layered structures. References

RWSi (nm)

1. B. N. Dev, A. K. Das, S. Dev, D. W. Schubert, M. Stamm, G. Materlik, Phys.Rev. B 61 (2000) 8462. 2. S. K. Ghose, B. N. Dev, A. Gupta, Phys. Rev. B 64 (2001) 233. 3. A. Gupta, C. Meneghini, A. Saraiya, G. Principi, D. K. Avasthi, Nucl. Instr. Meth. B 212 (2003)458 4. N. N. Novikova, E. A. Yurieva, S. I. Zheludeva, M. V.Kovalchuk, N. D. Stepina, A. L. Tolstikhina, R. V. Gaynutdinov, D. V. Urusova, T.A. Matkovskaya, A. M. Rubtsov, O. D. Lopina, A. Erko and O.V. Konovalov, J. Synchrotron Rad. (2005). 12, 511–516 5. S. I. Zheludeva, M. V.Kovalchuk, N. N. Novikova, et al. Crystallography Reports, 40, (1995), 132-144 6. A. Gupta, C. Meneghini, P. Rajput, G. Principi, to be published 7. A. Erko, I. Packe, W. Gudat, N. Abrosimov, A. Firsov Nuclear Instruments and Methods in Physics Research, A467–468 (2001) 358–361
q (nm )
-1

0.24 0.40

0.70

0.75

334

Effect of Zn and Ni impurities on the renormalization effects in Bi-2212
D. Inosov,1 V. Zabolotnyy,1 S. Borisenko,1 A. Kordyuk,1, 2 J. Fink,1 J. Geck,1 A. Koitzsch,1 M. Knupfer,1 B. B¨chner,1 H. Berger,3 A. Erb,4 C. T. Lin,5 B. Keimer,5 and R. Follath6 u
1

Institute for Solid State Research, IFW-Dresden, P.O.Box 270116, D-01171 Dresden, Germany

2

Institute of Metal Physics of National Academy of Sciences of Ukraine, 03142 Kyiv, Ukraine
3

Institute of Physics of Complex Matter, EPFL, CH-1015 Lausanne, Switzerland
4

Walther-Meißner-Institut, Bayerische Akademie der Wissenschaften, Walther-Meißner Strasse 8, 85748 Garching, Germany

5

Max-Planck Institut f¨r Festk¨rperforschung, D-70569 Stuttgart, Germany u o
6

BESSY GmbH, Albert-Einstein-Strasse 15, 12489 Berlin, Germany

The anomalies in the single-particle spectral function of a superconductor are commonly believed to be crucial for understanding HTSC. Along the Brillouin zone (BZ) diagonal the renormalization effects
k
x

a
E

h n = 5 0 e V

c

are represented by the unusual dispersion, the so called “kink”. In the vicinity of (π, 0) point of the BZ the renormalization is noticeably stronger and makes itself evident even in the shape of a single spectral line measured for a fixed momentum. It is widely accepted that these anomalies are naturally explained by coupling to a collective mode, while the origin of this mode remains a current controversy between phonons and spin fluctuations. Some of our recent experiments, which were carried out at the U125/1-PGM beam line with an angle-multiplexing photoemission spectrometer (SCIENTA SES100) at BESSY, showed that substitution of Cu in Bi2 Sr2 CaCu2 O8−δ with Zn and Ni essentially influences the renormalization anomalies in ARPES spectra both along the nodal and antinodal regions of the BZ. Moreover, the changes in ARPES spectra are consistent with the behaviour of the spin resonance mode as seen by inelastic neutron scattering in YBCO. This provides evidence that it is the magnetic excitations that are responsible for the unusual renormalization features in the quasiparticle excitation spectrum of cuprates.
0 .2 0 .1

k
y

Z n -B S C C O

b
N i- B S C C O B ro a d m o d e P u re B S C C O

N a rro w m o d e

0 .0

0 .2

0 .1

0 .0

B in d in g E n e r g y , e V

B in d in g E n e r g y , e V

Fig. 1. (a) Cut of a kω-space at the Fermi level. The top of the cube contains two sheets of the Fermi surface, red solid line corresponds to the dispersion of the antibonding band, blue dashed one – to the bonding band. (b) Model line shape of the (π, 0) EDC, when coupled to a narrow (2 meV) and broad (20 meV) collective mode. (c) Experimental EDC’s taken at (π, 0) point.

335

B in d in g e n e r g y ( e V )

In the antinodal region, the “dip” in a single band EDC can be a good indicative of the mode width and energy. Comparing the EDC’s, measured at (π, 0) point (Fig. 1), we see differences in the line shapes caused by the impurity substitution: while for the pure sample the spectrum has a pronounced dip, for the Zn substituted sample the dip vanishes, and for the Ni substituted sample its “strength” is intermediate. In the nodal region, the raw distributions of photoemission intensity as a function of energy and momentum are presented in the Fig. 2(a-c) together with the experimental dispersions. To compare the renormalization effects, one can use the area under the “kink” as shown in Fig. 2(d-f). It is easy to see that renormalization in case of the Zn substituted sample is decreased in comparison to the pure sample. Subtracting a bare particle band dispersion from the experimental dispersion the real part of the self-energy can be obtained, see Fig. 2(g-i). On Fig. 3 we also show experimental spectra at 20 meV below the FL. These data have been used to determine the exact doping level of each of the samples. Panels (a – c) correspond to the samples under consideration, and for comparison, in the panel (d) we give the identical map for an overdoped sample. Grey scale images below the experimental data represent tightbinding fits to the antibonding sheets of the FS. Summing up, we have shown that the substitution of Cu atoms in Cu-O plane changes renormalization features in ARPES spectra both in nodal and antinodal parts of the Brillouin zone, which can be well explained by coupling to a magnetic resonance mode. The effect of Zn and Ni substitution on the antinodal ARPES spectra agrees with the influence of these impurities on magnetic resonance mode seen by INS experiments. This provides another strong evidence that the mode, coupling to which causes observed renormalization effects, has rather magnetic than phononic origin. For details see Zabolotnyy et al., Phys. Rev. Lett. 96, 037003 (2006).

a

b 0 .0 0 .1 0 .2
-0 .1 0 k -k 0 .1 0 , (1 /Å )
F

c 0 .0 0 .1 0 .2

0 .0 0 .1 0 .2

B in d in g e n e r g y ( e V )

0 .0 0 .1 0 .2

d

0 .0 0 .1 0 .2

e

-0 .1 0 k -k
F

0 .1 0 , (1 /Å )

0 .0 0 .1 0 .2

f

-0 .1 0 k -k
F

0 .1 0 , (1 /Å )

S e lf e n e r g y Ó ' ( m e V )

8 0 4 0 0

g

-0 .1 0 k -k
F

, (1 /Å )

0 .0 0 8 0 4 0 h

-0 .1 0 k -k
F

, (1 /Å )

0 .0 0 8 0 4 0 i

-0 .1 0 k -k
F

, (1 /Å )

0 .0 0

P u re

0 .0 0 .3 B in d in g e n e r g y ( e V )

0

N i 2 %

0 .0 0 .3 0 .0 0 .3 B in d in g e n e r g y ( e V ) B in d in g e n e r g y ( e V )

0

Z n 1 %

Fig. 2.

(a)-(c) Photoemission intensity

distributions along the BZ diagonal. (a) Pure BSCCO, (b) Ni substituted and (c) Zn substituted BSCCO. Middle row shows corresponding experimental dispersion. In the last row the real part of the self-energy is displayed.
0 .5 0 .0 (1 /Å ) -0 .5 -1 .0 -1 .5 -2 .0 0 .5 -1 .0 -0 .5 0 .0 0 .5 1 .0

a

0 .5 0 .0 -0 .5 -1 .0 -1 .5 2 .0 1 .5 1 .0 0 .5

b

k

y

-0 .5

0 .0

0 .5

1 .0

1 .5

c

d

0 .0 (1 /Å ) k -1 .0 -0 .5 0 .0 k
x y

-0 .5

L

0 .5 1 .0 (1 /Å )

1 .5

-0 .5

0 .0 k

x

0 .5 (1 /Å )

1 .0

Fig. 3. Cuts of kω-space made 20 meV below the Fermi level and tight binding fits. (a) Ni substituted sample. (b) Zn substituted sample. (c) Pure BSCCO close to optimal doping. (d) Overdoped BSCCO, δ ≈ 0.20.

336

!
#

"" " "

# &

$ ( ) *+

% ,

"

&

"

'

! & ( ( " % "" % ' & %" 2
polarizer
i Ep

" ( " $ %" 0 % " %
"

# " $

" % %"

$ ) -( % !# ) +

"% %" " ( % " " 1 ( # # $ " . / )

$

' ! ! *) + ( $ $ ,

#

% ( !# %

# (

( # & "

#

"%

%

ρ=rp/rs=tanψe

i∆

analyzer
Ep
r

45°

Es

r

E si
0

linearly polarized light

sample

elliptically polarized light

'

* , #

% "

& %

) +

"

"

*#, 3

%

" "

"

$

%

%

4 % !# 5 ( ( $ ! %" " (

%" % %" 2 # " " ' 2 %" $

" "

( & ! % % ( % " " " " %

! ! % " $ % % ( " " %"

337

# #$ ( ( " ( ! (2 " ( ' % ( 6 %" % ( (

" % & $

% $ # % % ( (2 ( ( # ( #( " ( # ( 6 & 0 !%% , 4 % 7" $ %# " & ( " # ! ( $ #

& & $ ' #$ ( " % $ $ " ! ! 7" *'

'

# (

% & " " % " % # ( . /

(

" # % %" ( "

" $ ( % #! ( %" # # & , # + $# ( " * " ! % 5&

& ( ! # +# " $ & ( # 0, 8 $ ' %( #( ! %" %" 5 % & %( ( * % 2 %" ! *# . / $ %( " % ! & %( $ ( % %" , & & # ' , # & $ " # " % 2 # ( ) +

" (" " *' 5 ( # " % #, '

!%% $# 5 * (

. / % ( ! ( 2 % ( $ # ( ! ( %( 6 & # ! %( %% 5 % 2 # " ( " " % 4 $ 2 & (% # & & $ $ ! ( ' % 5 # ( $ $ ! ( -( & ! # ( & $ (

" 9 " % # %" & % % ( % %# % ( & " " 5 ( * % % 8 %( ( ( %( " ( , " $ " ! ( :; %% %" & $ ""

4 ) ( $

! %( #, & ! ! # *' " , ! "

$ $ ! 5 " % %( " 5

# * #

"

338

% %(

& $ #
0
1.00 0.99 0.98 0.97 1.6 1.5 1.4 1.3 3000 2750

" 2# (
4

& $ " " ( % &

# " " "
' % ( 5

. / 5 & % ( "

En

PPh

E
' %

Wavenumber /cm-1
" ! ! > * ( ,

2500

2250

2000

1750

1500

9 ( & &

"

<

=% # *# +

+ , (

" ! " # +

! $ ! 5

% * ,> %( *

$ $ ! ,* ,

2* , 2

) " ( % % 0
30 5 7 + %" ' 3 & (

(%% ! " ( ( % % ( # % $ %" 6
( ! 7 " " (% @A ("" 5 4( ( " @ 8 8 ("" % & %

# % & 8 $ 2 # ( &

) ) $ # % " (

4

? ! %" $ ( "

! " $

&

( % 2 "

% %

. / 7 -( $ !

!

" %

"

&

%

("" & 9 #! 0

% $ " B $ & ( ( ( 7

$

( $ * # 7, # %" %%( A 8 ( & ' #! ;;;: ( &

9 0 4 # @ ( C, ( 4 # ) ' ( ( 4 4'

% %

$ ( # ( '

4 < 54

(% A 4 (""

;: ( !

0 *90 )

0"" 4 ) !0 ( " 7 ! 5

" ;; : " * ;; ,

D 0 * ;;:, * ;;C,

339

Can magnetic flux lines be detected in optimally doped YBa2Cu3O7-δ δ by magnetic oxygen K-edge spectroscopy? J. Albrecht, M. Djupmyr and E. Goering Max-Planck-Institut für Metallforschung, Heisenbergstr. 3, D-70569 Stuttgart, Germany In the last 20 years cuprate-based high-temperature superconductors have attracted an enormous amount of interest. Nevertheless, all attempts failed so far to explain the origin of the superconductivity completely. It has been shown that not only the superconducting properties are very unusual, also the normal conducting state exhibits a variety of unique properties. In this context, the electronic structure of the high-temperature compound YBa2Cu3O7-δ (YBCO) has been successfully investigated by tunneling spectroscopy inside and outside a flux line. Figure 1 shows the differential conductance of an YBCO crystal at low temperatures, extracted from Ref. [1]. A clear superconducting gap is observable in these conductance measurements, as shown in the lower part of Fig. 1. Positioning the tunneling tip at the center of a flux line, where superconductivity is not existent, it is found that the intensity in the gap region is significantly enhanced. In contrast to conventional BCS superconductors these structures are not flat, a finite density of states is found in this gap region. It is obvious that the understanding of the origin and the symmetry of these “in-gap” states is crucial for a complete description of the mechanisms of high-temperature superconductivity.
broadened with 50meV superconducting state vortex state

1,6

dI/dV (arb.units.)

