Thin Solid Films 413 (2002) 212­217 Correlation between structural and magnetic properties of thin Fe Co x 1yx(1 1 0) films on sapphire J. Swerts*, K. Temst, N. Vandamme, B. Opperdoes, C. Van Haesendonck, Y. Bruynseraede Laboratorium voor Vaste-Stoffysica en Magnetisme, Katholieke Universiteit Leuven, Celestijnenlaan 200 D, B-3001 Leuven, Belgium Received 16 July 2001; received in revised form 26 January 2002; accepted 28 February 2002 Abstract Fe and Fe40Co60 thin films (5­60 nm) with a bcc structure have been prepared on a-axis oriented sapphire substrates by molecular beam epitaxy. The structural properties have been characterized in situ by reflection high-energy electron diffraction and ex situ by X-ray diffraction and atomic force microscopy. A two-dimensional magneto-optical Kerr effect set-up has been used to determine the in-plane magnetization components and to investigate the magnetic anisotropy and the orientation dependence of the magnetization reversal process. The Fe films and Fe40Co60 alloy films both display a uniaxial in-plane anisotropy. They also exhibit a comparable increase of the coercive field along the easy axis with increasing thickness. We have evaluated this dependence using the results of the structural characterization, indicating that the enhancement of the coercive field is linked to the growing surface roughness and decreasing structural coherence. 2002 Elsevier Science B.V. All rights reserved. Keywords: Fe; FeCo alloys; Structural properties; Magnetic properties and measurements; Anisotropy 1. Introduction room temperature using suitable substrates e.g. fct Fe can be stabilized by epitaxial growth on the Cu(1 0 0) There is at present considerable interest in the mag- surface w6x, even hcp Fe has been reported w7x. In netic properties of the 3d transition metals and metal contrast to the stable bcc Fe, the structural bcc Co phase alloys, in particular in thin ferromagnetic films with is metastable. Only on suitable substrates bcc Co can controllable magnetic anisotropy, due to promising tech- be stabilized e.g. on Cr(1 0 0) surfaces ultra-thin Co nological applications w1x. The Fe films could be grown in the metastable bcc phase w8x. xCo1yx alloy is a fundamentally and technologically interesting ferromag- The stability of bcc FexCo1yx alloys is dependent on netic material because of its large variations in magnetic the composition and the substrate. FexCo1yx alloys are properties with composition. In bulk bcc Fe in general only thermodynamically stable as bcc struc- xCo1yx alloys, the first crystal anisotropy constant K ture for 0.25-x(1 w9x. Gutierrez et al. achieved a 1 changes sign as the Co content increases w2x. Also in thin films stable bcc phase of FexCo1yx over the entire range of this behaviour has been reported for Fe alloy concentrations using ZnSe-epilayered GaAs sub- xCo1yx alloy films on ZnSe and MgO substrates w3,4x. strates w3x. The epitaxial growth and magnetic properties of bcc In this paper, we will focus on the ferromagnetic Fe are widely investigated and can be influenced by properties of the bcc Fe and bcc Fe40Co60 system grown using different deposition techniques and various sub- on a-axis oriented sapphire substrates. When bcc Fe was strates w5x. Bulk Fe exists in the stable bcc phase under grown onto sapphire to look into the magnetic properties ordinary conditions. Related phases can be stabilized at of Fe(1 1 0) thin films, a buffer layer has always been used w5,10,11x. To our knowledge, only Metoki et al. *Corresponding author. Tel.: q32-16-327195; fax: q32-16- looked at the structural and magnetic properties of Fe 327983. on a-axis (1 1­2 0) sapphire without using a buffer E-mail address: johan.swerts@fys.kuleuven.ac.be (J. Swerts). layer w12,13x. It was found that Fe(1 1 0) films on 0040-6090/02/$ - see front matter 2002 Elsevier Science B.V. All rights reserved. PII: S0040-6090Z02.00247-X J. Swerts et al. / Thin Solid Films 413 (2002) 212­217 213 sapphire form three domains and that they exhibit a strong in-plane uniaxial anisotropy. In FexCo1yx bulk alloys and thin films, the particular composition xs40 leads to a negative K1 value and this influences profoundly the in-plane magnetic anisotropy. We show that in our case the structural and magnetic properties of the Fe films and the Fe40Co60 films grown on a-axis (1 1­2 0) sapphire are similar to each other. The orientation of the bcc Fe and Fe40Co60 films on Al O 2 3(1 1­2 0) is (1 1 0), they both display a strong uniaxial in-plane anisotropy and they exhibit a compa- rable thickness dependence of the coercive field along the easy axis i.e. when the magnetization reversal process is dominated by domain wall nucleation and motion. We have evaluated this dependence using the results of the structural characterization. Hence in this article, the interplay between the structural and magnetic properties of thin Fe and Fe40Co60 films on a-axis (1 1­ 2 0) sapphire without buffer layer is discussed. 