PHYSICAL REVIEW B VOLUME 53, NUMBER 9 1 MARCH 1996-I Structure and magnetism of Fe/Si multilayers grown by ion-beam sputtering A. Chaiken,* R. P. Michel, and M. A. Wall Materials Science and Technology Division, Lawrence Livermore National Laboratory, Livermore, California 94551 Received 10 August 1995 Ion-beam sputtering has been used to prepare Fe/Si multilayers on a variety of substrates and over a wide range of temperatures. Small-angle x-ray-diffraction and transmission electron microscopy experiments show that the layers are heavily intermixed although a composition gradient is maintained. When the spacer layer is an amorphous iron silicide, the magnetic properties of the multilayers are similar to those of bulk Fe. When the spacer layer is a crystalline silicide with the B2 or DO3 structure, the multilayers show antiferromagnetic interlayer coupling like that observed in ferromagnet/paramagnet multilayers such as Fe/Cr and Co/Cu. De- pending on the substrate type and the growth temperature, the multilayers grow in either the 011 or 001 texture. The occurrence of the antiferromagnetic interlayer coupling is dependent on the crystallinity of the iron and iron silicide layers, but does not seem to be strongly affected by the perfection of the layering or the orientation of the film. Since the B2- and DO3-structure FexSi1 x compounds are known to be metallic, antiferromagnetic interlayer coupling in Fe/Si multilayers probably originates from the same quantum-well and Fermi surface effects as in Fe/Cr and Co/Cu multilayers. I. INTRODUCTION are crystalline but that the spacer layers are amorphous, simi- lar to the situation in other transition metal/silicon systems. Multilayer films formed from transition metals and semi- When the Si spacer layer thickness is less than about 20 Å conductors have long been studied because of their unusual thick, the iron silicide spacer layer forms a crystalline silicide superconducting properties1 and because of possible applica- with either the B2 or DO3 structure. The B2 structure con- tion as x-ray optical elements.2 Many unusual phenomena sists of two interpenetrating simple cubic sublattices and is have been produced, ranging from the observation of dimen- identical to the CsCl structure for a 1:1 ratio of Fe and Si,15 sional crossover in weakly coupled superconducting Nb lay- while the DO 3 structure is an fcc lattice with two inequiva- ers in Nb/Ge multilayers1 to the occurrence of bcc Ge in lent Fe sites.16 Extensive growth experiments, described be- short-period Mo/Ge multilayers.3 Unusual magnetic proper- low, suggest that crystallinity of the spacer layer is crucial ties have recently been observed in Fe/Si multilayers by for occurrence of the antiferromagnetic interlayer coupling, workers at ETH Ref. 4 and Argonne.5 A large antiferromag- in keeping with previous suggestions.5 Since both the B2 netic AF interlayer coupling in these multilayers manifests and DO3 phases are metallic,15,16 the fact that crystallinity is itself in hysteresis loops as a high saturation field and a low required for antiferromagnetic coupling suggests that the remanent magnetization. Similar magnetization curves are coupling in Fe/Si has a common origin with that observed in associated with large interlayer coupling in metal/metal mul- metal/metal multilayers. tilayers like Fe/Cr and Co/Cu.6,7 Much consideration has been given to whether the coupling in the Fe/Si system has II. EXPERIMENTAL METHODS the same origin as in the metal/metal multilayers.8,9 There- fore the question of whether the spacer layer in the Fe/Si The Fe/Si multilayers are grown in the ion-beam sputter- multilayers is a metal or semiconductor is of particular inter- ing IBS chamber whose layout is shown schematically est. in Fig. 1. The system base pressure is typically about Previous work on Nb/Si,10 Co/Si,11 Ni/Si,12 and Mo/Si 2 10 8 torr. The ion gun is a 3 cm Kauffman source with Ref. 13 multilayers have shown that there is a strong ten- focusing optics.17 The energy of the ions leaving the gun can dency towards compound formation at the metal/silicon in- be modulated by raising and lowering the voltage on the terface. In general these multilayers consist of polycrystalline acceleration grids, creating in effect an electrical shutter. The metal layers separated by an amorphous silicon layer which Ar ions are incident on the sputter target at 1000 V at an is bounded on either side by a layer of intermixed material. angle of about 45°. The Ar pressure is maintained in the 2­3 The intermixed silicide layers in these films were amorphous 10 4 range by a flow-controller coupled to a capacitance unless they were annealed at several hundred °C.11,14 These manometer.18 Four 3 in. diameter sputter targets are mounted previously studied multilayers were therefore likely in their on a tray which can be rotated by a stepper motor.19 Layer as-grown state to have metal/semiconductor character be- thickness is monitored by a quartz-crystal oscillator which is cause of the presence of the amorphous silicon layer. placed in close proximity to the substrates. The substrates are In order to investigate the character of the spacer layer in about 25 cm above the targets, clamped to a copper tray. The the Fe/Si multilayer system, we have grown a large number temperature of the tray is monitored by a thermocouple and of films with different substrate temperatures, substrate can be varied between 150°C and 200 °C.20 Three films types, and layer thicknesses. When the Si spacer layer thick- are grown per chamber pumpdown. ness is greater than about 20 Å, we find that the metal layers The thickness monitor, the controller for the stepper mo- 0163-1829/96/53 9 /5518 12 /$10.00 53 5518 © 1996 The American Physical Society 53 STRUCTURE AND MAGNETISM OF Fe/Si MULTILAYERS . . . 5519 FIG. 1. Schematic plan of the ion-beam sputtering system. FIG. 2. Magnetization curves for Fe 30 Å/Si 20 Å 50 and Fe tor and the ion-beam power supply are all interfaced to a 30 Å/Si 14 Å 50 multilayers grown on glass substrates at nominal personal computer which has been programmed using the RT during the same deposition run. Plotted on the y axis is the ASYST instrument control package.21 When the system is observed magnetization of the films divided through by the calcu- depositing a multilayer, the computer sends the material pa- lated magnetization of an equivalent thickness of bulk Fe. The Fe rameters to the thickness monitor, rotates the stepper motor 30 Å/Si 20 Å 50 multilayer has soft magnetic properties much to its new orientation, and turns the ion gun on. When the like bulk Fe, while the Fe 30 Å/Si 14 Å 50 multilayer exhibits desired thickness is reached, the thickness monitor turns the AF interlayer coupling. ion gun off and prompts the computer for the next layer. The basic design of the system is similar to one previously de- saturation fields were observed.27 The differences between scribed by Kingon et al.22 the previous IBS-grown films and ours may be related to the The