PHYSICAL REVIEW B VOLUME 57, NUMBER 1 1 JANUARY 1998-I Magnetic phase transition in epitaxial Ni1 xFex alloy thin films J. W. Freeland,* I. L. Grigorov, and J. C. Walker Department of Physics and Astronomy, The Johns Hopkins University, 3400 N. Charles Street, Baltimore, Maryland 21218 Received 22 August 1997 An investigation of fcc Ni1 xFex thin films utilizing a combination of Mo¨ssbauer spectroscopy and magne- tometry presents clear evidence for the emergence of a low-spin antiferromagnetic state in the region of high Fe concentration, which coexists with a high-spin ferromagnetic state. The emergence of this state is observed by magnetometry as a drop in the saturation moment while Mo¨ssbauer spectroscopy reveals the appearance of a low-spin antiferromagnetic state. Evidence of superparamagnetic behavior indicates that the portion of the sample in the high-spin state may exist as clusters in a matrix of the low-spin state. S0163-1829 98 05302-8 Ever since the discovery of the Invar effect in fcc Ni-Fe posited on the mica. The sample may then be removed from alloys,1 these alloys have been a subject of intensive study.2,3 the mica substrate using a small amount of ultrapure water. However, study of the rich magnetic behavior of these alloys For the 100 orientation the 1500 Å Cu base is grown di- in the region of high Fe concentration (x 0.6) has been rectly on cleaved NaCl 100 at a substrate temperature of complicated by the presence of a structural phase transition 350 °C. to the energetically favorable bcc phase.3 In order to better The multilayer structures have the general form understand the role of the Fe in the fcc Ni 57 1 xFex alloys and Ni1 x Fex(4 ML)/Cu(100 Å) N , where the 4-monolayer its relation to the magnetic order, it is necessary to explore ML film thickness was 7.2 Å for the 100 films and 8.4 Å the order in the region of high Fe concentration while sup- for the 111 films. The number of repetitions N, was 10 for pressing the structural phase transition. Using molecular the 111 orientation and 4 for the 100 orientation. The beam epitaxy MBE , this can be accomplished since fcc concentrations of Ni and Fe were varied by changing the Ni Ni-Fe has a close lattice match to Cu and can be prepared in cell temperature and thus the Ni growth rates and were con- the fcc phase by pseudomorphic growth on Cu substrates firmed after film growth by electron microprobe analysis. To over the entire Fe concentration range.4­7 Studies using in reduce intermixing with Cu at the interface, the substrate situ magneto-optic Kerr effect MOKE have found that thin temperature was held at 300 K during growth using a l-N2 pseudomorphic Ni1 xFex films on Cu 100 present a ferro- cooling finger. Since we are interested in studying indepen- magnetic signal over the entire concentration range,4,5 in dent magnetic layers, the 4 ML Ni1 xFex layers were sepa- contrast to the behavior of bulk alloys. The behavior of the rated by a 100 Å Cu layer, which has been demonstrated Curie temperature as a function of film thickness also pro- previously to be sufficient to magnetically isolate the layers.8 vides an indication of a magnetic transition in the region of The crystallinity and orientation of the Cu substrates were high Fe concentration.6 However, since MOKE is not a mea- verified by in situ reflection high-energy electron diffraction sure of absolute total moment and only sensitive to the pres- RHEED prior to the multilayer growth. RHEED was also ence of a ferromagnetic signature, additional techniques are utilized during the multilayer growth to monitor the sample then necessary to determine the behavior of the total moment crystallinity. It was only possible to monitor one orientation and whether or not magnetic transition involves the emer- during the multilayer deposition due to the presence of the gence of an antiferromagnetic phase. l-N2 cooling finger. However before capping the films the In this study we utilized a combination of Mo¨ssbauer cooling finger was removed and sample was rotated to con- spectroscopy and superconducting quantum interference de- firm that the surface had the expected symmetry fourfold for vice SQUID magnetometry to provide direct evidence for a the 100 and sixfold for the 111 . In addition, the RHEED magnetic phase transition in pseudomorphic fcc Ni1 xFex streak spacing, which is directly related to the in-plane lattice thin films. Our