Enhancement of L1 transformation in Fe/Pt multilayer by Cu addition

Magnetic storage media has marketed a remarkable increase in the storage capacity of hard disk drives with a critical role of high anisotropic materials in advancing magnetic recording.^1,2 High magnetic anisotropy is crucial as the size reduction of a bit is constrained by the thermal instability of magnetization whereas high anisotropy ensures the thermal stability and thus makes the device suitable for high density recording media.¹ In this context, L1₀ FePt with huge uniaxial magnetic anisotropy and high saturation magnetization is a promising candidate for future generation recording media.^1,2 In FePt alloy, fabricated at ambient conditions, iron and platinum atoms are randomly distributed at fcc sites and magnetic anisotropy is very low. Post deposition treatment like thermal annealing at temperatures in the range 500^oC – 600^oC is required in order to achieve ordered and high anisotropic L1₀ phase.² As a result grain size increases which is detrimental for application in high density recording media. Enormous efforts have been done in literature to enhance kinetics of order disorder transformation in FePt system so as to reduce the order-disorder transition temperature. Most common among them are chemical process,³ deposition of FePt on heated substrate,^4–6 ion irradiation,^7,8 multilayer structure,⁹ and addition of ternary element to FePt alloys like N, Au, Ag and Cu.^10–15 

Among the FePt based ternary alloys Cu is the most popular choice as additive element which is very well highlighted in the literature. For example, Gilbert et.al.¹⁶ have achieved a T_c of 400^oC and anisotropy of 3x10^-7 erg/cm³ for the optimal concentration of Cu. Whereas, Li et.al.¹⁷ obtained a T_c of 350^oC and coercive force of about 605 kAm⁻¹ in FePtCu thin films. Apart from ternary alloy approach, another interesting way to reduce the order disorder transition temperature of FePt is through solid state reaction between Fe and Pt in multilayer structure. It has been widely investigated that multilayer of Fe and Pt exhibits lower ordering temperature.^18–20 

This is to be noted that both the multilayer and ternary element approach are found to be highly effective in reducing the ordering temperature. In the present work an alternative approach is presented to lower the L1₀ transformation temperature of FePt system in which ternary element Cu is alloyed with Fe and Pt in the multilayer stack. Detailed structural and magnetic investigations have been done and the results are compared with FePtCu alloy system.

A lot of focus has also been there on the role of the additive element in accelerating the order disorder transition. Moreover understanding the mechanism by which such additives influence the ordering process is vital for the use of FePt in recording media. In alloy systems various mechanisms has been suggested in literature that accounts for the role of Cu in FePt alloy.^{13–16,21,22} While in multilayer lowering of ordering temperature is commonly attributed to enhanced long range diffusion. Thus in the present work, two multilayers viz. Cu at the Fe/Pt interface and Cu alloyed with Fe and Pt in multilayer structure are fabricated. Diffusion studies are done on both the multilayer films so as to have an insight on to the mechanism of transformation kinetics.

Two multilayer films viz. Fe(Cu)/Pt(Cu) and Fe/Cu/Pt designated as M1 and M2 respectively, and one FePtCu alloy film were deposited using dc magnetron sputtering technique on Si (100) substrates using a AJA Int. Inc. make ATC Orion-8 series sputtering system. In M1 Cu is alloyed with Fe and Pt and in M2 Cu is deposited at one interface of Fe/Pt. The nominal structure of the M1 was Si(substrate)/[Fe₈₅Cu₁₅(19Å)/Pt₈₅Cu₁₅(27Å)]_x10, M2 was Si(substrate)/[Fe(16Å)/Cu(6 Å)/Pt(24 Å)]_x10 and that of alloy film was [(Fe₅₀Pt₅₀)_0.85]Cu_0.15 respectively. The thicknesses of the Fe and Pt were so chosen that their relative atomic percent ration remained 50:50 and an optimized concentration of Cu i.e. 15 at.% was maintained in all three films. Base pressure of the chamber was ∼1×10^-7 Torr, while during deposition pressure was ∼3 mTorr. Post-deposition annealing of the samples was done in a vacuum of 2×10^-6 Torr. The x-ray diffraction (XRD) measurements were carried out using a Bruker D8 Advance x-ray diffractometer with Cu Kα radiation. The magnetic hysteresis loops were measured by high resolution magneto-optical Kerr microscope (M/s Evico Magnetics, Germany). The measurements are carried out in longitudinal mode. The x-ray reflectivity measurement of the multilayers was done using Bruker D8 Diffractometer.

