FePt-TiO2exchange coupled composite mediawith well-isolated columnar microstructure for high density magnetic recording
C. J. Jiang1*, J. S. Chen1, J.F. Hu,2G. M. Chow1
1Department of Materials Science and Engineering, NationalUniversity of Singapore, 117574, Republic of Singapore
2Data Storage Institute, Singapore, 117608.
We reported the fabrication of (001) textured FePt-TiO2exchange coupled composite (ECC)media including hard/soft bilayer and multilayerwith well isolated columnar microstructures. The magnetic anisotropy of FePt-TiO2was adjusted by applying various substrate bias during film deposition. The cross-sectional TEM images showed isolated granular microstructures in single hard layer, bilayer and multilayermedia. For the bilayer media, it was observed that both the coercivity and magnetization squareness of composite media decreased with increasing thickness of the soft layer. A soft layer with the thickness of 4 nm was more effective to significantly reduce the switching field and maintain ahigher thermal stability factor than that of others. Incoherent switching behavior was observed asthesoft layer thicknesswas increased to 6 nm. For multilayer media, it was found that the out-of-plane coercivity decreased to 6.5 kOe, which was closeto half of that of the single hard layer. However, the thermal stability factor of the multilayer media slightly decreased compared with the single hard layer and bilayer media due to finite thickness. The results suggested a way to obtainthe adjustable anisotropy forECC media in high density magnetic recording application.
*Corresponding author:
I. INTRODUCTION
With the increase in demand for high recording areal density, L10 phase FePt with high anisotropy (~7×107 ergs/cc) have received considerable attention in recent years due to its potential application for ultrahigh density magnetic recording1-5.In order to realize its practical application, it is a key challenge to obtain well isolated small grainsof FePt with the (001) texture. Due to the high anisotropy energy constant, FePt media is not writable using the conventional recording head. Therefore the exchange spring and ECC mediacomprising hard layer and soft layer have been proposed to reduce the switching field (Hsw) of hard layer while maintaining similar thermal stability6-9. Micrcomagnetic simulations indicated that a continuous variation in the anisotropy in the multilayer recording media can reduce the coercive field more effectively than that of bilayer media10-12. However, the preparations of ECC bilayer and multilayer media are challenging. Experimentally, FePtC ECC media consisted of L10 FePt/fcc FePt has been recently reported13. It was found that the coercivity reduced with the increase of magnetically soft layer thickness. L10-FePt/FeAu graded media were also reported, in which the graded anisotropy was controlled by interdiffusion at interface after postannealing14. However, in order to decrease the lateral exchange coupling among the grains, the ECC media with well isolated columnar microstructure is required. It is difficult to meet this requirementby varying the deposition temperature or controlling the post-deposition annealing. In this work, the anisotropy of L10 FePt-TiO2 media was controlled by varying the substrate bias during film deposition without changing the deposition temperature or employing post-deposition annealing. The magnetic properties and microstructuresof the ECCbilayer and multilayer media were investigated.
II. EXPERIMENTAL DETAILS
FePt-TiO2thin films with a structure of glass/Cr90Ru10/MgO/FePt-TiO2 (10 nm, 20 vol%) were prepared by AJA sputtering system(con-focal target sputtering) with a base pressure better than 2×10-8 Torr. The CrRu and FePt-TiO2 targets were dc sputtered while MgO target was rf sputtered in Ar.The film thickness of CrRu and MgO were fixed at 30 nm and 2 nm, respectively. The Ar working pressures were 1 and 10 mTorr for CrRu and MgO layers, respectively. The deposition temperatures for CrRu and MgO layers were fixed at 320 °C and 100 °C, respectively. The working pressure of depositing FePt-TiO2 layer was 15 mTorr. The sputtering power was 100 W and deposition temperature was 500°C. Different substrate bias power varying from 0 to 40 W was applied during deposition. The crystallographic texture of the films was examined by x-ray diffraction (XRD) using a Cu Kα radiation. The microstructures of the films were characterized using transmission electron microscopy (TEM). The magnetic properties of FePt-TiO2 films were measured using vibrating sample magnetometry (VSM) and alternating gradient force magnetometry (AGFM).
