3b2v7 MAGMA : 8417 Prod:Type: com ED:Jolly = BG pp:123ðcol:fig::NILÞ PAGN: sandhya SCAN: Shubha ARTICLE IN PRESS 1 3 Journal of Magnetism and Magnetic Materials ] (]]]]) ]]]­]]] 5 7 Interface alloying in the metallic magnetic heterostructures 9 with BCC lattice 11 U. Micka,*, V. Uzdinb, E. Kiskera 13 a Institute f.ur Angewandte Physik Universitatsstr. 1, Heinrich-Heine-Universit.at D.usseldorf, D-40225 D.usseldorf, Germany b St. Petersburg State University, ICAPE, 14 linia V.O. 29, 199178, St. Petersburg, Russia 15 17 Abstract 19 Analysis of STM images of Fe grown on Cr and Cr grown on Fe with the exploitation of the differences of the Fe and 21 the Cr surface state energies demonstrates strong intermixing during the epitaxial growth. We suggest a theoretical approach for modelling the epitaxial growth with subsequent self-consistent calculations of electronic and magnetic 23 structure. On the basis of these calculations together with the experimental data obtained by complementary experimental methods, the structure of the interface on the atomic scale and the correlation between the chemical and 25 magnetic roughness are investigated. We show that interface alloying is not symmetrical from both sides of the interface and suggest a scenario of the epitaxial growth that leads to this asymmetry. r 2001 Published by Elsevier Science B.V. 27 Keywords: Magnetization; Thin films; Superlattices; Epitaxy 29 31 Metallic magnetic superlattices with BCC structures confirmed strong intermixing at the interfaces. Davis 57 33 present a wide class of low-dimensional magnetic et al. [1] show that layer-by-layer growth at 3001C leads systems, demonstrating a number of new phenomena to the formation of a Cr­Fe alloy that is observed as a 59 35 important for fundamental magnetism and for applica- distribution of single atomic Cr impurities dispersed in tions. Interdiffusion and interface roughness strongly the Fe substrate in the submonolayer-coverage regime. 61 37 affect all macroscopic properties of these systems. In the low-coverage regime where the individual Cr Accordingly, the control of the epitaxial growth and atoms can be resolved, the spatial correlation can be 63 39 the investigation of the interface structure are important evaluated from the experimental data. Suppression of problems. Scanning tunneling microscopy (STM) is a nearest-neighbor occupation is indicative of an effective 65 41 very powerful tool allowing to determine the position of repulsive interaction between the Cr impurities. Accord- individual atoms and, consequently, to perform direct ing to Choi [2], the surface-alloy formation can also 67 43 measurement of the surface structure during the occur at the low Fe coverage on the Cr(1 0 0) surface. epitaxial growth. Generally, STM information is not The Fe was deposited at room temperature and 69 45 element specific but for metallic BCC-systems, even in subsequently, the sample was annealed at temperatures the case of close lattice constants, imaging at the bias between 2001C and 3001C. In contrast to the case when 71 47 voltages near the corresponding surface states can Cr diffuse into the Fe matrix and form a disordered differentiate elements and it also provides microscopic isomorphic alloy, observed Fe atomic rows indicate an 73 49 information 75 51 77 53 UNCORRECTED PROOF about alloying and the chemical structure ordered alloy formation for Fe grown on a Cr(1 0 0) of overlayers. substrate. Our own STM study of the structure of Cr on The results of STM investigations of alloying at the Fe(0 0 1)-films grown on an Ag(0 0 1)-substrate also interfaces Fe/Cr (Fe on Cr [1]) and Cr/Fe (Cr on Fe [2]) confirmed the intermixing at Fe/Cr interface. The differ very quite essentially although both studies difference between Fe/Cr and Cr/Fe interfaces was also detected by means of M.ossbauer spectroscopy [3]. 