Journal of Magnetism and Magnetic Materials 191 (1999) 61-71 Enhanced perpendicular magnetic anisotropy in chemically long-range ordered (0 0 0 1)CoVPt\V films M. Maret *, M.C. Cadeville , A. Herr , R. Poinsot , E. Beaurepaire , S. Lefebvre , M. Bessie re Institut de Physique et Chimie des Mate&riaux de Strasbourg, Groupe d'Etude des Mate&riaux Me&talliques, UMR 46 CNRS-ULP, 23 rue du Loess, F-67037 Strasbourg, France Laboratoire pour l'Utilisation du Rayonnement Electromagne&tique, CNRS-CEA-MEN, BaLt 209d, F-91405 Orsay, France Received 18 March 1998; received in revised form 18 September 1998 Abstract Chemical long-range ordering along the growth direction has been observed in mostly HCP CoPt films grown on a Ru(0 0 0 1) buffer by the molecular beam epitaxy technique. Such ordering, strongly dependent on both the growth temperature and the crystalline quality of the alloy film, is revealed to enhance perpendicular magnetic anisotropy in HCP films. Pt segregation effect at the advancing surface, allowed by dominant surface diffusion, would be the driving force for promoting such uniaxial long-range ordering during the MBE process. Yet, ex situ annealing treatments performed at temperatures similar to the growth temperature have shown that the MBE-promoted chemical ordering, different from those found in the bulk L1- or DO-type ordered AB compounds, is metastable in agreement with theoretical calculations. 1999 Elsevier Science B.V. All rights reserved. PACS: 61.10.F; 61.55.H; 75.30.G; 75.60; 75.70 Keywords: Co-Pt alloy films; Structure; Chemical order; Magnetic anisotropy 1. Introduction rials, the CoVPt\V films are among the most studied ones, since they are potential magneto-optical re- The growth of epitaxial alloy films is a promising cording media for blue laser recording [1]. For the way of producing chemically ordered phases, with study of magnetic anisotropy and its relation to respect to the conventional metallurgical tech- structure, Co-Pt is also a well-appropriate system. niques for bulk alloys. Among this class of mate- The Co-Pt phase diagram displays at high temper- atures a disordered FCC solid solution over the whole composition range, and at lower temper- atures two FCC L1 * Correspondence address. Universita¨t Konstanz, Fakulta¨t  ordered phases around the fu¨r Physik, Jakob-Burckhardt-Strabe 29, Postfach X915, D- composition CoPt and CoPt and one tetrag- 78457 Konstanz, Deutschland. Tel: #49-7531-88-2036; fax: onal L1 ordered phase around the equiatomic #49-7531-88-3895; e-mail: mireille.maret@uni-konstanz.de. composition. The FCC CoVPt\V alloys are 0304-8853/99/$ - see front matter 1999 Elsevier Science B.V. All rights reserved. PII: S 0 3 0 4 - 8 8 5 3 ( 9 8 ) 0 0 3 7 3 - 4 62 M. Maret et al. / Journal of Magnetism and Magnetic Materials 191 (1999) 61-71 ferromagnetic at room temperature for Pt content On the Co-rich side, chemical long-range order- up to 85 at% and interestingly, the ordered phases, ing (LRO) along the growth direction was found in CoPt and CoPt, have Curie temperatures lower epitaxial films. The first evidence was in 1000 As than those of the corresponding disordered phases. thick Co For example, the Curie temperature of the CoPt Pt films deposited on sapphire(0 0 0 1)  with a 10 As Pt buffer; such ordering was correlated ordered compound is 170 K lower than that of with the emergence of a new peak at 3.2 eV photon the disordered phase [2], which is attributed to the energy in magneto-optical spectra [9]. More re- absence of CoCo nearest neighbor pairs in the cently, ordering effects were found in 500 As thick ordered L1 compound [3]. Co In epitaxial (1 1 1) CoPt Pt films deposited on mica(0 0 1) with a  films, it was shown that 150 As Ru buffer, in which the stacking sequence is the development of perpendicular magnetic anisot- predominantly HCP; such ordering along the sur- ropy, strongly dependent on the growth temper- face normal clearly increases perpendicular mag- ature, is correlated with the existence of anisotropic netic anisotropy [10]. structural order effects, characterized by preferen- In this paper, we present a detailed structural tial heteroatomic pairs out of the film plane, study of the Co-rich films grown on (0 0 0 1)Ru promoted by the molecular beam epitaxy (MBE) buffer, from X-ray diffraction measurements. Their technique [4]. The highest uniaxial anisotropy