III. RESULTS AND DISCUSSIONS
A. Effect of substrate bias power
Figure 1(a) shows the XRD patterns of FePt-TiO2 nanocomposite thin films with different substrate bias power.All the peaks were attributed to the CrRu and the FePt. Note that the amorphous glass substrate rendered the differentiation of amorphous TiO2 matrix difficult in XRD spectra. Without applying substrate bias power, the FePt shows a (001) preferred orientation. With increasing the substrate bias power, the integrated intensity of the FePt (001) peak decreased, whereas FePt (002) /(200) peaks shifted toward lower angles. This indicated that the chemical ordering of FePt films decreased with increasing substrate bias power. Increased substrate bias power to 30 and 40 W, the supperlattice (001) peak of FePt film almost disappeared and the (002)/(200) peaks shifted to lower angle further, suggesting the transformation of FePt film to the undesirable fcc phase.The inset of Fig. 1 (a) shows the rocking curve of fct FePt (001) peak in different substrate bias power, respectively. It indicated that the full width half-maximum of the rocking curves were increased with increasing substrate bias power.The chemical ordering parameter (S) is defined according to relation I(001)/I(002) =1.87S2, where I (001) and I(002) are the intergrated peak intensities of the superlattice (001) peak and fundamental (002) peak8, 15. The variation of Swith the substrate bias power is shown in Fig. 1 (b). The chemical ordering decreased monotonically with increasing substrate bias power. This can be interpreted as follows: With increasing the substrate bias power the bombardment energy of Ar ions on the film surface increased16. The energy of Ar ions was transferred to surface atoms that could cause either re-sputtering of the already deposited film and/or an increase of atomic mobility along both the film plane and film normal. The deposition rates for different substrate biases remained constant, indicating the re-sputtering effect was small as there was no loss of deposited film mass.The atomic mobility along the film plane would promote the formation of chemical ordered FePt film. The atomic mobility along the film normal however would deteriorate the crystal structure and promoted the formation of fcc FePt. The experimental results indicated that the substrate bias caused the enhancement of atomic mobility along the film normal. This behavior was similar to the effect of implantation.
Figures 2(a) and (b) show the typical out-of-plane and in-plane hysteresis loops of FePt-TiO2films deposited at substrate bias powerof 0 and 40 W, respectively.It was observed that FePt-TiO2 films without substrate biasshowed the perpendicular anisotropy. At the substrate bias powers of 40 W, the film was magnetically soft with in-plane magnetic anisotropy. These results were consistent with the XRD results. Figure 2(c) shows the out-of-plane coercivityHcand magnetocrystalline anisotropy, Ku,as a function of the substrate bias power.For FePt films with perpendicular anisotropy, taking demagnetizing field into account, the anisotropy of films can be roughly estimated by Ku= MsHk/2+2πMs2, where Hkwas the magnetic anisotropy field that can be estimated by extrapolating the hard axis loop; while Ku= MsHk/2 can be obtained for FePt films with in-plane anisotropy.The out-of-plane coercivity monotonically decreased from 11 kOe to 0.5 kOe with increasing substrate bias power from 0 to 30 W.Further increase of substrate bias power, the coercivity remained almost the same as the film deposited with 30 W substrate bias power. Without applying substrate bias, Kuwas as large as 1.7×107 erg/cc., Ku decreased linearly with the increase of the substrate bias power up to 20 W. The FePt films still retained the perpendicular anisotropy.With further increase of the substrate bias power to 30 and 40 W, the FePt films showed in-plane anisotropy with Ku as low as 1-2×106erg/cc.Comparing with the chemical ordering of FePt films at different substrate bias power, the change of FePt films in coercivity and Ku were mainly attributed to the decrease of the chemical ordering of FePt film caused by the bombardment of high energy Ar ions under the substrate bias.