79 55 *Corresponding author. Fax: +49-0-211-81-13658. Conversion electron M.ossbauer spectra (CEMS) of Fe/ E-mail address: uwe.mick@uni-duesseldorf.de (U. Mick). Cr(0 0 1) superlattices with 2 monolayers thick 57Fe 81 0304-8853/01/$ - see front matter r 2001 Published by Elsevier Science B.V. PII: S 0 3 0 4 - 8 8 5 3 ( 0 1 ) 0 0 8 4 2 - 3 MAGMA : 8417 ARTICLE IN PRESS 2 U. Mick et al. / Journal of Magnetism and Magnetic Materials ] (]]]]) ]]]­]]] 1 probe layers placed at Fe/Cr and Cr/Fe interfaces, For the interpretation of experimental data within the 57 respectively, gave the different distribution of the terms of local atomic environment and atomic magnetic 3 hyperfine fields (hff). In particular, for the Fe-on-Cr moments at each site, we developed the theoretical 59 interface (as compared with Cr-on-Fe), a larger con- approach, which includes the modelling of the alloyed 5 tribution of the bulk hff (33 T) was obtained, whereas interfaces and subsequent calculation of magnetic 61 the satellite peaks with lower field were more narrow structure within a periodic Anderson model [7,5]. 7 and gave less contribution to the total spectra. The Interface alloying was introduced into the system by 63 amplitude of the low field peak (20 T) often was several random algorithms. For Fe/Cr systems with bcc 9 associated with atoms at the ideally smooth interface structure, which will be discussed in the following, these 65 [4], but our calculations did not confirm this assumption algorithms place atoms into the sites of ideal BCC lattice 11 [5]. Correlation between the amplitude of 20 T peak and inside the prism. Out of the prism we used periodic 67 giant magnetoresistance effect (GMR) leads to the boundary conditions. The simplest routine, which leads 13 conclusion about the role of interface and bulk to intermixing at the interface, is the algorithm of 69 scattering for GMR in Fe/Cr systems [4]. This conclu- ballistic deposition. This algorithm adds single atoms to 15 sion and assumption, that alloying at the interfaces is the top level of the prism in a random procedure and lets 71 driven by the melting points of bulk Fe and Cr [6], have them descend through empty sites until further descend- 17 to be revized in accordance with our calculations of Fe/ ing is blocked by occupied sites. The bottom layer 73 Cr superlattices with rough interfaces. initially is blocked. The procedure of ballistic deposition 19 75 21 77 23 79 25 81 27 83 29 85 31 87 33 89 35 91 37 93 39 95 41 97 43 99 45 101 47 103 49 105 51 107 53 Fig. 1. AmountUNCORRECTED PROOF 109 of Fe-atoms with adifferent number of the nearest neighbors (n1) and second neighbors (n2) Cr-atoms for the 55 superlattice Fe4/Cr41 (left panel) and Fe6/Cr41 (right panel) and for different values of intermixing parameter B: Bottom axis is 111 graduated by 10n1 þ n2: MAGMA : 8417 ARTICLE IN PRESS U. Mick et al. / Journal of Magnetism and Magnetic Materials ] (]]]]) ]]]­]]] 3 1 gives a relatively thin interface region, where only 2­3 growth or the deposition ratio (which determines the monolayer contain atoms of different elements simulta- parameter B in our model) one can manipulate the 47 3 neously. Such a scenario, however, cannot reproduce the distribution of Fe atoms in the local configuration as different structures of the Fe/Cr and Cr/Fe interfaces. well as the magnetic structure. 