magnetic properties, studied by means of SQUID about 1 MJ/m was found in a film grown around magnetometry and polar Kerr effects, are also re- 690 K [5,6]. In contrast, the occurrence of the L1- ported and discussed in relation with their struc- type isotropic chemical ordering, favored by an ture. Futhermore, the contributions of surface and increase of the deposition temperature around volume diffusion during the co-deposition process 800 K and characterized by (1 1 1) planes of identi- to the formation of anisotropic chemical ordering cal CoPt composition, gives rise to a disappear- and associated perpendicular magnetic anisot- ance of perpendicular magnetic anisotropy. Re- ropy are discussed, together with the decrease of cently, angle-dependent X-ray magnetic circular chemical ordering observed after ex situ anneal dichroism measurements in these films have shown treatments. that the microscopic origin of the magneto-crystal- line anisotropy is related to 3d and 5d orbital moment anisotropies, such as the out-of-plane 2. Experimental procedures components of the orbital moments are higher than the in-plane components [7,21]. Co The growth of CoPt films on an MgO(0 0 1) VPt\V films of 400-500 As thickness were de- posited at different temperatures ranging from 500 substrate at 780 K leads to the formation of a L1 to 750 K in a 10\ Torr vacuum onto a 150 As tetragonal ordered phase with the c-axis along the Ru(0 0 0 1) buffer grown at 900 K on a mica(0 0 1) surface normal, which presents strong perpendicu- substrate. These alloy films were covered by a 20 As lar anisotropy [8]. Let us recall that in a fully protective Pt layer deposited at room temperature. L1-type ordered film the stacking sequence along Electron gun sources were used for both Co and Pt the [0 0 1] easy direction of magnetization consists with deposition rates in the range 0.05-0.2 As/s of alternate pure Co and pure Pt planes, while monitored by two quartz balances. In our UHV along the hard axes [1 0 0] and [0 1 0] all the deposition chamber, up to six mica substrates, 3 cm planes have the equiatomic composition; therefore, in diameter each, can be fixed to a holder. First the all nearest-neighbor CoPt pairs are oriented out of six Ru buffer layers were prepared simultaneously. the (0 0 1)CoPt film plane. As also found in CoPt Then, using an appropriate disk of 60° aperture films, the existence of strong perpendicular mag- masking five substrates, alloy films could be grown netic anisotropy stems from the hybridization be- at different temperatures successively, by starting tween Co atoms of large magnetic moment and Pt with the sample grown at the highest temperature. atoms of strong spin-orbit coupling, preferentially Nevertheless, the advantage of preparing a series along the surface normal. of six samples is somewhat attenuated. On the one M. Maret et al. / Journal of Magnetism and Magnetic Materials 191 (1999) 61-71 63 hand annealing effects can occur in the first depos- Basically, symmetric reflection measurements ited films. On the other hand, though the same give information on the spacing between planes deposition rates are used, the film composition can stacked along the growth direction and the asso- vary from one mica position to another with respect ciated coherence length, while symmetric transmis- to both Co and Pt sources; such composition gradi- sion measurements provide the in-plane lattice ent observable in the first series, was subsequently parameters and the lateral coherence length. eliminated by increasing the oscillation angle of the The magnetic properties of the alloy films sample holder from 180 to 330° during co-depo- were studied at 300 K by measuring the polar and sition. The substrate temperature was measured longitudinal Kerr effects using a laser diode with a K-type thermocouple located close to the ( "670 nm) with the applied field up to 1.3; holder leading to an uncertainty estimated at 20 K. 10 kA/m, and at 30 K using a SQUID mag- The RHEED patterns observed along the azi- netometer with the applied field up to 5.6; muths [1 0 1 0] and [1 1 2 0] revealed the crystal- 10 kA/m. For the SQUID measurements, the di- line quality of the alloy films on the Ru(0 0 0 1) mensions of the samples were 4;4 mm and buffer with the following epitaxial relationships: cleaved down to a thickness of &10-30 m to minimize the contribution of