B. Bilayer media
According to the above experimental results, Kudecreased with increasing substrate bias power and approached 2×106erg/cc at 40 W substrate bias. Therefore the bilayer media comprising 10 nm hardest layer of FePt-TiO2 (no substrate bias) and different thickness(2, 4, 6, and 8 nm) of softest layer deposited at substrate bias of 40 W were fabricated.Figure 3 shows the XRD spectra of the bilayermedia with different thickness of the soft layer. The intensity of superlattice FePt (001) peaks remained almost unchanged with different soft layer thicknesses.With increasing soft layer thickness, combined FePt (200) and (002) peaks broadened, which suggestedan increase in the fcc FePt (200) component of the films. These results can be explained as follows: the deposition parameters of the hard layer in all samples were kept constant, which maintained the chemical ordering of the bottom L10 FePt layer. The grains in the overlayer deposited at the substrate bias 40 W were formed with fcc (200) texture which grew epitaxially on FePt (001)of hard layer.
Figure 4 (a) shows the out-of-plane hysteresis loops of the bilayer media with different soft layer thickness. The coercivity and magnetization squareness Mr/Ms, (ratio of remanent magnetization to saturation magnetization)with various thickness of soft layer are summarized in Fig. 4b. With increasing thickness of the soft layer, the coercivity decreased monotonically, whileMr/Ms decreased slightly with the soft layer thickness less than 4 nm, and then decreased hugely at 6 nm. With the soft layer thickness less than 4 nm, the magnetization of soft layer was almost parallel with that of hard layer due to strong exchange coupling between them, which resulted in slight decrease of the squareness. The reduced squareness with increasing soft layer thickness to 6 nm was attributed to the canting of magnetization direction in the soft layer. Due to the high Msoft~900 emu/cc of the soft FePtlayer, the demagnetization energy was as large as~5×106 erg/ cc. Such large demagnetization energy forced the magnetizationof the soft FePt layer to be in the film plane direction.In order to identify the mechanism of the coercivity reduction, we calculated the domain wall width of the soft layer by the following formula17:
, (1)
where Happl was the applied field, Asoft and Msoftwere stiffness constant and saturation magnetization of the soft layer, respectively. For our samples with the soft thickness of 2, 4, 6, and 8 nm and having respective switching fields of 9.0, 8.2, 7.1, and 6.1 kOe, the domain wall width were calculated as 5.0, 5.2, 5.6, and 5.9 nm, respectively. Initially for soft layer thickness 4 nm less thandomain wall width, the switch mechanism would be coherent rotation due to strong exchange coupling between hard and soft layer, which was testified by following recoil loop. In this case, the effective anisotropy of the bilayer can be described by Keff=(KhardVhard +KsoftVsoft)/(Vhard+Vsoft) due to almost same magnetization of two layers. Khard, Ksoft were the anisotropies of the hard and soft layers, Vhard and Vsoft were the volume of hard and soft grain. Due to KsoftKhard, Keffof bilayer media was less than Khard, which resulted in the reduction of coercivity after coupling with a soft layer. When the thickness of soft layer exceeded 6 nm, more than domain wall width, it was enough to form a full domain wall. Therefore, the switching mechanism could become incoherent switching18.The magnetic reversal process with applied magnetic field could be described in thefollowing: the domain wall nucleated in the soft layer, andpropagated to soft/hard interface where it was pinned. Therefore, the coercivity was dependent on the pinning field of the domain wall at the interface between soft and hard layer.In order to confirm the switching mechanismwith the soft layer thickness around domain wall width, the recoil loops of bilayer media with 4 and 6 nm soft layers were measured.The recoil loops were obtained from removal and reapplication of a demagnetization field to a magnetically saturated material, which can provide information about the reversible and irreversible components of the magnetization, as shown inFig. 5.In the case of 4 nm soft layer, the magnetization was slightly increased after removing the applied field. This indicated that the switching mechanism was dominated by coherent rotation. However,with increasing thickness of soft layer to 6 nm,themagnetization after removing the applied field increased. This was due to the increase of reversible magnetization switching, whichresulted from incoherent rotation.