1,4

1,2

-100

-50

0

50

100

Voltage (mV)

Figure 1: Left: Differential conductance spectra dI/dV(V) of an YBCO single crystal performed at T = 4.2 K from Ref. [1]. The upper curve shows the spectrum at the center of a vortex core. The lower curve shows the quasiparticle excitation spectrum of the superconducting state. Right: Differential conductance spectra broadened by 50 meV. We use high energy resolution x-ray absorption spectroscopy to investigate epitaxially grown thin films of optimally doped YBCO. It is tried to resolve these “in-gap” states at the threshold of the O K edge as a function of external applied magnetic field. For this purpose XMCD-like (X-ray magnetic circular dichroism) spectra between the zero-field cooled nonmagnetic superconducting state and a flux-lattice state at B = 2 T are obtained. The experiment is performed over the temperature range between T = 25 K and T = 90 K to provide the required information about the normal conducting unoccupied density of states of the highly renormalized quasiparticles. Some of these structures should be observable with an

340

energy resolving power better than 15000 at the O K edge [2]. The 50 meV broadened dI/dV curves (from the left part of Fig.1) are shown in the right part of Fig. 1. Strong and clear differences between core and superconducting spectra are observable. This is strongly encouraging, that the proposed features are observable at the O K edge. If so, this could at the same time offer a contrast mechanism for magnetic x-ray microscopy of flux lines in superconductors. Note, that an important property of the high-Tc flux lines is their very small diameter, which prevents a direct observation of the flux lines by conventional optical microscopy. Fig. 2 shows the XAS (X-ray absorption spectrum, black) and the obtained XMCD-like effect (red) of an YBCO film at the oxygen K-edge at a temperature of T = 80 K. The XAS spectrum clearly shows two peaks, the Zhang-Rice singlet (left) and the lower Hubbard band (right). The high intensity of the Zhang-Rice peak indicates the high quality of the superconducting film [3]. The magnetic effect is measured as an average of about 30 spectra, each determined as difference between the x-ray absorption spectrum at zero field and at B = 2 T, respectively. The red curve is scaled by a factor 100 for better visibility.

Figure 2: XAS (black) and XMCD-like (red) spectra of the oxygen K-edge of YBCO at T = 80 K. The red curve is obtained by subtracting the XAS spectra at B = 0 and B = 2 T, respectively, averaging over about 30 experiments and multiplying it by a factor of 100. Figure 2 clearly shows that no significant magnetic effect can be seen in the experimental data. We can state that the modification of the absorption signal due to the presence of flux lines is smaller than 0.1 %. Analogue experiments have been performed at different temperatures but no significant signal referring to a magnetic field induced modification of the absorption edge could be obtained. It is worth to say that a time-dependent decay of the absorption spectra intensities has been noticed during the illumination of the superconducting samples. The used high photon intensities could have introduced a significant damage of the superconducting state. In conclusion, we state that the experimental set-up that has been used in our experiment does not allow the detection of quasiparticle states inside of magnetic flux lines in hightemperature superconductors. Up to now it is not clear if this is in general impossible with magnetic fields of B = 2 T and a resolving power of 15000 or if the incident photon beam at UE49 might be harmful for the superconducting samples. [1] I. Maggio-Aprile et al., Phys. Rev. Lett 75, 2754 (1995). [2] M. Coreno et al., Chem. Phys. Lett. 306, 269 (1999); A. S. Schlachter et al., J. Phys. B 37, L103 (2004); E. McGuire, Phys. Rev. 185, 1 (1969). [3] M. Merz et al., Phys. Rev. B 55, 9160 (1997).

341

Magnetic anisotropy of triangular shaped, hexagonally arranged Co nanostructures
P. Imperia1, W. Kandulski2, A. Kosiorek2, H. Głaczyńska2, D. Schmitz1, H. Maletta1 and M. Giersig2
2

Hahn Meitner Institut, Glienicker Strasse 100, 14109 Berlin, Germany CAESAR Research Center, Ludwig-Erhard-Allee 2, 53175 Bonn, Germany

1

The quest for new materials and technical solutions able to match the requirements of modern performances for data storage devices produced in the last few years a large number of studies about the magnetic properties of nanostructured materials. The control of the magnetic properties by means of the shape could be decisive in improving the performances of newly designed devices [1, 2]. The ways 30 devised to induce the shape anisotropy 0 have especially tampered with adjustments of the substrate topology or have deployed the power of electron beam lithography. Here we report on the in plane magnetic anisotropy observed in nanostructured Co Fig. 1 thin films prepared by nanosphere lithography [3]. This relatively young technique allows the simple and economical preparation of thin films of magnetic metals laterally structured in a number of different topologies [4]. Polystyrene (PS) latex particles of 1710 nm diameter were deposited from a water surface on the polished surface of a chemically cleaned Si substrate by a methodology related to the preparation of Langmuir-Blodget films [4]. Then, a Co layer of 32 nm thickness was deposited by electron beam evaporation on top of the PS mask in high vacuum (HV) conditions. After the chemical removal of the nanosphere mask the final result is a matrix of polycrystalline triangular elements arranged in a highly ordered hexagonal symmetry. Atomic force microscopy (AFM) has been used to 6 investigate the samples homogeneity and to 5 determine the orientation of the evaporated 4 triangular nanostructures with respect to the 3 substrate (fig. 1). The magnetic properties were 2 studied by means of x-ray magnetic circular 1 dichroism (XMCD) at the beam line UE 46 PGM 0 at BESSY, Berlin. The experimental technique 1.0 0.4 allows to determine separately the orbital and spin 0.5 0.2 moment. 0.0 0.0 The relative simplicity of the substrate preparation -0.2 constitutes the main advantage of the sample -0.5 -0.4 preparation technique. However, the necessity to -1.0 -0.6 chemically remove the mask ex situ produces Co -1.5 -0.8 structures covered with a natural thin layer of CoO, -1.0 -2.0 thus the X-ray absorption spectra (XAS) show the 770 780 790 800 810 Photon Energy (eV) typical splitting of the Co L3 edge. A way designed + Fig. 2 to remove the CoO layer is H ion sputtering. This method has been already successfully used to remove the oxide shell of chemically synthesised Co nanoparticles [5]. The samples were etched for 210 minutes at a relatively low sputtering energy (700 eV, H+ pressure of 2.4x10-5 mbar). The sputtering parameters were carefully adjusted to allow an effective removal of the CoO layer without destroying the
XMCD (arb. u.) norm. absorption (arb. u.) Integral (arb. u.)

342

0.70

0.65

0.60

0.55

0.50

0.45

0.40 -60 -40 -20 0 20 40 60

φ (deg)

sample’s regular pattern. Longer sputtering time at lower energy as well as shorter sputtering time at higher energies destroy partially or completely it. The sputtering energy and the time necessary to remove the CoO layer depend on the sample thickness and on the diameter of the PS lattex spheres. Each combination of Co thickness and triangles lateral dimension has a different sputtering time. Fig. 1 shows a typical AFM picture of our samples, the triangular structures are well

defined with a very low density of defects, 0.055 the shape and dimension of the Co pattern 0.050 are given by the curvature of the latex spheres. The AFM scan shows that the 0.045 samples retain their well-ordered geometry after the correct H+ sputtering procedure. 0.040 Small debris, resulting from the cleaning action of the H+ ions and having a diameter 0.035 between 2 and 10 nm, are present on the Si 0.030 surface in between the metal triangular structures. -60 -40 -20 0 20 40 60 φ (deg) The XMCD measurements were done in remanent magnetisation at room temperature (RT). A pulsed field of 800 Oe was applied alternating its direction planar to the sample surface at each energy scan point. After each scan the sample was rotated around its axis. The arrows imposed on the AFM picture of fig. 1 show the geometry of the experiment. The angle ϕ defines the direction along the triangle pattern; the angle ϕ = 0° is the direction along the spherical voids, while ϕ = 30° defines the direction along the triangles edges. The geometry of the experiment was carefully planned to make sure the position of the centre of the samples remained unaltered after changing the azimuthal angle, the correct position of the sample with respect to the incident synchrotron light beam was checked at each measurement step. The XAS spectra were recorded in total electron yield, collecting the sample drain current. The spectral curves were normalised toward the incident incoming light by means of the last mirror current. The measurements were performed in grazing incidence, the angle of the sample surface with respect the incoming circularly polarised light was θ = 20°. In the data analysis no self absorption effects were taken into account. The upper panel of fig. 2 shows the typical line shape of the absorption spectra after sputtering. Three lines are drawn: the sum of the positive and negative magnetised spectra (black), the step function representing the L2,3 absorption edge jump (blue), and the isotropic spectrum resulting from the subtraction of the previous two (black). During the data analysis the orbital and spin moments, µl and µs respectively, were calculated assuming the 3d electron occupation number as for bulk hcp Co n3d = 7.5. The shoulders at the lower and higher energies of the main peak at the Co L3 edge (fig. 2) reveal that despite the H+ etching procedure a small amount of oxide is still present. The separation of the metallic and non metallic contributions to the magnetic properties of the Co crystal cannot be easily achieved; the presence of a small amount of CoO on the surface of the samples has an impact on the results obtained by the surface sensitive XAS technique. It leads primarily to a reduced remanent magnetisation of the samples. The calculated values of the orbital and spin moment
orbital moment (µB/atom)

spin moment (µB/atom)

343

are lowered by an unknown factor with respect to a perfectly oxide free surface. The lower panel of fig. 2 shows the dichroic signal as a result of the subtraction of the positive and negative magnetised XAS spectra, the XMCD signal (black line), together with its integral value (red line). Fig. 3 shows the spin moment, µs versus the azimuth angle ϕ, while fig. 4 shows the orbital moment, µl. Clearly, the patterned samples have an angular dependency with a period of ϕ = 60 degrees. Two sets of data are shown in the graph. The black line represents the first run from ϕ = –30° to ϕ = 60° while the red line the way back from ϕ = 60° to ϕ = -60°. The µs and µl maxima were both found at azimuth angles of ϕ = –30 and ϕ = 30 degrees while the minima at ϕ = –60, ϕ = 0 and ϕ = 60 degrees, respectively. According to the geometry of the experiment this means that the magnetization easy axis lies along the edge to edge direction of the triangle pattern, while the hard magnetization axis lies along the sphere to sphere center direction. The orbital and spin moment variation is about 15% for both values. The maximum value calculated for the spin moment is µs = 0.68 µB/atom while the minimum value is µs = 0.43 µB/atom. The maximum and minimum value for the orbital moment are µl = 0.051 µB/atom and µl = 0.031 µB/atom. The absolute values of µl and µs calculated by means of the sum rules are lower with respect to the calculated or measured values of the orbital and spin moments for hcp bulk Co. Several factors play a role in such a result. Probably the applied pulsed field of 800 Oe leads to an incomplete magnetisation of the samples measured in remanence. Static magnetization loops show that the saturation for non H+ sputtered samples is achieved at about 2 kOe and the cycle becomes reversible at about 800 Oe. However, for samples having their oxidised surface removed the magnetic hysteresis cycles are different and 800 Oe should be enough to nearly fully saturate them. Turning back the angle ϕ the minima/maxima get shifted of about 10 degrees like if an energetic barrier must be overcome in order to rotate the magnetic ordering of the triangles. Theoretical calculation have been planned to explain such behaviour. The dipolar interaction probably plays a role in the magnetic alignment of the triangular elements and the geometrical arrangement of the patterned elements should play a significant role in the observed anisotropic effect. The sixfold magnetic anisotropy shown by the Co patterned samples is consequence of the pattern shape. The effect probably can be ascribed to configurational anisotropy [6]; a deviation from the uniform magnetization driven by the shape of the pattern. With respect to previous studies on triangular nanostructures in the present case the sixfold anisotropy cannot be ascribed only to the shape of a single triangular element, but is consequence of the symmetry of the whole Co pattern. The variation of the magnetic moment as a function of the azimuthal angle obtained by XMCD was also independently confirmed by the angular dependence of the coercive field obtained from magnetic hysteresis loops using a vibrating sample magnetometre. We thank S. Rudorff for the technical support during measurements at BESSY, Dr. H. Rossner, Dr. E. Holub-Krappe and Prof. A. Tennant for useful scientific discussion and continuos encouragement. References
[1] C. Chappert, H. Bernas, J. Ferré, V. Kottler, J.-P. Jamet, Y. Chen, E. Cambril, T. Devolder, F. Rousseaux, V. Mathet, and H. Launois, Science 280, 1919 (1998). [2] A. Ney, C. Pampuch, R. Koch , and K. H. Ploog, Nature 425, 485 (2003). [3] J. Rybczynski, U. Ebels, and M. Giersig, Colloids Surf. A: Physicochem. Eng. Aspects 219, 1 (2003). [4] A. Kosiorek, W. Kandulski, P. Chudzinski, K. Kempa, and M. Giersig, Nano Lett. 4, 1359 (2004). [5] P.Imperia, D. Schmitz, H. Maletta, N. Sobal, and M. Giersig, Phys Rev. B 72, 014448 (2004) [6] R. P. Cowburn, D. K. Koltsov, A. O. Adeyeye, M. E. Welland, and D. M. Tricker, Phys. Rev. Lett. 83, 1042 (1999).