substrates for multilayers growth include glass lower ion-beam voltage used by Inomata et al.27 Compari- coverslips, oxidized silicon wafers, MgO 001 and sons on the basis of layer thickness are made here only be- Al tween films grown during the same deposition run in order to 2O 3 02 ¯11 . The first two substrates, which are used for growth of polycrystalline films, were rinsed in solvents be- insure that the relative layer thicknesses are meaningful. fore loading into the vacuum chamber. The second two, Films with similar layer thicknesses have been grown many which are used for epitaxial growth, are cleaned according to times to establish reproducibility of the observed trends. a recipe reported by Farrow and co-workers.23 The typical deposition rate for Fe is 0.2 Å/s while that for Si is about 0.3 A. Layer-thickness dependence of properties Å/s. All films are capped with a 200 Å Ge oxidation barrier. The magnetic and structural properties of the films are stable Forty- and fifty-repeat multilayers have been grown with for at least one year. Ge is used for capping instead of Si in tFe 14, 20, 30, 40, and 50 Å and tSi 14 and 20 Å. order to prevent interference with element-specific soft x-ray Magnetization curves for 50-repeat Fe 30 Å/Si 20 Å and fluorescence measurements, which will be reported Fe 30 Å/Si 14 Å multilayers grown on glass at nominal RT elsewhere.24 about 60 °C are shown in Fig. 2. On the y axis of this X-ray-diffraction characterization has been performed us- plot is the magnetic moment of the multilayer normalized to ing a 18 kW rotating anode system outfitted with a graphite the moment of an equivalent volume of bulk Fe. The mag- monochromator. All spectra are taken using the Cu K netization curve of the 30/20 multilayer looks much like wave- length. Conventional high-resolution electron microscopy that of an Fe film, while the magnetization curve of the 30/ and electron diffraction have been performed in order to 14 multilayer shows the high saturation field and low rema- characterize the microstructure of the as-deposited films in nence which characterize AF interlayer coupling. At its satu- cross-section. Magnetization curves are obtained using a vi- ration field the magnetization of the 30/14 multilayer is brating sample magnetometer. All the data shown here were about the same as for the 30/20 multilayer. Both of these taken at room temperature. films have moments only about half as large as an equivalent volume of bulk Fe. Our observation of AF coupling for Si thicknesses between 10 and 20 Å and the disappearance of III. RESULTS coupling for Si thicker than 20 Å confirm previous observa- Overall the magnetic properties of the Fe/Si multilayers tions on magnetron-sputtered films.5 made by IBS are similar to those made previously by mag- X-ray-diffraction spectra for these multilayers are shown netron sputtering.5,25 Definitive confirmation of AF interlayer in Figs. 3 and 5. Figure 3 shows the small-angle x-ray- coupling in our multilayers has been obtained by polarized scattering SAXS data with peaks at angles neutron reflectivity measurements.26 For some unknown rea- n2 2 4 2sin2 2 , 1 son the magnetic properties of our multilayers are closer to those of the magnetron-sputtered multilayers than those re- where is the x-ray wavelength, tFe tSi is the ported on in a previous study using IBS, where much lower multilayer bilayer period and is the index of refraction for 5520 A. CHAIKEN, R. P. MICHEL, AND M. A. WALL 53 FIG. 3. X-ray-diffraction spectra at small angle for the same films whose magnetization curves are shown above. Broader peaks show that there is more disorder in layering for the AF-coupled film with t FIG. 5. High-angle spectra for two Fe/Si multilayers showing Si 14 Å. Using Eq. 1 , these data give bilayer periods 41.82 Å for the nominal Fe 30 Å/20 Å 50 film and the Fe 011 and 002 peaks. The tSi 20 Å film is predominantly 38.10 Å for the nominal Fe 30 Å/Si 14 Å 50 film. 011 -textured, while the AF-coupled film with tSi 14 Å has mixed 011 and 001 textures. No x-ray-diffraction peaks which could be indexed to crystalline silicon or silicide spacer layer x rays.28 This grazing incidence data gives information about phases have been observed in any Fe/Si multilayer. A superlattice the quality of the multilayer interfaces. Figure 3 shows four satellite just below the Fe 002 peak is labeled `` 1.'' low-angle peaks for both films, indicating a reasonably strong composition modulation along the growth direction. the Fe 30 Å/Si 14 Å 50 film 38.10 0.04 Å. is The higher frequency oscillations between 1° and 4° are 8.2 Å shorter than the nominal value for the tSi 20 finite-thickness fringes from the Ge cap layer. The most Å film, and 5.9 Å shorter than nominal for the tSi 14 notable difference between the two spectra is that the Å film. Although some of the discrepancy between the multilayer peaks are broader for the AF-coupled tSi 14 nominal and observed bilayer period may be due to calibra- Å film, indicating more fluctuations in bilayer period and tion inaccuracies, most is undoubtedly due to intermixing of probably more interface roughness. Using the spacing be- the Fe and Si layers, in keeping with observations in the tween peak positions to eliminate the unknown from Eq. other metal/Si multilayers.10,14 Throughout this paper we will 1 gives values of the bilayer period for the two films. For continue for convenience to refer to the films in terms of the multilayer with nominal layering of Fe 30 Å/Si 20 Å their nominal layer thicknesses. 50, the derived value for is 41.82 0.07 Å, while for Comparison of the magnetization data to the x-ray data can give some further insight into the question of intermix- ing. Because of the presumed interdiffusion of the Fe and Si layers, the magnetic moment of the Fe layers is also reduced from the nominal value. The missing magnetic moment can be expressed as an equivalent thickness of Fe. Figure 4 shows a plot of missing moment in units of Å of Fe versus missing bilayer period determined from multilayer peak po- sitions in SAXS for films grown at room temperature RT . The plot shows that while the diffusion-induced reduction in bilayer period varies between 1 and 8 Å, the missing Fe moment per bilayer for both interfaces is consistently be- tween 10 and 12 Å. The one outlier in Fig. 4 is for a film which had tFe 20 Å, the thinnest Fe for which we have ever observed interlayer coupling. Other groups have previ- ously observed a moment reduction of 12­14 Å per bilayer in polarized neutron reflectivity measurements on uncoupled Fe/Si multilayers with thick Si layers.29,30 The disparity between the magnetic moment reduction and the bilayer period reduction numbers may at first appear FIG. 4. Missing Fe magnetic moment expressed as an equiva- lent thickness of Fe plotted versus missing bilayer period as ob- to be puzzling. This disparity occurs because the moment tained from fits to small-angle x-ray-diffraction data. Symbols indi- and bilayer period are affected by different aspects of the cate different nominal Si layer thicknesses and different film structure. In calculating the moment reduction in Å the as- textures. The film labeled ``LN'' was grown on a liquid-nitrogen sumption has been made that the Fe layer has the magneti- LN -cooled substrate; all others were grown at nominal RT. All zation of bulk Fe. This is equivalent to assuming that there is multilayers have 40 or 50 repeats and were grown on either glass or no Si in the Fe layer, which is undoubtedly false. In calcu- oxidized Si substrates. lating the missing bilayer period, the assumption has also 53 STRUCTURE AND MAGNETISM OF Fe/Si MULTILAYERS . . . 