results clearly show the emergence of a low- parameter, was monitored during growth using a K-Space spin antiferromagnetic AFM state in the region of high Fe CCD RHEED camera system.9 In all cases the streak spacing concentration, which coexists with a high-spin ferromagnetic varied by less than 0.05%. This indicates that the Ni1 xFex FM state. layers are growing pseudomorphically on the Cu substrate. All of the samples were prepared by MBE in a Perkin- X-ray diffraction was used as an additional check of the crys- Elmer PHI 430 B MBE system, which has a base pressure of tallinity and to confirm that no bcc phase was present in 1 10 10 Torr. The Cu 111 substrates were evaporated in these films. situ on V-1 quality natural mica sheets at a substrate tem- Figure 1 displays the saturation moment normalized to the perature of 200 °C. The mica sheets were cleaned in ethanol volume of the film as measured by a SQUID magnetometer. for 15 min, air-cleaved just prior to introduction into the As seen in Fig. 1, the samples display a ferromagnetic sig- UHV chamber, and then outgassed at 400 °C for several nature over the entire Fe concentration range. Below an Fe hours prior to growth. To facilitate removal of the sample concentration, x, of 0.6 the behavior is similar to that of the from the mica substrate, which is required for transmission bulk alloys as shown by the solid line in Fig. 1 . Above x Mo¨ssbauer spectroscopy, a 300 Å epilayer of NaCl was de- 0.6, there is a linear continuation of the increase in mo- 0163-1829/98/57 1 /80 4 /$15.00 57 80 © 1998 The American Physical Society 57 BRIEF REPORTS 81 FIG. 1. The saturation magnetic moment normalized to the sample volume vs Fe concentration. The drop in the moment is due FIG. 3. Central feature width vs temperature for the pure Fe to the emergence of an antiferromagnetic state in the region of high (x 1) film. This figure shows the drastic reduction in the central Fe concentration. The solid line is the data for bulk Ni1 xFex alloys feature width indicative of a Nee´l transition around 35 K. The solid from Ref. 3 for comparison. line denotes the central feature width at 300 K for comparison. ment into the Invar region, where a precipitous drop in mo- well fit by high-spin magnetic sextet s with the inclusion of ment is seen in the bulk alloys caused by a transition to an a broad singlet in the region of high Fe concentration. In the antiferromagnetic state.3 Presence of a high moment phase in region of low Fe concentration the two high-spin site fit is this region has been previously observed in both sputtered10 consistent with that observed in bulk random alloy data.10,13 and epitaxial4­6 thin films and is believed to be due to strain The hyperfine field associated with the high-spin state re- induced by the growth process. Expansion of the lattice pa- mains constant around 300 kOe ( 2 rameter has been theoretically predicted to stabilize the alloy B) with a variation of less than 10%. Since the Fe local moment probed by the in the high-spin ferromagnetic state.11,12 Moving to higher Fe hyperfine field is not changing enough to explain the drop in concentrations there is a drop in the moment from a value of the saturation moment, the drop must be associated with the 2.3 B for x 0.65 to a value of 1 B for x 1. The pure emergence of the broad central feature in the region of high Fe (x 1) saturation moment appears to be reproducible and Fe concentration. This feature grows in intensity and can be independent of crystalline orientation as shown by three dif- well correlated with the drop in the saturation moment. To ferent measurements for films with 100 and 111 orienta- determine that this broad central feature is indeed caused by tions. As will be seen the Mo¨ssbauer spectroscopy data, the Fe in an antiferromagnetic state and not due to relaxation drop in the magnetic moment at high Fe concentrations ap- phenomena, the central feature width was studied as a func- pears to be caused by the emergence of a low-spin antiferro- tion of temperature for the pure Fe film (x 1). Figure 3 magnetic state coexisting with the ferromagnetic state. Since shows the dramatic reduction in the width of the central fea- this antiferromagnetic state is invisible to the magnetometry ture with increasing temperature consistent with a Nee´l tran- measurements, as more of the sample becomes AFM, the sition at T magnetic moment drops. The