344

Reading magnetism of one layer of Single Molecule Magnets
D. Gatteschi,1 M. Mannini,1 R. Sessoli,1 A. Cornia,2 L. Zobbi,2 C. Cartier dit Moulin,3 P. Sainctavit,4 and P. Imperia.5
1. INSTM & Department of Chemistry, University of Florence, via della Lastruccia 5, I-50019 Sesto Fiorentino, Italy 2. INSTM & Department of Chemistry, University of Modena and Reggio Emilia, via G. Campi 183, I-41100 Modena, Italy 3. Laboratoire de Chimie Inorganique et Matériaux Moléculaires, Case 42, Université Pierre et Marie Curie, 4, place Jussieu F-75252 Paris cedex 05, France 4. Laboratoire de Minéralogie Cristallographie de Paris, Tour 16-26, Université Pierre et Marie Curie, 4, place Jussieu F-75252 Paris cedex 05, France 5. Hahn-Meitner-Institut Berlin Glienicker Str. 100 D-14109 Berlin, Germany

Nowadays chemists are able to produce molecules having, individually, properties of bulk materials or even capable to execute a function. Thus molecules can constitute the building blocks for the growing little world of nanotechnologies. Magnetic materials make no exception. Actually since the early nineties of the last century a new kind of molecules has been investigated that individually behaves like a magnet and for this reason they have been called Single Molecule Magnets (SMMs).[1] They show a magnetic hysteresis that, contrarily to traditional materials is not a cooperative effect but rather a feature of the molecule on its own. SMMs are in fact polynuclear metal complexes comprising paramagnetic transition metal centres that, thanks to exchange interactions inside the molecule, originate a high spin ground state. When a large spin in the ground state is associated with a strong easy axis magnetic anisotropy the magnetization freezes at low temperature as its reversal requires the overcome of an energy barrier. Due to their reduced dimensions quantum effects are important and in principle SMMs could be exploitable in quantum devices.[2] The first and most investigated SMM is a cluster constituted by twelve Mn MnIII atoms (called Mn12, Fig.1) organized in an MnIV internal tetrahedron of four MnIV and an O external ring of eight MnIII atoms; all Mn C atoms are in an octahedral coordination, R connected by oxygen atoms creating a sort of mixed valence manganese oxide in which the growth is systematically blocked by an external organic shell (16 carboxylic acids) creating a perfectly monodisperse set of magnetic particles. Each MnIV owns a magnetic moment antiparallel to those of the MnIII atoms, resulting in a ferrimagnetic structure that determines a ground state Fig1. Schematic view of the structure of the Mn12 cluster. characterized by a total spin S=10. The interesting potentiality of this kind of materials resides in the magnetic memory of the individual molecule, but, up to now, single molecule properties have been extrapolated from the those of bulk samples (single crystal, powder, and diluted solution). The simplest strategy to reach individual addressing of the clusters is to organize SMMs on surfaces and use techniques such as Scanning Probe Microscopy (SPM) that indeed have the high spatial resolution required for single–molecule addressing. Recently different methods of depositing Mn12 derivatives on a surface have been reported.[3] For instance we have suggested some chemical modifications of the original

345

external organic shell to graft the clusters on a gold surface. A crucial step is to evidence that during the deposition procedure the complex molecule still retains its structure and, above all, its peculiar magnetic behaviour. Traditional techniques for magnetic characterization lacks the necessary sensitivity while the x-ray absorption spectroscopy (XAS), through the x-ray magnetic circular dichroism (XMCD) method, has shown to be able to detect the magnetism of a submonolayer of magnetic atoms.[4] Moreover, by recording the magnetic dichroic signal at the manganese L2,3 edges in Mn12 clusters, one has immediate access to the magnetic structure of the cluster, because the contribution from MnIII is easily distinguished from that of the antiparallel aligned MnIV ions, as shown for the simplest derivative of Mn12.[5] We have used XMCD technique to characterize the bulk properties of Mn12TE = ([Mn12O12(OOCPhSCH3)16(H2O)4]), a more complex Mn12 derivative (Fig. 2) that is able to bind gold surfaces through sulphur atoms; XAS and XMCD characterizations, obtained recording signals in total electron yield at 4K and using a field of 4 T evidence the persistence of the characteristic ferrimagnetic spin structure. When transferring this kind of characterizations to the analysis of a single layer of molecule
0.07 0.06 0.05 0.04 0.03 0.002 0.001 0.000 -0.001 635 640 645 650 655 660 665

σ+ σ-

XM CD signal

Energy (eV )

Fig2. Structure of the Mn12TE cluster (left) and bulk XAS and XMCD spectra (right).

problems related to photo-reduction can be easily encountered due to the high energy of third generation synchrotron radiations. The necessity of a lower flux appears as counter-intuitive but indeed position and energy stability play a key role in our experiment. For this reason BESSY, and in particular UE 46 PGM beamline, represents the ideal facility for this kind of characterizations. In fact, playing with various parameters we reduced the number of photons by a factor of 500 and an extra factor of 400 concerning the photon density was obtained thanks to the UE46 parallel beam mirror system. In this way we have been able to characterize samples prepared ex situ starting from a solution o f Mn12TE that is adsorbed on a gold (111) surface. The Mn12TE clusters form a submonolayer in which each molecule is isolated from the others, as shown in Fig. 3, leaving a wide clean gold surface in which each molecule is well identifiable and addressable.[6] This kind of sample was inserted inside the liquid He cryostat of the XMCD setup and during one day of measurements no evidences of significant photo-reduction have been noticed. Preliminary results (see Fig. 3) evidence that XAS spectra strongly resemble those of the bulk sample with a slight increase of a signal due to spurious MnII contributions. Surprisingly the dichroic signal is substantially modified after the deposition on the Au surface.

346

0.40 1.2 1.0 0.8 0.6 0.4 0.2 0.0 0.00 -0.2 -0.4 -0.6 -0.05 -0.10 -0.15 635 640 645 650 655 660 665

σ+ σ-

0.35 0.30 0.25 0.20 0.15 0.10 0.05

XMCD signal

5nm

m

Energy (eV)

Fig3. Constant–current STM image of Mn12TE assembled on the Au(111) surface (left) and XAS and XMCD spectra (right) of this submonolayer.

A qualitative analysis, to be confirmed by a series of detail calculations now in progress, suggests that in addition to MnIII and MnIV also a MnII component is present. Moreover the typical fingerprint of antiparallel alignment of MnIII and MnIV is not detected from the monolayer. Interactions with the gold substrate, as well as a reduction during the grafting reaction, might be at the origin of this observation. This intriguing result does not however reduce the interest in this kind of innovative materials but suggests us to continue this investigation by playing on the deposition conditions and by exploiting different substrates. References
[1] D. Gatteschi, R. Sessoli, J. Villani Molecular Nanomagnets, Oxford University Press, Oxford 2006. [2] H. Park, J. Park, A. K. L. Lim, E. H. Anderson, A. P. Alivisatos and P. L. McEuen Nature 407, 57 (2000). [3] A.Cornia, A. C. Faretti, L.Zobbi, A. Caneschi, D. Gatteschi, M. Mannini, R. Sessoli Structure & Bonding, (2006) in press. [4] P. Gambardella, A. Dallmeyer, K. Maiti, M. C. Malagoli, W. Eberhardt, K. Kern, C. Carbone Nature 416, 301 (2002). [5] R. Moroni, Ch. Cartier dit Moulin, G. Champion, M.-A. Arrio, Ph. Sainctavit, M. Verdaguer, and D. Gatteschi, Phys. Rev. B 68, 064407 (2003). [6] L. Zobbi, M. Mannini, M. Pacchioni, G. Chastanet , D. Bonacchi, C. Zanardi, R. Biagi, U. Del Pennino, D.Gatteschi, A. Cornia and R. Sessoli Chem. Comm. 12, 1640 (2005). Acknowledgments The authors are grateful to the BESSY staff, for support before and during measurement time. Financial support from the EC through BESSY IA-SFS Access Programme Contract N° R II 33-CT2004-506008, "Magmanet" NMP3-CT-2005-515767, and "QuEMolNa" MRTN-CT-2003-504880 projects is gratefully acknowledged.

347

IR - Synchrotron Ellipsometry for the Characterisation of nanostructured organic and biomolecular films: DNA Bases and Polymer Brushes K. Hinrichs1, S. D. Silaghi2, U. Schade3, K.-J. Eichhorn4, M. Stamm4, L. Ionov4, D. R. T. Zahn2, N. Esser1 and M. Gensch1 Institute for Analytical Sciences- ISAS, Albert-Einstein-Str. 9, D - 12489 Berlin Chemnitz University of Technology, Semiconductor Physics, D- 09107 Chemnitz 3 BESSY GmbH, Albert-Einstein-Str. 15, 12489 Berlin, Germany 4 Leibniz Institute of Polymer Research Dresden e.V., Hohe Str. 6, 01069 Dresden, Germany
2 1

Nanostructured and biomolecular films were characterized by IR ellipsometry [1-4] and complementary methods (VUV ellipsometry, XRD and AFM). For the laterally resolved ellipsometric measurements the recently upgraded IR synchrotron mapping ellipsometer at the IRIS beamline was used and it could be shown that the molecular structure of nanopatterned organic films can be derived from the measurements. Organic films are of technological interest for development of electronic, optical and sensoring devices. For the design of such organic devices it is often important to know the structural and optical properties of organic films and how they are bonded to the substrate. Optical methods are well suited for the contactless and non-invasive characterization. Infrared and VUV optical properties of thin films are correlated to vibrational and electronic excitations, respectively. Therefore the corresponding spectra are well suited for structural analysis, while the application of VIS ellipsometry for this purpose is often limited by the similarity of refractive indices for many organic materials. Ellipsometric results for two thin organic films on silicon, a DNA base and a polymer brush gradient film are discussed in the following. Guanine film on silicon In cooperation with Chemnitz University of Technology different films of DNA bases were studied and it could be shown that the average molecular orientation and the anisotropic optical constants can be determined from synchrotron ellipsometric spectra. Fig.1 shows measured and calculated ellipsometric spectra for a 84 nm thick guanine film together with schematic drawings for the molecular orientations used for the calculation. The shape of molecular bands, which are assigned to characteristic molecular vibrations, is characteristic for the average molecular orientation. From simulations within optical models the thickness and the molecular orientations can be derived [1-4]. However, even from the raw data qualitative assertions concerning the molecular orientation can be drawn from the observed spectral line shapes.

Fig.1: Thin guanine film on silicon: Measured tanΨ spectrum is shown in comparison with calculated spectra for two different molecular orientations. In the shown spectra dip-up features indicate predominant in-plane orientation of the corresponding transition dipole moments, dip-down features indicate substantial out-of plane components. A single tanΨ spectrum was measured within less than 25 s (including the time needed for rotation of polarizers). The structure of the guanine molecule is shown as inset.

348

Polymer brush film on silicon During the 2nd semester of 2006 the existing IR ellipsometric set up was modified to allow micro-focus mapping experiments for scanning areas up to 50 x 50 mm. The upgrading of the ellipsometer was successfully tested with the investigation of mixed polymer brush gradient films. These ultra thin stimuli responsive films are studied in collaboration with IPF Dresden with the aim to analyze the dependence of their functionality on the brush composition. As a proof of principle 1D polymer gradient brushes of PS/PBA and PS/P2VP were mapped (see figure 2).

Fig. 2: Schematic of the geometry of a polymer brush gradient film and maps perpendicular to the gradient (top left) and along the gradient (top right) of a PS /PBA brush on silicon. The laterally varying composition of the brushes can easily be followed by the IR ellipsometric parameter tanΨ. In conclusion the presented results show that mapping ellipsometric measurements for structural analysis of nanostructured thin organic films are now possible within suitable time scales (a single spot with a size of 400 x 400 µm2 can be investigated in 25 seconds). Based on these results further mapping experiments for analysis of biodiagnostic arrays and stimuliresponsive polymer brush gradients are scheduled in 2006. For example the binding of biomolecules to a specific linker will be investigated. Support by the EU through SSA DASIM (ctr. Nr. 00055326) and through the EFRE program (ProFIT grant, contract nr. 10125494) is gratefully acknowledged. [1] K. Hinrichs, M. Gensch, N. Esser, Applied Spectroscopy 59 (2005) 272A [2] K. Hinrichs, S. D. Silaghi, C. Cobet, N. Esser and D. R. T. Zahn, phys. stat. sol. b 242 (2005) 2681 [3] L. Ionov, A. Sidorenko, K.-J. Eichhorn, S. Minko, K. Hinrichs, Langmuir 21 (2005) 8711. [4] M. Gensch, E.H. Korte, N. Esser, U. Schade and K. Hinrichs, Infrared Phys. Technol. (2006) in print.

349

Polarization of oxygen in Co-doped TiO2
A. Nefedov1 , N. Akdogan1 , R.I. Khaibullin2 , L.R. Tagirov2,3 , H. Zabel1
1 Institut 2 Kazan

¨ ¨ fur Experimentalphysik/Festkorperphysik, Ruhr-Universitat Bochum, Germany ¨ PhysicalTechnical Institute of RAS, Kazan 420029, Russian Federation 3 Kazan State University, Kazan 420008, Russian Federation

Diluted magnetic semiconductors (DMSs), in which a portion of atoms of the nonmagnetic semiconductor hosts are replaced by magnetic ions, are key material for spintronics. Recently several oxide-based DMSs (and in particular, the Co-doped TiO2 system) have been reported to be robust, room temperature ferromagnets. Recently, we have reported room temperature ferromagnetism and in-plane magnetic anisotropy of single-crystalline rutile structures after Co implantation [1]. We concluded that ferromagnetism in this system results from incorporation of Co ions in the TiO2 lattice, but co-existence with Co nanoclusters could not be excluded. To clarify this situation we studied in a detail magnetic properties of Co-doped (100)-oriented rutile TiO2 single crystals for different implantation doses. The investigations were made by using x-ray resonant magnetic scattering (XRMS) at room temperature and below. In this contribution, we report on a polarization of oxygen ions in TiO2 matrix, which can be one of possible origins of ferromagnetism in oxide-based DMSs. The (100)-oriented single-crystalline rutile TiO2 plates were implanted using by Co+ ions with the energy of 40 keV and with the current density of 8 mA/cm2 . The implantation dose varied in the range of 0.25-1.50·1017 ions/cm2 . In order to understand nature of magnetism in oxide-based

Figure 1: Hysteresis curves measured at Co L (closed symbols) and O K (open symbols) edges at T=30 K. DMSs we carried out XRMS studies at Co and Ti L3,2 edges as well as at O K edge. The XRMS experiments were carried out using a UHV-diffractometer ALICE at the undulator beamline UE56/1 of BESSY. For measurements at Co L edges the scattering angle was fixed at position of 2θ = 8.2◦ , but at Ti L3,2 and O K edge the scattering angle was fixed at 2θ = 12◦ , which corresponds to the same scattering vector in reciprocal space. Within our sensitivity no magnetic signal was found at Ti edges, however, in addition to a strong magnetic signal on Co L edges a small, but clearly visible magnetic signal was observed at O K edge.The hysteresis curves measured at 30 K at Co L edge

350

(E=780 eV) and O K edge (E=533 eV) are depicted in Fig. 1 (closed and open symbols, respectively). In the figure it is seen the shape of both hysteresis curve is the same, but with the opposite sign. It means that oxygen atoms neighbouring to Co atoms are polarized antiferromagnetically and ferromagnetic behaviour of Co-doped TiO2 samples can be explained by exchange mechanism through oxygen atoms. In order to distinguish ferromagnetic behaviour from superparamagnetism field cooling (FC) and zero field cooling (ZFC) measurements were carried out for sample implanted with the dose of 0.25·1017 ions/cm2 . This sample demonstrates a pure paramagnetic behaviour at room temperature after subtracting diamagnetic contribution from the substrate. In Fig. 2 the XRMS results are presented after cooling down this sample to 4.2 K with a magnetic field H=2700 Oe and without magnetic field. In Fig. 2a) asymmetry ratio data measured in remanent state after FC (closed circles) and ZFC (open circles) are depicted. Both curves are the same within experimental errors and clearly demonstrate the same value of the remanence magnetization. The asymmetry ratio measured after field cooling to T=30 K, i.e. temperature above TC (TC = 12 ± 3 K) is depicted by star symbols. In Fig. 1b) the hysteresis curves measured at 4.2 K after FC (closed circles) and ZFC (open circles) are shown. In a conclusion, magnetic signal at O K edge was observed, i.e. after implantation of Co ions into

Figure 2: Asymmetry ratio measured after FC (closed circles) and ZFC (open circles) at T=4.2 K. Star symbols corresponds to a data measured after FC to T=30 K. b) Hysteresis curves measured after FC (closed symbols) and ZFC (open symbols) at T=4.2K.

