PHYSICAL REVIEW B VOLUME 53, NUMBER 4 15 JANUARY 1996-II Interfacial mixing of ultrathin Cr films grown on an Fe whisker D. Venus* and B. Heinrich Department of Physics, Simon Fraser University, Burnaby, Canada V5A 1S6 Received 11 September 1995 12-ML Cr films grown on an Fe whisker have been studied using angle-resolved Auger electron forward scattering as a function of the substrate temperature at which the Cr was deposited. The angular scans of the peak-to-peak Cr Auger intensity show pronounced peaks due to forward scattering, which indicate significant intermixing of Cr atoms into the second layer for films grown at 100, 180, and 246 °C, and into the third layer for the film grown at 296 °C. Calculations based on a single-scattering theory show that half of the Cr deposited at 296 °C is below the surface. Since previous studies of exchange coupling through Cr films used samples grown at this elevated temperature, the alloying may be related to the earlier finding that the exchange coupling through Cr 001 is much weaker than, and of opposite phase to, that predicted by first-principles calculations. Ultrathin magnetic metallic structures are studied inten- has been acknowledged so far. Layer-by-layer growth of the sively both to provide insight into the basic aspects of low- Cr on the whisker was observed only in a small range of dimensional magnetism, and for the development of different substrate temperatures near 300 °C.3,4,10 However, the atomi- materials based on nanostructures. One thrust of this research cally smooth growth may be chemically rough. Interface al- has been to investigate the magnetic exchange coupling be- loying due to an atomic exchange mechanism has been seen tween two ferromagnets separated by a nonferromagnetic for Fe growth on Ag 001 and Cu 001 , even at room film. In this regard, the Fe whisker/Cr/Fe 001 system has temperature.11,12 As was demonstrated using Mo¨ssbauer played an important role, particularly as the 001 surface of spectroscopy,11 the atom exchange mechanism is an asym- the whisker provides a nearly perfect, atomically smooth metric effect which favors a configuration where the atom template for the epitaxial growth of the Cr layer. Studies type with the lower bulk melting temperature is at the sur- using secondary electron microscopy with polarization face. In the whisker/Cr/Fe studies, this will favor Fe at the analysis SEMPA and the magneto-optical Kerr effect surface. Furthermore, the growth of the Fe film on top of the MOKE Refs. 1 and 2 have revealed that the exchange Cr was performed at a much lower temperature 30 °C than coupling through the Cr film oscillates as the Cr film thick- the Cr growth on the whisker. This will act to decrease fur- ness is increased. This oscillation has two characteristic ther the interchange of atoms at the second interface. Taken wavelengths: 2.11 and 12 ML. Quantitative studies of the together, these effects might create two inequivalent inter- exchange coupling have been subsequently performed using faces, and alter the phase of the overall coupling compared to Brillouin light scattering BLS .3,4 The strength of the ex- two equivalent interfaces. change coupling through the Cr 001 spacer was found to be In order to investigate the mixing at the interface of a Cr extremely sensitive to small deviations in growth conditions, film grown on an Fe whisker, the angular distribution of and was reproducible only in those samples which exhibited Auger electrons emitted by the Cr atoms was studied. Be- a nearly perfect layer-by-layer growth. Short-wavelength os- cause of the strong forward scattering of electrons with a cillations in the exchange coupling were confirmed, with the kinetic energy of several hundreds of eV, the number of Au- coupling strength Jmax 1 erg/cm2. The phase of the short- ger electrons emitted within a small solid angle is markedly wavelength component, in agreement with the SEMPA peaked when the emission direction lies along the line from results,5,6 was ferromagnetic for an even number of Cr ML, the emitting atom to a nearest neighbor or next-nearest- and antiferromagnetic for an odd number of Cr layers, for neighbor atom.13 These forward focusing effects are most thicknesses between 5 and 13 ML. pronounced when caused by a single scattering event, that is, The presence of the short-wavelength oscillatory compo- when only one atom intervenes between the emitting atom nent in the exchange