5521 FIG. 6. Cross-sectional TEM images a and b and selected area diffraction patterns c and d for the same Fe 30 Å/Si 20 Å 50 multilayer and a Fe 40 Å/Si 14 Å 50 multilayer grown that shows strong AF coupling. a and b show that the Fe/Si multilayers have layers which are continuous for large lateral dis- tances. There is no sign of propagating roughness or columnar growth. c The 30/20 multilayer shows only an Fe 011 ring. d The 40/14 film shows 011 and 002 spots plus a faint spot at the 001 position indicated by an arrow . been made that the spacer layer is pure Si, also clearly false. observed for polycrystalline multilayers grown on glass, in The fact that the missing magnetic moment is almost con- agreement with observations by Foiles et al.31 stant irrespective of the reduction in bilayer period suggests The thin-Si multilayers which have AF coupling usually that the spacer layer is nonmagnetic independent of Si thick- show a mixed 001 and 011 orientation when grown on ness. The lack of variation of the missing moment is then glass substrates at RT. Occasionally tSi 14 Å films with a explained by the diffusion of a constant number of iron at- pure 011 orientation are obtained at RT. The variation in oms into the silicon layer, irrespective of its thickness. The texture may be due to changes in film stress under slightly wide variation of the measured bilayer period is most likely different deposition conditions. Stress induced during depo- related to the varying orientation and crystallinity of the sition has been postulated to explain the mixed Mo texture spacer layer, neither of which affects the magnetic moment if found in Mo/Ge multilayers.3 In contrast to the thin-Si case, the spacer itself is nonmagnetic. the thicker-Si Fe/Si multilayers which do not show interlayer Figure 5 shows the high-angle x-ray spectra where peak coupling always have a pure 011 texture. Since the 011 positions give information about the orientation and crystal- plane is close-packed for the bcc crystal structure, one would linity of the films. The intense peak near 70° in this plot is expect the 011 orientation to be energetically favored for due to the Si substrate. Included are data for an Fe 40 Å/Si the Fe in a multilayer with amorphous Si. Films grown at 14 Å 40 antiferromagnetically coupled multilayer and for nominal RT on glass or oxidized Si substrates typically had an Fe 30 Å/Si 20 Å 40 uncoupled multilayer, both grown rocking curves about 10° wide indicating a moderate amount on oxidized Si 001 at RT. The peaks for the 40/14 film are of orientation. narrower than for the 30/20 . The Scherrer formula gives 78 Transmission electron microscopy TEM has been used Å or about two bilayer periods for the coherence length of to further investigate the morphology of the films. TEM the 40/14 film and 34 Å or about one bilayer period for the cross-sectional images of an Fe 30 Å/Si 20 Å 50 coherence length of the 30/20 film. Coherence lengths in multilayer and an Fe 40 Å/Si 14 Å 50 multilayer grown IBS-sputtered antiferromagnetically coupled films are often during the same deposition run are shown in Figs. 6 a and as long as 200 Å. Fullerton et al. have inferred that the 6 b , respectively. The most salient features of the 30/20 spacer layer in thin-Si Fe/Si multilayers must be crystalline multilayer are the long lateral continuity of the layers and the based on their observation of coherence lengths longer than a smoothness of the interfaces. Since there is no interlayer co- bilayer period.5 In keeping with its superior crystallinity, the herence in the 30/20 film, the crystalline grains have a high 40/14 multilayer has one superlattice satellite on the low- aspect ratio. The 40/14 multilayer also has long, continuous angle side of the Fe 002 peak. Typically only one satellite layer planes but has rougher interfaces, consistent with the on the low-angle side of the Fe 011 or 002 x-ray peak is SAXS data. 5522 A. CHAIKEN, R. P. MICHEL, AND M. A. WALL 53 FIG. 7. High-resolution TEM images of the same films whose low-resolution images are shown above. a Fe 30 Å/Si 20 Å 50 multilayer image showing amor- phous silicide layers between polycrystalline Fe layers. b Fe 40 Å/Si 14 Å 50 multilayer im- age showing crystalline coherence between the polycrystalline Fe layers and iron silicide spacer lay- ers. There is no amorphous layer present. Transmission electron selected-area diffraction patterns High-resolution TEM images of the 30/20 and 40/14 for the 30/20 and 40/14 films are shown in parts c and multilayers are displayed in Fig. 7. The 30/20 film is shown d of Fig. 6. The 30/20 films show only a Fe 011 ring, in Fig. 7 a to have a crystalline Fe layer and amorphous consistent with the high-angle x-ray-diffraction scans. The spacer layer, similar to the morphology seen before in Mo/Si 40/14 film, on the other hand, displays spots corresponding Refs. 13 and 14 and Co/Si multilayers.11 The 40/14 to the 011 and 002 reflections seen using x rays. The multilayer in Fig. 7 b on the other hand is made up entirely presence of spots rather than rings in the 40/14 image im- of crystalline layers. The coherence between the Fe and sil- plies the presence of large, oriented crystallites in the film. icide spacer is clearly evidenced by the continuity of atomic Most interestingly, the 40/14 image includes a faint spot layer planes from the Fe layer into the spacer. Some crystal- near what would be the Fe 001 position were the Fe 001 lites in the 40/14 film extend all the way from the substrate peak not forbidden by symmetry in the bcc crystal structure. to the surface of the film. The small coherence lengths ob- The 001 peak is allowed in the B2 and DO3 crystal struc- served in x-ray-diffraction data for the uncoupled thicker-Si tures. The B2 structure is found in the equilibrium phase films are explained by the presence of the amorphous layers. diagram only at 10­22 % Si range of composition,32 but The lack of crystallinity in the spacer layer of tSi 20 workers at ETH have grown this crystal structure throughout Å films is presumably due to insufficient time for full inter- the range of composition on Si substrates using molecular- diffusion and ordering in the thicker layers. A kinetic mecha- beam epitaxy MBE .33 The DO3 phase found in the equi- nism for the lack of crystallization is supported by experi- librium