TiO2 substrates the polarization of oxygen atoms is taken place. We believe that this polarization can be responsible for nature of ferromagnetism in DMSs. We gratefully acknowledge the BMBF for the financial support through Contracts No. 03ZA6BC2 (ALICE diffractometer) and No. 05ES3XBA/5 (travel to BESSY). N. Akdogan acknowledges a fellowship through the International Max-Planck Research School ”SurMat”.

References
[1] N. Akdogan et al., J. Phys.: Condens. Matter. 17, L359 (2005).

351

Energy Dispersive Small Angle X-ray Scattering
Tushar P. Santa, Wolfram Leitenbergerb, Tobias Panznera, Ullrich Pietscha
b a

Institute of Physics, University of Siegen, D-57068 Siegen, Germany Institute of Physics, University of Potsdam, D-14469 Potsdam, Germany.

We report on energy dispersive small angle X-ray scattering (EDSAXS) experiments using white synchrotron radiation performed at Energy Dispersive Reflectivity (EDR) beamline at BESSY II. The experimental set-up is shown schematically in Fig.1.
Detector Slits

Sample 30 m Slits

Flight tube 2θ

Energy Dispersive Detector

White Synchrotron Source E = 5 to 25KeV

30cm

130 cm

Figure 1 : Schematic sketch of experimental set-up at EDR beamline at BESSY II

Fig. 2 shows the measured spectrum at a fixed scattering angle of 0.18 deg for pure water in a glass capillary tube with 1.5 mm diameter and 100 micron wall thickness. Compared to emission spectrum from the bending magnet the measured spectrum is modified by absorption due to air and sample in low energy range. This spectrum is used as reference standard for further measurements.
10
3

10
Intensity

2

10

1

10

0

6

8

10 12 14 16 18 20 22 24
E(KeV)

Figure 2 : Measured spectrum of water at fixed scattering angle of 0.18 deg

We measured EDSAXS spectra for Gold nanoparticles with diameters of about 12 nm and 40 nm solved in water. Fig. 3 shows the scattering for Au nanoparticles with 12 nm diameter as a function of energy. The nanoparticle solution was filled in glass capillary of 1.5 mm diameter. The sample to detector distance was 1.3 m. The measurement time for each spectrum is 10 min for emission at 15 mA current from the storage ring. The initial spectrum as seen in Fig. 2 becomes modified by the form factor of the particles to the spectrum shown in Fig.3. We could observe a weak minimum and a subsequent maximum in the scattering profile. To confirm that the origin of both is the form factor of the particle we performed the measurements at different scattering angles 0.42, 0.47, 0.50, 0.52, 0.57 deg. One can clearly observe that the position of the peak maximum (also the minimum) shifts from higher to lower energies as the scattering angle is increased (see arrows). In Fig 4 the SAXS spectra are plotted as a function of scattering vector q. These spectra are obtained by normalizing

352

spectra of nanoparticles (Fig. 3) to that of water (Fig. 2) at the corresponding scattering angles. This procedure removes all experimental influences from the data.
10 10 10
Intensity
6 5 4 3 2 1 0

θ = 0.42 θ = 0.47 θ = 0.50 θ = 0.52 θ = 0.57

ο ο ο ο ο

10

2
Θ=0.42 Θ=0.47 Θ=0.50 Θ=0.52 Θ=0.57

10 10 10 10 10

I(Sample/Water)

10

1

Monochromatic

10

0

-1

10

-1

6

8

10 12 14 16 18 20 22 24

0.06 0.08 0.10 0.12 0.14 0.16 0.18 0.20

E(keV)

q(Å-1)

Figure 3 : Au nanoparticle 12 nm spectra as a function of energy at different scattering angle.
10 10
Intensity
4

Figure 4 : Au Nanoparticle 12nm spectra normalised to water spectra as a function of scattering vector q.
Θ=0.18 Θ=0.20 Θ=0.21 Θ=0.22 Θ=0.24

3

10 10 10 10

2

I(Sample/Water)

Θ=0.18 Θ=0.20 Θ=0.21 Θ=0.22 Θ=0.24

10

2

Monochromatic

10

1

1

0

10

0

-1

6

8

10 12 14 16 18 20 22 24

0.02

0.03

0.04

0.05
q(Å-1)

0.06

0.07

0.08

E(KeV)

Figure 5 : Au Nanoparticle 40 nm spectra as a function of energy.

Figure 6 :Au nanoparticle 40 nm spectra as a function of q.

The position of the first minimum occurs at qminR = 4.5. For the above sample (see red line in Fig.4) with qmin = 0.066 Å-1 one obtains R = 6.8 nm. This agrees well with the results of monochromatic SAXS measurement. Fig. 5 shows the scattering from other sample Au nanoparticles with diameter of 40nm. Again the measurements are taken at different scattering angles 0.18, 0.20, 0.21, 0.22, 0.24 degrees. The time for measurement was 3 min for normal emission from storage ring. Fig. 6 is the spectra for Au nanoparticles 40 nm normalized to water spectra as a function of scattering vector q. qmin occurs at 0.026Å-1 which gives diameter of nanoparticles as 34 nm. This again agrees with the monochromatic SAXS measurements but is smaller then expected (2R=40nm). Comparing monochromatic SAXS measurements in home laboratoty with EDSAXS at BESSY we were able to reduce the measurement from 3 hours to 3 minutes. This fact can be useful in systematic investigation of large series of samples which is essential for optimising the synthesis procedure of nanoparticles. Also we can use EDSAXS for studying the processes with slow dynamics. But the specific experimental conditions at BESSY II limit the accessible q range of investigation. This in turn puts the limit for the measurable size of the particles. References [1] Yu, K.L., Lee, C.H., Hwang, C.S., Tseng, H.C., Tseng, P.K., Lin, T.L., Chang, S.L., Sheu, R.J., Chen, S.H. (1999) Rev. Sci. Instrum. 70, 3233-3238. [2]Portale, G., Longo, A., D´llario, L., Martinelli, A., Caminiti, R. (2004) Appl. Phy. Lett. 85, 4798-4800. [3] Pietsch, U., Panzner, T., Leitenberger, W. (2003) Physica B 357, 45-52.

353

Relaxation of lattice distortion in Creep-deformed Single Crystal Superalloy SC16 at 1173 K
G. Schumacher1, N. Darowski1, I. Zizak1, H. Klingelhöffer2, W. Chen3 and W. Neumann3
2

Hahn-Meitner-Institut Berlin GmbH Bundesanstalt für Materialforschung und –prüfung 3 Institut für Physik, Humboldt-Universität zu Berlin

1

Single crystal superalloys show excellent creep and fatigue properties at high temperatures and are widely used as structure materials in gas turbines. Up to now, few is known about thermal stability of the microstructure in creep-deformed single crystal superalloys. Therefore, we have measured positions and profiles of 001 and 002 reflections in moderately creep-deformed single crystal superalloy SC16 (Ni-16Cr-3Mo-3.5Ti-3.5Al-3.5Ta; in wt %) as a function of time at 1173 K. The initial two-phase microstructure of the alloy consisted of cuboidal γ’ precipitates with L12 structure of Ni3Al-type which were coherently embedded in the γ fcc solid solution matrix phase. The precipitates had an average edge length of 450 nm and a volume fraction of about 40 % [1]. Prior to X-ray analysis the specimens were creep deformed at 1223 K and 150 MPa parallel to the [001] direction under tensile load. The maximum creep strain ε was 0.5 0,038 %. XRD measurements were carried out using the 6-circle diffractometer at the KMC-2 beamline at BESSY at an X-ray 0,036 energy of 8 keV. Symmetrical Bragg geometry was used on samples with surface perpendicular to the load axis. 0,034 Measurements were performed in a high 0,E+00 5,E+04 1,E+05 vacuum chamber at 1173 K. The widths time [s] and positions of the 001 reflections were determined by fitting Gaussian functions to Fig.1 Line width of 001 reflection in creepdeformed SC16 (ε = 0.5%, T = 1223K ) as a the data. 001 superlattice reflection provided information on the crystal lattice function of time measured at 1173 K. of the γ’ precipitates only, while measurements of 002 profiles contained information on both phases, γ and γ’. The procedure to determine the lattice misfit from the 002 profile is described in previous work [2-4].
0,040

Fig. 1 shows the line width of the 001 peak as a function of time measured at 1173 K. Though variation of data is considerable the line width shows a clear trend to lower values with increasing time. The line is a linear least squares fit to the data. With increasing time the position of the 001 reflection shifted to higher values (see Fig. 2) indicating a decrease in lattice spacing within the γ’-precipitates parallel to the [001] direction. The 002 profiles split into two maxima which can be assigned to the lattice parameters of the γ and γ’ phase (Fig. 3). From the position of the 001 reflection of γ’ phase it can be concluded that the maximum at larger angles is predominantly due to the 002 reflection of the γ’-phase. The maximum at lower angles can therefore be ascribed to the 002 reflection of the γ-matrix phase. After thermal treatment for 8.5⋅104 s the distance between the two maxima has increased. The shift

FWHM [degree]

354

of the 002 γ’ peak is consistent with the observed shift in 001 peak position shown in Fig. 2. A more detailed analysis showed that the difference in 002 peak position between γ and γ’ phase increased by about 25 % after t = 8.5x104 s with respect to the value at t = 0.
24,564 24,562 24,560 24,558 24,556 24,554 24,552 0,E+00

peak position [degree]

5,E+04 time [s]

1,E+05

Fig. 2 : Position of 001 reflection in creep-deformed SC16 (ε = 0.5%, T = 1223K ) as a function of time measured at 1173 K. In the present experiment, the temperature was kept constant. The measured decrease in 001 line width and the change in lattice misfit can, therefore, not be ascribed to thermoelastic behavior caused by the different thermal expansion coefficients of the two phases. The observed changes are rather due to relaxation of the deformed state to a different structural state which corresponds the state prior to deformation. Prior to deformation the microstructure is free of dislocations. The arrangement of dislocations after deformation prior to thermal treatment is schematically depicted in Fig. 4. During tensile creep-deformation dislocations have arranged in a way that the half-planes lie within the γ’-precipitates at the interfaces which are aligned perpendicular to the [001] load axis while the half planes lie within the γ matrix phase at the interfaces which are parallel to the load axis [3,4]. This configuration has been shown to change the misfit in the positive direction compared to the non-deformed state
35000 30000 25000

t = 8.5E4 t=0

20000 15000 10000 5000 0 -5000 50.0

Fig. 3: Profiles of the 002 X-ray reflections measured at 1173 K in SC16 immediately after tensile-creep deformation (t=0, dots) and after heating (t = 8.5 x 104 s , squares).

Intensity [counts]

50.1

50.2

50.3

50.4

50.5

2 Theta [degree]

355

[3,4]. The changes in γ and γ’ lattice parameters measured in the present work suggest changes in dislocation substructure during thermal treatment. The dislocations at the γ/γ’ interfaces rearrange in a way to approach the configuration prior to deformation. This state has a lattice misfit which is more negative than the one prior to heat treatment.

γ‘ γ

Fig. 4: Schematic illustration of the side view of a γ’ raft in creep-deformed superalloy SC16 directly after creep-deformation

Support by the DFG (NE646/5-3 and Schu1254/3-4) is gratefully acknowledged. References [1] [2] [3] T. Malow, J. Zhu and R.P. Wahi, Z. Metallkd. 85 (1994) 9. N. Darowski, I. Zizak, G. Schumacher, H. Klingelhöffer, W. Chen and W. Neumann, J. Phys. D: Appl. Phys. 38 (2005) A200-A203. W. Chen, N. Darowski, I. Zizak, G. Schumacher, H. Klingelhöffer and W. Neumann Proc. Spring Meeting of the EMRS, May 31 – June 3, 2005, Strasbourg (France), Symposium O “Synchrotron Radiation and Materials Science”, Nucl. Instr. Meth. B, in press. W. Chen, N. Darowski, I. Zizak, G. Schumacher, H. Klingelhöffer, N. Wanderka and W. Neumann, Mat. Sci. Forum 426 (2003) 4555.