coupling is in agreement with theoreti- and the detector.14 The technique is therefore ideally suited cal predictions.7,9 Bulk Cr has a spin-density wave with a to the study of the intermixing of monolayer and submono- wavelength of 2.11 ML, which should be pinned by the layer films with a crystalline substrate. Figure 1 a demon- strong antiferromagnetic exchange coupling with Fe at both strates the qualitative expectation for Cr atoms deposited on of the interfaces. However, the phase of the short-wavelength the 001 face of bcc Fe. The substrate is rotated by 45° oscillations is predicted to be opposite to that observed. Fur- about the normal vector n, so that a 110 plane lies in the thermore, the strength of the exchange coupling obtained plane of emission formed by n and the detected outgoing from first-principles calculations,8,9 Jmax 30 erg/cm2, is Auger electron beam k. The rotation axis for lies normal to much greater than the experimental results. This represents a the plane of emission, and is measured between n and k. A significant disagreement between experiment and theory. Cr atom in the top vacuum layer has no atoms intervening These discrepancies may indicate that the detailed forma- between itself and the detector, and should therefore produce tion of the Fe whisker/Cr/Fe sample is more complex than an isotropic Auger intensity. A Cr atom in the second layer 0163-1829/96/53 4 /1733 4 /$06.00 53 R1733 © 1996 The American Physical Society R1734 D. VENUS AND B. HEINRICH 53 should produce a forward scattering peak at 1 54.7°, whereas one in the third layer should produce forward scat- tering peaks along both the normal, 2 0°, and along 1. This last peak will, however, have a significant contribution from multiple scattering, as two atoms lie between the emit- ting atom and the detector.14 A Cr atom in the fourth layer would contribute to the peaks at 2 and 3 25.2° through single scattering, and to the peak at 1 via multiple scatter- ing.It is evident that Auger intensity scans in the 45° azi- muth will indicate the mixing of Cr into the Fe whisker. There is, however, a complication with the measurement near 0° arising from the forward scattering of the exciting primary beam q.15 In the apparatus used for these experi- ments, q makes a fixed angle of 54° with the emission direction k. As can be seen in Fig. 1 b , when 45° and 0°, q coincides almost precisely with the line connecting a first- and second-layer atom. Under these conditions, the incoming beam undergoes forward scattering for a small range of about 0°, and Cr atoms in the second layer expe- rience an increased flux of exciting primary electrons, thus producing proportionately more Auger electrons. This results in an increase in Auger intensity near 0°, which may be misinterpreted as due to emission from atoms in deeper lay- ers. The solution to this geometric difficulty is to measure the FIG. 1. The experimental geometry. a The plane of emission includes angular distribution at two azimuthal angles. Measurements the surface normal n and the direction at which Auger electrons are detected. at 45° display the peak at 1 54.7° with maximum sen- When 45°, this is a 110 plane of the whisker. The axis of rotation lies sitivity, and measurements at another azimuth 34.2° normal to the page, and 1 ­ 3 are emission angles where forward scattering which does not meet the forward focusing condition for the increases the emission from atoms in the top four atomic layers. b The bcc incoming beam, give an unambiguous measure of any peak structure of the whisker is shown in perspective. The incident primary elec- tron beam q makes a fixed angle with the emission direction k. at normal emission. Fe whiskers are small samples. A typical whisker is a These data demonstrate the crystalline origin of the intensity rectangular bar about 100 100 10000 m3, with its long variations in Fig. 2 a , and also characterize the angular pro- axis oriented in the 001 direction, and displaying 001 file of the measurement technique. facets. For this study an uncommon ``blade''-shaped whisker Having demonstrated the ability to perform Auger was selected which had a 10-mm length, a 500- m width, forward-scattering measurements on an Fe whisker, a series and a thickness of about 100 m. The sample was mounted of12 -ML Cr films were grown on the whisker at different on an polycrystalline Mo holder using a spring clip. The temperatures, and the angular