the mica substrate. CoVPt\V (0 0 0 1) [1 0 1 0] # Ru(0 0 0 1) [1 0 1 0], The effective magnetic anisotropies (K " Co K VPt\V (1 1 1) [1 1 2] # Ru (0 0 0 1) [1 0 1 0]. !0.5 M) were deduced from the area en- closed between the parallel and perpendicular mag- The X-ray diffraction measurements were per- netization hysteresis curves measured at 30 K. formed: (i) on a high-resolution(HR) X-ray Philips diffractometer using a four-Ge(2 2 0) crystal mono- chromator providing a pure Cu K? parallel beam; 3. Results  three types of measurements were collected, -2 symmetric reflection scans, rocking curves in sym- 3.1. X-ray diffraction measurements metric and asymmetric geometries, (ii) on a D500 Siemens diffractometer with monochromatic Co The main diffraction peaks of the alloy layer, K beam of higher intensity but of lower resolution, buffer and substrate are shown in Fig. 1 for four for measuring weak superstructure peaks in -2 samples among three different series. The chosen geometry. samples were the most relevant for studying the X-ray measurements were also carried out at the relationships between their structure and magnetic laboratoire pour l'utilisation du rayonnement elec- properties. The solid curves were measured in -2 tromagne´tique (LURE) in Orsay, France, on the reflection geometry on the high-resolution diffrac- DIF4C diffractometer below the Co-K absorption tometer and the dashed curves are the rocking edge at an energy of 7501 eV ( "1.653 As). Mea- curves around the (0 0 0 2) or (1 1 1)CoPt and surements in both reflection and transmission geo- (0 0 0 2)Ru peaks. For the samples a and b of a first metries were done in samples previously oriented series and the sample c of a second series, the alloy and whose substrates were cleaved down to a thick- films were grown at 690, 500 and 650 K, respective- ness of 30-80 m. Furthermore, diffraction rods ly, their thickness controlled by low-angle X-ray were recorded by measuring the diffracted intensity reflectivity measurements is close to 500 As and as a function of the normal component of the scat- their Co composition checked by electron probe tering vector, with the in-plane component fixed at microanalysis is around 80 at%. For the sample 4 /d, (d, is the in-plane nearest-neighbor dis- d of a third series, the alloy film grown at 700 K is tance). Along this rod, the Bragg peaks character- only 400 As thick and somewhat poorer in Co with istic of both HCP and FCC stacking sequences 75 at% Co content. The main peak of the alloy film are well separated. From their integrated intensi- refers to a 0 0 0 2 or 1 1 1 reflection depending on ties, the volume fractions of each phase can be its majority stacking sequence (HCP or FCC). The determined. normal coherence lengths, ¸," /2 F I J cos F I J, 64 M. Maret et al. / Journal of Magnetism and Magnetic Materials 191 (1999) 61-71 Fig. 1. -2 X-ray diffraction scans (-) and -scans around the 0 0 0 2 (or 1 1 1)CoVPt\V and 0 0 0 2 Ru reflections (- - -) measured on a HRXRD Philips diffractometer ( "1.5406 As) for four CoVPt\V films of three different series: (a and b) are two samples of series 1 grown at 690 and 500 K, (c) one sample of series 2 grown at 650 K; and (d) one sample of series 3 grown at 700 K. h k l"0 0 2 or 1 1 1, together with the widths of can be decomposed into two Gaussian compo- the rocking curves are reported in Table 1. nents. Based on the h-rods diffraction rods de- The most interesting feature observed for some scribed later, the smallest one refers to the 2 2 2 of these alloy films is the existence of a superstruc- reflection of minority FCC stacking and the largest ture peak in the -2 patterns around 1.5 As\ one to the 0 0 0 4 reflection of majority HCP stack- which is a fingerprint of chemical LRO along the ing. Such a decomposition shows that HCP stack- growth direction. As shown in Fig. 2, the intensity ing is about 0.015 As denser than FCC stacking. In of this peak, referred to as a 0 0 0 1 reflection, Table 1, are thus reported the out-of-plane para- increases with the growth temperature T , from 500 meters for both HCP and FCC stackings when the to 690 K. As precised further by the calculation of two components can be solved reliably. a chemical LRO parameter, S, the chemical modu- The 1 0 1 diffraction rod measurements lead to lation along the growth direction is favored by a complete separation between