Surface Science 414 (1998) 209­220 Growth modes of vanadium and iron on V(110) single crystals T. Nawrath *, H. Fritzsche, H. Maletta Hahn-Meitner-Institut Berlin, Glienicker Strasse 100, D-14109 Berlin, Germany Received 2 March 1998; accepted for publication 10 June 1998 Abstract In this paper we present investigations on the growth of the bcc structured metals vanadium and iron on V(110) single crystals in the thickness range 0­20 A . For the analysis we used low energy electron diffraction (LEED) and Auger electron spectroscopy (AES). The growth was performed by molecular beam epitaxy (MBE) at 320 K and 570 K for vanadium and at 320 K and 470 K for iron. For both materials we observe a strong change in the growth modes from 320 K to 470 K and 570 K, respectively. For vanadium we observe well-ordered surfaces at T=320 K for tV>10 A with a different island size of 89 A and 50 A in the [001] and [11:0] direction, respectively. For T=570 K a change of the growth mode is observed, represented by a quasi-periodic sequence of up and down staircases in the [11:0] direction with inclinations of 50° with respect to the film plane and the ridges orientated along the [001] direction. For iron we find a quasi-Frank­van der Merwe growth at 320 K with an anisotropic island size for the [001] and [11:0] direction, with larger values by a factor of 1.5­2.0 in the [001] direction. The island size is smaller than that for V on V(110) in the whole thickness range and we observe a minimum of island size at tFe=4 A . At T=470 K, the growth also changes to a faceted growth mode, with the facets in the same orientation as for vanadium, but with an inclination of ±40° with respect to the film plane. © 1998 Elsevier Science B.V. All rights reserved. Keywords: Auger electron spectroscopy; Epitaxy; Iron; Low energy electron diffraction (LEED); Magnetic films; Single crystal surfaces; Surface structure; Vanadium 1. Introduction preceding the study of the thin film magnetism itself. This includes the growth of the magnetic It is well known that the surface and interfaces film, as well as its non-magnetic neighbour. of magnetic thin films play a fundamental role in In this article we want to focus on the growth explaining the observed behaviour of properties of iron and vanadium on V(110) single crystals. like magnetization, magnetic anisotropy or cou- Concerning the growth of thin films, these systems pling between two magnetic films through a non- are not very well studied, although it might be of magnetic spacer layer. Furthermore, both experi- interest in comparison to the very well known mental [1­5] and theoretical [6­8] contributions growth of Fe on W(110) [9­11] as both vanadium emphasize the importance of the surface (interface) and tungsten have a larger lattice constant than topology to these properties. According to this, iron (Fe: 2.87 A ; V: 3.02 A ; W: 3.16 A ) with the growth studies seem to be a compulsory topic same crystalline structure (bcc) for all metals. In contrast to vanadium, W has a much higher free * Corresponding author. Fax: +49 30 8062 2523; surface enthalpy than Fe (Fe: 2.939 J/m2; V: e-mail: nawrath@hmi.de 2.876 J/m2; W: 3.468 J/m2 [12]). According to the 0039-6028/98/$ ­ see front matter © 1998 Elsevier Science B.V. All rights reserved. PII: S0039-6028 ( 98 ) 00513-5 210 T. Nawrath et al. / Surface Science 414 (1998) 209­220 arguments of Bauer and van der Merwe [13,14], surface, the crystal must be annealed at a temper- the interplay between the lattice constant and the ature of 1200 K for 20 min. However, this annea- free surface enthalpy gives rise to the growth mode ling process leads again to a surface contamination. of the particular system. For W/Fe the larger free Therefore, the sputter annealing cycle must be surface enthalpy of W can explain the pseudomor- repeated for approximately 120 h to get a smooth phic growth of Fe within the first two monolayers and clean V(110) surface. Similar cleaning pro- [9,10]. As the free surface enthalpy of vanadium cedures for vanadium are also described in Refs. is slightly smaller than that of Fe, it is probable [26­28]. After this initial cleaning procedure of that the growth differs from the W/Fe system. A the surface, new experiments could be performed closer comparison between these two systems could after an entire cleaning time of 12 h. give some further hints on the importance of lattice All measurements were performed under UHV mismatch to the growth modes for metallic conditions with a base pressure less than systems. 