2. Experimental details Fe and Fe40Co60 films were grown by molecular beam epitaxy (MBE) onto single crystal a-axis (1 1­2 0) sapphire substrates at various temperatures. The base pressure during deposition in the MBE chamber is f10y10 mbar. The substrates were degassed by anneal- Fig. 1. XRD profiles for the Fe (upper curve) and Fe40Co60 (lower ing them for 30 min at Ts950 8C. The Fe films were curve) films at (a) low angles and (b) high angles. The curves have deposited by e-beam evaporation and the Fe40Co60 alloys been offset vertically for clarity. were prepared by co-deposition from two independently controlled e-beam guns. The e-beam guns are controlled by quadrupole mass spectrometer systems. The evapo- The magnetic properties of the samples were charac- ration rates were calibrated by quartz crystal oscillators terized ex situ by using the longitudinal magneto-optical and X-ray reflectivity measurements. The Fe films were Kerr effect (MOKE). The MOKE system consists of a grown at a rate of 0.45 A ys and the evaporation rates polarized HeNe 10 mW laser source operating at 632.8 of Fe and Co during co-deposition were 0.18 and 0.27 nm, two polarizers, a Faraday modulation coil, and a A ys, respectively. The film thickness was varied from 5 photodetector. Magnetization measurements in the lon- to 60 nm. Protective Ag layers with a thickness of 2 gitudinal geometry were carried out in two configura- tions. In the first configuration the magnetic field is nm were deposited on top of the Fe and Fe40Co60 films. applied parallel to the film surface and to the plane of The quality of the films has been checked in situ by incidence in order to determine the in-plane component reflection high-energy electron diffraction and ex situ of magnetization parallel to the applied field. In the by X-ray diffraction (XRD) and was optimised by second configuration the applied field is parallel to the varying the substrate temperature during deposition. We film surface but perpendicular to the plane of incidence find that Fe and Fe40Co60 films deposited on sapphire in order to measure the in-plane component of magnet- at a substrate temperature Tss300 8C develop a good ization perpendicular to the applied field. crystalline quality. The X-ray measurements presented in this paper are obtained with u­2u scans on a Rigaku diffractometer 3. Structural characterization with a 12 kW rotating anode and using a pyrolitic graphite single crystal monochromator to select Cu Ka 3.1. X-ray diffraction radiation (ls0.154 nm). The step size of the measured data was 0.018. After preparation, the Fe and Fe40Co60 films were The surface study is performed on a commercial taken out of the vacuum system for the XRD measure- atomic force microscope (M5, Park Scientific ments. For both compositions, Fe and Fe40Co60, a typical Instruments). film is considered. 214 J. Swerts et al. / Thin Solid Films 413 (2002) 212­217 Information about the perpendicular crystalline coher- ence length can be obtained with Scherrer's law. This coherence length, in general, is limited to the perpendic- ular grain size. It should be noted that Scherrer's law might lead to a small overestimate, especially for thinner films. In Fig. 2b the perpendicular coherence length vs. film thickness for the Fe and Fe40Co60 films is plotted. The full line corresponds to a grain size equal to the film thickness. It is clear that below 30 nm the films have a structural coherence length of almost 100% of the total film thickness. For the thicker films the grain size tends to saturate. The structural coherence length will be less than the total film thickness. Hence, there will be relatively more grain boundaries present in the thicker films when compared to the thinner films. These grain boundaries can act as pinning sites for magnetic domain walls. 