[4]

356

Temperature Dependence of Lattice Distortion in Strongly Creep-deformed Single Crystal Superalloy SC16 Measured by Means of X-Ray Diffraction
G. Schumacher1, N. Darowski1, I. Zizak1, H. Klingelhöffer2, W. Chen3 and W. Neumann3
2

Hahn-Meitner-Institut Berlin GmbH Bundesanstalt für Materialforschung und –prüfung 3 Institut für Physik, Humboldt-Universität zu Berlin

1

In previous work we investigated lattice distortion of γ’-precipitates after tensile creep deformation to 0.5 % strain by means of X-ray diffraction (XRD) [1-3]. While creep-straining to 0.5% occurs within stage I of creep-deformation, creep-straining to fracture proceeds over three stages, stage I to stage III characterized by different deformation processes. We studied therefore lattice distortion after severe plastic deformation (15% strain) of single crystal superalloy SC16 and compared the results to those obtained after straining to 0.5%. The single crystal superalloy SC16 (Ni-16Cr-3Mo-3.5Ti-3.5Al-3.5Ta; in wt %) was used for our study. The initial two-phase microstructure of the alloy consisted of cuboidal γ’ precipitates with L12 structure and a γ fcc matrix phase. Prior to deformation the precipitates had an average edge length of 450 nm and were coherently embedded in the γ matrix phase. The volume fraction of the γ’ phase was about 40 %.
0,08

FWHM [degree]

0,06

Specimens were creep-deformed at 1223 K and 150 MPa parallel to the [001] direction under tensile load to maximum creep strain of 15 %. The measurements were performed at various temperatures during heating from room temperature to 1073 K and after cooling back to room temperature.

X-ray diffraction (XRD) measurements were made at X-ray 0,04 energy of 8 keV. The 6-circle diffractometer at the KMC-2 200 700 1200 beamline of BESSY was used for Temperature [K] the measurements. The 001 superFig. 1. Line-width (FWHM) as a function of lattice reflection and the 002 temperature measured on creep-deformed single fundamental reflection were crystal superalloy SC16. measured in symmetrical Bragg geometry on samples with surface perpendicular and parallel to the load axis, respectively. The measurements made on the 001 reflection provide information on the crystal lattice of the γ’ precipitates only, while measurement of the 002 profiles reflects changes in the crystal lattice of both the precipitate phase and the matrix phase. The widths and positions of the 001 reflections were determined by fitting a Gaussian function to the data. The difference in lattice paramters was deduced from the 002 profiles.

357

Fig. 1 shows the line width (FWHM) as a function of temperature. The line width decreases as a function of temperature indicating relaxation of micro-strains within the γ’-precipitates. The relative decrease in line width is about 35 %. This relative change is comparable to that measured on the 0.5 % creep deformed specimen [2]. After the heat treatment the specimen was cooled down to room temperature and the line-width was measured again. This value of the line width (indicated by a triangle in Fig. 1) is lower compared to the value measured prior to heat treatment. The measurements performed after 0.5% creep strain did not indicate different values before and after heat treatment and were therefore discussed in terms of thermo-elastic effects [2-3]. The decrease in line width measured in the present work might therefore not exclusively be due to thermo-elastic effects. Inelastic structural relaxation at high temperatures might also affect the line width. The position of 001 reflection as a function of temperature is shown in Fig. 2. The values of the 001 reflections were used to determine the position of the 002 reflection of the γ’ precipitates.

24,8 2 Theta [degree]
Fig. 2: Position of the 001 reflection as a function of temperature after tensile creep deformation to 15 % strain.

24,6

24,4 200 700 T [K] 1200

The 002 profiles measured at room temperature and at 1073 K are shown in Fig. 3a and 3b, respectively. The profile shown in Fig. 3a has two separate maxima. The first maximum coincides with the position of the 002 reflection of the γ’-phase calculated by use of the 001 peak position. The maximum at higher angles is, therefore, ascribed to the γ matrix phase indicating a positive lattice misfit δ = 2⋅[a(γ’)−a(γ)]/[a(γ’)+a(γ)] at room temperature. A rough estimation of the misfit yields a value at room temperature of about +0.34 %. This value is large compared to the value δ = 0.10 % measured at room temperature on non-deformed specimens [1,4], but it is only slightly larger than the value δ = 0.28 % measured on the specimen creep-deformed to 0.5% strain [1]. The changes in lattice misfit are therefore large in stage I of creep-deformation, while further deformation obviously does not cause comparable changes in lattice parameters. This is a remarkable result, as further deformation to 15 % strain causes further appreciable changes in the γ’-raft morphology. The measured changes in lattice misfit are, therefore, rather ascribed to changes in the arrangement of dislocations at the differently aligned interfaces than to the evolution of γ’-rafts.

358

16000
50000

RT
Intensity [counts]

14000 12000

1073 K

40000

10000 8000 6000 4000 2000 0

Intensity [counts]

30000

20000

10000

0 50,4 50,6 50,8 51,0 51,2 51,4

2 Theta [degree]

-2000 49,6

49,8

50,0

50,2

50,4

50,6

50,8

2 Theta [degree]

a)

b)

Fig. 3: 002 profile measured by means of X-ray diffraction on creep-deformed superalloy SC16 at RT (a) and at 1073 K (b) At 1073K the 002 profile reveals only one maximum without any pronounced shoulder. This indicates that the lattice misfit at 1073 K is close to zero. Using the date shown in Fig. 2 the position of the 002 peak of the γ’ phase is determined to be 2θ = 50.153 °. This position is on the left hand side of the profile maximum of Fig. 3b. The lattice misfit is therefore still slightly positive. The decrease in lattice misfit with increasing temperature indicates a larger thermal expansion coefficient of the γ-phase compared to the γ’-phase in agreement with previous measurements [ 4,5]. This work was supported by DFG (NE646/5-3 and Schu1254/3-4). Help in specimen preparation by C. Förster (HMI) and H. Kropf (HMI) is gratefully acknowledged. [1] [2] W. Chen, N. Darowski, I. Zizak, G. Schumacher, H. Klingelhöffer, N. Wanderka and W. Neumann, Mat. Sci. Forum, 426 (2003) 4555. W. Chen, N. Darowski, I. Zizak, G. Schumacher, H. Klingelhöffer and W. Neumann Proc. Spring Meeting of the EMRS, May 31 – June 3, 2005, Strasbourg (France), Symposium O “Synchrotron Radiation and Materials Science”, Nucl. Instr. Meth. B, in press N. Darowski, I. Zizak, G. Schumacher, H. Klingelhöffer, W. Chen and W. Neumann, J. Phys. D: Appl. Phys. 38 (2005) A200-A203. G. Bruno, G. Schumacher, H. C. Pinto, C. Schulze, Metallurgical and Materials Transactions A 34 (2003) 193. G. Schumacher, W. Mieckeley, W. Chen, R.P. Wahi, G. Frohberg and W. Wever, Z. Metallkd. 89 (1998) 661.

[3] [4] [5]

359

Analysis of EXAFS and MEXAFS above the L3-edge of Fe
H.H. Rossner, D. Schmitz, P. Imperia, H. Maletta, and H.J. Krappe, Hahn-Meitner-Institut Berlin, Glienicker Str. 100, D-14109 Berlin, Germany, and J.J. Rehr, Department of Physics, University of Washington, Seattle, Washington 98195-1560
Spectra of spin-averaged and spin-polarized extended x-ray absorption fine structure (EXAFS and MEXAFS) data above the L3 -edge of Fe have been measured at temperatures of 180 K and 296 K. The data were taken at the elliptical undulator beamline UE46-PGM using the gap-scan technique in the energy range of 690 eV ≤ E ≤ 1160 eV. At each energy step the absorption measurements were performed with magnetization vector parallel (µ+ ) and anti-parallel (µ− ) to the photon beam direction, resulting in the spin averaged and spin polarised absorption coefficients µ and µM : µ= µ+ + µ− 2 , µM = µ+ − µ− .

For the analysis of the EXAFS data we applied the Bayes-Turchin procedure described in Ref. [1], using the results of the code FEFF8.2 [2] for 97 scattering paths within a cluster radius of 8 ˚. The fitting A procedure is applied to the measured absorption coefficient µexp , which has been normalized to FEFF results. Possible deviations of the atomic like background absorption component µ0 are taken into account by the introduction of correction functions δµ0,Ls (k) (s = 1, 2, 3), which were determined by cubic splines at fixed support points kp [1]. For all three edges the same correction function has been used, just scaled with the proper normalization factor of FEFF. The FEFF code further provides the E0 reference energies for the three L-edges and the k-dependent values for amplitude reduction factor, scattering amplitudes, phases, and the effective mean free path lengths which also include the core hole life times. In addition to 2 these parameters EXAFS Debye-Waller (DW) parameters σj are needed for all paths j. The total crystal 2 2 disorder σj for each path was separated into the two components of structural disorder σj,struct and thermal 2 disorder σj,therm : 2 2 2 σj = σj,struct + σj,therm . It has been assumed that the structural disorder is independent of temperature and proportional to the number of atoms in the path, and that the thermal disorder is related to an ideal bcc-crystal structure. 2 The thermal components σj,therm were calculated for each scattering path starting from the elements of the dynamical matrix that describe the lattice vibrations, and applying the recursion method proposed by Poiarkova and Rehr [3], which has been described in Ref. [4]. Different from Ref. [4], where the elements were determined by spring constants, they are now constructed by Born-von Karman (BvK) parameters taking into account the crystal symmetry. The 13 BvK parameters used for the first five shells around the absorbing atom had been extracted from phonon dispersion relations measured by inelastic neutron scattering [5]. These parameters determined the thermal disorder at both measured temperatures of 180 K and 296 K, assuming that the BvK parameters do not depend on temperature. For further reduction of the fit parameters we fixed the amplitude reduction factor to 0.9, took the energy differences between the L-edges from the FEFF code, set the third cumulants C3,j for all paths to zero, and calculated the variations of the scattering half path lengths Rj from the variation of the lattice parameter a, assuming an ideal bcc-crystal structure and using the relation δχ = δa δχ δRj . δRj δa

j

The reference energy E0 for the wave vector k has been fixed to E0 = 713.08 eV. The energy scales of experiment and model have been adjusted such that the centers of the first EXAFS oscillation coincide.

360

During the fitting procedure we observed significant oscillations of the model function µmod above the L1 edge, which were not present in µexp . To take into account possible variations of the core-hole life-times for the 2s and 2p electrons corresponding to the L1 and L2,3 edges, we introduced fitting factors fλ,L1 and fλ,L2,3 for the effective mean free paths lengths λL1 (k) = fλ,L1 λ0,L1 (k), λL2,3 (k) = fλ,L2,3 λ0,L2,3 (k),

where λ0,Ls (k) were taken from FEFF8.2. The number of equally distributed support points kp on the considered k range 3.6 ˚−1 ≤ k ≤ 10.6 ˚−1 was chosen to be 10. This is the smallest number of points A A necessary to reach the condition that the minimum of χ2 with respect to the model parameters is lower than the number of degrees of freedom.

4 2 0 -2 -4x10 2x10
-2 -2

180 K

296 K

χ

exppost post

exppost post

180 K
1 0 -1 4 6 8
prior post

296 K

χµ , L

0

3

prior post

10

4

6

8

10

k [Å ]

-1

k [Å ]

-1

Figure 1: Upper part: Comparison of a posteriori experimental EXAFS oscillations (green dots with error bars) and the corresponding a posteriori model functions. Lower part: Comparison of a priori (blue dashed line) and a posteriori (red line with error band) background oscillations. In a preliminary fitting procedure the correction functions δµ0 (kp ), (kp = 1 . . . 10), the factors fλ,L1 and 2 fλ,L2,3 , the structural component σstruct , and the lattice parameter a were determined for both absorption 2 spectra measured at 180 K and 296 K. Assuming that fλ,L1 , fλ,L2,3 , and σstruct are independent of temperature, the corresponding values of the two temperatures were averaged and kept constant in the final fit. 2 2 The values used were S0 = 0.9, E0 = 713.08 eV, fλ,L1 = 0.49, fλ,L2,3 = 0.81, σstruct = 0.005 ˚2 , resulting in A ˚ for both temperatures. The EXAFS and L3 -background oscillations of the experimental a = 2.869±0.003 A data and the corresponding model functions are shown in Fig. 1. The background oscillations were defined by the smoothed L3 component of the atomic-like background µ0,L3 : χµ0 ,L3 = (µ0,L3 − µ0,L3 )/µ0,L3 . The amplitudes of the background oscillations χµ0 ,L3 and of the EXAFS oscillations χ are of similar size, which underlines the importance of the background determination. We further note that the resulting background correction functions for the two independent measurements at 180 K and 296 K coincide within their error bands and that their enhancement relative to the a priori FEFF function has been interpreted as an atomic EXAFS effect [6]. The spin polarized absorption coefficients measured at the two temperatures 180 K and 296 K are 2 shown in Fig. 2. These coefficients are given in ˚ due to the normalization of the measured data to FEFF A results. Application of a simplified rigid band model picture results in a correlation between the EXAFS

361

5x10

-5

296 K

µ

M exp

[Å ]

2

µ

M

0 Eq. (1), ∆E = -1 eV Eq. (2), ∆E = -1 eV

-5
-5

5x10

180 K
0

µ

M exp

µ

M

[Å ]

2

-5 100 200

Eq. (1), ∆E = -1 eV Eq. (2), ∆E = -1 eV 300 400

E - E0 [eV]
Figure 2: Experimental MEXAFS oscillations µM (green dots) measured at the temperatures of 180 K and 296 K are compared with the approximations of Eq. (1) (red line) and Eq. (2) (blue line).

and magnetic EXAFS (MEXAFS) oscillations: µM (E) + µM (E) ≈ L3 L2 µ0,L3 (E) dχL3 (E) dχL2 (E) − ∆E, 4 dE dE µ0,L3 (E) d2 χL3 (E − ESO /2) ∆E. 4 dE 2 (1) (2)

≈ ESO

Whereas this rigid band model picture has been verified above the separated L edges of Gd [6], a phase shift of roughly 1800 corresponding to a negative exchange correlation potential ∆E is observed for the 3d transition metal Fe. We conclude that either the simple band model picture or the exchange correlation potential have to be modified in this case.