phase diagram is Fe 3Si, which is ferromagnetic.32 ments which show that intentional placement of Fe in the Si Clearly a ferromagnetic spacer phase is not consistent with layer allows thicker spacer layers to crystallize.25,35 the observation of antiferromagnetic interlayer coupling, al- Another striking feature of the image in Fig. 7 b is the though a nonstoichiometric DO3-structure phase might have periodic modulation that occurs in the silicide spacer layer. different magnetic order. The B2 and iron silicide phases The modulation originates from scattering by inequivalent have both been previously suggested as possible candidates planes of atoms. Simulation of this image using a multiple- for the spacer layer in AF-coupled Fe/Si multilayers.5,31,34 scattering computer calculation may be helpful in positively The position of the 001 TEM spots is not consistent with identifying the crystal structure of the spacer layer phase. the d spacings of the phase. Dark-field images of the 40/14 multilayer can help an- According to the powder-diffraction files for the B2 and swer questions about the texture of the film as well. Figure DO 3 structures, only the 111 peak of the fcc-family DO3 8 a shows the same bright-field image as in Fig. 6 b . Dark- does not coincide with a B2 peak. The 111 peak would be field images were formed using 001 , 002 , and 011 spots expected to be very weak in the diffraction patterns formed from the diffraction pattern shown in Fig. 6 d . The resulting from cross-sectional specimens of the film. The reason is that micrographs are shown in Figs. 8 b , 8 c , and 8 d , respec- a small number of grains contributes to the cross-sectional tively. Panels a and b of this figure show the same region image, and the probability of sampling a grain with its 111 of the 40/14 multilayer. The brightness of the spacer layers planes in the observable direction is small because of the in this dark-field image demonstrates that the 001 reflection random in-plane orientation. Future work will include elec- does indeed come from the spacer layer and is not the for- tron diffraction studies of a 40/14 specimen prepared in the bidden 001 spot of bcc Fe. Figures 8 c and 8 d also show plan-view geometry, where the number of grains which are the same region although a different region than panels a sampled is considerably larger and the odds of observing the and b . The bright areas in these two images are the fcc 111 peak are improved. complement of one another; where one is bright, the other is 53 STRUCTURE AND MAGNETISM OF Fe/Si MULTILAYERS . . . 5523 FIG. 8. a The same bright field TEM micrograph of the Fe 40 Å/Si 14 Å 50 multilayer as is shown in Fig. 6 b . b A dark- field image of the same region of the 40/14 multilayer. This dark- field image was formed using the 001 reflection. Comparison with the bright field image shows that the 001 reflection originates from the Si substrate and the spacer layers. c and d Dark- field images formed from 002 and 011 reflections. Image c shows that planes with 002 ori- entation predominate near the film surface. Image d shows that planes with 011 orientation pre- dominate near the substrate. The film surface is on the top of all these images. dark and vice versa. The dark-field images in panels c and 14 Å 40 multilayers grown at 150 °C, 60 °C nominal d of Fig. 8 demonstrate convincingly that the orientation of RT , and 200 °C are shown. The data show that as the the film evolves from predominantly 011 to predominantly substrate temperature increases the saturation field increases 002 as the thickness increases. The reason for the change in indicating larger AF coupling. The saturation magnetization orientation with film thickness is not obvious; it may be re- also decreases, suggesting a larger degree of interdiffusion in lated to the bilayer-period-number dependence discussed in the films grown at higher temperatures. Sec. III C. The suspicion that more interdiffusion occurs at higher The effect of varying the Fe thickness has also been stud- substrate temperatures is confirmed by examination of the ied. Magnetic properties for films with 20 Å tFe 50 SAXS spectra for the three films, shown in Fig. 10. The film Å are found to change only slightly in keeping with the grown at reduced temperature has 7 peaks while the film expected inverse proportionality of the saturation field with grown at nominal RT has 5 and the film grown at tFe .6 SAXS peaks tend to broaden and even split with in- creasing Fe thickness, indicating increased disorder in the 200 °C has only 4. Quantitative modeling of low-angle layering. The splitting of these peaks may indicate different x-ray data has shown that the suppression of higher-order bilayer periods in areas of the film with the 011 and 001 peaks may be due to either interdiffusion or cumulative textures. When the Fe is made less than 20 Å thick, the Fe roughness.28,37 Certainly larger cumulative roughness could high-angle diffraction peaks disappear and so does the AF also occur at higher growth temperatures, but one would ex- coupling. The disappearance of crystalline Fe peaks near pect very rough growth to suppress AF coupling due to an t increased number of pinholes and larger magnetostatic inter- Fe 20 Å is consistent with previous results on evaporated Fe/Si multilayers.29 Thus poor crystallinity of the Fe layers layer coupling.38 Since higher growth temperatures seem to appears to suppress the interlayer coupling even when the Si enhance rather than suppress the coupling, it seems more thickness is favorable. The lack of AF coupling in films with likely that high substrate temperatures are promoting inter- poorly crystalline Fe may be related to the lack of a template diffusion rather than roughness. Studies of Mo/Si multilayers for the crystalline iron silicide spacer to grow on. showed that a growth temperature of 150 °C gives maximum SAXS reflectivity, which the authors attribute to greater in- terface smoothness than for RT deposition.39 Smaller bilayer B. Dependence of properties on growth temperature periods in multilayers grown at higher temperatures support and post-growth annealing the claim of increased interdiffusion. Fitting Eq. 1 to peak Depositing the multilayers at different substrate tempera- positions from Fig. 10 gives 52.7, 49.3, and 43.8 Å, tures is an obvious way of influencing the composition and respectively, for the 150 °, 60°, and 200° multilayers crystallinity of the spacer layer phase in the Fe/Si multilay- versus the nominal value of 54 Å. ers. Fullerton has suggested that the interlayer of Fe/Si mul- Higher substrate temperatures may also promote ordering tilayers is improved by high-temperature growth.36 We have of the Fe and Si atoms in the crystalline spacer layer. In the grown films on glass substrates at various temperatures be- fully ordered B2 phase, the Fe and Si atoms sit on different tween 150 °C and 200 °C. The effect of substrate tem- simple cubic sublattices. The sublattice order can occur irre- perature on the