The cross-sectional TEM images of single hard layer and bilayer media with 4 nm thickness of soft layer are shown in Fig. 6.Both single hard layer and bilayer media showed only one layer of well-isolated FePt grains, whereas the CrRu and MgO layers showed the continuous microstructures. It is worth noting that there was no double layer or multilayer structured FePt found in bilayer media, which had been observed in previous report13. This indicated that the soft layer was epitaxially grown on the underlying hard FePt layer. The epitaxial growths of soft layer on hard layer caused a strong vertical exchange coupling between these two layers, which effectively decreased the out-of-plane coercivity.
Thermal stability factor (TSF) at zero magnetic field were investigated by the Sharrock equation19
, (2)
Here, H0 is the intrinsic coercivity without thermal fluctuation; ∆E0 the energy barrier at zero external field, f0 the attempt frequency for an individual switching event, typically of the order of lattice vibrations (109-1010 Hz); kBthe Boltzmann constant, and t the duration of applied field. The effect of the exponent n on the thermal stability factor was under investigation. According to the results of simulation20, the exponent n=1.5 for the single layer and bilayer media and n=1 for the multilayer was used.Figure7 shows the measured energy barrier versus soft layer thickness.It was found that the thermal stability factor (KuV*/kBT) was 208 for the single hard layer media and increased to 259 for the bilayer media with soft layer thickness increasedto 4 nm. The increase of TSF for bilayer media with the 4 nm soft layer may be attributed to anincreased switching volume due to the strong exchange interaction between the hard and soft layers. However, the TSF decreased when the soft layer exceeded 4 nm.This was due to the presence of demagnetization energy in the soft layer resulting in the reduction of energy barrier. The thicker the soft layer, the larger thereduction in the energy barrier.The figure of merit, ξ, which is the ratio of the energy barrier and the switching energy21, was also calculated based on the experimental data, as shown in figure 7 (right y-axis). ξ reached its maximum value of 1.3 when the soft layer thickness increased to 4 nm. Then, ξ decreased slightly to 1.1 at the soft layer thickness of 8 nm.
C.Multilayer media
According to experimental results with various substrate bias powers, Kudecreased gradually with increasing substrate bias power. The multilayer media was thus fabricated by first depositing the 10 nm FePt-TiO2 hard layer without substrate bias, thereafter, successive FePt-TiO2 layers of 2 nm each were deposited with substrate bias power 10, 20, 30 and 40 W, respectively. Figure 8(a) showsthe XRD patterns of the multilayer media. The FePt film showed a (001) preferred orientation and broad (002) + (200) peaks comparing with single hard layer, which was attributed to the capping soft layer. The inset of Fig. 8 (a)shows the rocking curve of fctFePt (001) peak in the single hard layer and the multilayer media, respectively. Both the full width half-maximum of the rocking curves were about 7°, indicating a good FePt fct (001) texture. The out-of-plane and in-plane hysteresis loops of the multilayer media are shown in Fig. 8b. The out-of-plane multilayer media reduced drastically to 6.5 kOe which was close to the half of that of the hardest layer.However, it was slightlylarger compared with that of the bilayer media with the same thickness.This was due to that the thickness of each soft layer inthe multilayer media was only 2 nm and total thickness of graded soft layer was 8 nm. At the saddle point,the nucleated domain wall sweep to interface of hard layer, it was difficult to form a full 180° domain wall.In order to clarify the mechanism of the coercivity in the single hard layer, bilayer and multilayer media, their recoil loops were measured, as shown in Fig. 9.For the single hard layer, the magnetization remained almost unchanged after removing the applied field, indicating coherent magnetization rotation behavior. For bilayer and multilayer media, both showed increased magnetization upon removal of applied field.Moreover, the slope of bilayer media was larger than that of multilayer media, suggesting a more incoherent switching behavior in bilayer media.