10-10 mbar and a pressure during evaporation less Concerning its magnetic behaviour, the than 2×10-10 mbar. The vanadium and iron films interlayer coupling of Fe in Fe/V multilayers [15­ were prepared by an electron beam evaporator at 17] became an important subject of comparison a rate of 0.07 A /s. All films as well as the annealed to the very prominent and strongly RKKY coupled crystal were checked for surface contaminations Fe/Cr multilayers [18­20]. with AES. The detection limit of this method is Further on, the predicted and partly measured 2% of a monolayer for Auger transitions that have antiferromagnetic polarization of vanadium in the no overlap with transitions of the film or the vicinity of the Fe(110) interface [21,22] (for the substrate. For oxygen the main transition ( KLL) (100) surface see also Refs. [23,24]) is of interest at 514 eV is relatively close to that of the V-LMM for basic research programs. Here, vanadium is of transition at 509 eV. However, as the transition particular interest as it is an ideal substrate for the probability of the V-LMM peak is small, oxygen newly developed field of in situ magnetization contaminations can be detected at a modulation measurements of ultrathin films with polarized amplitude of the Auger analyser of 2 eV due to an neutrons (in situ PNR) [25]. energy shift and an increase of the V-LMM peak. The sensitivity for oxygen contaminations is about 5%. For all experiments presented in this article, 2. Experimental the surface contaminations were beyond the detec- tion limit of the Auger spectroscope. All samples were evaporated on a vanadium The thickness of the films was monitored by a (110) single crystal, which had dimensions of quartz balance which had been calibrated by meas- 30×12 mm2 and was orientated with an accuracy urements and simulations of the X-ray reflectivity of 0.25° with respect to the surface normal. It (low angle X-ray diffraction) of thicker films (e.g. could be heated by a graphite layer embedded in V(110)/10 A V/6 A Fe/300 A Cr). From these sim- boron nitride, which was positioned behind the ulations, which treat the interface reflectivity as single crystal, up to a temperature of 1400 K. The known from optics, described by the Fresnel sample temperature was measured with a reflectivity that is multiplied by a Debye­Waller chromel/alumel thermocouple, which was in direct factor for the interfacial roughness [29­31], the contact with the sample holder. mean square surface roughness of the interfaces To clean the V single crystal from its impurities, can be deduced, too. From this a typical value for mainly sulphur, carbon and oxygen, repeated the roughness of the V/Fe and Fe/Cr interfaces cycles of sputtering with a 400 eV argon beam and was 3 A . annealing at 1200 K were performed. To clean the To perform our measurements in dependence on surface, 2 min sputtering at a beam current of the thickness, the films were evaporated as wedges 10 mA was sufficient, as could be verified by AES. on the single crystal. The data presented later in In order to recrystallize and smooth the crystal Figs. 3 and 6­8 were obtained from wedges with T. Nawrath et al. / Surface Science 414 (1998) 209­220 211 an inclination of 0.6 A /mm, whereas the data of the nth layer is completely filled, the Auger inten- Figs. 9 and 10 were acquired from wedges with an sity at a non-integer layer thickness r, with inclination of 1 A /mm. n10 A a ments as AES is not suitable for homoepitaxial saturation value for the island size of 89±8 A in growth. the [001] direction and 50±6 A in the [11:0] direc- Fig. 1a shows the LEED pattern of the pure tion is