Figure 1 shows the XRD pattern of multilayer M1 as a function of isochronal annealing at indicated temperatures. XRD pattern of the pristine film consists mainly of Pt(111) reflections surrounded by satellite peaks that arise due the periodicity in multilayer structure. Hence, presence of these peaks in the present multilayer indicates the uniformity of individual layer thickness throughout the multilayer structure. The separation between the satellites peaks is equal to the bilayer period of the multilayers. The same has been calculated using the formula

where i and j the orders of satellite and ω is the half of scattering angle of that satellite.²³ The bilayer period calculated by using the above formula comes out to be 50Å which is consistent with x-ray reflectivity data (Table II). Now as the annealing temperature increases neither Pt(111) peak shows any shifting nor the intensity of satellite peak varies significantly upto 200^oC. This infers that upto this temperature the multilayer structure is preserved and there is no significant intermixing taking place. However at 300^oC there are many changes evolving in XRD spectra which are summarized as follows, (i) satellite peaks completely attenuate, which is attributed to destruction in multilayer periodicity due to intermixing of Fe(Cu) and Pt(Cu) layers in M1 multilayer. (ii) secondly a strong superlattice peak emerges at 2θ=24^o which is present only in alloy exhibiting some degree of order. The intensity of this peak depends upon the degree of ordering. Emergence of this superlattice peak in the present multilayer is because of formation of L1₀ ordered fct phase of FePt at this temperature. (iii) Apart from that, Pt (111) peak shifts to higher angle (2θ=41.2^o) and additional peaks appears at 2θ=47.3^o and 49.6^o that can be indexed to (200), (002) reflections of ordered fct phase of FePt, respectively. It is to be noted that splitting of (002) and (200) peaks are signature of tetragonal L1₀ phase formation.

Figure 2 shows the XRD pattern of multilayer M2 at indicated annealing temperatures. In M2 also, satellite peaks are observed similar to M1. The bilayer period calculated for M2 using equation (1) comes out to be 47Å which again matches with the bilayer period obtained from x-ray reflectivity data (Table II). One may note that similar to multilayer M1 there is no significant variation in the XRD pattern up to 200^oC. Moreover at 300^oC, the intensity of super lattice peak and fundamental peaks are lower in comparison to M1 multilayer, however the position of Pt(111) peaks is same as that for M1 indicating the same Fe/Pt ratio in the two multilayers.²⁴ Now if we take a look at the XRD pattern of FePtCu alloy film (Fig. 3) one may notice that in the as-deposited film there is only one Bragg peak at 41.19^o corresponding to disordered fcc phase of FePt. This peak stays at the same position till 200^oC while at 300^oC we observed the appearance of superlattice peak corresponding to fct phase along with its other fundamental reflections viz. (111) at 41.3^o, (200) at 47.2^o and (002) at 49.6^o. However, the intensity of the superlattice peak in this case is weaker than that in the Fe(Cu)/Pt(Cu) multilayer sample after same annealing treatment.

For the quantitative estimate of degree of ordering from the XRD data exhibiting broad and overlapped peaks, the order parameter is calculated from the intensity ratio of superlattice peak and fundamental peak.²⁵ For the sake of clarity, fitted (001) superlattice peak and (111) fundamental peak are shown in figure 4 for all the three films and relative intensity ratio has been calculated. The results of fitting are summarized in Table I. One may clearly see that the area ration comes out to be 0.45, 0.13 and 0.33 for M1, M2 and alloy films respectively. From these results one can infer that after annealing at 300^oC almost complete intermixing of Fe and Pt layers takes place and an ordered fct FePt phase is formed in all the films but the degree of L1₀ ordering in Fe/Cu/Pt multilayer and alloy film is much smaller than Fe(Cu)/Pt(Cu) multilayer, indicating that alloying Cu with Fe and Pt in multilayer structure leads to the enhanced transformation kinetics than insertion of Cu at the Fe/Pt interface or simply alloying Cu with Fe and Pt. The grain size estimated using Debye-Scherrer formula²⁶ for all the three films comes out to be around 19 nm.

On-set of L1₀ fct ordering in FePt is marked by increase in magnetic coercivity (H_c) when the magnetic measurements are carried out in-plane. High values of coercive field ensure thermal stability of magnetization in this phase and hence make it suitable for device application. Thus H_c may be considered as a measuring parameter of L1₀ ordering. In order to measure the H_c in our samples, MOKE measurements have been done in longitudinal geometry. Figure 5 shows the magnetic hysteresis curve for multilayers and alloy film annealed at indicated temperatures, recorded from MOKE measurements. It is to be noted that for all the three systems the H_c is small till 200^oC. At 300^oC H_c shoots up to very high value which is of 8200 Oe, 7200 Oe, 5300 Oe for multilayer M1, M2 and alloy film respectively. Again, H_c is found to be the highest in multilayer M1 that attributes to higher degree of L1₀ ordering in multilayer M1. Detailed comparison of H_c between alloy and multilayers can be seen from Fig. 6. These results are consistent with the XRD results which also indicate that degree of ordering is larger in M1 in comparison to M2 and FePtCu alloy.

Although the free energy of L1₀ Phase is lower than that of fcc phase, transformation to L1₀ phase is constrained kinetically because of low diffusivity of the constituent species.²⁷ Therefore diffusion studies in these systems can throw some light on the atomic level mechanism responsible for enhanced transformation process. Figures 7 and 8 show the x-ray reflectivity data of multilayer M1 and M2 fitted using Parratt’s formalism. The fitting parameters for both the multilayers have been summarized in Table II.

