Magnetic properties of Co/Ni grain boundaries after annealing Open Access

There is a large interest in magnetic multilayers for fundamental studies and applications. To obtain desired magnetic properties, magnetic multilayers are often grown along a preferential orientation. This is achieved by depositing magnetic multilayers on top of single-crystal substrates or seed layers with a pre-set crystallographic orientation. The latter method is utilized for fabrication of textured magnetic layers in devices such as spin transfer torque random access memory (STT RAM), hard drives and magnetic sensors. In nanostructures consisting of magnetic multilayers, the seed layers also serve as a bottom contact. In this case the seed layers are desired to have low resistance.

In hard drives, magnetic recording perpendicular media,¹ and tunnel magnetoresistance sensors^2–4 magnetic multilayers are annealed above 573 K to optimize their magnetic and transport properties. To fabricate nanostructures, magnetic multilayers are usually heated at temperatures above 550 K. Thus, it is important to understand how microstructures and magnetic properties of magnetic multilayers are affected by annealing.

Co/Ni multilayers (ML) grown along the $111$ crystallographic orientations exhibit large perpendicular magnetic anisotropy (PMA), high spin polarization, and low intrinsic damping that make them a promising candidate for spintronic devices.⁵ To achieve strong texture along $111$⁠, Co/Ni ML are grown on top of metallic Ta/Cu seed layers and annealing is performed to optimize magnetic properties.⁶ It was shown that annealing above 523 K reduces the saturation magnetization, M_s and uniaxial magnetic anisotropy K_u, but increases the coercivity μ₀H_c of Co/Ni ML.^7–9 Co and Ni are known to interdiffuse at the interface with Cu leading to a reduction in M_s.^10,11 A reduction in M_s with annealing can then be expected from increased mixing of Cu and Co/Ni at the interface. Furthermore, a strong decrease of the slope of the M(H) curve at μ₀H_c is observed if Co/Ni ML are grown on top of a Cu seed layer.⁸ However, it is not well understood why annealing causes the observed changes in magnetic properties of the Co/Ni ML.

In the present work, the magnetic properties and microstructure of $111$ Co/Ni ML grown on top of a Cu seed layer are studied after annealing. The goal is to correlate the change in magnetic properties with the change in microstructure after annealing.

Following solvent cleaning, Si (100) wafers were etched using RCA-1 method.¹² Deposition was performed at room temperature using DC (Cu) and RF (Co, Ni, Ta) magnetron sputtering 2.0 × 10⁻³ Torr Argon pressure. Base pressure was kept below 5 × 10⁻⁸ Torr. Deposited samples were annealed at temperatures of 473, 498, 523, 548 and 573 K for 15 minutes in vacuum.

Ta(3 nm)/Cu(5 nm)/N×[Co(0.2 nm)/Ni(0.6 nm)]/Ta(3 nm) where N = 4, 6, 8, 10, 16, 32 were deposited on Si wafers. Out-of-plane XRD measurements were performed with scattering wave vector normal to the film surface. XRD data indicate that all the Co/Ni ML have $111$ texture with the full width at half maximum of the c-axis distribution, FWHM, below 4°.⁶ 

High-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) imaging, as well as element mapping based on energy-dispersive X-ray spectroscopy (EDXS), were performed at 200 kV with a Talos F200X microscope equipped with a Super-X detector system (FEI). Prior to STEM analysis, to remove organic contamination, the specimen was mounted in a high-visibility low-background holder and placed for 10 s into a Model 1020 Plasma Cleaner (Fischione).

The field dependence of the magnetization is measured using a SQUID magnetometer in magnetic fields up to 7 T and by Polar magneto-optic Kerr effect (Polar MOKE) in fields up to 1 T. For details on measurement of perpendicular anisotropy with SQUID magnetometer see Arora et al.⁶ 

Dependence of the magnetic anisotropy, K_u, and saturation magnetization, M_s, on the number of bilayer repeats N in as-deposited and annealed Ta/Cu/N×[Co/Ni]/Ta are shown in Figure 1. For as-deposited films both K_u and M_s increase with N for N ≤ 10 and then stay practically constant for higher N. K_u is not affected by annealing up to 573 K for N ≥ 16. For N ≤ 8, K_u decreases if annealing temperature exceeds 523 K. Annealing has much stronger effect on M_s than on K_u. M_s is unchanged if the annealing temperature is equal or below 473 K and decreases with increasing annealing temperature above 473 K.