observed. vanadium crystal at an energy of 70 eV. One can see the {10} spots, as indicated in the picture. The orientation of the substrate is indicated by two 4.2. Growth of vanadium at 570 K arrows for the [11:0] and [001] direction. The two profiles of the spots along these directions are also When increasing the evaporation temperature shown. It can be seen that DK is approximately to 570 K, the shape of the LEED patterns changes the same in both crystalline directions. Fig. 1a can entirely. This is shown in Fig. 4, where the LEED also be seen as proof of the absence of oxygen on pictures of a 50 A thick V film on V(110) are the surface, as this results in a (6×2) superstruc- shown with electron energies of 56, 68, and 76 eV. ture [28]. At an energy of 68 eV the {10} peaks are also Fig. 4. LEED patterns of a 50 A vanadium grown on V(110) at electron energies E=56, 68, and 76 eV. The evaporation temperature was 570 K. E=68 eV corresponds to the in-phase condition of the {10} spots. T. Nawrath et al. / Surface Science 414 (1998) 209­220 215 Fig. 6. The Fe and V Auger intensities of Fe prepared on V(110)/10 A V at 320 K versus the iron thickness tFe. The lines Fig. 5. Position of the {10} spot satellites in reciprocal space of are fits as mentioned in the text. 50 A V on V(110) at 570 K. The angle a=50±2° corresponds to the inclination of the facets with respect to the film plane. In Fig. 6 the Auger intensities of the Fe-L3M45M45 (703 eV) edge and V-L3M23M45 elliptically shaped with the longer side in the (473 eV ) edge are presented. The Fe intensity IFe, [11:0] direction. However, if the electron energy is represented by up triangles, shows an increase due increased up to 76 eV (Fig. 4c), one can see an to the increasing Fe thickness tFe, whereas the V energy dependent splitting of the peaks into two intensity IV (down triangles) decreases for the same satellites. This behaviour is not observed for the reason. Both intensities are normalized to the V films grown at 320 K. The observation of such intensity IV0 of the pure vanadium surface. One a pattern as a function of energy can be explained can see that the saturation value for the Fe peak by a faceted surface, as discussed in Section 2. The intensity is smaller than that of the V peak of the up and down staircases are orientated along the uncovered surface (47% in the fit). This is due to [11:0] direction and the ridges along the [001] the smaller transition probability for the pure iron direction. The height of the facets Hd peak compared to that of vanadium (I z can be Fe/IV=0.52 estimated due to the disappearance of the central [37]). Also shown is the intensity ratio (Fe-LMM spike at energies of 66 eV and 74 eV, according to intensity/V-LMM intensity), which has a strong Eq. (5), to 20±6 A . uprise because of the increasing (decreasing) inten- In Fig. 5 the position of the {10} satellites is sities of Fe (V ), respectively. The data are fitted plotted in reciprocal space. A clear analogy to the (straight lines) due to Eqs. (1) and (2), assuming linear dependence of the satellites in Fig. 2a can an electron mean free path of 13.4±1.3 A for the be observed. Here Ky corresponds to the [11:0] 703 eV iron Auger electrons and 9.4±1.3 A for direction. A fit to the data gives an angle a= the 473 eV vanadium Auger electrons. 50±2°. This is a slightly higher inclination than The LEED patterns for this growth temperature a=45°, which corresponds to alternating (100) are presented in Fig. 7 for Fe thicknesses of 6.2 A , and (010) planes. 10 A , and 16.5 A . The cross-sections of the spots in the [11:0] and [001] directions are also plotted. 