the reflections increasing T from 500 to 690 K. However, such coming from HCP and FCC stackings, and there- a peak was observed neither in the film d grown fore to a better determination of the volume frac- at 700 K (not shown in Fig. 2), nor in three other tions of two stackings. Fig. 4 displays the 1 0 1 samples of the third series grown at 600, 650 and diffraction rods referred to the HCP phase along 750 K and this result will be discussed hereafter. which are superimposed the 0 0 2 and 1 1 1 reflec- Fig. 3 shows the high-order 0 0 4 reflections for tions of FCC stacking and the 1 1 1 and 0 0 2 sample a which has the most intense superstructure reflections of twinned FCC. The volume fractions peak. The CoVPt\V alloy peak is asymmetric and of HCP and FCC(#twin) stackings, reported in M. Maret et al. / Journal of Magnetism and Magnetic Materials 191 (1999) 61-71 65 Table 1 Main structural and magnetic properties of the four CoVPt\V alloy films grown on a Ru(0 0 0 1) buffer. ¸, and ¸, normal and in-plane coherent lengths;     !. and     0 full-widths at half-maximum of the rocking curves around the 0 0 0 2(1 1 1)CoVPt\V and 0 0 0 2 Ru reflections; %HCP volume fraction of HCP stacking; c&!. and a&!. lattice parameters of HCP stacking; S chemical order parameter; a$!! , and a$!! , lattice parameters of FCC stacking deduced from the 1 1 1 and 0 2 2 peak positions; (c/a)* deviation from the ideal ratio of HCP stacking equal to (8/3; a$!! , /a$!! , deformation of FCC stacking; M saturation magnetization; K effective magnetic anisotropy and K"K #0.5 M uniaxial magnetic anisotropy. Alloy Growth ¸, (As)     !. (°) % HCP c&!. (As) a$!! , (As) (c/a)* M K (MJ/m) Comp. Temp. (K) ¸, (As)     0 (°) S a&!. (As) a$!! , (As) a$!! , /a$!! , (kA/m) K (MJ/m) CoPt 690 185 1.1 86 4.18 &3.645 0.985 1245 1.6 (a-series1) 230 1.2 0.54 2.6 3.675 0.992 2.6 CoPt 500 190 1.4 7 - 3.66 - 1215 !0.2 (b-series1) 140 1.2 0 &2.61 3.685 0.993 0.7 CoPt 650 205 1.4 42 4.18 3.635 0.988 1260 0.05 (c-series2) 210 1.1 0.1 2.59 3.66 0.993 1 CoPt 700 280 1.9 90 4.205 - 0.985 1105 0.8 (d-series3) 190 1.8 0 2.615 3.7 - 1.5 The missing lattice parameters are related to the insufficient contributions of the minority phase to the diffracted intensities. Fig. 3. X-ray intensities around the 0 0 0 4, 2 2 2 CoVPt\V and Fig. 2. X-ray diffracted intensities as a function of the scattering 0 0 0 4 Ru reflections measured on a D500 diffractometer for the vector q"4 sin / around the 0 0 0 1 reflection measured, on CoPt alloy grown at 690 K (samplea). The dotted curves are a D500 Siemens diffractometer, for three samples grown at the two Gaussian components attributed to the majority HCP 690 K (-), 650 K (- - -) and 500 K (....) described in Fig. 1. and minority FCC phases. Table 1, are deduced from the integrated intensities The crystalline quality of the HCP stacking, of these peaks, normalized by the square of their characterized by the width of the 1 0 1 reflections is respective structure factors. Up to 700 K, an in- clearly better in sample a than in sample d. The crease of the deposition temperature favors HCP monocrystalline quality of the alloy film is directly stacking, while beyond 700 K, as shown by Harp et dependent on the Ru buffer, as indicated by the al. [9], FCC stacking becomes predominant in widths of the rocking curves around the 0 0 0 2 these films. CoPt and Ru reflections given in Table 1. 66 M. Maret et al. / Journal of Magnetism and Magnetic Materials 191 (1999) 61-71 Fig. 4. 1 0 l diffraction rods referred to the HCP phase measured on the DIF4C diffractometer at LURE(Orsay) for the four samples presented in Fig. 1; along this rod are superimposed the 0 0 2 and 1 1 1 for the FCC phase and 1 1 1 and 0 0 2 for the twinned FCC. Chemical long-range order along the growth The in-plane lattice parameter extracted from these direction can be characterized by a parameter S reflections, which for the HCP phase is in good deduced from the ratio of the I    and agreement with that deduced from HCP 1 0 0 re- I    diffracted intensities, corrected for absorp- flection, also holds for the FCC phase (labelled tion and Lorentz polarization factor. It can be a$!! written as S"( , in Table 1). ! )/2x., where  and  From Table 1, the deviations (c/a)* from the are the occupancy rates of Pt atoms on the Pt-rich ideal HCP stacking smaller than 1 reveal that the and Pt-poor