interlayer coupling of the films is illustrated spective of whether or not the Fe to Si ratio is 1:1. It is in Fig. 9, where magnetization curves for three Fe 40 Å/Si interesting to speculate whether the AF coupling is depen- 5524 A. CHAIKEN, R. P. MICHEL, AND M. A. WALL 53 FIG. 9. Magnetization curves for three Fe 40 Å/Si 14 Å 40 multilayers grown on glass sub- strates at 150 °C, 60 °C, and 200 °C. The increase of the saturation field with increasing substrate temperature indicates an increase in AF coupling. Note that the saturation magnetization also decreases slightly with increasing substrate temperature. dent on the degree of ordering in the spacer layer. An while those grown at high temperatures have long crystalline ordering-dependent coupling seems plausible in light of the coherence lengths. The reasons for the strange temperature Fermi-surface theories of coupling in metal/metal dependence of growth texture are not understood, although multilayers.40,41 A well-ordered B2 or DO3 phase would one presumes that they have to do with the kinetics of have more well-defined Fermi surface features than a random growth. It is not clear why the 001 texture should appear at solid solution. Unfortunately the 001 silicide peak has only all, although it has also been seen in Mo/Ge multilayers.3 An been observed by TEM, making experimental attempts to oscillatory dependence of film texture on spacer layer thick- address this issue difficult. Further studies with x-ray- ness and deposition conditions has been reported for diffraction and soft x-ray fluorescence are underway. NiFe/Cu multilayers grown by IBS.42 The 001 texture has The crystallinity of the films also varies with growth tem- not been reported in polycrystalline magnetron-sputtered perature. Surprisingly, films grown at both low and high tem- Fe/Si multilayers, and may be due to some peculiarity of IBS peratures on glass substrates always have only the 011 tex- ture, while films grown at nominal RT often have mixed growth. 001 and 011 textures. The multilayers deposited on A logical extension to the growth temperature studies is to heated and cooled substrates do differ greatly in that those try annealing the Fe/Si multilayers grown at lower substrate grown at low temperature have amorphous spacer layers, temperatures to see if their properties evolve towards those of the multilayers grown at higher temperatures. As far as the magnetic properties are concerned, the answer is ``no.'' An- nealing the uncoupled RT-grown Fe 30 Å/Si 20 Å 40 and low-temperature-grown Fe 40 Å/Si 14 Å 40 multilayers at 200 °C for two hours had almost no effect on their mag- netic properties beyond a slight magnetic moment reduction. A subsequent 300 °C anneal for two hours once more pro- duced a moment reduction and a decrease in coercive field in the uncoupled multilayers. A very low coercive field for an- nealed Fe/Si films is not surprising given the well-known softness of Fe-Si alloys. A 300 °C anneal even eliminated the interlayer exchange coupling of a RT-grown Fe 40 Å/Si 14 Å 50 film used as a control. For this 40/14 multilayer, the 300 °C anneal caused the SAXS peaks to narrow and re- duced their number from 5 to 4. At the same time the bilayer period decreased from 49.4 Å to 46.0 Å. High-angle x-ray spectra not shown indicated that the Fe lattice constant FIG. 10. Small-angle x-ray-diffraction spectra for three Fe 40 slightly decreased, which is consistent with increased diffu- Å/Si 14 Å 40 multilayers grown on glass substrates at 150°C, sion of Si in the Fe layer.31 These x-ray and magnetization 60 °C, and 200 °C. The disappearance of higher-order peaks at results imply that annealing primarily promotes interdiffu- higher substrate temperatures is an indication of greater interdiffu- sion of the Fe and silicide layers. With sufficient interdiffu- sion. sion the spacer layer may become ferromagnetic, which 53 STRUCTURE AND MAGNETISM OF Fe/Si MULTILAYERS . . . 5525 FIG. 11. Magnetization curves for 2-, 12-, and 25-repeat Fe 40 Å/Si 14 Å multilayers grown during the same deposition run at FIG. 12. Magnetization curves of three Fe/Si/Fe trilayers. The nominal RT on glass substrates. The 2-repeat multilayer really an open circles are data for an Fe 100 Å/Si 14 Å/Fe 100 Å film Fe/Si/Fe trilayer shows no signs of AF coupling. The 12-repeat grown directly on glass at 200 °C. The filled circles are data on a multilayer appears to have a smaller coupling than the 25-repeat Fe 100 Å/Si 14 Å/Fe 100 Å film grown at 200 °C on a 500 Å one. a-Si buffer layer on glass. The solid curve is for a Fe 100 Å/Si 14 Å/Fe 100 Å film grown at nominal RT on a 500 Å a-Si buffer layer would explain the suppression of antiferromagnetic inter- on glass. The coupling is stronger in the film grown at high tem- layer coupling. These Fe/Si multilayers show less thermal perature on a buffer than in either of the other two films. stability than Mo/Si multilayers with comparable layer thick- nesses, which do not show changes in SAXS spectra until One would not expect interlayer coupling that is quantum- 400 °C.39 There was no sign of the solid-state amorphization mechanical in nature to be affected much by total film thick- previously observed in Fe/Si multilayers with thicker ness. The unusual thickness dependence therefore raises the layers.43 question of whether there is quantum-mechanical coupling at Whatever process occurs during annealing, it does not all, or whether some other mechanism might determine the enhance the interlayer coupling the way that 200 °C shape of the magnetization curves. Disordered magnetic ma- growth does. This is hardly surprising given that annealing terials such as small amorphous Fe particles can have low will tend to drive the multilayer towards its equilibrium state, remanence and high saturation fields without any layering at presumably a mixture of different iron silicide phases. There all. The magnetization curves of these Fe particles are in fact is no reason to think that the crystalline Fe/Fe quite similar to those of the Fe/Si multilayers.46 This resem- xSi 1 x multilayer should be an intermediate phase during the an- blance might lead to speculation that the topmost Fe layers in nealing. In the future the kinetics of Fe/Si multilayer growth Fe/Si multilayers are discontinuous and that the magnetic at different substrate temperatures will be investigated fur- properties are dominated by particle shape. However, the ex- ther by employing an ion-assist gun to improve atomic sur- istence of half-order peaks in polarized neutron reflectometry face mobility. measurements in the IBS-grown Fe/Si multilayers26 and the magnetron-sputtered multilayers44 gives unambiguous evi- dence that the magnetic properties are due to magnetic order C. Dependence of properties on number of bilayers rather than structural disorder. In addition, TEM pictures One puzzling aspect of the interlayer exchange coupling such as Fig. 6 show that the Fe layers are continuous in films in the Fe/Si system has been the