4.3. Growth of iron on V(110)/10 A V at 320 K It can be seen that DK at tFe=10 A and 16.5 A is larger in the [11:0] direction. This indicates a larger Iron was grown on V(110) with a cap layer of average island size in the [001] than in the [11:0] 10 A vanadium, also grown at 320 K. The topology direction, as for V on V(110). of this substrate was described in Section 4.1. All In Fig. 8a this island size is plotted versus the experiments presented in this section are performed Fe thickness for the [001] direction (down trian- on wedges. gles) and for the [11:0] direction (up triangles). It 216 T. Nawrath et al. / Surface Science 414 (1998) 209­220 Fig. 7. LEED patterns at 70 eV of Fe on V(110)/10 A V prepared at a growth temperature of 320 K. The spot profiles are also shown. This collapse to values of 21±2 A and 9±1 A for the [001] and [11:0] directions is much smaller than the island size of the vanadium surface, which has typical values of 89 A (50 A ) along [001] ([11:0]) for 10 A V on V(110). In the range 6­8 A the Fe surface gets more ordered again in both directions, reaching a maximum value at tFe=12 A . For higher tFe values there is a slow decrease in the island size visible. It is remarkable that, despite the different island sizes in the main crystalline directions, their charac- teristic behaviour with respect to tFe remains the same in the whole thickness range, i.e. a factor of 1.5­2.0 bigger terrace width in the [001] than in the [11:0] direction. In Fig. 8b the peak intensities of the {10} spots at an energy of 70 eV are plotted versus tFe. The intensity also rises in the range of 6­8 A , with a maximum at about 15 A . In accordance with the island size the peak intensity also decreases above 17 A . Fig. 8. Fe on V(110)/10 A V at an evaporation temperature of 320 K. (a) Island size in the [11:0] and [001] directions in depen- dence on the Fe thickness tFe, in (b) the peak intensity of the 4.4. Growth of iron at 470 K on V(110)/10 A V {10} spots at 70 eV is plotted versus tFe. (320 K) can be seen that in both crystalline directions the For the investigations on the growth of Fe on island size decreases in the range from 0 to 3 A Fe V, a lower temperature was chosen to avoid an thickness, reaching a minimum at about t intermixing of Fe and V in the low thickness range Fe =4 A . T. Nawrath et al. / Surface Science 414 (1998) 209­220 217 due to diffusion. Again, all experiments described in this section were performed on wedge samples. The LEED patterns of iron on V(110)/10 A vanadium grown at 470 K are very similar to those of vanadium on V(110) at 570 K. One can see a splitting of the LEED spots in the [11:0] direction in dependence on the electron energy, indicating facets with the same orientation as for the V films grown at 570 K. In Fig. 9 the splitting of the {10} spots for a 30 A thick Fe film on V(110) is illustrated, where the position of the satellites is plotted in reciprocal space. As the angle a is 40±2°, one does not get exactly repeated (001) and (010) surfaces in the Fig. 10. Dependence of the Auger intensities of Fe and V on tFe at an evaporation temperature of 470 K. The lines are guides [11:0] direction. Here the height Hdz of the facets to the eye. can be estimated as 22±6 A in the same way as described in Section 4.2. The Auger intensities of the V- and Fe-LMM This indicates a larger deviation from the averaged peaks show a different behaviour compared to the iron thickness for the growth at 470 K compared growth at 320 K (Fig. 10). The increase (decrease) to the growth at 320 K. of the Fe-L3M45M45 (V-L3M23M45) peak as a function of tFe is slower than for the growth at 320 K. As a consequence, the ratio of 5. Discussion Fe-L3M45M45 to V-L3M23M45 also has a much smaller increase (the lines are guides to the eye). First we want to discuss the growth on V(110) at 320 K. There is a similar behaviour of the Fe and V films concerning the anisotropic island size, as for both systems the island size is much bigger in the [001] direction than in the [11:0] direction. This can be seen impressively for the growth of vanadium on V where the system starts at an island size of 55±5 A and 63±6 A for the [001] and [11:0] direction, respectively. Above tV=10 A the system reaches its saturation values of 89±8 A and 50±6 A in the different crystalline directions. The island sizes which were obtained from a V film with tV=10 A are in accordance with the data of the wedge sample within the error limits. The ratio between the island size in the [11:0] and [001] direction