3.2. Atomic force microscopy Magnetization reversal in thin films is strongly influ- enced by the surface roughness w15,16x. Therefore the surface roughness of the Fe40Co60 alloys for different thickness was evaluated ex situ by atomic force micros- copy (AFM) in order to correlate their morphology to the magnetization measurements. Scanning probe Fig. 2. (a) Lattice constant for Fe and Fe40Co60 films vs. film thick- microscopy is well suited to obtain quantitative infor- ness; (b) Perpendicular coherence length vs. film thickness for Fe and Fe mation about surface roughness and its dependence on 40Co60 films. lateral length scale w17,18x. For the purpose of these roughness measurements, Fig. 1 shows the low (a) and high (b) angle XRD three samples of different thickness were prepared simul- scans of a 27 nm thick Fe film (upper curve) and a 20 taneously by sliding a mask across the sample holder. nm Fe40Co60 (lower curve) film. For clarity, the curves Fig. 3a shows a typical AFM top view of a 500 nm by have been shifted vertically with respect to each other. 500 nm area of a 5 nm thick Fe Co 40 60 thin film. The The low-angle profiles allow calculating the film thick- vertical interface width, determined by the root mean ness. The high-angle diffraction profile consists of 2 square (rms) roughness s, characterizes the surface peaks: the left peak is caused by the (1 1­2 0) plane of roughness along the vertical direction. The Fe40Co60 rms the sapphire substrate, while the right one corresponds roughness at different length scales (0.5­5 mm) is to the main reflection from the (1 1 0) planes of the plotted vs. thickness in Fig. 3b. As is seen on this graph, magnetic film. The presence of finite size peaks in the the rms roughness value increases as the thickness of high-angle profile confirms that both films have a rather the samples increases for all length scales. limited roughness. The thickness calculated from the high-angle profiles is consistent with the calculated 4. Magnetic characterization thickness using the low-angle profiles. Fig. 2a shows the average lattice parameter vs. film 4.1. Magnetic anisotropy thickness obtained from the high-angle XRD measure- ments. The lattice constant of the Fe40Co60 alloys exhib- The in-plane magnetic anisotropy of the bcc Fe and its a relaxation towards a value 2.835 A corresponding bcc Fe40Co60 films has been studied in detail by varying to the calculated value for the bulk alloy using Vegar ´ d's the orientation of the magnetic field with respect to the law. Our experimental value, 2.835 A, is also in agree- crystallographic directions in steps of 58 or 158. Fig. 4 ment with the value obtained for bulk alloys of the shows for both compositions the angular dependence same composition w14x. Hence we can assume that there MRyMS of the normalized remanent parallel magnetiza- is a larger strain present in the thinner films whereas in tion in the longitudinal geometry. The Fe and Fe40Co60 the thicker films a relaxation of the lattice constant films are, respectively, 60 and 53 nm thick. The data occurs. Internal strain may influence the magnetic ani- for both samples exhibit cosine-like polar plots, showing sotropy of the films w11x. that a simple uniaxial anisotropy is consistent with the J. Swerts et al. / Thin Solid Films 413 (2002) 212­217 215 Fe Co 40 60 films are similar to each other due to the presence of a dominant uniaxial anisotropy. Hence, a change of sign of the first anisotropy constant K1 could not be observed. The precise orientation of the easy axis with respect to crystallographic directions of the sub- strate is not clear yet. The easy axis is not oriented consistently 808 (or y108 for samples thicker than 26 nm) away from the Al2O3w0 0 0 1x axis as determined by Metoki et al. This inconsistence can be due to the different preparation technique and growth conditions. Moreover, the fact that Fe films form three domains probably enhances the sensitivity to the experimental conditions. Further investigation is needed to clarify this issue. 