References
[1] H.J. Krappe and H.H. Rossner, Phys. Rev. B 70, 104102 (2004). [2] A.L. Ankudinov, B. Ravel, J.J. Rehr, S.D. Conradson, Phys. Rev. B 58, 7565 (1998), A.L. Ankoudinov, Ph.D. thesis, Univ. of Washington (1996). [3] A.V. Poiarkova and J.J. Rehr, Phys. Rev. B 59, 948 (1999). [4] H.J. Krappe and H.H. Rossner, Phys. Rev. B 66, 184303 (2002). [5] V.J. Minkiewicz, G. Shirane, and R. Nathans, Phys. Rev. 162, 528 (1967). [6] H. Wende, Rep. Prog. Phys. 52, 2105 (2004).

362

Stress distributions in finite structures
Karen Pantleon Technical University of Denmark, Department of Manufacturing Engineering and Management IPL Building 204, DK – 2800 Kgs. Lyngby, Denmark

Electrochemical deposition has become the key technology in manufacturing functional thin films with finite structures, e.g. for microsystems and microcomponents. For instance, electroplating through a mask results in free-standing line patterns, which can either directly be used as metallic patterns in micro-systems or act as mould for the specially developed injection moulding process. The two most important materials in this field are i) Cu, which has become the dominating material for interconnects in integrated circuits in microelectronics, and ii) Ni, a promising material to realize movable structures for micro-electro-mechanical (MEMS). Recently, the influence of the geometry of free-standing line patterns of electrochemically deposited Cu- and Ni-films on their microstructure and crystallographic texture was studied by means of conventional X-ray diffraction [1]. Supplementary to X-ray diffraction averaging over hundreds of lines, finite element modelling (FEM) of the strain distribution within a single line was carried out [2]. FEM calculation indicate fairly inhomogeneous strain distributions; both in depth as well as over the line width. The stress state of line patterns has a dramatic effect on failure of the electrodeposited lines. An experimental verification of the strain (stress) distribution in such line patterns could not be obtained so far. Stress analysis by means of X-ray diffraction using conventional X-ray radiation was found to be unsuitable, because of insufficient diffraction intensities related to the small diffracting volume. Experiments at the beamline MagS were aimed to check the possibilities for analyzing internal stress in such line patterns using synchrotron radiation.

III II

I

IV

Fig. 1: Electrodeposited line patterns (SEM-image). Individual arrays (I…IV) contain several hundreds of parallel lines. Synchrotron experiments at the beamline MagS at BESSY were carried out on Ni-line patterns, which have been electrodeposited on a glass wafer. Four different arrays of parallel Ni-lines with varying line width (10 mm, 20 µm), interline distance (10 µm, 20 µm, 50 µm) and line length (50

363

µm, several millimeters (infinite)) were arranged on one and the same wafer, see Fig. 1. For comparison, it was additionally measured on a non-patterned Ni-film. The diffracted intensity of the 111 lattice plane was recorded as a function of the sample tilt angle ψ. During the experiments the sample was mounted and adjusted by appropriate rotation around the sample normal such that lines were oriented either vertically or horizontally with respect to the primary beam, i.e. stress was determined across (rotation angle ϕ = 0º) and along (ϕ = 90º) the lines, respectively. As a result, the dependence of the position 2θ of the diffraction peak, the integrated intensity and the corresponding full-width-half-maximum on sin2ψ was obtained. In principle, this should lead to a straightforward calculation of stresses from the measured lattice strain (from changes of the peak position, i.e. the corresponding d-spacing) according to the sin2ψ-method [3]. However, the obtained 2θ dependence on sin2ψ revealed interesting peculiarities, see Fig. 2.
28,52 28,51
°2θ

Ni - range I

28,50 28,49
ϕ=0: across the line ϕ=90: along the line

range I: W = D = 20 µm, L infinite
0,6 0,8

0,0

0,2

0,4 2 sin ψ

28,52 28,51 28,50

Ni - range IV

°2θ

28,49 28,48 28,47
ϕ=0: across the line ϕ=90: along the line

range IV: W = D = 10 µm, L = 50 µm
0,6 0,8

0,0

0,2

0,4 2 sin ψ

28,53

Ni - range V
28,52

°2θ

28,51 28,50 28,49
ϕ=0 ϕ=90

continuous film
0,4 sin ψ
2

0,0

0,2

0,6

0,8

Fig. 2: The diffraction angle 2θ in dependence on the sample tilt angle ψ (sin2ψ) for various line geometries (W – line width, D – interline distance, L – line length) as well as a non-patterned film. A pronounced non-linearity was observed in the sin2ψ-plot by a sudden decrease of the measured position of the 111-peak, which becomes lowest at a tilt angle of about 54.5º. The measured intensity distribution as well as previous studies indicate that the Ni-films have a pronounced <100> fibre texture and indeed, the angle between the measured 111 lattice plane and the 200

364

amounts to 54.7º. It was found that the extent of the non-linearity depends strongly on the pattern geometry and the adjustment of the sample during the measurement, i.e. whether the lines are oriented vertically or horizontally. Comparison to the non-patterned Ni-film indicates that not the texture solely, rather the specific geometry of the various line patterns is responsible for the observed effect. Stress relaxation is expected to be different for the two stress components (across and along the lines) and to depend on the line geometry. [1] [2] [3] K. Pantleon, M.A.J. Somers, Acta Materialia 52 (2004) 4929-4940. K. Pantleon, H. M. Jensen, M.A.J. Somers, Mat. Res. Soc. Proc. vol. 812 (2004) F3.20.1F3.20.6. Structural and residual stress analysis by nondestructive methods, ed.: V. Hauk, Elsevier Science, The Netherlands 1997.

The excellent support by E. Dudzik and R. Feyerherm (HMI Berlin) during the experiments is gratefully acknowledged.

365

High-resolution tomography investigations of micro-cracks in hard rocks
K. Thermann1, B. Kremmin1, S. Zabler2, I. Manke1, J. Tiedemann1
1

TU Berlin, FG Ingenieurgeologie,13355 Berlin 2 Hahn-Meitner-Institut, 14109 Berlin

To investigate fracture propagation in hard rocks in response to applied loads it is essential to know the existence and orientation of pre-existent microcracks. Because nucleation, growth and interaction of microcracks are considered to be the dominant, controlling mechanisms of macroscopic failure. Nevertheless, grain boundaries, low-aspect ratio cavities or interfaces of two different minerals can also function as stress concentrators and be responsible for crack initiation. In former publications, transparent materials (resin, glass, PMMA) or rock type materials (gypsum, cements, mortar) that simulate brittle failure of rock were investigated to study the different influences and their interaction [2, 3, 4, 5, 6]. Synchrotron tomography at BESSY provides the possibility to investigate microscopic features of natural rock samples. Two different types of sedimentary rocks were investigated. A Carboniferous greywacke as a clastic sedimentary rock and a Triassic limestone as a chemical sedimentary rock. Whereas limestone is composed mainly of the mineral calcite (CaCO3) a greywacke consists of angular grains of quartz, feldspar, and small rock fragments (e.g. quartzite, slate, various schists, or gneiss) set in a compact, clay-fine matrix. Because of the large sample height of 10 mm three vertical sections were measured. After the reconstruction the sections were merged to a complete tomogram of the sample on which the analysis was finally done. Rock samples loaded to different stress stages were investigated. A noticeable crack initiation was not observed until the loading stage at approximately 90 % of the uniaxial compressive strength of the particular material. In Fig. 1 voids (pores, cavities, cracks) are coloured red whereas high absorbing particles have a yellow colour. In the greywacke tomogram (Fig. 1a) a quartzite healed fracture (yellow) can be seen. The new propagated crack has nearly the same orientation as the healed crack. The high absorbing phase in the limestone sample (Fig. 1c) is pyrite (FeS2). As a result of limiting resolution these cubic minerals appear spherical in the tomogram. In Fig. 1b) the fracture process zone in front of the crack tip can be seen.

a)

b)

c)

d)

Fig. 1: Different viewing options of the greywacke (a, b) and the limestone sample (c,d)

According to the site of origin microcracks can be subdivided into grain boundary cracks, intragranular cracks and inter- or transgranular cracks but often an unambiguous classification is not possible. Hence, crack type labels also depend to some extent on the resolution of the

366

observation [6]. Concerning crack tip displacement three basic modes exist: mode I - tensile, mode II – in-plane shear and mode III – anti-plane shear [1]. a) b) c) d)

Fig. 2: Section of the unloaded (a) and loaded (b) sample and sketches of the propagated cracks (c and d)

In Figure 2 a section of the unloaded (a) and loaded sample (b) is shown. In Fig. 2c) the cracks were characterized according to their type of mode whereas in Fig. 2d) the location was analyzed. As is shown in Figure 2, first tensile or mode I cracks at grain boundaries are generated. In the upper part of Figure 2b) can be seen that intragranular crack initiation is possible, too. The tensile cracks are oriented sub-parallel to the loading direction. At the tips of these cracks were wing cracks initiated. These wing cracks start at the tips of the preexisting cracks and propagate in a curvilinear path as the load is increased. According to [2] wing cracks grow in a stable manner since an increase in load is necessary to lengthen the cracks, and align with the direction of the most compressive load. In Figure 2 can also be observed that crack coalescence occurred when cracks (grain boundary, intragranular or transgranular) were sufficiently close to each other. In a next step the crack in the greywacke sample (Fig. 1b) was precisely analyzed. At first the crack was separated from the sample and the geometrical parameters were determined. A plane was fitted to the crack to investigate the orientation (dip direction, dip angle) of the crack. The distance of each crack voxel to the plane can be calculated and so information about the crack surface roughness can be gained. Further investigations on a larger number of samples are needed to specify the results and to accomplish a statistical assessment. References
[1] ATKINSON, B.K. (1987) Fracture mechanics of rock. Academic Press, London. [2] BOBET, A. EINSTEIN, H.H. (1998) Fracture coalescence in rock-type materials under uniaxial and biaxial compression. Int. J. Rock Mech. Min. Sci. Vol. 35, No. 7, 863-888. [3] DYSKIN, A.V., SAHOURYEH, E., JEWELL, R.J., JOER, H., USTINOV, K.B. (2003) Influence of shape and locations of initial 3-D cracks on their growth in uniaxial compression. Engineering Fracture Mechanics 70, 2115-2136. [3] HOEK, E. BIENIAWSKI, Z.T. (1965) Brittle fracture propagation in rock under compression. Int. J. Fract. Mech. 1, 137-155. [5] HORII, H., NEMAT-NASSER, S. (1985) Compression-induced microcrack growth in brittle solids: axial splitting and shear failure. Journal of Geophysical Research Vol. 90, No. B4, 3105-3125. [6] KRANZ, R.K. (1983) Microcracks in rocks: a review. Tectonophysics 100, 449-480. [7] LANDIS, E.N., NAGY, E.N., KEANE, D.T. (2003) Microstructure and fracture in three dimensions. Engineering Fracture Mechanics 70, 911-925.

367

Energy and time resolved coherent X-ray reflectivity of a smooth polymer film
G. Gleber1, T. Panzner1, W. Leitenberger2, A. Pucher2, U. Pietsch1
2

Institute of Physics, University of Siegen, D-57068 Siegen, Germany Institute of Physics, University of Potsdam, D-14469 Potsdam, Germany

1

The aim of the EDR coherence project consists in the use of coherent white radiation provides by a BESSY bending magnet for time correlation spectroscopy. As model system we used a light-sensitive polymer film covered with gold colloids. Under influence of external green light the polymer becomes soft which may induce the colloids to sink into the film. The expected dynamics is slow enough to detect variations of speckle pattern in a time scale of a few seconds. The experimental setup used is shown in fig.1. The coherent part of the radiation is selected by the incident pinhole of 15 µm in diameter. In order to extend the usable

Figure 1: Experimental set up at the EDR beam line at BESSY II

spectral range (5 <15 keV) we installed a flight tube between sample and energy-dispersive detector. Unfortunately the experiment was partially successful up to now, only, caused by technical reasons. The first time-correlation we observed was the evolution of a radiation damage at the polymer induced by the intense synchrotron radiation. Furthermore we suffered from some instabilities in speckle positions caused by thermal drift of the sample table, which causes the whole reflected beam to move. The first problem will be solved by installation of a fast shutter. It will reduce the radiation dose at the sample, also. Additionally we plan to optimize the conditions of green laser irradiation to the sample in order to tune the dynamics to the time window accessible by the experiment. The second has to be solved by extracting the thermal drift by means of numerical analysis of data. Nevertheless the first experiment has shown speckle modifications under influence of the external light. Figure 2 shows the spectra Figure 2: The spectra changes with time: The black line is the first measurement, the red one the last. The change in the spectra is clearly seen, especially at 9keV and 12keV.