dependence of the Fe and Cr cleaned blade face produced an excellent RHEED pattern Auger lines were measured for each. The 12-ML coverage consisting of pointlike diffracted beams without appreciable was chosen to minimize the occurrence of Cr deposition on streaking. The Auger spectra were recorded using a PHI 10- Cr. The Cr growth was monitored using a RHEED primary 360 180° hemispherical analyzer configured to give an esti- beam scattered at the second anti-Bragg condition for the mated resolution of 2° and 500- m diameter. The spatial substrate. Cr growth was stopped when the specular beam resolution was checked by measuring the Auger signals of intensity first reached a minimum, and a picture of the the Fe and the Mo holder as the sample was translated. The RHEED pattern was recorded. Four growth temperatures Fe 702 eV /Mo 186 eV peak-to-peak Auger ratio changed were investigated: 100, 180, 246, and 296 °C. These tem- by more than two orders of magnitude across the edge of the peratures span the range where the RHEED oscillations for sample, and the Fe signal showed a clear central plateau layer growth show a strongly damped, sinusoidal oscillation region. It was concluded that by positioning near the center characteristic of rough growth 100 °C , to where the oscil- of the whisker, the Auger signal from Cr deposited on the lations have a sharply peaked, parabolic shape characteristic Mo holder would constitute 1% of the Auger signal from of very smooth layer-by-layer growth 296 °C . It is impor- an equivalent thickness of Cr on the Fe whisker. The Fe tant to note that at all the above substrate temperatures the measurements were collected simultaneously with the Cr first RHEED intensity oscillation shows a large and sharply measurements for all of the samples, and were found to be cusped peak, indicating that the first ML of Cr is smooth highly reproducible. Typical peak-to-peak Fe 702 eV Auger with negligible pileup of the Cr atoms. The RHEED patterns data are shown in Fig. 2 a , and reveal features at the ex- at 12 ML for the sample growths showed long streaks in the pected angles. The relative sizes of these features are difficult scattering direction, because of the high density of atomic to interpret qualitatively because of the important effects of steps at this coverage. There was clear lateral splitting of the multiple scattering in the thick Fe sample. Figure 2 b pre- streak for 100 and 180 °C growths, indicating a mean sepa- sents measurements of the peak-to-peak Cr 529 eV Auger ration of the nucleation centers for the growth of approxi- signal from a nominal 1-ML film on the Mo sample holder. mately 100 and 200 Å, respectively. This is in agreement 53 INTERFACIAL MIXING OF ULTRATHIN Cr FILMS GROWN . . . R1735 FIG. 3. Cr 529 eV Auger intensity as a function of for 1 FIG. 2. a Peak-to-peak Fe 702 eV Auger intensity as a function of 2-ML Cr films grown at the indicated temperatures, for two values of azimuthal rotation. for the whisker. The angles 1­ 3 are those indicated in 1 a . b Peak-to- The lines are fitted using single-scattering theory as described in the text. peak Cr 529 eV Auger intensity as a function of for about 1 ML Cr on the The dotted lines show the variation in the fit in one case, when the layer Mo holder. abundances are varied by 0.04. with the findings of STM studies of this system.10 At 246 and SSC created by Friedman and Fadley17 for single elastic scat- 296 °C the lateral splitting was no longer evident, the streak- tering from an array of atoms. The scattering events are ing became progressively less pronounced, and the patterns treated using scattering phase shifts generated by the com- were very similar to those described in earlier studies.3,4 puter program FEFF-3 written by Rehr, Albers, and Mustre de The angular dependence of the Cr peak-to-peak Auger Leon.18 Single-scattering calculations should be entirely ad- signal is plotted in Fig. 3. The data for each successive tem- equate for these studies because i the absence of a peak at perature have been offset by an additional three units along 25.2° suggests that there is little Cr in the fourth layer or the y axis for clarity. All measurements were made at or deeper; ii in emission from the top three layers, there is close to room temperature. Integrating over , the average only one chainlike alignment of atoms where multiple scat- Cr/Fe Auger ratio is a factor of 1.3 times greater than that