4.2. Magnetization reversal As shown by Daboo et al. w19,20x the magnetization reversal process can be studied in more detail by comparing the magnetization components parallel and perpendicular to the applied field. In our films with uniaxial anisotropy the reversal process should be rather simple. Fig. 5 shows three MOKE loops of the transverse magnetization of a 39 nm thick Fe film. The inset in each graph shows the parallel component of magneti- zation. When the magnetic field is applied close to the easy axis (Fig. 5a), no transverse magnetization is Fig. 3. (a) AFM micrograph of the surface of a 5 nm Fe40Co60 film measured, corresponding to a `single-jump' switching for an area of 500 nm by 500 nm. The black to white contrast is 3 nm; (b) evolution of the rms roughness as a function of sample thick- process. This indicates that the switching behaviour is ness derived from the AFM measurements on the Fe dominated by domain wall motion. When the magnetic 40Co60 alloys on various length scales. field is applied perpendicular to the easy axis, the reversal is mainly achieved through rotation. This is magnetic anisotropy of our Fe and Fe40Co60 films. illustrated in Fig. 5b. When the field is applied in any Contrary to the expectations, we have observed that the other direction the reversal process will be a combination in-plane anisotropy of all the investigated Fe and of both mechanisms as is shown in Fig. 5c. In this case Fig. 4. Polar plot of the normalized in-plane remanent magnetization M y R MS for a 60 nm Fe thin film (d) and a 53 nm Fe4 C 0 o60 thin film (s). 216 J. Swerts et al. / Thin Solid Films 413 (2002) 212­217 The dotted line is a guide for the eye. Both the systems reveal comparable thickness dependence. The coercive field increases with film thickness except for the 5-nm thick films that show an unexpected high coercivity (50 Oe). 5. Discussion Domain wall motion in a thin metal film can be described as a domain wall that propagates in an energy landscape. The shape of this landscape is influenced by the presence of grain boundaries, inclusions, surface roughness and other defects. Hence, potential barriers and potential minima, so-called pinning sites, are being formed that inhibit the nucleation or propagation of the domain wall. To overcome the barrier or to remove the domain wall from such a pinning site, external energy needs to be supplied e.g. by applying a magnetic field. In this case the coercive field is a good measure of this extra energy. The difficulty is to distinguish which defect is responsible for an enhanced coercivity. The structural characterization with XRD and AFM led to the following results: (i) the lattice parameter shows a relaxation with increasing thickness towards a value that is consistent with bulk values; (ii) the grain size (in a direction perpendicular to the sample) scales with the thickness for thinner films but tends to saturate for thicker films. Consequently, relatively more grain boundaries will be present as the film thickness increas- es; (iii) the rms roughness of the samples increases as their thickness increases. The first result indicates that the thinnest films exhibit a large internal strain. If this strain is inhomogeneous, Fig. 5. Measurements of the in-plane transverse component of the the magnetic anisotropy will be influenced by this strain magnetization of a 39 nm Fe film; (a) with the applied field along and the resulting strain fields may deform the potential the easy axis; (b) perpendicular to the easy axis; and (c) 308 away landscape and enhance the nucleation field and conse- from the easy axis. The insets of the graphs show the corresponding quently the coercivity. Since the Fe(1 1 0) film is stiffer measurements of the parallel magnetization component. in the w1­1 0x direction than in the w0 0 1x direction, the field is applied 308 away from the easy axis. The magnetization first rotates away from saturation field direction towards the easy axis (A, D) then switches along the easy axis (B , E) and finally rotates again towards reversed saturation field direction (C, F). These measurements confirm the conclusion that a uniaxial anisotropy characterizes the anisotropy of our samples and give information about the orientational dependence of the magnetization reversal process. Similar results are obtained for the Fe40Co60 films. In the following we will concentrate on the magnetic reversal process when the applied field is along the easy axis of magnetization. The dominant switching mecha- nism is in this case reversal by domain wall motion. In view of the rectangular shape of the hysteresis loop along an easy axis, the coercive field Hc will be equal to the so-called switching field at which the reversal Fig. 6. Coercive field vs. sample thickness with the magnetic field process commences. In Fig. 6, a study of the coercive applied along the easy axis. Fe films (d) and Fe40Co60 films (s). field along the easy axis vs. sample