alternate (0 0 0 1) planes, with HCP stacking in the alloy films is under strain. # "2x.; S is then equal to 1 as one layer Similarly, as shown by the deviation from the FCC is occupied by only Co atoms (i.e. "0). Chemical ideal stacking characterized by a ordering in these films is really uniaxial since ,/a, ratio, the FCC stacking is also slightly compressed. The no  0 0 and  0 0 reflections, signatures of in-plane S parameters of samples a and c do not fit com- chemical ordering, were detected from synchrotron pletely with the growth temperature dependence of radiation measurements in symmetric transmission the chemical LRO parameter found in Ref. [9], geometry. The values of the in-plane lattice para- which can be attributed to the uncertainty of tem- meter a&!. and the lateral coherence length ¸, of perature measurements. However, as shown in the HCP phase deduced from the 1 0 0 CoPt peak Fig. 5, our data (except for series 3) display the are reported in Table 1. No separation is observed same increase of S with the HCP volume fraction, between the 0 2 2 FCC and 2 2 0 HCP reflections. as that reported by Harp et al. [9]. M. Maret et al. / Journal of Magnetism and Magnetic Materials 191 (1999) 61-71 67 3.2. Magnetic measurements Fig. 6 shows the parallel and perpendicular mag- netization hysteresis loops measured at 30 K for the four samples. The magnetic behavior of these sam- ples changes drastically with their structure, since for samples a and d, mostly HCP, the easy axis of magnetization is along the growth direction [0 0 0 1], while for sample b, mostly FCC, the easy axis lies in the film plane and for sample c, consist- ing of equivalent HCP and FCC volume fractions, an intermediate behavior is observed. The saturation magnetization and the magnetic anisotropy energies, K Fig. 5. Chemical long-range order parameter S ( ) and uniaxial and K, deduced from the magnetic anisotropy K SQUID hysteresis loops are listed in Table 1. The  ( ) as a function of the HCP volume fraction for the films a, b and c grown on Ru(0 0 0 1). For effective anisotropy found for the chemically long- comparison, the S values for CoPt films grown on sap- range ordered HCP film (sample a, 1.6 MJ/m) phire(0 0 0 1) measured by Harp et al. [9] are also displayed ( ). is comparable to the value measured in [Co(3 As)/Pt(10 As)] multilayers [12] and is twice larger than the value of the chemically disordered Nevertheless, even if the formation of HCP HCP film (sample d, 0.8 MJ/m). stacking and chemical ordering occur in the same Besides, as shown in Fig. 5, the uniaxial anisot- growth temperature range, chemical ordering in the ropies, K minority FCC phase cannot be excluded. This , display the same HCP volume fraction dependence as the chemical parameter question was recently elucidated from digital S. Therefore, the magneto-crystalline anisotropy processing of FCC areas in cross-section HRTEM increases with both the HCP volume fraction and images of sample a and 2D-Fourier transformation. the uniaxial chemical LRO. It is well established Chemical ordering in the minority FCC phase was that the origin of the magneto-crystalline anisot- indeed confirmed by the presence of an intermedi- ropy is the spin-orbit interaction which induces ate peak referring to a    reflection along the an orbital moment coupling the total (spin#or- [1 1 1] growth direction [11]. Chemical ordering bital) magnetic moment to the crystal axes. along the growth direction was also observed by The enhancement of the perpendicular anisot- X-ray diffraction in an epitaxial (1 1 1)CoPt ropy in the LRO HCP films results clearly from film grown at 690 K and adopting only FCC stack- the uniaxial symmetry of the HCP phase and from ing. the 3d-5d hybridization preferentially along the Coming back to the absence of a 0 0 0 1 peak for c-axis and the large 5d spin-orbit coupling of Pt the alloy films of series 3 grown in a temperature atoms. range, where chemical ordering was expected, we Also, it is worth noting that the K can emphasize the larger out-of-plane (1.9°) and  value of the chemically disordered HCP film (sample d) of in-plane mosaicities (3°) measured for this series 1.5 MJ/m is significantly larger than the value of compared to those measured for sample a (1.1 and pure HCP Co equal to 1 MJ/m. This result sug- 1.5°, respectively). Hence, the lateral and normal gests the existence of anisotropic chemical short- extents of alternate Co-rich and Co-poor plane range order, such that the CoPt nearest-neighbor stacking is reduced by the presence of a great num- pairs are preferentially oriented along the surface ber of steps, which would prevent constructive in- normal, leading to an enhancement of perpendicu- terference effects between second-neighbor (0 0 0 1) lar anisotropy compared to HCP Co. As already planes, which are at the origin of the 0 0 0 1 super- done for CoPt structure peak in the X-ray pattern.  