tering will be important; and iii this single chain alignment expected for 12-ML Cr on semi-infinite Fe Ref. 16 -this is for emission at 54.7° by atoms in the third layer, and small discrepancy is likely due to the fact that the angular previous investigations into the effect of multiple scattering scan is along a direction where the Cr Auger intensity is have shown how an approximate compensation can be made. concentrated more effectively by single-scattering events The compensation is based on calculations of the forward than the Fe intensity is concentrated by multiple scattering. scattering along the close-packed direction of the 3d metal Ignoring for the moment the solid lines, a number of quali- Cu,14 where the nearest-neighbor distance and scattering tative results are immediately evident. The data for 45° phase shifts at the energy of the Auger electrons are very on the left side of the figure show a clear peak in Auger close to those of bcc Fe. The calculations for Cu show that intensity near 54.7°, indicating that a significant number of the effect of multiple scattering along such a chain is to Cr atoms lie in the second layer or deeper in these 12-ML reduce the contribution of the third-layer atoms to 60% of films. Furthermore, the size of this peak relative to the inten- the contribution calculated in the single-scattering approxi- sity at 38° increases monotonically with increasing mation. Detailed results for emission from Cr in bcc Fe when growth temperature, showing that more Cr atoms are moving 45° show that scaling the contribution to the peak at below the top layer. At all growth temperatures, the data 57.4° from a third-layer atom by 0.60 makes it very point at 0° also shows a greater intensity than at 38°. nearly equal to that calculated for a second-layer atom-that This is primarily due to forward scattering of the primary is, second- and third-layer atoms are indistinguishable along beam onto the Cr atoms below the surface. This is evident this emission direction. The data at 45° are therefore fit to because a the peak is absent at the lower growth tempera- a linear combination of angular distributions calculated for tures for the data at 34.2°, where the primary beam effect the first- and second-layer atoms, giving a measure of the is absent, and b because the peaks at 54.7° and 0° in- relative Cr abundance at the surface and deeper. For the data crease in tandem. The data on the right side of Fig. 3 show a at 34.2°, 0°, there is no ambiguity due to multiple peak at normal emission due to forward scattering once the scattering, but the calculated angular profiles for both first- growth temperature reaches 296 °C, and it is evident that this and second-layer atoms are flat and indistinguishable- 12-ML Cr film has significant mixing into the third layer of neither shows forward scattering. These data are therefore fit the Fe substrate. to a linear combination of the angular distributions for a These conclusions can be made more quantitative using top-layer and third-layer atom, giving a measure of the rela- single-scattering calculations of Cr Auger electron forward tive abundances in the third layer and above the third layer. scattering. These calculations used the computer program Combining these estimates gives the relative abundances in R1736 D. VENUS AND B. HEINRICH 53 film was prepared, and Auger electron spectroscopy at nor- mal emission showed that the Fe substrate was covered very well. The Fe signal was diminished by a factor of 8.8 com- pared to that of a clean whisker. Assuming that intermixing occurs only near the interface, this decrease was due to two effects: the attenuation by the Cr on top of the Fe a factor16 of 3.7 , and the removal of the forward-scattering peak for normal emission Fig. 2 a shows a factor of 2.5 . This yields an overall reduction of 9.3 times, in approximate agreement with the measurement. In conclusion, interfacial alloying of Cr grown on an Fe whisker is significant. For films grown at 296 °C, where ex- cellent layer-by-layer growth of Cr is found, about half of the first ML of Cr is below the surface layer, but the mixing is likely confined to three layers. Theoretical calculations for Fe/Cr superlattices which have interfaces formed by 2 ML of FIG. 4. The relative abundances of Cr in the top three layers, as found an ordered alloy Fe\Fe 75% -Cr 25% ; Cr 75% -Fe 25% \ from the fit to the data shown in Fig. 3. The finding of a