is approximately the same for the V/V (1.8) and the V/Fe system (1.5­2.0). A point of consideration is whether the anisot- ropy of the 10 A V surface predetermines the observed anisotropy of the Fe islands. Keeping in Fig. 9. Reciprocal position of the {10} spot satellites of a 30 A mind that the island size of the substrate is about thick Fe film grown at 470 K on V(110)/10 A V. The angle a= 40±2° corresponds to the inclination of the facets with respect four times larger in both directions in the thickness to the film plane. range up to tFe=5 A , an influence of the substrate 218 T. Nawrath et al. / Surface Science 414 (1998) 209­220 structure on the film structure can be excluded. with a and b the lattice constants of the substrate For larger Fe thickness the island size of the film and the film, respectively. In the LEED patterns increases, but due to the Auger data the V surface this results in satellites around the {10} spots that is completely covered, so that there is no inter- have an energy independent reciprocal distance of action between the Fe atoms on the surface and g11:=K11:(b/p) parallel to the [11:0] direction and the V edges. g11=K11(b/p) for the [001] direction. For Fe on V The collapse of the Fe island size is only one gets p/b=20.1, from which it can be concluded observed for the V(110)/Fe system and not for the that the resolution of the CCD camera is not the V(110)/V system. That underlines the special limiting factor for the observation of these satel- behaviour of the Fe epitaxy on V. A reason for lites. To observe this superstructure, which looks this behaviour could be a hindered island coalesc- similar to a broad diffraction pattern of a (110) ence due to the lattice mismatch of V and Fe, as surface, a further condition for the FWHM of the also discussed for Fe on W(110) [11,38]. The {10} spots has to be fulfilled, i.e. the relative width of the spots has to be smaller than the relative result of this mismatch is that an Fe island that is distance b/p of the satellites: more or less pseudomorphic will always have a partly relaxed lattice at the edges, where the Fe atoms are shifted towards the centre. For further DK b < (7) Fe atoms the profit in energy will therefore K p decrease with increasing island size, as the misfit of the outer Fe atoms increases. This can lead to This condition is clearly fulfilled for the a situation where the formation of many small V(110)/10 A substrate in the [001] direction, but islands is energetically more advantageous than only scarcely along the [11:0] direction. Assuming the formation of a few big ones. As the free surface periodic lattice distortions for Fe on V in the enthalpy of vanadium is small, this formation thickness range up to 5 A , the satellites should be could occur at a relatively small island size, which observable at least as lines parallel to the [11:0] causes this rather uncorrelated structure up to direction. Further on, the observed spots for Fe t on V(110) are broader in the [11:0] direction for Fe=5 A . In order to decide whether there is pseudomor- the whole thickness range. In contradiction to this, phic growth in this thickness range, a direct meas- even a satellite pattern that is smeared out in both urement of the Fe lattice constant had to be directions would lead to rhomboid shaped {10} performed, which is not feasible as the FWHM of spots that are broader by a factor of 2 in the the LEED spots is too big to give a precise value [001] direction. From this we exclude a growth for the lattice constant. Here a limit for diffraction with periodic lattice distortions. The {10} peak intensity in Fig. 8b shows a methods is reached, which could only be overcome parallel behaviour to the island size for higher by real space studies, as e.g. STM (scanning tunnel values in tFe (Fig. 8a). This can be explained by microscopy). the FWHM of the LEED peaks being inversely In this context we want to discuss an alternative proportional to the island size and the peak inten- growth model, which also occurs in some cases, sity multiplied by the FWHM of the peaks