films [4], XAFS measurements at the Pt-L edge using the polarization of synchrotron 68 M. Maret et al. / Journal of Magnetism and Magnetic Materials 191 (1999) 61-71 Fig. 6. Magnetization hyteresis loops measured at 30 K for the four samples of Fig. 1, with the applied field perpendicular (-) and parallel to the film plane (- -). radiation have confirmed the existence of such an- advancing surface. This point is supported by two isotropic local order [13]. recent studies reported in the literature: (i) a crystal- lographic LEED analysis which has revealed at the 4. Discussion (1 1 1) surface of a random FCC CoPt single crystal the existence of a pure Pt segregated top The chemical LRO along the growth direction layer and an oscillatory composition profile observed in the CoPt film (sample a) results in through the second and third underlying planes a stacking of alternate Pt-rich and Pt-poor planes [15]; (ii) resonant surface magnetic X-ray diffrac- whose Pt compositions are an average equal to 28 tion measurements on the (1 1 1) surface of a ferro- and 8 at%. Such ordering is really different from magnetic Co that established in the FCC L1 Pt alloy which have shown a reduction -type ordered of Pt magnetic moment on the top layer also re- CoPt phase [2] or in the HCP DO-type ordered lated to a reduced Co concentration [16]. The AB compounds (such as CoMo or CoNb) [14], second mechanism is based on a dominant surface since along the [1 1 1] or [0 0 0 1] direction, all the diffusion process compared to bulk diffusion in the planes have the same AB composition. Therefore temperature range which favors LRO. This is sug- this uniaxial ordering is promoted by the MBE gested by the values of the relaxation times for technique and is established along the advancing LRO determined in CoPt surface during the co-deposition of Co and Pt  bulk alloy [17], which decrease from 170 h (i.e much longer than the de- atoms. position time) to 8 h, as the temperature increases We suggest that two effects would drive this from 690 to 800 K. Thus driven by these two effects, LRO. The first effect is the segregation of Pt at the the difference of composition between the two M. Maret et al. / Journal of Magnetism and Magnetic Materials 191 (1999) 61-71 69 uppermost atomic layers is successively frozen as the free surface advances, i.e the alloy film can be seen as a stacking of surface alloys. On increasing the growth temperature, the atomic diffusion in the bulk increases and tends to annihilate the atomic arrangements at the free sur- face as well as in the buried layers. Inversely if the growth temperature is relatively low, surface diffu- sion is then too weak for favoring surface arrange- ments between Co and Pt atoms. Therefore, there exists a growth temperature, found around 690 K, which maximizes the uniaxial chemical ordering. Such behavior was recently described by a simple model based on both bulk and surface effects (diffu- sion and interactions) which reproduces quite well the growth temperature dependence of LRO in both CoRu and &CoPt films [19]. From this Fig. 7. Effect of ex situ annealing on the 0 0 0 1 peak measured model, the activation energies for bulk and surface in the Co diffusion, E Pt film grown at 690 K (sample a) on the HRXRD and E, together with a maximum diffractometer: as-deposited (-), annealed at 690 K for 6 h (....) chemical long-range order parameter S and one week (- - -).  (corre- sponding to the equilibrium value near the free surface) were obtained for CoPt films: E " 2.5$1 eV, E"0.39$0.06 eV and S "0.65. progressively with annealing time. Together with The value of E is in relatively good agreement with the decrease of chemical ordering, we observe those found in L1-type ordered CoPt bulk alloys a small increase in the distance between (0 0 0 1) of 3.1 eV deduced from LRO kinetics [17] and in planes of about 5;10\ As. The atomic rearrange- ordered L1-type CoPt films of 2.1 eV [18]. S , ments, which take place during annealing, are now which is close to the maximum value of 0.63 found driven by the only bulk interactions and the alloy in the CoPt films grown on sapphire [9], is film evolves towards its equilibrium state. therefore not equal to 1 indicating that alternate This result is in agreement with the total energy pure Co and mixed (Co,Pt) planes cannot be calculations carried out for three different struc- achieved by the MBE process, due to weak bulk tures of the Co diffusion. Pt compound, namely L1, DO and DO The stability of the chemically ordered Co  [20]. The labelled DO structure is VPt\V a modified form of DO films has been investigated through ex situ anneal-  which yields chemical ordering along the c-axis observed in the epitaxial ing treatments applied to the most ordered film films, i.e in the two alternate planes A-B along c, (sample a) at the same temperature as its deposition one plane contains only Co atoms. From these temperature. Fig. 7 shows the change in the (0 0 0 1) calculations, based on a TB-LMTO method, the superstructure peak with annealing time, measured ferromagnetic DO on the HR Philips diffractometer. The chemical  phase is the least stable, while the most stable is the ferromagnetic DO long-range order parameters deduced from the ra-  phase. From long-time annealing, it should be possible to tio I   /I    are equal to 0.43 and 0.1 for confirm whether the Co annealing times equal to 6 h and one week. We VPt\V films evolve towards the DO have checked that there is no noticeable change in  structure foreseen by the theory, instead of the L1 the volume fractions of HCP and FCC stackings  structure already observed in a bulk Co in the annealed films. It appears clearly that Pt alloy [2]. Fig. 8 displays the decrease of S with annealing the uniaxial chemical ordering promoted by the time found in sample a. As indicated by the dashed MBE technique is metastable, since it disappears curve, such a decrease cannot be described by 70 M. Maret et al. / Journal of Magnetism and Magnetic Materials 191 (1999) 61-71 Fig. 8. Decrease of the chemical long-range order parameter as Fig. 9. Polar Kerr hysteresis loops measured at 300 K in the a function of the annealing time in the Co as-deposited Co Pt film grown at Pt film grown at 690 K (sample a) (-), and 690 K (sample a). For t"0, S"0.65, i.e. the equilibrium value after ex situ annealing at 690 K for 6 h (- - -) and one week (....). at the free surface from Ref. [19]. Here, the description by a simple exponential law (dotted curve) is not valid (see also text). for example the existence of Pt- or Ru-enriched regions localized at the grain boundaries where Ru buffer or Pt cap layer can diffuse preferentially. In a simple exponential law in the whole annealing fact, these (Pt,Ru)-enriched regions contribute to time range, such as S"S  exp(!t/ ), where is a shoulder on the left side of the (0 0 0 2)Ru peak in a relaxation time. Thus, this behavior differs from the X-ray pattern of the long-time annealed film the relaxation process observed in bulk CoPt and represent &1.5% of the alloy volume. The when returning to its equilibrium ordered state. uniaxial magnetic anisotropy for the film annealed The dashed curve, which fits relatively well the at 690 K for one week, deduced from the SQUID S values for short times, corresponds to a relaxa- magnetization loops is equal to 1 MJ/m, and be- tion time of 5;10 s, i.e. much shorter than that comes close to the value of pure HCP Co. There- obtained for LRO in bulk CoPt about 10 s. This fore, from these annealing treatments, a decrease of deviation could be related to the existence of a va- the chemical LRO parameter from 0.54 to 0.1 (cor- cancy supersaturation at the vicinity of defects responding to a decrease of the CoPt pairs oriented (such as grain boundaries or dislocations), which out of the film plane), leads to a strong decrease of disappears progressively during the anneal time. the uniaxial anisotropy from 2.6 to 1 MJ/m. Therefore, the change of S over long times should In conclusion, we have shown that the pre- be described at least by a longer relaxation time, dominantly HCP&Co that would be the equilibrium time. The decrease of Pt films grown on Ru(0 0 0 1) buffer using a mica