threshold level of Cr Ref. 8 predict that the strength of the exchange cou- Cr in the third layer at the three lowest temperatures is not believed to be pling across the Fe\Cr interface is greatly reduced, and that significant. the Cr in the alloyed atomic layer closest to the pure Fe is all three layers, and results in the calculated solid lines in antiferromagnetically coupled to it. Alloying changes the Fig. 3. An estimate of the uncertainty in the relative abun- number of atomic layers containing Cr by one, and the phase dances is 0.04. This magnitude of change generates the fits of the coupling, if plotted against equivalent ML of deposited shown by the dotted lines for the data at 100 °C, 45°. The Cr, becomes reversed compared to an abrupt interface. The fitted relative abundances are shown in Fig. 4 as a function level of interface alloying found at the Fe whisker/Cr inter- of the growth temperature of the film. This plot clearly face is comparable to that used in these calculations, and shows the diffusion of the Cr into the Fe whisker as the therefore could be responsible for the measured phase of the temperature increases. Significant movement of Cr into the short-wavelength oscillations in Fe whisker/Cr/Fe samples. third layer is first seen for growth at 296 °C. Although inclu- Quantitative BLS and MOKE studies show that the interface sion of a small contribution from the third layer is indicated mixing very strongly affects the strength of the bilinear ex- by the fit at lower temperatures, it does not vary systemati- change coupling.19 Studies are under way to find out whether cally with temperature, and is close to the estimated fitting the degree of the interface mixing can be limited to the point error. At and below 246 °C, the mixing is likely confined to that the measured phase of the short-wavelength oscillations the top two layers. At 296 °C, roughly half the Cr is in the are reversed and brought into agreement with first-principles second and third layers. Evidently, smooth film growth is calculations. accompanied by strong intermixing of Cr and Fe at the in- We are grateful to C. S. Fadley for providing the com- terface, and noticeable intermixing occurs even at 100 °C. puter program SSC and to C. M. Schneider for help in imple- The present data do not address directly the question of menting it. This work was performed with the financial sup- mixing in a thicker film, but there is evidence that mixing port of the Natural Sciences and Research Council of beyond the third layer will not be important. An 11-ML Cr Canada. *Permanent address: Dept. of Physics and Astronomy, McMaster 8 D. Stoeffler and F. Gautier, Phys. Rev. B 44, 10 389 1991 . University, Hamilton, Canada L8S 4M1. 9 M. van Schilfgaarde and F. Herman, Phys. Rev. Lett. 71, 1923 1 J. Unguris et al., Phys. Rev. Lett. 67, 140 1991 ; 69, 1125 1993 . 1992 . 10 J. A. Stroscio et al., J. Vac. Sci. Technol. B 12, 1789 1994 . 2 S. T. Purcell et al., Phys. Rev. Lett. 67, 903 1991 . 11 P. J. Schurer et al., Phys. Rev. B 51, 2506 1995 . 3 B. Heinrich et al., in Magnetic Ultrathin Films: Multilayers and 12 D. D. Chamblis et al., J. Vac. Sci. Technol. A 10, 1993 1991 . Surfaces/Interfaces and Characterization, edited by B. T. Jonker 13 W. F. Egelhoff, Jr. in Ultrathin Magnetic Structures I, edited by J. et al., MRS Symposia Proceedings No. 313 Materials Research A. C. Bland and B. Heinrich Springer, Berlin, 1994 . Society, Pittsburgh, 1993 , p. 119. 14 4 A. P. Kaduwalla et al., J. Electron Spectrosc. Relat. Phenom. 57, B. Heinrich and J. F. Cochran, Adv. Phys. 42, 523 1993 . 5 J. Unguris et al., in Magnetism and Structure in Systems of Re- 223 1991 . 15 duced Dimension, Vol. 39 of NATO Advanced Study Institute, S. Chambers, Adv. Phys. 40, 357 1991 . 16 Series B: Physics, edited by R. F. C. Farrow, B. Dieny, M. Do- For Fe, the inelastic mean-free path is 12 Å. Practical Surface nath, A. Fert, and B. D. Hermsmeir Plenum, New York, 1993 , Analysis, 2nd ed., edited by D. Briggs and M. P. Seah Wiley, p. 101. New York, 1990 , p. 209. 6 17 B. Heinrich et al., in Magnetism and Structure in Systems of Re- D. J. Friedman and C. S. Fadley, J. Electron Spectrosc. Relat. duced Dimension Ref. 5 , p. 175. Phenom. 51, 689 1990 . 7 D. Stoeffler and F. Gautier, in Magnetism and Structure in Sys- 18 J. J. Rehr et al., Physica B 158, 417 1989 . tems of Reduced Dimension Ref. 5 , p. 411. 19 B. Heinrich et al. unpublished and unpublished .