In M1, the thicknesses of FeCu and PtCu layers come out to be 22.5 Å and 29 Å respectively for pristine film. One may note that with thermal annealing scattering length density of FeCu and PtCu decreases till 200^oC, which may be attributed to movement of Cu in Pt layer as Fe and Cu, is an immiscible system with positive heat of mixing. At 300^oC, height of Bragg peak decreases substantially because of possible intermixing of Fe and Pt, which result in low scattering length density contrast of both the layers. The variation of scattering length density of FeCu and PtCu layers as a function of annealing temperature is shown in Figure 9 (top).

In M2, fitting of pristine sample gives the thicknesses of Fe, Cu and Pt layer as 17 Å, 6 Å and 24 Å respectively. With thermal annealing at 100^oC, no significant interdiffusion has been observed. At 200^oC, an intermixed layer of around 6Å is formed at Pt/Cu interface with intermediate scattering length density. At 300^oC, x-ray reflectivity has been fitted considering two interfaces of Pt, one mixed with Cu, as Fe and Cu is an immiscible system, and other mixed with Fe. The scattering length density contrast between two layers of Pt i.e. FePt and PtCu has been reduces significantly, which results in reduction in height of Bragg peak. Figure 9 (bottom) depicts the variation of scattering length density for Fe and Pt layers as function of annealing temperature.

Interdiffusion in multilayer can also be studied by the decay of first order Bragg peak in x-ray reflectivity.²⁸ Insets of figures 7 and 8 show the first order Bragg peak of multilayers M1 and M2 respectively as a function of annealing temperatures. Pristine samples exhibit a well-defined first order Bragg peak due to periodicity in the multilayers structure. With increasing annealing temperature, height of the Bragg peak decreases due to interdiffusion across interfaces. Variations in the height of the Bragg peak can be used to determine the diffusion constant for interdiffusion using the relation.²¹ 

where I(0) and I(T) are the intensities of the n^th-order Bragg peak in the as-deposited sample and after annealing for time t, respectively. D is the constant for interdiffusion and Λ is the periodicity of the multilayer.

Table III gives the diffusion constant for the multilayers M1 and M2 at different annealing temperatures. One can see that at 200^oC the diffusivity in both the multilayers is quite low. At 300^oC, diffusivity increases by almost two orders of magnitude. However, one can see that diffusivity in multilayer M1 is only marginally higher as compared to that in multilayer M2. It may be noted that Fe and Cu have a positive heat of mixing. Thus, a layer of Cu at one of the interfaces of Fe and Pt may partially deter the intermixing. This may be the reason for a lower inter-diffusivity in multilayer M2. On the other hand as seen from x-ray diffraction measurements the L1₀ transformation rate in M1 is significantly higher than in that in M2. These results can be understood as follows:

Earlier studies have shown that in case the starting structure is a multilayer then the fcc phase which is formed after intermixing of different layers has higher concentration of defects which results in an enhanced diffusivity as compared to the case when the initial structure is a fcc alloy.⁹ Further, it was shown that if a ternary element like nitrogen is incorporated in fcc film, during subsequent thermal annealing the ternary element comes out of the fcc structure leaving behind defects which further enhance the diffusivity of the constituent elements resulting in a faster L1₀ transformation.¹⁰ In the present case, L1₀ transformation rate is highest in the case of M1 multilayer since in this case both the above mentioned factors are operative. In the alloy film the transformation rate is slower as only the second factor is operative. In case of multilayer M2, in which Cu is in the form of a third layer between Fe and Pt layers, Cu does not get alloyed in the fcc phase due to immiscibility with Fe. In this case the transformation rate is the lowest. This suggests that creation of structural defects due to out diffusion of Cu from the fcc grains is a dominant factor in enhancing the diffusivities and thus the L1₀ transformation rate.

In the present work, an alternative approach is presented to enhance the L1₀ kinetics in FePt based system. Two multilayers M1 and M2 with (Cu alloyed with Fe and Pt in multilayer stack and Cu at Fe/Pt interface respectively) and one FePtCu alloy film have been studied. Detailed structural and magnetic studies were done on all the films.

Both XRD and MOKE measurements suggest that L1₀ transformation is the fastest in case of multilayer M1 in which an optimum amount of Cu is alloyed with both Fe and Pt layers. L1₀ transformation rate in the alloy film with similar overall composition is relatively slower. This shows that the alternative approach suggested in this work, which combines the two strategies, namely using a multilayer structure as the starting point and incorporating a ternary element which has limited solubility in the fcc FePt phase, indeed results in further enhancement of the L1₀ transformation. The transformation rate is found to be the lowest in multilayer M2 in which Cu is incorporated as a separate layer. This suggests that the creation of structural defects during the out diffusion of Cu atoms from the fcc phase during thermal annealing is the dominant cause for the observed enhancement in the L1₀ transformation.

KS, DST INSPIRE faculty thanks DST, New Delhi for financial assistance in the form of INSPIRE faculty award (DST/INSPIRE/04/2013/000772). The work was partly supported by BRNS project No. 37(3)/14/17/2015/BRNS.