being when the evaporated film has to equalize the lattice proportional to the number of coherent scattering mismatch of the substrate. This growth has been atoms. Therefore the peak intensity increases in observed e.g. for Fe on W(110) [9,11], where the proportion to the island size, when the layer is lattice mismatch results in periodic lattice distor- thick enough so that the number of scattering tions which is characterized by a Vernier period atoms remains unchanged with increasing layer thickness. ab There can be two reasons for the low intensity p= (6) a-b of the {10} spots at smaller Fe thicknesses. First T. Nawrath et al. / Surface Science 414 (1998) 209­220 219 there is the small island size as discussed above; a general phenomenon of (110) orientated metallic second reason may be a change in the scattering bcc surfaces. It has also been observed for W on phase between Fe and V. W [39], Fe on W [35], Fe on Cr [3,36], and Cr From AES we get information about the cover- on W [40]. To our knowledge it is observed for age of the V(110) surface. From the fits to the vanadium and iron on V(110) for the first time. data we get a mean free path of 13.7±1.3 A for A comparison of the Auger data taken at 320 K the Fe-L3M45M45 and 9.4±1.3 A for the and 470 K shows that there must exist regions V-L3M23M45 transition. The fitted ratio of the with a smaller iron thickness as the averaged saturated Fe intensity to the intensity of the pure thickness. This is in accordance with the faceting vanadium surface is 0.47, which is in good of the iron surface. agreement with the literature data [37]. The fitted This special faceted topology for the iron and mean free path for the Fe and V transitions is vanadium films grown at 470 K (570 K) could slightly smaller compared with the empirical curve also hint at an explanation for the results of Tomaz of Seah and Dench [33], which was derived from et al. [21], where the total magnetic polarization different elements and a significantly dispersed set of V and the total reduction of the Fe magnetic of data points. One can also try to compare these moment is measured for V/Fe(110) and V/Fe(100) data with the AES measurements of other systems, multilayers grown at #500 K. Here the magnetic e.g. the W/Fe system [9]. Here the experimental moments are identical for both orientations. If one mean free path is about 3.9 A (for the Fe low assumes faceted V/Fe interfaces for the (110) sur- energy Auger transition at 47 eV in a fit up to the face, the surprisingly high V polarization per atom first four monolayers), which is smaller by a factor for this orientation could be explained. However, of 1.1 than the value of the empirical curve intro- in a naive model, a perfect faceting of the (110) duced in Ref. [33]. For our system the averaged surface should give a larger total reduction of the factor is 1.15. From that we can conclude a total magnetic moment by a factor of 2 than coverage of the substrate similar to Fe on W(110) that of a flat (100) surface, as the interface surface in the first four monolayers. Therefore this growth increases by the same factor. can be called quasi-Frank­van der Merwe in the As a conclusion, the crucial point was the pre- sense discussed above. Here the missing Auger sentation of the V(110)/Fe and V(110)/V epitaxy kinks can be explained by the rather disturbed at different temperatures. We observed a faceting ordering for tFe<5 A , as observed in the LEED of the surface at higher and a formation of aniso- patterns. tropic islands at lower temperatures. There is a We want to mention that this special growth tendency to an unordered growth for Fe on V at for tFe<5 A could also be an explanation for the 320 K, followed by a more ordered surface in the absence of ferromagnetism for 6 A Fe films on V(110) at 80 K, observed with polarized neutron regime up to 17 A . For larger Fe thickness the reflectometry [22,25], if one assumes that the films surface becomes rougher again. This behaviour are still paramagnetic in this thickness range. cannot be seen for V on V, where a saturation For