technique of Mo¨ssbauer spec- N 35 K. We find that the width drops from 1.1 mm/sec to a value 0.5 mm/sec as the temperature is raised troscopy is well suited to study the appearance of such a state from 30 to 40 K. The width then remains relatively constant due to its sensitivity to both types of magnetic order. near 0.5 mm/sec above 40 K. This proves conclusively that The low-temperature 13 K Mo¨ssbauer data as a function this feature is associated with an unresolvable low-spin anti- of Fe concentration are shown in Fig. 2. These spectra are ferromagnetic state moment 0.5 B /atom . From a com- parison of the width and isomer shift of the central feature observed for the Ni15Fe85 film at both 13 and 300 K to the pure Fe film, we conclude that the central feature present here is also due to an AFM state. This Nee´l temperature is significantly lower than the value of 65 K measured for fcc Fe precipitates14 and thin films.15 However, since the film thickness is only 4 monolayers, the finite size of the sample can affect the magnetic order and cause a suppression of the transition temperature.16 In addition, the temperature-dependent Mo¨ssbauer study of the Ni15Fe85 and pure Fe films revealed the disappearance of the high-spin state as the temperature was raised to 300 K. This disappearance together with an observed ferromagnetic hysteresis at 300 K provides an indication of superparamag- FIG. 2. Low-temperature 13 K Mo¨ssbauer data as a function netic behavior in this film. For both films we observed irre- of Fe concentration. Note the emergence of the broad central fea- versible behavior of field cooled vs zero-field cooled mag- ture in the region of high Fe concentration that is correlated to the netic moment as a function of temperature, which is a clear drop in the saturation magnetic moment. Such a feature is due to the indication of superparamagnetic behavior. Additional confir- formation of an antiferromagnetic state. mation of this is made by measuring the Mo¨ssbauer spectra 82 BRIEF REPORTS 57 of the pure Fe film with a 5 kG applied field and observing structure. Since we cannot resolve the sextet for the low-spin the appearance of a sextet corresponding to a high-spin state. site, extraction of the quadrupole information is not possible, The applied field suppresses the relaxation of the magnetic but the isomer shifts of the two sites can be compared. If the clusters and causes the appearance of a magnetic site in the volume increases, the s-electron density in the nucleus drops absorption spectra. From both of these measurements we causing the isomer shift to become more positive. For spec- conclude that the high-spin FM state may exist as clusters in tra in the region of high Fe concentration the high-spin FM a matrix of the low-spin AFM state. site always has a more positive isomer shift than the low- This observation of coexisting magnetic states in the spin AFM site. This may indicate that the FM regions have a Ni-Fe alloys is in agreement with studies of bulk alloys in larger volume than the low-spin AFM state. This is consis- the invar region 65% Fe , where there is strong evidence for tent with both experimental observations19,20 and theoretical coexisting AFM and FM order.17,18 In these studies it was predictions12,25 that the lower volume favors the low-spin postulated that there were Fe-rich clusters ordered AFM and state. coupled to a Ni-rich FM matrix. The spectrum of the Information concerning the structure of the pure Fe film Ni15Fe85 film is well described by two high-spin states simi- can also be inferred by comparing our results to those of lar the sites in the Ni63Fe37 film Ni rich and a broad central uncovered fcc Fe 100 films studied in situ by both Mo¨ss- feature corresponding to a low-spin AFM site as seen in the bauer spectroscopy20 and magnetometry.26 For uncovered pure Fe film Fe rich . This argues for Ni-rich and Fe-rich 4-ML fcc Fe 100 films the magnetic order is ferromagnetic regions with different magnetic order. For the case of the while the structure is fct. But for the case of our 4-ML Fe pure Fe film, previous studies of fcc Fe 100 have seen evi- films, the order is predominantly antiferromagnetic. Both dence for coexisting magnetic order,19,20 however the mode studies of the uncovered films indicate that the AFM order is by which the two types of order coexist is not clear. One associated with the fcc structure. This