thickness is shown. The dotted line is a guide for the eye. J. Swerts et al. / Thin Solid Films 413 (2002) 212­217 217 inhomogeneous strain is favoured. The second result Acknowledgments leads to the following conclusion: the thicker the sam- ples, the smaller the structural coherence, the larger the This work has been supported by the Fund for number of grain boundaries becomes. Since each grain Scientific Research-Flanders (FWO), the Belgian IUAP, boundary can act as a local barrier, the coercive field and the Flemish GOA programs. K.T. is a Post-Doctoral will increase as the thickness of the samples increases. Research Fellow of the FWO. The third result is also consistent with an increased coercivity with increasing thickness, since enhanced surface roughness can also inhibit domain wall nuclea- References tion and motion. Our interpretation is consistent with the measured w1x J. de Boeck, G. Borghs, Phys. World (1999) 27. thickness dependence of the coercive field in the Fe as w2x R.C. Hall, J. Appl. Phys. 31 (1960) 157S. well as in the Fe w3x C.J. Gutierrez, J.J. Krebs, G.A. Prinz, Appl. Phys. Lett. 61 40Co60 thin films. Although the rough- (1992) 2476. ness is small and the structural coherence length is large w4x Th. Muhge, ¨ Th. Zeidler, Q. Wang, Ch. Morawe, N. Metoki, H. in the thinnest films (5 nm), these films exhibit a rather Zabel, J. Appl. Phys. 77 (1995) 1055. unexpected high coercive field. This could be caused by w5x Yu.V. Goryunov, I.A. Garifullin, Th. Muhge, ¨ H. Zabel, JETP the large internal strain. Another effect that could be 88 (1999) 377. involved, is the oxidation of the metals on sapphire w6x J.H. Dunn, D. Arvanitis, N. Martensson, J. Phys. IV 7 (1997) substrates w12,21x. Oxidation of the Fe atoms close to 383. the sapphire interface can cause strong interface aniso- w7x M. Maurer, J.C. Ousset, M.F. Ravet, M. Piecuch, Europhys. Lett. 9 (1989) 803. tropy and consequently can enhance the coercivity. Since w8x F. Scheurer, B. Carriere, J.P. Deville, E. Beaurepaire, Surf. Sci. the oxidation process is an interface effect and since the Lett. 245 (1991) 175. internal strain is reduced in the thicker films, surface w9x T. Nishizawa, K. Ishida, Bull. Alloy Phase Diagrams 5 (1984) roughness and structural coherence start to dominate the 250. domain wall nucleation and motion processes. This w10x A.D. Kent, U. Ruediger, J. Yu, S. Zhang, P.M. Levy, Y. Zhong, implies in our case an increased coercivity with increas- S.S.P. Parkin, IEEE Trans. Magn. MAG-34 (1998) 900. w ing film thickness. 11x B.M. Clemens, R. Osgood, A.P. Payne, B.M. Lairson, S. Brennan, R.L. White, W.D. Nix, J. Magn. Magn. Mat. 121 (1993) 37. 6. Conclusions w12x N. Metoki, M. Hofelich, Th. Zeidler, T. Muhge, ¨ Ch. Morawe, H. Zabel, J. Magn. Magn. Mat. 121 (1993) 137. We have grown Fe(1 1 0) and Fe40Co60(1 1 0) thin w13x T. Muhge, ¨ A. Stierle, N. Metoki, H. Zabel, U. Pietsch, Appl. films on single crystal a-axis (1 1­2 0) sapphire sub- Phys. A 59 (1994) 659. strates without any buffer layer. The structural properties w14x W.C. Ellis, E.S. Greiner, Trans. Am. Soc. Met. 29 (1941) 415. are investigated with XRD and AFM measurements. w15x Y.-L. He, G.-C. Wang, J. Appl. Phys. 76 (1994) 6446. MOKE measurements show that the Fe films as well as w16x M. Li, Y.-P. Zhao, G.-C. Wang, H.-G. Min, J. Appl. Phys. 83 the Fe (1998) 6287. 40Co60 alloy films have a uniaxial anisotropy. Both w systems reveal a strong thickness dependence of the 17x J. Krim, I. Heyvaert, C. Van Haesendonck, Y. Bruynseraede, Phys. Rev. Lett. 70 (1993) 57. coercive field with the applied field along the easy axis. w18x K. Temst, M.J. Van Bael, B. Wuyts, C. Van Haesendonck, Y. We have evaluated this dependence using the results of Bruynseraede, D.G. de Groot, N. Koeman, R. Griessen, Appl. the structural characterization. The results show that in Phys. Lett. 67 (1995) 3429. the thicker films larger coercivity is due to the enhanced w19x C. Daboo, J.A.C. Bland, R.J. Hicken, A.J.R. Ives, M.J. Baird, roughness and increased number of grain boundaries. M.J. Walker, J. Appl. Phys. 73 (1993) 6368. w The thinnest films exhibit an unexpected large coercivity 20x C. Daboo, R.J. Hicken, E. Gu, M. Gester, S.J. Gray, D.E.P. Eley, E. Ahmad, J.A.C. Bland, Phys. Rev. B 51 (1995) 15964. that is probably caused by large internal strain or an w21x N. Metoki, Th. Zeidler, A. Stierle, K. Brohl, ¨ H. Zabel, J. interfacial oxidation process. Magn. Magn. Mater. 118 (1993) 57.