49 5 The second part of the algorithm presupposes the Different distributions of hff for the samples with floating up of some atoms after deposition of the next probe 57Fe layer only at Fe/Cr or Cr/Fe interface can be 51 7 layer on the surface. It assumes that site exchange of easily explained using the algorithm we developed. For atoms and their diffusion take place only at the surface the probe layer at Fe/Cr (Fe on Cr) interface, 57Fe atoms 53 9 during the epitaxial growth and that there is no internal will flow up into the Fe slab and will increase the bulk- bulk diffusion. Exchange of atoms during deposition of like hff. At the Cr/Fe interface, the Fe atoms will flow up 55 11 the next layer leads to the asymmetry of the interface: into the Cr spacer and it will increase the low-field atoms could flow up on several layers but did not move contribution. It is such a behaviour of the hff that was 57 13 down due to suppression of diffusion into the internal observed in the experiment [3]. Note that this scenario of layers below the surface. Modelling of such a scenario epitaxial growth is very general and gives a natural 59 15 was organized as follows: we start from multilayers explanation of the change of the hff distribution on constructed by the algorithm of simple ballistic deposi- 119Sn atoms in V/Cr superlattices versus the position of 61 17 tion. Then in every layer we chose a definite fraction ðBÞ 119Sn probe layer inside the Cr spacer [8] as well. of atoms using a random procedure, and layerwise, This work was partially supported by the Russian 63 19 starting from the bottom, we exchanged this fraction of Ministry of Higher Education (Grant E00-3.4-547) and atoms in every pair of neighboring layers. The value of by the program ``Universities of Russia: fundamental 65 21 B ¼ Nexch=Ntot (where Nexch is the number of exchanged researches'' (Project 015.01.01.083). V.U. appreciates atoms and Ntot is the total number of atoms in the base the Alexander von Humbolt foundation and the 67 23 layer of the prism) is the parameter of the model. MML'01 symposium organizers for financial support. Fig. 1 shows the distribution of the Fe atoms, which 69 25 have a given number of the nearest neighbors (n1) and second neighbors (n2) Cr-atoms for the superlattice Fe4/ 71 27 Cr41 (a; c; e) and Fe6/Cr41 (b; d; f ). All structures were obtained using an algorithm with the floating of atoms References 73 29 during the deposition and with different parameters B (B ¼ 0 for Figs. 1a and b; B ¼ 0:5 for Figs. 1c and d and [1] A. Davies, J.A. Stroscio, D.T. Pierce, R.J. Celotta, Phys. Rev. Lett. 76 (1996) 4175. 75 31 B ¼ 0:75 for the Figs. 1e and f). The bottom axis is [2] Y.J. Choi, I.C. Jeong, J.-Y. Park, S.-J Kahng, J. Lee, graduated by the values 10n1 þ n2: Distributions in Y. Kuk, Phys. Rev. B 59 (1999) 10918. 77 33 Fig. 1a and b with B ¼ 0 correspond to the simple [3] T. Shinjo, W. Keune, J. Magn. Magn. Mater. 200 (1999) ballistic deposition algorithm. Increase of parameter B 598. 79 35 leads to the more uniform distribution of Fe-atoms on [4] R. Schad, P. Belien, G. Verbanck, K. Temst, the configuration, and to the filling of the state with a V.V. Moshchalkov, Y. Bruynseraede, B. Bahr, J. Falta, 81 37 larger number of n J. Dekoster, G. Langouche, Europhys. Lett. 44 (1998) 379. 1 and n2: Especially a large difference was found in the number of atoms with n [5] V. Uzdin, W. Keune, H. Schr.or, M. Walterfang, Phys. Rev. 1 ¼ 8 and 83 39 n B 63 (2001) 104407. 2a0; i.e. for atoms inside the Cr spacer but not far from the interface. Increasing the thickness of the Fe [6] B. Heinrich, J.F. Cochran, T. Monchesky, R. Urban, Phys. Rev. B 59 (1999) 14520. 85 41 slab leads to a larger contribution of bulk-like Fe atoms [7] V. Uzdin, D. Knabben, F.U. Hillebrecht, E. Kisker, Phys. and Fe atoms with small numbers of Cr neighbors. Rev. B 59 (1999) 1214. 87 43 Therefore, via the changing of the thickness of the Fe [8] M. Almokhtar, K. Mibu, A. Nakanishi, T. Kobayashi, slabs and substrate temperature during the epitaxial T. Shinjo, J. Phys.: Condens. Matter 12 (2000) 9247. 89 45 UNCORRECTED PROOF