substrate can exhibit chemical ordering in the annealed film modifies chemical LRO along the growth direction, charac- significantly its magnetic properties, as illustrated terized by alternate Co-rich and Co-poor planes. In by the polar Kerr effect loops measured at 300 K contrast, no chemical ordering in the (0 0 0 1) (Fig. 9). Thus, a regular increase of the saturation planes was observed. This ordering, strongly de- field is observed together with a significant increase pendent on the growth temperature and also on the of the coercivity after long-time annealing. Such an crystalline quality of the film, enhances perpendicu- increase of coercivity would be due to a separation lar magnetic anisotropy, characterized by uniaxial of magnetic domains by non-magnetic regions, as anisotropy energies up to 2.6 MJ/m as found in M. Maret et al. / Journal of Magnetism and Magnetic Materials 191 (1999) 61-71 71 a highly chemically ordered CoPt film grown [5] M. Maret, M.C. Cadeville, R. Poinsot, A. Herr, E. Be- at 690 K. Yet, after ex situ annealing performed at aurepaire, C. Monier, J. Magn. Magn. Mater. 166 (1997) the same temperature as the growth temperature, 45. the uniaxial ordering strongly decreases together [6] P.W. Rooney, A.L. Shapiro, M.Q. Tran, F. Hellman, Phys. Rev. Lett. 75 (1995) 1843. with a large decrease of perpendicular anisotropy. [7] W. Grange, J.P. Kappler, M. Maret, J. Vogel, A. Fontaine, This result indicates that in agreement with theo- F. Petroff, G. Krill, A. Rogalev, J. Goulon, M. Finazzi, N. retical calculations [20], the MBE-promoted low- Brookes, J. Appl. Phys. 83 (1998) 6617. temperature chemically ordered alloy is metastable. [8] G.R. Harp, D. Weller, T.A. Rabedeau, R.F.C. Farrow, R.F. Both Pt segregation at the advancing surface and Marks, Mater Res. Soc. Symp. Proc. 313 (1993) 493. [9] G.R. Harp, D. Weller, T.A. Rabedeau, R.F.C. Farrow, associated dominant surface diffusion during co- M.F. Toney, Phys. Rev. Lett. 71 (1993) 2493. deposition at temperatures close to 690 K are sug- [10] M. Maret, M.C. Cadeville, W. Staiger, E. Beaurepaire, R. gested to be the driving effects of such unexpected Poinsot, A. Herr, Thin Solid Films 275 (1996) 224. chemical ordering effects in alloy films close to the [11] M. Maret, C. Ulhaq-Bouillet, W. Staiger, M.C. Cadeville, Co S. Lefebvre, M. Bessie re, Thin Solid Films 319 (1998) Pt composition. On the contrary, in equiatomic 191. (0 0 1)CoPt films grown on a (0 0 1) orientated sub- [12] C.J. Lin, G.L. Gorman, C.H. Lee, R.F.C. Farrow, E.E. strate, a similar mechanism would favor the forma- Marinero, H.V. Do, H. Notarys, C.J. Chien, J. Magn. tion of the tetragonal L1 Magn. Mater. 93 (1991) 194. -type ordered equilibrium phase, as found in Ref. [8]. [13] C. Meneghini, M. Maret, V. Parasote, M.C. Cadeville, J.L. Hazemann, R. Cortes, S. Colonna, Eur. Phys. J. B (1998), in press. Acknowledgements [14] P. Villars, L.D. Calvert, Pearson's Handbook of Crystallo- graphic Data for Intermetallic Phases, vol. 2, American Society for Metals, Cleveland, OH, 1985. We thank J. Arabski for the MBE growth of the [15] Y. Gauthier, R. Baudoing-Savois, J.M. Bugnard, U. Bardi, samples and G. Schmerber for the annealing treat- A. Atrei, Surf. Sci. 276 (1992) 1. ments. [16] S. Ferrer, P. Fajardo, F. de Bergevin, J. Alvarez, X. Tor- relles, H.A. Van der Vegt, V.H. Etgens, Phys. Rev. Lett. 77 (1997) 747. [17] C.E. Dahmani, M.C. Cadeville, V. Pierron-Bohnes, Acta References Metall. 33 (1985) 269. [18] D.M. Artymowicz, B.M. Lairson, B.M. Clemens, J. Crystal [1] D. Weller, in: H. Ebert, G. Schu¨tz (Eds.), Spin Orbit In- Growth 169 (1996) 83. fluenced Spectroscopies of Magnetic Solids, Lecture Notes [19] L. Bouzidi, V. Pierron-Bohnes, O. Haemmerle´, C. Ulhaq- in Physics, Vol. 466, Springer, Berlin, 1996, p. 1. Bouillet, M.C. Cadeville, Thin Solid Films 318 (1998) [2] C. Leroux, M.C. Cadeville, V. Pierron-Bohnes, G. Inden, 215. F. Hinz, J. Phys. F 18 (1988) 2033. [20] G. Morai¨tis, J.C. Parlebas, M.A. Khan, J. Phys.: Condens. [3] J.M. Sanchez, J.L. Moran-Lopez, C. Leroux, M.C. Ca- Matter 8 (1996) 1151. deville, J. Phys.: Condens. Matter 1 (1989) 491. [21] W. Grange, J.P. Kappler, M. Maret, J. Vogel, A. Fontaine, [4] C. Meneghini, M. Maret, M.C. Cadeville, J.L. Hazemann, F. Petroff, G. Krill, A. Rogalev, J. Goulon, M. Finazzi, N. J. Phys. IV France 7 C2 (1997) 1115. Brookes, Phys. Rev. B 58 (1998) 6298.