dependence of its strength with both high and low saturation fields. on the number of bilayers in the multilayer. This trend is How then does the number of bilayer periods influence illustrated in Fig. 11, where magnetization curves for Fe 40 the AF coupling strength? It has been suggested that the Å/Si 14 Å N multilayers with 2, 12, and 25 repeats are difference between thin and thick multilayers grown at nomi- displayed. The 2-repeat multilayer is just an Fe/Si/Fe nal RT is that the substrates of thick multilayers have time to trilayer. Although the trilayer has magnetic properties like rise to a higher temperature about 60 °C for our system bulk Fe, the 25-repeat multilayer data has a magnetization during the longer growth.36 This idea seems reasonable in curve similar to the 40-repeat multilayer data shown above. light of the larger coupling in samples grown on heated sub- The magnetization curve for the 12-repeat multilayer falls in strates as described above. In order to investigate this idea, a between that for the thicker and thinner films. Evidence for Fe 100 Å/Si 14 Å/Fe 100 Å film was grown on glass at AF coupling which is stronger near the top of an Fe/Si 200 °C. The magnetization curve for this film is shown in multilayer than near the substrate has previously been de- Fig. 12. Also shown in this figure are data for a Fe 100 Å/Si scribed by Fullerton et al.44 Presumably the increase of cou- 14 Å/Fe 100 Å trilayer deposited at nominal RT and for a pling with bilayer-number is a manifestation of the same Fe 100 Å/Si 14 Å/Fe 100 Å trilayer deposited at 200 °C, phenomenon. The interlayer coupling in Co/Cu multilayers both grown on a 500-Å-thick a-Si buffer. The trilayer depos- also increases with the number of bilayer periods up to about ited directly on glass at elevated temperature has only 25 bilayers.45 slightly less remanence and higher saturation field than the 5526 A. CHAIKEN, R. P. MICHEL, AND M. A. WALL 53 trilayer grown at RT whose data are shown in Fig. 11. This higher for the epitaxial samples and that magnetocrystalline result implies that it is not substrate temperature alone which anisotropy effects are observed. The magnetocrystalline an- causes bilayer-number effects. The magnetization curves of isotropy energies of epitaxial trilayers grown on MgO and the trilayers grown on buffer layers, on the other hand, look Ge are similar to bulk Fe.50 much more like typical tFe 40 Å 40-repeat multilayer re- The shape of the high-angle peaks plus superlattice satel- sults. An epitaxial Fe 100 Å/Si 14 Å/Fe 100 Å trilayer lites are described by a theory due to Fullerton et al.28 Ap- grown directly on an MgO 001 at 200 °C substrate also plication of this theory to the Fe/Si multilayers is difficult has strong AF coupling data not shown . Undoubtedly the because the silicide lattice constant, the thickness of the re- strong AF coupling of the trilayer grown directly on the maining pure Fe, and the thickness of the silicide spacer can MgO is due the superior surface quality of the single-crystal be estimated only roughly. A precise determination of the substrate. silicide lattice constant should make a quantitative analysis The take-away lesson from all of these results is that sub- of these satellite features possible. strate roughness is probably responsible for the reduced in- terlayer coupling in Fe 40 Å/Si 14 Å multilayers with a low number of bilayers. Conformal growth may propagate this IV. DISCUSSION roughness up from the substrate into the multilayer. Parkin Fe and Si appear to be the only known transition-metal/ et al. have found that the interlayer coupling in MBE-grown semiconductor combination in which the two elements inter- Co/Cu multilayers is very sensitive to the substrate and the diffuse to form a crystalline spacer layer with coherent inter- buffer layer type, perhaps due to pinholes through the Cu faces. The reasons why this unusual morphology occurs in layers.47 Presumably thin Fe layers grown directly on glass the Fe/Si system are unknown but likely involve a high rate are so wavy that pinhole and magnetostatic coupling domi- of Fe diffusion into a-Si and a low heat of crystallization of nate the interlayer interactions for the first few bilayer peri- the iron silicide compound. A detailed discussion of these ods. Recent calculations show that magnetostatic effects as- issues is beyond the scope of this paper. sociated with propagating roughness can give interlayer Three different crystal structures have been proposed for ferromagnetic coupling of the same order of magnitude as the crystalline spacer layer of the Fe/Si multilayers. The the coupling derived from quantum-well effects.38 Ongoing phase can be eliminated on the basis of the electron diffrac- polarized neutron reflectivity experiments may give more in- tion patterns and TEM dark field images presented here. The formation on the variation of the coupling with position in B2 and DO the thicker multilayers.26 3 crystal structures are better lattice-matched to Fe than -FeSi or - and -FeSi2 . The lattice constant of the B2 phase was reported by Ma¨der and co-workers to be D. Growth on single-crystal substrates 2.77 Å, only 3.1% different from Fe.51 The lattice constant of the phase is 4.46 Å,51 which matches the Fe 110 plane That Fe films can be grown epitaxially on MgO and only in the energetically unfavorable 210 direction.35 Al2O3 substrates is well known.48 One might therefore ex- Recent conversion-electron Mo¨ssbauer data are inter- pect to be able to grow high-quality Fe/Si superlattices on preted in support of the B2 crystal structure, although the these substrates. Figure 13 a shows high-angle x-ray- possibility of the DO3 phase was not considered in that diffraction spectra for a purely 001 -oriented Fe 40 Å/Si Å study.34 It is plausible that the B2 or DO 3 structures form in 60 multilayer grown on MgO 001 . The spectrum in Fig. rapid, far-from-equilibrium growth conditions because of 13 b is data for a highly 011 -oriented Fe 40 Å/Si 14 their small unit cells. Since silicon deposited at low substrate 46 multilayer grown on Al2O3. Both multilayers were de- temperatures is amorphous, the most likely scenario is the posited at 200 °C. Figure 13 c shows a scan for the following. Silicon deposited on a crystalline Fe layer goes MgO 110 and Fe 110 peaks for the film on the MgO down amorphous and diffuses only slightly into the Fe. Sub- substrate. These sets of peaks are offset from one another by sequently deposited Fe atoms diffuse rapidly into the amor- 45° in , confirming the well-known epitaxial relation phous Si, analogous to what happens during the growth of Fe 001 MgO 001 and Fe 110 MgO 100 .48 The Mo/Si multilayers.13,14 During the diffusion of Fe into Si, scans for the Al2O3 substrate show that this film is only crystallization of the silicide occurs, possibly driven by the weakly oriented in-plane. Mattson et al. have previously heat of mixing or by the kinetic energy of the incident Fe grown Fe/FeSi multilayers on Al2O3, but