Enhancement of Spin-Orbit Torques through Material Engineering: The Role of Boron

Abstract

The conversion of charge current into pure spin current based on the spin Hall effect provides a new way of spin injection and manipulation in novel spintronic architectures. The seeking for materials with large spin Hall angle (SHA) values is important for both scientific interest and device applications. Compared to a rather small value in GaAs when it was first observed a decade ago, much larger SHAs have been observed in heavy metals such as Pt, Ta and W [1, 2]. Ta and W have the benefits such as the economic advantage as well as the largest SHA reported to date. However, the Joule heating due to the rather high resistivity of both $\beta-\mathrm {W}$ and $\beta-$Ta becomes of a great concern for the applications, combined with the excessive challenges in achieving a stable β form of W and Ta. Pt possesses low resistivity and high stability, but the high cost and the rather small spin diffusion length hinders its further development in this field. With a similar structure to Pt, but a much larger spin diffusion length [3], Pd has anomalously small SHA values reported, and has little been explored. Understanding the underlying mechanism plays an important role in the enhancement of SHA in different materials. It could be of intrinsic origin connected with the band structure or device geometry, or extrinsic origin due to impurities [4, 5]. As a common element and doping source in a spintronic device/structure, the role of boron in the enhancement of spin Hall effect is largely unexplored, when discussed into the mechanisms involved. Here we report a giant enhancement of SHA in palladium through boron engineering. We measure a large SHA of 0.16 in palladium, with perpendicular magnetic anisotropy in the Pd/CoFeB-based structure compared to the Pd/CoFe-based structure ($\mathrm{SHA}=0.02)$. Combined with theoretical calculations, it is found that both intrinsic and extrinsic spin Hall effects have been significantly enhanced through the introduction of boron. The incorporation of boron in the thin films results in significant microscopic and electronic changes in the Pd host metal. Our result provides a further understanding of the spin Hall effect in metals, leading to a new exploration of material engineering for large SHA sources. Together with high conductivity and long spin diffusion length, this work makes Pd a promising candidate for spintronic devices with magnetization switching by current pulse injection. The film stacks Pd(5 nm)/Co 40 Fe 40 B 20 (1 nm)/MgO(2 nm)/Ta(5 nm) and Pd(5 nm)/Co 50 Fe 50 (1 nm)/MgO(2 nm)/Ta(5 nm) are sputtered on thermally oxidized Si substrates. The stacks are patterned into Hall bar devices using photolithography (Fig. 1a and 1b). Fig. 1c and 1d shows the longitudinal and transverse effective field as a function of the injected current density for the CoFeB-based sample, with the magnitude of the former about three times larger than that of the latter, indicating a strong Slonczewski-like torque. Based on the data, the SHA determined is 0.16. It is the largest SHA reported so far for Pd based PMA structures. Fig. 1f shows the magnetization switching loop measured by the magneto-optical Kerr effect (MOKE) from a $5 \mu \mathrm {m}\times 20 \mu \mathrm {m}$ device. To explore the possible origins, we carry out synchrotron X-ray photoelectron spectroscopy (XPS) study of the samples for depth profiling. Fig. 2 shows that the introduction of B results in a highly pure metallic Pd state, as well as a high concentration of metallic contents of Co and Fe in the Pd layer for the CoFeB based sample, which is strikingly different to that observed in the CoFe based sample. The consumption of oxygen by B, leads to the highly metallic Pd state as well as reduced Co/Fe oxidation in Pd. We perform ab initio calculations of spin Hall conductivity (SHC) of B-, Fe- and Co-doped palladium as well as pure palladium metal based on the density functional theory. The calculated SHCs for pure Pd, B 1 Pd 31 , Fe 1 Pd 31 and Co 1 Pd 31 are 2852, 1818, 3114 and 3758 (h/2e)(S/cm), respectively. Interestingly, doping Pd with ∼3% B reduces the SHC by ∼36%. In contrast, doping Pd with ∼3% Co (Fe) increases the SHC by ∼32% (∼9%). Pure palladium metal has a large SHC which is further enhanced by substitutional Fe- and Co-doping. The experimental results comply well with the theoretical modelling.