368

taken at an incidence angle of 0.2 deg. The intensity strongly varies as function of energy, indicating the existence of speckles. The various speckle changes as function of time which is the proof for the appearance of a dynamical process induced by the external light. This process must be quantified in future experiments. In case of success white beam correlation spectroscopy could provides the advantage of stable illumination conditions during the whole experiment and the potential for detecting the q-dependence of the speckle dynamics from one and the same experiment. Reference: [1] T. Panzner, W. Leitenberger, J. Grenzer, U. Pietsch, J. Phys. D Appl. Phys.2003, 36, A93 [2] Dirk O. Riese, Willem L. Vos, Gerard H. Wegdam, Frank J. Poelwijk, Phys. Rev. E 2000, 61, p.1676-1680 [3] O. Henneberg, et al., Z. Kristallogr. 2004, 219, 218-223

369

Study of segregation process in Ni94Pd6 foil using XPS G. A. Dosovitskiy,1 L. I. Burova,1 M. Fonin,2 Yu. S. Dedkov,3 U. Rüdiger,2 A. R. Kaul4 1 Department of Material Sciences, Lomonosov Moscow State University, 119899 Moscow, Russia 2 Fachbereich Physik, Universität Konstanz, 78457 Konstanz, Germany 3 Institut für Festkörperphysik, Technische Universität Dresden, 01062 Dresden, Germany 4 Chemical Department, Lomonosov Moscow State University, 119899, Moscow, Russia The discovery of high temperature superconductivity (HTSC) in 1986 [1] have atttracted sufficient interest to materials possessing HTSC properties, among them YBa2Cu3O7δ. A number of technological applications, such as power cables, motors, transformers, fault current limiters [2] require long current-carrying wires with superconducting properties. Several approaches to produce long superconducting cables[3], and the second generation HTSC cables (coated conductors) [3,4] are of the highest interest at the present moment. A promising way to obtain coated conductors is so-called rolling assisted biaxially textured substrates (RABiTS) technology [5], which is based on the deposition of multilayer structures on biaxially textured tapes of Ni-alloys as substrates. The main point of this concept is translating the texture of a metal tape to a buffer layer and then to an HTSC layer. The Ni-alloy tape should be highly textured with sharp (100) cubic texture in oreder to ensure good quality of HTSC layer. RABiTS technology is very sensitive to the surface quality of the Ni-tape. For example Ni oxidation can significantly influece the texture of the buffer layer. Noble metal, such as Pd, are used as dopants in order to increase the tape stability against oxidation. In the catalytic chemistry Ni-Pd system is widely used. Recently structure and chemical composition of Ni92Pd8 single crystal surface was investigated [6]. In this study intensities of Ni 2p an Pd 3d peaks were measured by x-ray photoelectron spectroscopy (XPS) as a function of emission angle. These measurements have shown that Pd segregates on the single crystal surface. Pd concentrations of about 28 at.% and 38% were observed on the (111) and the (110) surface, respectively. According to the LEIS data Pd concentration in the surface layerwas even higher (up to 81 at.%). High Pd concentration in the surface layer makes the crystal surface more stable against oxidation in comparisson with the pure Ni surface. Until now Pd segregation process have been investigated on NiPd single crystals, so it is still unknown how the additional factors such as polycrystallinity and high degree of mechanical deformation may influence the segregation process. The main goal of the present study was to investigate segregation process in Ni-Pd textured polycrystalline samples. XPS experiments were carried out at room temperature (RT) at the RGBL-PGM beamline at the BESSY II storage ring. The RGBL-PGM dipole beamline provides a tunable source of photons over a 20-1500 eV energy range with a linear polarization of the light. The UHV system located at the Russian-German Laboratory (base pressure of 1×10-10 mbar) was equipped with a 127° CLAM4 analyzer. The total energy resolution in the XPS measurements was set to 100 meV. The position of the Fermi energy was determined form the valence-band spectrum of a polycrystalline Au foil in the electrical contact with the sample. Ni-Pd samples were prepared by cold rolling of rods of Ni94Pd6 alloy (60 cycles, ~99% deformation) followed by texturing annealing. The sample Ni94Pd6 was annealed in Ar/H2 atmosphere at 1100oC for about 1 hour (texturing annealing) and subsequentally at 600 oC for 6 hours (texturing annealing).

370

a

b

c

d

Fig. 1. Photoemission spectra taken on the Ni94Pd6 sample at different photoemission angles: (a) Ni 2p peak,( b) Pd 3d peak, and as a function of sputtering: ( c) Ni 2p peak, (d) Pd 3d peak.

Fig. 2. Pd concentration depth profiles obtained by photoemission intensity analysis of the

Ni 2p and Pd 3d peaks: as a function of the photoemission angle (left-hand panel) and as a function of the sputtering time (right-hand panel). After the introduction into the UHV S, C and O ipurity peaks were found on the sample surface. O and C are common surface impurities, and S is an impurity, which is usually present in Ni. After the Ar+ sputtering the intensities of impurity peaks decreased significantly. Ni 2p and Pd 3d peak intensities depending on electron emission angles were measured to study the concentration in-depth profile [Fig. 1 (a) and (b)]. The series of spectra after differnet Ar+ sputtering times were taken for the same purpose. Pd concentration was calculated from Ni 2p3/2 and Pd 3d5/2 peaks intensities[Fig. 1 (c) and (d)]. Fig. 2 shows Pd concentration depth profiles obtained by photoemission intensity analysis of the Ni 2p and Pd 3d peaks: as a function of the photoemission angle (left-hand panel) and as a function of the sputtering time (right-hand panel). It could be seen from the concentration depth profiles, that Pd segregates on the Ni surface. Inelastic mean free path calculated from the Tanuma, Powell, and Penn formula in [7] for electrons with energies 140 eV (Ni2p electrons) and 660 eV (Pd3d electrons) are 4,58 Å and 11,35 Å respectively. It means, that information, obtained from angle dependences,

371

refers to 2 atomic layers. It could be supposed, that the data from fig. 2 (a) approves the model of oscillatory distribution of Pd on the surface of Ni-Pd alloy [8]. 1. J. G. Bednorz and K. A. Muller, Physica B 64, 189 (1986). 2. M. Chen, L. Donzel, M. Lakner, W. Paul, J. Eur. Ceram. Soc. 24, 1815 (2004). 3. L. J. Masur et al., Physica C 392–396, 989 (2003). 4. T. Watanabe et al., IEEE Transactions on Appl. Superconductivity 13, 2445 (2003). 5. D. K. Finnemore et al., Physica C 320, 1 (1999). 6. A. C. Michel et al., Surf. Sci. 416, 288 (1998). 7. S. Tanuma, C. J. Powell, D. R. Penn, Surf. Interf. Anal. 21, 165 (1993). 8. G. N. Derry, C. B. McVey, P. J. Rous, Surf. Sci. 326, 59 (1995).

372

Computed tomography experiments at BAMline
G. Weidemann, J. Goebbels, H. Riesemeier, Th. Wolk, M. Bartscher*, U. Hilpert* Bundesanstalt für Materialforschung und -prüfung (BAM) Unter den Eichen 87, 12205 Berlin, Germany * Physikalisch-Technische Bundesanstalt (PTB) Bundesallee 100, 38116 Braunschweig, Germany Photo resist negatives of micro gears have been studied to evaluate the geometrical shape. The measurements were performed with 15 keV X-ray energy and voxel sizes of (3.6 µm)3 and (1.5 µm)3. From the resulting 3D voxel data set a surface mesh was extracted and stored in the stereo-lithography file format (‘stl’ – format), which is commonly used for surface representations. Fig. 1 shows a cross section (3.6 µm voxel size) from the 3D image data set.

Fig. 1: Cross section (left) of a micro component, outer diameter about 6.5 mm. Conversion of CT data to stl-data format and segmentation of the four cylinders (right). From the surface mesh data set the four cylinder geometries are segmented. To study the deviation of geometry the cylinder geometries are compared each with each other. As an example fig. 2 shows the deviation of the geometry of cylinder 1 from cylinder 2. The deviations are colour coded. The colour scale extends from –10 µm (dark blue) to +10 µm (dark red). Smaller details of the geometry were studied with improved spatial resolution (voxel size (1.5 µm)3).

373

Fig. 2: Evaluation of cylinder geometry; comparison of cylinder 1 with cylinder 2.

Fig. 3: Iso-surface of a detail with improved spatial resolution

The spatial resolution of the CT set-up was further improved using new scintillators together with other lens optics in cooperation with HMI. Now a voxel size of (0.6 µm)3 is available. As an example fig. 4 shows a spore studied with 9 keV in cooperation with TU Leipzig.

Fig. 4: Two cross sections and an iso-surface visualisation of a spore.

374

Inorganic contact formation on oriented organic films S. Berkebile, J. Ivanco, G. Koller, M.G. Ramsey Institut für Physik, Karl-Franzens-Universität Graz, A-8010 Graz, Austria HPRI-CT-1999-00028 Oligomers such as sexiphenyl (6P) and sexithiophene (6T) are attracting renewed interest not only as models for their related polymers, but also in their own right as active materials in organic devices such as FETs, LEDs and solar cells. The bonding interaction at the organicinorganic interface has been shown to be important to the electronic properties via the band alignment [1], and there is much interest regarding molecular geometry in this interfacial region as it is crucial to many aspects of organic devices, from charge injection and transport to thin film growth. The corresponding interface of inorganic on organic, the formation of a top contact, however, remains less explored. Given that the surface free energies of organics are between one to two orders of magnitude lower than for metals it should be all but impossible to form metallic wetting layers on organic films. In earlier work we have shown that Al (a common contact in organic devices) does not wet vertically oriented sexiphenyl. However, when evaporated in oxygen environment (10-7 mbar) ultra-thin, conducting wetting layers could be produced [2]. Our goal was to explore the formation of contact layers on organics in a controlled manner for a number of different materials on sexithiophene and sexiphenyl films with known molecular orientations.[3][4] Here we report an investigation into the formation of ultra-thin metallic films of Ti on pristine 6P(001) and Al on 6P(001) using a layer of LiF as a surfactant. Ti is increasingly used in the semiconductor industry as a contact material; and layers of LiF often proceed the metal evaporation to improve device characteristics, yet it is unclear as to what is occuring at this interface. The 6P(001) orientation is such that the molecular axes are tilted 17° w.r.t. the surface normal of the underlying substrate [3] and organic film thicknesses were chosen such that the underlying substrate is spectroscopically covered (i.e. a closed film was produced). Molecular orientation was observed with NEXAFS prior to and after metal evaporation. The surfaces were prepared in-situ under UHV and investigated in the MUSTANG end station attached to the Russian-German beamline. 6P was evaporated using a home-made triple source Knudsen cell type evaporator. Ti, Al and LiF were evaporated using an Oxford e-beam evaporator. SXPS, UPS and CK-edge Auger yield NEXAFS spectra were obtained using a SPECS Phoibos 150 electron energy analyzer with an overall resolution of <0.1eV. The incremental deposition of Ti on a 6P(001) substrate was monitored by SXPS and NEXAFS. Figure 1 displays the development of the C1s and Ti2p core levels as a function of

375

the Ti coverage. The C1s intensity decreases rapidly with increasing Ti coverage. At an initial coverage of 4 Å, it has about half the intensity with a tail at lower BE, indicative of a chemical reaction with the Ti. On increasing coverage, the quanitity of reacted components increases to a coverage of 16 Å
Fig 1: The development of the C1s and Ti2p core levels as a function of Ti coverage on 6P(001) as observed by SXPS.

of Ti and becomes increasingly one species at 282 eV, suggestive of the formation of

TiC. The Ti2p intensity increases and shifts to lower BE with coverage implying a transition from TiC to metallic Ti. At a coverage of 16 Å, the broad Ti2p peak suggests both TiC and metallic Ti are present. This transition to metallic Ti after 32 Å of Ti is also seen in the decrease in intensity of the carbidic peak of the C1s spectrum (282 eV), from which we infer that the TiC remains buried at the interface and is covered by a metallic contact layer, also indicated by the sharp Ti2p peak at the metallic BE. Dramatic changes in the NEXAFS spectra, particularly in the π* resonances, also indicate a chemical reaction with the 6P film. The deposition of 2.5 Å of LiF and the subsequent incremental deposition of Al on a 6P(001) substrate was also monitored by SXPS and NEXAFS. In Figure 2, the development of the C1s, Al2p, Li1s and F2s core levels are shown as a function of LiF and Al deposition. After 2.5 Å of LiF is deposited on a 6P film, the C1s peak shifts significantly by 1.1 eV to higher BE, suggesting the Li is diffusing into the film and doping the 6P at the interface. The shoulder remaining at the pristine 6P C1s position hints at deeper layers of undoped 6P. A doping effect is further supported by the appearance of states in the bandgap of the 6P just below the Fermi level (not shown). In addition to this electronic effect, the NEXAFS also suggests a disturbance in molecular orientation which becomes more marked after Al deposition. As Al is deposited on top of the LiF/6P(001), the development of a metallic Al2p peak is steady and gradual. At a small coverage of 3 Å Al, the Al2p peak is at a higher BE than metallic Al and the spin-orbit splitting is unresolved, suggestive of charging in an electrically discontinuos film. As the Al coverage is further increased, the Al2p peak gradually shifts to the metallic position and the spin-orbit splitting becomes more apparent until, at a coverage of 38 Å Al, it is well resolved, indicating that a continuous conductive

376

metallic layer has been formed. This is supported decrease by in a the concurrent

intensity of the C1s, Li1s and F2s core levels. We note that a continuous metallic layer could not be achieved if Al was directly 6P(001).[2] Here our studies have shown that LiF, which is often used in device production to improve contacts, acts in two ways: the lithium dopes the 6P causing a large band offset favourable for e-injection and the layer of LiF allows the Al evaporated on this surface to form a continuous layer. Also, Ti evaporated onto
Fig 2: The development of the C1s, Al2p, Li1s and F2s core levels as a function of LiF and subsequent Al coverage on 6P(001) as observed by SXPS.

evaporated

on

6P(001) forms a TiC capping layer on which a continuous metallic Ti film grows.

Acknowledgments: This work has been supported by the Austrian Science Foundation (FWF) and the EU via HPRI-CT-1999-00028. The assistance of Mike Sperling with the MUSTANG chamber and the beamline staff at the RG-BL is gratefully acknowledged. References: 1) G. Koller, R.I.R. Blyth, A. Sadar, F.P. Netzer, M.G. Ramsey, Appl. Phys. Lett. 76 (2000) 927. 2) J. Ivanco, B. Winter, F.P. Netzer, L. Gregoratti, M. Kiskinova, M.G. Ramsey, Appl. Phys. Lett. 85 (2004) 585. 3) G. Koller, S. Berkebile, G. Tzvetkov, M.G. Ramsey, Bessy Report 2003; G. Koller, S. Berkebile, F. Pfuner, M.G. Ramsey, Bessy Report 2004. 4) B. Winter, S. Berkebile, J. Ivanco, G. Koller, F.P. Netzer, M.G. Ramsey, Appl. Phys. Lett. Submitted 2006.

377

Exotic growth mode in molecular-beam epitaxy of GaMnSb on GaSb
Wolfgang Braun , Bernd Jenichen, Achim Trampert, Dillip Kumar Satapathy, Klaus H. Ploog Paul-Drude-Institut f¨ r Festk¨ rperelektronik, Hausvogteiplatz 5-7, D-10117 Berlin, Germany u o email: braun@pdi-berlin.de, phone: +49-(0)30-20377-366, fax: +49-(0)30-20377-366.