the growth at T=470 K (570 K), both value in the island size is observed in the studied systems show faceted surfaces with the ridges along thickness regime. the [001] direction and the staircases in the [11:0] direction. For both systems a tendency to build {100} surfaces is clearly visible, as also reported for Fe/Cr multilayers [2]. For vanadium the angle of the staircases is 50±2°, for iron it is 40±2°. Acknowledgements This faceting can be explained in a model with an anisotropic sticking probability of atoms at steps, This work was supported by the Verbund- being high on steps with the edges along [11:0] and forschung of BMFT through Grant No. 03-MA4 low on steps along [001] [35]. It seems to be a HMI-1. 220 T. Nawrath et al. / Surface Science 414 (1998) 209­220 [20] M. van Schilfgaarde, F. Herman, S.S.P. Parkin, J. References Kudrnovsky, Phys. Rev. Lett. 74 (1995) 4063. [21] M.A. Tomaz, W.J. Antel, Jr., W.L. O'Brien, G.R. Harp, [1] M. Albrecht, U. Gradmann, T. Furubayashi, W.A. Har- J. Phys.: Condens. Matter 9 (1997) L179. rison, Europhys. Lett. 20 (1992) 65. [22] H. Fritzsche, T. Nawrath, H. Maletta, H. Lauter, Physica [2] W. Folkerts, F. Hakkens, J. Appl. Phys. 73 (1993) 3922. B 241­243 (1998) 707. [3] R. Coehoorn, J. Magn. Magn. Mater. 151 (1995) 341. [23] T.G. Walker, H. Hopster, Phys. Rev. B 49 (1994) 7687. [4] J. Schwabenhausen, T. Du¨rkop, H.J. Elmers, Phys. Rev. [24] P. Fuchs, K. Todtland, M. Landolt, Phys. Rev. B 53 B 55 (1997) 15119. (1996) 9123. [5] S. Miethaner, G. Bayreuther, J. Magn. Magn. Mater. 148 [25] T. Nawrath, H. Fritzsche, F. Klose, J. Nowikow, C. (1995) 42. Polaczyk, H. Maletta, Physica B 234­236 (1997) 505. [6] D. Stoeffler, F. Gautier, J. Magn. Magn. Mater. 147 [26] C.M. Kim, B.D. deVries, B. Fru¨hberger, J.G. Chen, Surf. (1995) 260. Sci. 327 (1995) 81. [7] A. Vega, C. Demangeat, H. Dreysse´, A. Chouairi, Phys. [27] T. Valla, P. Pervan, M. Milun, Surf. Sci. 307­309 (1994) Rev. B 51 (1995) 11546. 843. [8] M.E. Elzain, D.E. Ellis, J. Magn. Magn. Mater. 65 [28] D.L. Adams, H.B. Nielsen, Surf. Sci. 107 (1981) 305. (1987) 128. [29] L.G. Parratt, Phys. Rev. 95 (1954) 359. [9] U. Gradmann, G. Waller, Surf. Sci. 116 (1982) 539. [30] G.P. Felcher, R.O. Hilleke, R.K. Crawford, J. Haumann, [10] M. Przybylski, I. Kaufmann, U. Gradmann, Phys. Rev. B R. Kleb, G. Ostrowski, Rev. Sci. Instrum. 58 (1987) 609. [31] V.O. de Haan, G.G. Drijkoningen, Physica B 198 (1994) 40 (1989) 8631. 24. [11] H. Bethge, D. Heuer, Ch. Jensen, K. Resho¨ft, U. Ko¨hler, [32] H.J. Elmers, Ph.D. Thesis, Technische Universita¨t Surf. Sci. 331­333 (1995) 878. Clausthal, 1989, p. 14. [12] L.Z. Mezey, J. Giber, Jpn. J. Appl. Phys. 21 (1982) 1569. [33] M.P. Seah, W.A. Dench, Surf. Interface Anal. 1 (1979) 2. [13] E. Bauer, Z. Kristallogr. 110 (1958) 372. [34] M. Henzler, Surf. Sci. 132 (1983) 82. [14] E. Bauer, J.H. van der Merwe, Phys. Rev. B 33 (1986) [35] M. Albrecht, H. Fritzsche, U. Gradmann, Surf. Sci. 294 3657. (1993) 1. [15] P. Isberg, P. Grandberg, E.B. Svedberg, B. Hjo¨rvarsson, [36] H. Fritzsche, U. Gradmann, Mater. Res. Soc. Symp. Proc. R. Wa¨ppling, P. Nordblad, submitted to Phys. Rev. B. 312 (1993) 321. [16] P. Poulopoulos, P. Isberg, W. Platow, W. Wisny, M. Farle, [37] L.E. Davies, N.C. MacDonald, P.W. Palmberg, G.E. B. Hjo¨rvarsson, K. Baberschke, J. Magn. Magn. Mater. Riach, R.E. Weber, Handbook of Auger Electron Spectro- 170 (1997) 57. scopy, Physical Electronics Division, Perkin-Elmer Corpo- [17] A. Vega, A. Rubio, L.C. Balbas, J. Dorantes-Davila, S. ration, MN, 1976. Bouarab, C. Demangeat, A. Mokrani, H. Dreysse´, J. Appl. [38] H.J. Elmers, J. Hauschild, H. Ho¨che, U. Gradmann, H. Phys. 69 (1991) 4544. Bethge, D. Heuer, U. Ko¨hler, Phys. Rev. Lett. 73 (1994) [18] M.N. Baibich, J.M. Broto, A. Fert, F. Nguyen Van Dau, 898. F. Petroff, P. Etienne, G. Creuzet, A. Friedrich, J. Cha- [39] P. Hahn, J. Clabes, M. Henzler, J. Appl. Phys. 51 (1980) zelas, Phys. Rev. Lett. 61 (1988) 2472. 2079. [19] J. Unguris, R.J. Celotta, D.T. Pierce, Phys. Rev. Lett. 69 [40] H. Fritzsche, Ph.D. Thesis, Technische Universita¨t (1992) 1125. Clausthal, 1995.