Normalized M(H) loops of Ta/Cu/N×[Co/Ni]/Ta (N = 4 and 16) measured by Polar MOKE are presented in Figure 2. For N = 4, the coercivity, μ₀H_c, increases and the slope of the M(H) curve at μ₀H_c decreases with increasing annealing temperature. A similar increase in μ₀H_c is observed for N = 16 after annealing. However, for N = 16 the slope of the M(H) curve at μ₀H_c does not change significantly. For both N = 4 and 16, μ₀H_c increases about 0.22 T, more than ten times, after annealing at 573 K.

Coercivity depends on K_u/M_s and the exchange coupling between magnetic grains, μ₀H_ex. Figure 1 shows that while K_u does not change much with annealing, M_s decreases with increasing annealing temperature. Thus, the increase of μ₀H_c in Figure 2 is in part due to the increase of K_u/M_s with annealing temperature. However, this can not explain the ten fold increase in μ₀H_c.

The dependence of μ₀H_ex on annealing temperatures can be estimated from the following equation¹³ 

where H_n is the nucleation field. From Equation 1, μ₀H_ex of Ta/Cu/4×[Co/Ni]/Ta is reduced from 0.45 T for as-deposited to about 0.23 T after annealing at 573 K, and for Ta/Cu/16×[Co/Ni]/Ta from 0.86 T for as-deposited to about 0.74 T after annealing at 573 K. To understand these results EDXS analyses were performed in STEM mode.

Figure 3 and 4 show cross-sectional HAADF-STEM images with corresponding Co, Ni, and Cu element maps of as-deposited and annealed Ta(3 nm)/Cu(5 nm)/16×[Co(0.2 nm)/Ni(0.6 nm)]/Ta (3 nm) films, respectively. The elemental maps indicate that the diffusion of Cu into Co/Ni is not visible for the as-deposited films. After annealing at 573 K for 15 min, Cu diffuses into the grain boundaries of Co/Ni ML to about two-thirds of the thickness of the 16×[Co/Ni] ML. In Ta/Cu/4×[Co/Ni]/Ta films, the Co/Ni ML thickness is 4 times lower. Thus, it can be expected that Cu diffuses into the 4×[Co/Ni] ML grain boundaries to greater extent. The reduction in M_s observed in the 4×[Co/Ni] ML but not the 16×[Co/Ni] ML further supports this assertion.

Diffusion of Cu into grain boundaries of Co/Ni ML changes the reversal mechanism in the films by inducing more uniform magnetization reversal of individual grains. This explains the decreases in both the μ₀H_ex and the slope of the M(H) curve at μ₀H_c (Figure 2) in Ta/Cu/4×[Co/Ni]/Ta after annealing. Additionally, the diffusion of Cu along grain boundaries is the main reason for the increase in the coercivity of films with N = 4 and 16 as shown in Figure 2.

EDXS analyses performed in STEM show that in Ta/Cu/N×[Co/Ni]/Ta annealing causes Cu diffusion from the seed layer into the grain boundaries of Co/Ni. This reduces the exchange interaction between magnetic grains and alters the magnetic properties of Ta/Cu/N×[Co/Ni]/Ta; M_s decreases for ML with N ≤ 15 while μ₀H_c increases due to the presence of uniform magnetization reversal of individual grains. The slope of the M(H) curve at H_c decreases for N = 4 but not for N = 16. Cu diffusion along the grain boundaries more reduces the exchange interaction more significantly between magnetic grains in the N = 4 than the N = 16 ML.

Financial support for this project was provided by the Natural Sciences and Engineering Research Council of Canada (NSERC). The use of HZDR Ion Beam Centre TEM facilities and the support by its staff is gratefully acknowledged. In particular, we acknowledge the funding of TEM Talos by the German Federal Ministry of Education of Research (BMBF) under Grant No. 03SF0451 in the framework of HEMCP.