is a good indication possible answer may be found in the complicated structures that the multilayers are stabilizing the 4-ML Fe films in a formed by the competition between structural and magnetic more pure fcc state. Such a conclusion is consistent with an energies. Low-energy electron diffraction21,22 and medium x-ray-absorption study of fcc Fe 100 on Cu 100 , which energy ion scattering23 of fcc Fe 100 on Cu 100 have pre- found that a Cu cap layer relaxes fct Fe into fcc Fe.27 sented clear indications of complex strain waves forming in In summary, our investigation of fcc Ni1 xFex thin films fcc Fe thin films on Cu 100 . gives clear evidence for the emergence of a low-spin antifer- In addition to the magnetic information contained in the romagnetic state in the region of high Fe concentration in Mo¨ssbauer data, there is a great deal of structural informa- coexistence with a high-spin ferromagnetic state. The emer- tion. By comparing the low Fe concentration spectra quali- gence of this state is observed by magnetometry as a drop in tatively with those of bulk single phase fcc Ni-Fe random the saturation moment in this region while Mo¨ssbauer spec- alloy spectra, it is inferred that the alloy films are indeed troscopy reveals the appearance of a low-spin antiferromag- randomly ordering during codeposition of Fe and Ni and netic state. Together with the evidence of superparamagnetic have not formed any ordered alloy phases.10,24 The high-spin behavior, there is strong evidence that high-spin state exists site has zero quadrupole splitting expected for a pure fcc as clusters intermixed with the low-spin state. *Present address: Naval Research Laboratory, NSLS Bldg. 725A/ S. Murayama, and Y. Miyako, ibid. 67, 55 1987 . U4B, Brookhaven National Lab., Upton, NY 11973. 11 R. J. Weiss, Proc. R. Soc. London, Ser. A 82, 281 1963 . 1 Ch. E´d. Guillaume, C. R. Hebd. Seances Acad. Sci. 125, 235 12 I. A. Abrikosov, O. Eriksson, P. So¨derlind, H. L. Skriver, and B. 1897 . Johansson, Phys. Rev. B 51, 1058 1995 . 2 Physics and Applications of Invar Alloys, edited by Hideo Saito 13 B. Window, J. Appl. Phys. 44, 2853 1973 . Maruzen, Tokyo, 1978 . 14 U. Gonser, C. J. Meechan, A. H. Muir, and H. Wiedersich, J. 3 E. F. Wassermann, in Ferromagnetic Materials, Vol. 5, edited by Appl. Phys. 34, 2373 1963 . K. H. J. Buschow and E. P. Wohlfarth Elsevier, Amsterdam, 15 W. A. A. Macedo and W. Keune, Phys. Rev. Lett. 61, 475 1988 . 1990 . 16 M. E. Fisher and A. E. Ferdinand, Phys. Rev. Lett. 19, 169 4 G. J. Mankey, S. Z. Wu, F. O. Schumann, F. Huang, M. T. Kief, 1967 ; D. S. Ritchie and M. E. Fisher, Phys. Rev. B 7, 480 and R. F. Willis, J. Vac. Sci. Technol. A 13, 1531 1995 . 1973 ; D. P. Landau, ibid. 13, 2997 1976 . 5 F. O. Schumann, S. Z. Wu, G. J. Mankey, and R. F. Willis, J. 17 H. Za¨hres, M. Acet, W. Stamm, and E. F. Wassermann, J. Magn. Appl. Phys. 79, 5635 1996 . Magn. Mater. 72, 88 1988 . 6 F. O. Schumann, S. Z. Wu, G. J. Mankey, and R. F. Willis, Phys. 18 H. Maruyama, R. Pauthenet, J. C. Picoche, and O. Yamada, J. Rev. B 56, 2668 1997 . Phys. Soc. Jpn. 55, 3218 1986 . 7 J. Dresselhaus, M. Mo¨ller, Th. Kleeman, and E. Kisker, J. Magn. 19 D. J. Keavney, D. F. Storm, J. W. Freeland, I. L. Grigorov, and J. Magn. Mater. 148, 172 1995 . C. Walker, Phys. Rev. Lett. 74, 4531 1995 . 8 J. W. Freeland, D. J. Keavney, D. F. Storm, I. L. Grigorov, J. C. 20 R. D. Ellerbrock, A. Fuest, A. Schatz, W. Keune, and R. A. Walker, M. G. Pini, P. Politi, and A. Rettori, Phys. Rev. B 54, Brand, Phys. Rev. Lett. 74, 3053 1995 . 9942 1996 . 21 P. Bayer, S. Mu¨ller, P. Schmailzl, and K. Heinz, Phys. Rev. B 48, 9 D. Bartlett et al., Rev. Sci. Instrum. 62, 1263 1991 . 17 611 1993 . 10 G. Dumpich, E. Becker, K. Schletz, W. Stamm, W. Keune, W. 22 S. Mu¨ller, P. Bayer, C. Reischl, K. Heinz, B. Feldmann, H. Kiauka, and S. Murayama, J. Magn. Magn. Mater. 74, 237 Zillgen, and M. Wu¨ttig, Phys. Rev. Lett. 74, 765 1995 . 1988 ; G. Dumpich, E. F. Wassermann, V. Manns, W. Keune, 23 J. V. Barth and D. E. Fowler, Phys. Rev. B 52, 1528 1995 . 57 BRIEF REPORTS 83 24 S. Tomiyahsi, H. Yamamoto, and H. Watanabe, J. Phys. Soc. Jpn. 26 D. Li, M. Frietag, J. Pearson, Z. Q. Qiu, and S. D. Bader, Phys. 30, 1605 1971 . Rev. Lett. 72, 3112 1994 . 25 V. L. Moruzzi, P. M. Marcus, and J. Ku¨bler, Phys. Rev. B 39, 27 H. Magnan, D. Chandesris, B. Villette, O. Heckmann, and J. 6957 1989 . Lecante, Phys. Rev. Lett. 67, 859 1991 .