they did not com- atoms. Growth of the crystalline phase may proceed upward ment on the orientation of the multilayer.49 Rocking curves from the lattice-matched Fe template, or downward from the widths for both films are about 1° wide, indicating a consid- atomically bombarded film surface. If the growth of the crys- erably smaller mosaic than for the multilayers grown on talline silicide phase proceeds downward from the film sur- glass. SAXS data for the multilayers on single-crystal sub- face, one might expect to see some crystalline silicide in the strates are comparable to the data for films grown on glass. high-resolution TEM image for the tSi 20 Å film Fig. The films grown on MgO are the only purely 001 - 7 b . The lack of any evidence for crystalline silicide in this textured Fe/Si multilayers produced by IBS so far. Dekoster image suggest that the crystallization proceeds upward from et al. have grown epitaxial Fe/FeSi multilayers on MgO 001 the iron/silicide interface, not downward from the film sur- by MBE, but they do not present any x-ray-diffraction data face. or magnetization curves.34 Magnetization curves of films It is difficult to determine how realistic this model for grown on single-crystal substrates not shown are qualita- growth of the crystalline silicide is since the Fe/Fe-Si and tively similar to those grown on glass or oxidized Si sub- Si-Fe/Fe interfaces appear identical in Fig. 7 b . In contrast, strates. The only differences are that the saturation fields are the Mo/Si and Si/Mo interfaces in Mo/Si multilayers appear 53 STRUCTURE AND MAGNETISM OF Fe/Si MULTILAYERS . . . 5527 quite different from one another.13,14 In the Mo/Si multilay- ers, an amorphous MoSi 2 region appears which is thicker at the Mo/Si interface than at the Si/Mo interface. Detailed TEM studies of multilayers with tSi larger than 20 Å may help to answer whether amorphous silicides can occur in IBS-grown Fe/Si multilayers. Using the B2 phase lattice constant reported by the Zu¨r- ich group,51 we can estimate the expected bilayer period of a nominal Fe/Si multilayer in which Fe atoms diffuse into the Si layer up to a 1:1 stoichiometry. The spacing between the Fe and FexSi1 x layers is taken as the average of the inter- planar spacings of the two materials. The result of this rough calculation is that an Fe 40 Å/Si 14 Å multilayer which interdiffuses up to the 1:1 stoichiometry should form a Fe 33.2 Å/FeSi 16.3 Å multilayer with a bilayer period of 49.4 Å. The missing bilayer period predicted from this model is 4.6 Å, in the middle of values on the x axis of Fig. 4. One can also calculate the expected magnetic moment reduction assuming that Fe atoms in the silicide layer have no moment and those in the Fe layer have their full moment. Under this assumption a calculation predicts 8.2 Å of missing Fe mo- ment, slightly lower than indicated in Fig. 4. This calculation neglects the possibility that some Fe atoms in the Fe layer with Si near neighbors may have reduced magnetic mo- ments. In the discussion above the possibility has not been men- tioned that the missing bilayer period and magnetic moment are due to an inaccurate thickness calibration. This explana- tion is contradicted by magnetization and x-ray-diffraction measurements on Fe/Ge multilayers, where measured mag- netic moments and bilayer periods are in much closer agree- ment with nominal values than for Fe/Si.50 The improved agreement in the case of Fe/Ge multilayers suggests that in- terdiffusion is less important in multilayers with Ge spacer layers than in multilayers with Si spacers. The main point is that the formation of the B2 silicide does qualitatively explain the bilayer period reduction ob- served in the Fe/Si multilayers. The underlying reason for the bilayer period reduction is that the silicide which forms is denser than both Fe and Si. This situation is similar to that observed in other metal/Si multilayers10,14 except that in the other multilayers the silicide remains amorphous. Confirmation that the spacer layer phase has the B2 or DO3 structure is important for understanding the coupling mechanism in these compounds. Both the B2 and DO3 phases are known to be metallic for some ranges of composition.15,16 Thus the present results and those of other workers5,34 suggest that Fe/Si is really a metal/metal multilayer. The origin of the interlayer coupling is then likely to be described by the same theories as describe coupling in Co/Cu and Fe/Cr multilayers.40,41 Fe/Si multilayers may FIG. 13. High-angle x-ray-diffraction spectra from Fe/Si multi- therefore not be a good test case for theories which model layers grown on single-crystal substrates. a Data for a Fe 40 Å/Si interlayer exchange coupling across insulators.8,9 14 Å 60 multilayer grown on MgO 001 . The Fe 002 peak is In the discussion above the possibility has been neglected shown with 5 satellites centered at 64.77°. b Data for a Fe 40 that the amorphous spacer layer in the thick-Si films may Å/Si 14 Å 46 multilayer grown on Al2O3 02¯11 . Visible in the also be metallic. If both the thick amorphous spacers and the spectrum are the Al2O3 02¯11 peak at 37.79° and the Fe 011 peak thin crystalline spacers are metallic silicides, then it must be centered at 44.99° with its 4 satellites. c scans plotted on a the crystallinity that is the essential feature for the existence logarithmic scale for the MgO and Fe 110 peaks of the Fe 40 Å/Si 14 Å 60 multilayer grown on MgO. The Fe 100 direction is of AF interlayer coupling. Up to now there have been no parallel to the MgO 110 , as expected, but a small amount of ma- reports of AF coupling across amorphous metallic spacer terial with a secondary orientation is also visible. layers. Toscano et al. have reported AF coupling across 5528 A. CHAIKEN, R. P. MICHEL, AND M. A. WALL 53 amorphous silicon spacer layers.4 These Fe/a-Si/Fe trilayers position modulation along the growth direction is maintained were prepared at low temperature so as to suppress as evidenced by SAXS measurements. interdiffusion.4 The character of AF coupling in the a-Si There are two surprising results from this study. One is spacer trilayers is likely quite different than in the multilay- that the films grow on glass with a mixed 011 and 001 ers described in this study, where substrate heating increases texture near nominal RT and with a pure 011 texture at the strength of coupling. higher and lower temperatures. The other surprise is that the At the moment there is no direct evidence regarding the strength of the interlayer coupling depends strongly on the metallic or insulating nature of the amorphous spacer layers number of bilayer periods in films with thin Fe layers. This found in the Fe 30 Å/Si 20 Å multilayers. Temperature- latter result is explained on the basis of substrate surface dependent current-in-plane resistivity measurements suggest roughness. that both crystalline and amorphous spacer layers in Fe/Si Unraveling the behavior of the Fe/Si multilayer system multilayers are poorly