To make use of the spin in addition to the charge in future spintronic devices, the preparation of a defined spin state of the charge carriers is required. One option to achieve this is to inject spin-polarized carriers into a semiconductor, an operation that presumably requires the fabrication of a highly perfect interface between a ferromagnetic injector and a semiconductor. MnSb offers the advantage of a much higher Curie temperature (851 K) compared to MnAs (313 K). Just as in the MnAs case, the use of a joint element between film and substrate material offers a convenient way to fabricate heterostructures by MBE. We have therefore studied the deposition of MnSb on GaSb to investigate the interface configuration of this heterostructure. The results are unexpected and fascinatingly different from the MnAs/GaAs system. MnSb was deposited on a high-quality GaSb(001) buffer layer by opening the Mn shutter under constant Sb flux. The sample temperature was initially 330 ◦ C during MnSb nucleation, and then was increased during the growth to 470 ◦ C. The surface morphology of the resulting structure measured by AFM is shown in Fig. 1. Instead of a continuous film, we observe rectangular patches with strongly rounded corners. These patches usually have a flat upper surface which is either coplanar with the substrate surface or slightly inclined. Some crystals tower above the surface in diferent orientations, exposing hexagonal shapes. The origin of this peculiar morphology can be clarified by cross-sectional transmission electron microscopy. Images taken from the same sample are shown in Fig. 2. The image reveals homogeneous single crystals of sub-micrometer

Figure 1: Atomic force microscopy scan revealing the surface morphology of the GaMnSb microcrystals grown into GaSb. Note that the z scale for the 3D rendering is unity, therefore giving an accurate representation of the angles.

378

size embedded in the surface of the GaSb substrate matrix. Atomic resolution images like the one shown in the top of Fig. 2 allow the analysis of the lattice constants parallel and perpendicular to the surface of both the microcrystals and the substrate matrix. Within the accuracy of such a graphical evaluation, the lattice match perpendicular to the surface is perfect. Laterally, the microcrystal lattice constant perpendicular to the viewing direction is around 5 % smaller than the one of the substrate. The shape of the buried section is close to hemispherical to minimize the interfacial energy. On the other hand, facets defined by the GaSb (111) planes are also discernible. The difference between the 35.26 ◦ the (111) planes make with the surface normal and the 30 ◦ of a possible hexagonal microcrystal structure may explain the tilt angles of ± 5 ◦ of some grains as well as the rare atypical orientations in which the a or c axis of the microcrystal is parallel to the GaSb (111) planes. The TEM shows remarkably little strain, within the area sampled by TEM both the microcrystals and the GaSb matrix are free of dislocations. In situ synchrotron x-ray diffraction performed at the PHARAO beamline during the growth reveals the formation process of the phase-separated embedded microcrystals. Figure 3 shows two ω − 2θ in-plane scans around ¯ the GaSb 2 2 peak in reciprocal space along the ¯ [1 1] direction of the GaSb substrate. During the nucleation at low growth temperature, the deposited material develops an epitaxial relationship with one of the broad film peaks closely ¯ matching the narrow GaSb 2 2 peak. Additional reflections are present. Upon continuation of the growth at elevated temperatures, the migration process of the deposited material into the substrate matrix becomes evident by the fusion of ¯ the substrate and film peaks at 2 2. The growth at this stage can be performed arbitrarily close to thermodynamic equilibrium, as we have observed the reduction of the deposited material peak when further increasing the substrate temperature during deposition. The dominant peak at 152.3 ◦ (a distance 3.3 % smaller than 2 2 in real space agreeing well with the TEM analysis) does not match a peak

Figure 2: Atomic resolution (top) and overview (bottom) TEM micrographs of the epitaxial microcrystals forming during Mn and Sb codeposition. The microcrystals are homogeneous, single crystal and lattice-matched to GaSb.

379

1e+06

nucleation hi-temp growth

GaSb 2 2 0

1e+06

[2 2] [2 2]

Intensity

100

Intensity 148

10000

10000

100

1 146 150 152 Angle [omega] 154

1 162 164 166 168 Angle [omega] 170 172

Figure 3: In-situ x-ray diffraction scans along the surface containing reflections from the substrate and the microcrystals. During the nucleation (black curve), an epitaxially oriented structure with additional peaks due to other orientations forms. During growth at elevated temperatures, the substrate and layer peaks merge and other orientations vanish (blue curve). of MnSb. This suggests that the deposited Mn forms a Ga-Mn-Sb alloy of a composition that minimizes the strain between microcrystals and substrate. A comparison of ω − 2θ scans along the two orthogonal <2 2> in-plane directions is shown in Fig. 4. It reveals clear differences that cannot be induced by the fourfold symmetry of the substrate matrix. This means that the orientation of the final microcrystals is determined by the initial nucleation on the surface reconstruction that has a twofold symmetry. If the microcrystal orientation were determined by the bulk properties only, no such preference would be expected. Measurements of the magnetization vs. temperature show strong ferromagnetism of the system even well above room temperature. Between 20 K and 370 K (the measurement range of our magnetometer), the magnetization decreases by a mere 20 %. Both the microcrystals and the surrounding matrix are highly perfect crystals. To form an alloy, atoms therefore have to migrate into the bulk, unlike the typical MBE growth mode where only surface atoms are considered mobile enough to contribute to the growth. It seems pos-

Figure 4: In-plane anisotropy of the microcrystal structure. The different peaks present in the two orthogonal directions indicate a preferential alignment along one of the azimuths. sible that the microcrystals form by migration of Mn along the interface with the GaSb instead of by bulk diffusion through the volume of the microcrystals or the host lattice. Obviously, such a migration could take place along the more stable GaSb(111) planes. The exotic growth mode observed here has intriguing and far-reaching implications. Both bulk migration and the growth of highly perfect epitaxial crystals into the substrate are concepts that are not common in MBE. A system that selfadjusts its lattice constant to minimize defects at the interface is ideal for the formation of heterostructure devices. Since one of the materials involved is ferromagnetic and the other is a high mobility semiconductor, the growth mode presented here has a high potential for spintronic device technology. Using the mechnism demonstrated here, it may even be possible to form very high quality epitaxial contacts by mere alloying of prepatterned Mn into GaSb under an Sb flux following a photolithographic process.

380

Analysis of binding media in cross sections using synchrotron infrared radiation
Oliver Hahn1, Jens Bartoll2, Ulrich Schade3
Bundesanstalt für Materialforschung und –prüfung (BAM),12200 Berlin Stiftung Preußische Schlösser und Gärten Berlin-Brandenburg (SPSG), 14414 Potsdam 3 BESSY GmbH, Albert-Einstein-Straße 15, 12489 Berlin
1

2

The present study is the continuation of a project dealing with the analysis of binding media in cross sections. Cross sections are small samples extracted from art objects embedded in a matrix of resin. The characterisation of different organic compounds in different layers of the cross sections is of special interest in conservation science and art history. Usually binding media are identified by infrared spectroscopy in transition mode. For this purpose it is necessary to separate each layer from the other mechanically which is very difficult or often impossible because of the smallness and complexity of the samples. Usually only microscopic methods are applied to study the inorganic pigment grains within the cross section. Until now there are no satisfying methods for the analysis of organic compounds in cross sections. The method described here presents a direct investigation without any mechanical separation in high local resolution by using infrared spectroscopy in reflection. The measurements were performed at the synchrotron infrared beamline IRIS. A FTIR spectrometer (Bruker 66/v) and an IR microscope (Thermo Nicolet Continuum and Nexus) were used. Selected samples of typical binding media such as dry oil, resins, and glues were measured in order to collect reference data. In addition, the half of each sample was sputtered with gold. The golden surface was measured for background correction. The results indicate that it is possible to distinguish different classes of binding media using this method. Two examples are shown in Figures 1a/b and 2a/b. Spectra of gelatine and shellac in original furniture coatings could be detected. They fit well to the spectra of the reference samples. However, the reliable identification of oil coatings causes difficulties (Fig. 3). The specific surface properties of binding media containing oil might influence the process of reflection. All in all these results indicate that this technique provides a suitable method for organic analyses with high lateral resolution in the field of archaeometry. Acknowledgements The authors thank Alexander Firsov for his versatile support.
References S. Wülfert, Der Blick ins Bild, Lichtmikroskopische Methoden zur Untersuchung von Bildaufbau, Fasern und Pigmenten, Bücherei des Restaurators Bd. 4, U. Schießl (Hrsg.) Ravensburger Buchverlag 1999. M.R. Derrick, D. Stulik und J.M. Landry, Infrared Spectroscopy in Conservation Science, The Getty Conservation Institute Los Angeles (ed.) 1999. J. Bartoll, O. Hahn, U. Schade: Analysis of binding media in cross sections using synchrotron infrared radiation, in preparation.

381

Figures

sample gelatine (reference)

7,5

Reflection / a.u.

5,0

2,5

3500

3000

2500

2000
-1

1500

1000

Wave number, ν/cm

Fig. 1a/b: Cross section / Identification of gelatine in a furniture coating (spot size: 20 µm x 20 µm)

40
shellac sample 1 sample 2

Reflection a.u.

30

20

10 3500 3000 2500 2000
-1

1500

1000

Wave number; ν/cm

Fig. 2a/b: Cross section / Identification of shellac in furniture coatings (spot size: 20 µm x 20 µm)

sample oil (reference)

6 Reflection / a.u.

4

2

3500

3000

2500

2000
-1

1500

1000

Wave numbers ν/cm

Fig. 3: FTIR-spectra of oil in comparison with an unknown sample (containing probably oil)

382

Magnetic linear dichroism in the reflectivity of iron and cobalt in the vicinity of the 2p edges
A. Kleibert, P.M. Oppeneer* and J. Bansmann** Institut fur Physik, Universit¨t Rostock, D-18051 Rostock a ¨ *Department of Physics, Uppsala University, Box 530, S-75121 Uppsala, Sweden **Abteilung Oberfl¨chenchemie und Katalyse, Universit¨t Ulm, D-89069 a a Ulm Supported by BMBF 05KS4HRA2/4. During the last decade magnetic soft x-ray spectroscopy techniques have experienced a rapid development. Among the various magneto-optical effects – occurring in absorption, transmission as well as reflection of light by a magnetic sample – particularly the x-ray magnetic circular dichroism (XMCD) in absorption received considerable attention. This is mainly due to the so-called ‘sum rules’ relating the magnetization dependent absorption of circularly polarized radiation to detailed information on electronic and magnetic ground state properties [1, 2]. In combination with these sum rules, the XMCD allows to study both the spin and the orbital magnetic moments of a material separately in an elementspecific manner. Since the magneto-crystalline anisotropy energy (MAE) is connected to the anisotropy of the spin-orbit interaction, XMCD has been used as a valuable tool to investigate the microscopic origin of this fundamental magnetic quantity [3]. Recently, a new ‘sum rule’ was published connecting the x-ray magnetic linear dichroism (XMLD) directly to the anisotropy of the spin-orbit interaction, thus offering a new opportunity to study the magneto-crystalline anisotropy [4]. Moreover, in contrast to magnetic circular dichroism, the XMLD effect is even in the magnetization and therefore well suited for the study of antiferromagnetic materials. Since the magneto-optical properties of a magnetic sample (in transmission, absorption, and reflection) are determined by one and the same dielectric tensor, the information that can be obtained from XMCD/XMLD absorption experiments, is also accessible by means of corresponding reflectivity based techniques. Reflectometry-based experiments benefit especially from the large probing depth being solely limited by the penetration depth of the radiation. They are therefore well suited to study capped samples or buried layers in multilayers, respectively. Moreover, measurements of the reflectivity are not disturbed by large external magnetic fields [5, 6]. Recently, we demonstrated the possibility to use the transverse magneto-optical Kerr-effect in the vicinity of the 2p core levels of ultrathin Co films as an alternative to XMCD photoabsorption experiments [7, 8]. In this contribution we present recently obtained experimental results concerning the XMLD and its manifestation in corresponding reflectivity measurements. The XMLD effect usually occurs in absorption experiments where the polarization plane is aligned first parallel and then perpendicularly to the magnetization of a sample. In the case of 3d transition metals the measurable XMLD effect is quite weak (in the order of 3%), thus making high demands on experimental investigations when quantitative analysis is desired [9].

383

The experiments have been carried out at the PM3 dipole beamline with the BESSY polarimeter chamber [10]. Sputtered iron and cobalt films with a thickness of about 200 ˚ A ˚ Absorphave been investigated, both samples were capped with a chromium layer of ∼30 A. tion and reflectivity spectra have been recorded simultaneously by means of total electron yield detection and a photodiode, respectively. An external magnetic field could be applied parallel as well as perpendicularly to the plane of incidence. In the upper panel of figure 1 reflectivity spectra of s-polarized light for both orientations of the magnetization are shown as red (transverse magnetization) and black dashed (longitudinal) lines. The complex shape of the reflectivity spectra is generally determined by interference effects. Note that reversing the magnetization in the transverse or longitudinal direction will not result in any intensity change in this geometry. However, rotating the magnetization from longitudinal to transverse orientation obviously gives rise to a remarkable intensity change, especially in the vicinity of the Fe 2p3/2 edge (cf. inset in the upper panel). The corresponding asymmetry given by the blue line in the lower panel of figure 1 shows values of up to 40%, and thus is an order of magnitude larger than the XMLD in absorption. Moreover, the observed intensity change depends strongly on the angle of incidence as well as on the polarisation (not shown here). Similar effects have been observed when investigating the cobalt films. The data therefore underline the promising potential of magneto-optical reflectometry when compared to the weak effects in absorption experiments. Moreover, they will probe recent predictions on the shape of XMLD spectra [11]. A detailed, quantitative analysis of the data and an explanation of the magnitude of the observed effects will follow in the near future. References: [1] P. Carra et al.