conducting.50 Fe70Si30 and Fe65Si35 has proven to be a considerably more complex task than amorphous alloys have a temperature-independent resistivity, understanding the Fe/Cr or Co/Cu multilayer systems. The suggesting nonmetallic behavior.52 Overall the evidence sug- reason is that compound formation at the Fe/Si interface is gests that the amorphous spacer layers in Fe 30 Å/Si 20 Å crucial to understanding the AF interlayer coupling. Identifi- multilayers are not metallic, but spectroscopic measurements cation of possibly disordered phases in the spacer layer of a like soft x-ray fluorescence24 are needed for confirmation. multilayer continues to be an experimental challenge. The interesting question as to whether there can be AF inter- Mounting evidence suggests that the spacer layer in the AF- layer coupling across an amorphous metal spacer layer must coupled Fe/Si multilayers is metallic and crystalline and that then be left for another study. the Fe/Si interlayer coupling therefore has the same origin as in metal/metal multilayers. V. CONCLUSIONS ACKNOWLEDGMENTS An extensive study of the growth of Fe/Si multilayers by ion-beam sputtering has been performed. The crystalline We would like to thank P.E.A. Turchi, T.W. Barbee, Jr., quality of the films is better when they are grown with thick T.P. Weihs, E.E. Fullerton, Y. Huai, and E.C. Honea for help- Fe layers, with thin Si layers, at high temperature, and on ful discussions, and B.H. O'Dell and S. Torres for technical single-crystal substrates. Improved growth conditions lead to assistance. Further thanks go to C.-T. Wang of Stanford for higher saturation fields and lower remanence in magnetiza- the four-circle x-ray diffractometry and to Sandia National tion curves. Measured bilayer periods are consistently shorter Lab for use of their electron microscope for HREM work. in these multilayers than the nominal value, suggesting for- Part of this work was performed under the auspices of the mation of a dense silicide phase in the spacer layer. Despite U.S. Department of Energy by LLNL under Contract No. considerable interdiffusion in the multilayers, a strong com- W-7405-ENG-48. *Electronic address: chaiken@llnl.gov 13 D.G. Stearns, R.S. Rosen, and S.P. Vernon, J. Vac. Sci. Technol. A Electronic address: michel@cmsgee.llnl.gov 9, 2662 1991 . Electronic address: mark_wall@internetqm.llnl.gov 14 K. Holloway, K.B. Do, and R. Sinclair, J. Appl. Phys. 65, 474 1 S.T. Ruggiero, T.W. Barbee, Jr., and M.R. Beasley, Phys. Rev. B 1989 . 26, 4894 1982 . 15 H. von Ka¨nel, K.A. Ma¨der, E. Mu¨ller, N. Onda, and H. Sir- 2 T.W. Barbee, S. Mrowka, and M.C. Hettrick, Appl. Opt. 24, 883 ringhaus, Phys. Rev. B 45, 13 807 1992 . 1985 . 16 J. Kudrnovsky´, N.E. Christensen, and O.K. Andersen, Phys. Rev. 3 L.C. Wilson, Ph.D. thesis, Stanford University, 1993. B 43, 5924 1991 . 4 S. Toscano, B. Briner, H. Hopster, and M. Landolt, J. Magn. 17 Made by Commonwealth Scientific Corporation, 500 Pendleton Magn. Mater. 114, L6 1992 . St., Alexandria, VA 22314. 5 E.E. Fullerton, J.E. Mattson, S.R. Lee, C.H. Sowers, Y.Y. Huang, 18 MKS Instruments, 6 Shattuck Rd., Andover, MA 01810. G. Felcher, S.D. Bader, and F.T. Parker, J. Appl. Phys. 73, 6335 19 MDC, 23842 Cabot Blvd., Hayward, CA 94545. 1993 . 20 A. Chaiken, E.C. Honea, W.S. Rupprecht, S. Torres, and R.P. 6 S. Demokritov, J.A. Wolf, and P. Gru¨nberg, Europhys. Lett. 15, Michel, Rev. Sci. Instrum. 65, 3870 1994 . 881 1991 . 21 Keithley Instruments, 440 Myles Standish Blvd., Taunton, MA 7 D.H. Mosca, F. Petroff, A. Fert, and P.A. Schroeder, J. Magn. 02780. Magn. Mater. 94, L1 1991 . 22 A.I. Kingon, O. Auciello, M.S. Ameen, S.H. Rou, and A.R. 8 P. Bruno, Phys. Rev. B 49, 13 231 1994 . Krauss, Appl. Phys. Lett. 55, 301 1989 . 9 Z.-P. Shi, Rajiv R.P. Singh, and B.M. Klein, Europhys. Lett. 29, 23 R.F.C. Farrow, G.R. Harp, R.F. Marks, and T.A. Rabedeau, J. 585 1995 . Cryst. Growth 133, 47 1993 . 10 E.E. Fullerton, J. Pearson, C.H. Sowers, S.D. Bader, X.Z. Wu, 24 J. A. Carlisle, A. Chaiken, R. P. Michel, L. J. Terminello, J. J. Jia, and S.K. Sinha, Phys. Rev. B 48, 17 432 1993 . T. A. Calcott, and D. L. Ederer unpublished 11 H. Miura, E. Ma, and C.V. Thompson, J. Appl. Phys. 70, 4287 25 C.L. Foiles, M.R. Franklin, and R. Loloee, Phys. Rev. B 50, 1991 . 16 070 1994 . 12 W.H. Wang, H.Y. Bai, and W.K. Wang, Mater. Sci. Eng. 179A, 26 R.P. Michel, A. Chaiken, J.W. Dykes, J.F. Ankner, and H.J. Kaiser 229 1994 . unpublished . 53 STRUCTURE AND MAGNETISM OF Fe/Si MULTILAYERS . . . 5529 27 K. Inomata, K. Yusu, and Y. Saito, Jpn. J. Appl. Phys. 33, L1670 39 H.-J. Stock, U. Kleineberg, B. Heidemann, K. Hilgers, A. Kloidt, 1994 . B. Schmiedeskamp, U. Heinzmann, M. Krumrey, P. Mu¨ller, and 28 E.E. Fullerton, I.K. Schuller, H. Vanderstraeten, and Y. Bruynser- F. Scholze, Appl. Phys. A 58, 371 1994 . aede, Phys. Rev. B 45, 9292 1992 . 40 P. Bruno and C. Chappert, Phys. Rev. B 46, 261 1992 . 29 C. Dufour, A. Bruson, G. Marchal, B. George, and Ph. Mangin, J. 41 M.D. Stiles, Phys. Rev. B 48, 7238 1993 . Magn. Magn. Mater. 93, 1991. 42 R. Nakatani, T. Dei, and Y. Sugita, J. Appl. Phys. 73, 6375 1993 . 30 J.F. Ankner, C.F. Majkrzak, and H. Homma, J. Appl. Phys. 73 43 A. Gupta, G. Principi, R. Gupta, A. Maddalena, F. Caccavale, and 6436 1993 . C. Tosello, Phys. Rev. B 50, 2833 1994 . 31 C.L. Foiles, M.R. Franklin, and R. Loloee, in Magnetic Ultrathin 44 E.E. Fullerton, J.E. Mattson, S.R. Lee, and C.H. Sowers, J. Magn. Films, Multlayers and Surfaces, edited by E. E. Marinero, B. Magn. Mater. 117, L301 1992 . Heinrich, W. F. Egelhoff, Jr., A. Fert, H. Fujimori, G. 45 H.A.M. van den Berg and G. Rupp, IEEE Trans. Magn. 30, 809 Gu¨ntherodt, and R. L. White, MRS Symposia Proceedings No. 1994 . 382 Materials Research Society, Pittsburgh, 1995 . 46 32 Binary Alloy Phase Diagrams, edited by T.B. Massalski ASM M.W. Grinstaff, M.B. Salamon, and K.S. Suslick, Phys. Rev. B International, Materials Park, OH, 1990 . 48, 269 1993 . 47 33 N. Onda, H. Sirringhaus, E. Muller, and H. von Ka¨nel, J. Cryst. S.S.P. Parkin, R.F. Marks, R.F.C. Farrow, and G.R. Harp, Phys. Growth 12, 634 1993 . Rev. B 46, 9262 1992 . 34 48 J. Dekoster, A. Vantomme, S. Degroote, R. Moons, and G. N. Metoki, M. Hofelich, T. Zeidler, and T. Muhge, J. Magn. Langouche, in Magnetic Ultrathin Films, Multilayers and Sur- Magn. Mater. 121, 137 1993 . faces Ref. 31 . 49 J.E. Mattson, E.E. Fullerton, S. Kumar, S.R. Lee, C.H. Sowers, 35 J.E. Mattson, S. Kumar, E.E. Fullerton, S.R. Lee, C.H. Sowers, M. Grimsditch, S.D. Bader, and F.T. Parker, J. Appl. Phys. 75, M. Grimsditch, S.D. Bader, and F.T. Parker, Phys. Rev. Lett. 71, 6169 1994 . 185 1993 . 50 R.P. Michel and A. Chaiken unpublished . 36 E.E. Fullerton private communication . 51 K.A. Ma¨der, H. von Ka¨nel, and A. Baldereschi, Phys. Rev. B 48, 37 A.P. Payne and B.M. Clemens, Phys. Rev. B 47, 2289 1993 . 4364 1993 . 38 D. Altbir, M. Kiwi, R. Rami´rez, and I.K. Schuller, J. Magn. 52 G. Marchal, Ph. Mangin, and Chr. Janot, Solid State Commun. Magn. Mater. 149, L246 1995 . 18, 739 1976 .