Get 20M+ Full-Text Papers For Less Than $1.50/day. Subscribe now for You or Your Team.

Learn More →

Spin–orbit torque engineering in β-W/CoFeB heterostructures with W–Ta or W–V alloy layers between β-W and CoFeB

Spin–orbit torque engineering in β-W/CoFeB heterostructures with W–Ta or W–V alloy layers between... The spin–orbit torque (SOT) resulting from a spin current generated in a nonmagnetic transition metal layer offers a promising magnetization switching mechanism for spintronic devices. To fully exploit this mechanism, in practice, materials with high SOT efficiencies are indispensable. Moreover, new materials need to be compatible with semiconductor processing. This study introduces W–Ta and W–V alloy layers between nonmagnetic β-W and ferromagnetic CoFeB layers in β-W/CoFeB/MgO/Ta heterostructures. We carry out first-principles band structure calculations for W–Ta and W–V alloy structures to estimate the spin Hall conductivity. While the predicted spin Hall conductivity values of W–Ta alloys decrease monotonically from −0.82 × 10 S/cm for W at% as the Ta concentration increases, those of W–V alloys increase to −1.98 × 10 S/cm for W V at% and then gradually decrease. 75 25 Subsequently, we measure the spin Hall conductivities of both alloys. Experimentally, when β-W is alloyed with 20 at% V, the absolute value of the spin Hall conductivity considerably increases by 36% compared to that of the pristine β-W. We confirm that the W–V alloy also improves the SOT switching efficiency by approximately 40% compared to that of pristine β-W. This study demonstrates a new material that can act as a spin current-generating layer, leading to energy- efficient spintronic devices. Introduction interface generate a SOT. The spin current, in turn, is In recent decades, there has been tremendous known to be generated by the spin–orbit interaction at 1,2 9,10 advancement in spintronics. Spin–orbit torque (SOT) , the NM or NM/FM interface . Recently, FMs (e.g., which is significantly more rapid and energy-efficient NiFe) have been shown to generate a spin current. 3,4 than spin-transfer torque (STT) ,has generated Other material classes, such as topological insulators 12–14 interest from technological and scientific perspectives. and Weyl semimetals, have also been studied . SOT is a critical part of magnetization switching, facil- However, their application in device manufacturing 5–8 itating devices with magnetic memory and logic .Ina appears to be limited due to the complexity of the typical nonmagnetic transition metal (NM)/ferromagnet growth method and low thermal stability. Therefore, it is (FM) heterostructure, polarized spin moments carried essential to design NMs with manufacturing process- by the spin current and accumulated at the NM and FM friendly materials, such as W, commonly used in the current semiconductor industry. It is also equally essential to identify material combinations that possess Correspondence: Sonny H. Rhim (sonny@ulsan.ac.kr) or Young perpendicular magnetic anisotropy (PMA), which is the Keun Kim (ykim97@korea.ac.kr) 1 15,16 Department of Materials Science and Engineering, Korea University, Seoul key to achieving high bit density . Two orthogonal 02841, Korea torques contribute to the SOT: damping-like (DL) and Department of Physics, University of Ulsan, Ulsan 44610, Korea field-like (FL) SOTs. Note that the DL-SOT efficiency Full list of author information is available at the end of the article These authors contributed equally: Gyu Won Kim, Do Duc Cuong © The Author(s) 2021 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to theCreativeCommons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/. 1234567890():,; 1234567890():,; 1234567890():,; 1234567890():,; Kim et al. NPG Asia Materials (2021) 13:60 Page 2 of 9 60 per unit current density (ξ ) is correlated to the spin follows DL Hall angle (θ ), which is the charge-to-spin conversion SH ratio, as follows: ξ = T θ ,where T is the inter- DL int SH int σ ¼ f Ω ðkÞ ð1Þ SH kn kn facial spin transparency. Unless we know the exact value of T ,the valueof θ is not equal to ξ . int SH DL Of the various NMs, β-W is a good candidate, showing where f is the Fermi–Dirac function for the nth band at kn PMA with CoFeB and a high DL-SOT efficiency ξ of k; therefore, the Berry curvature of the n-th band at k is DL 17,18 −0.33 to −0.40 . There have been several efforts expressed as made to improve the ξ of β-W derivatives, e.g., the DL 0 0 deposition of W by sputtering under an oxygen atmo- hj kn jji kn hj kn vbc kn x y Ω ðÞ k ¼ 2 Im 19  ð2Þ sphere and multistep W deposition to form a thick 0 e  e 0 n ≠n k;n k;n β-W layer of up to 16 nm . Both methods achieved ξ DL values of approximately −0.5. In the former study, the where j ¼ fg Σ; v is the spin current and Σ is the spin x x PMA strength weakened due to oxygen incorporation, operator in the full relativistic formalism. In this study, the 27,28 whereas the processing time increased in the latter. interpolation technique using Wannier90 was Moreover, Ta substitution into β-W was theoretically employed with s, p, and d orbitals of W, Ta, and V predicted to achieve a ξ of −0.5 due to intrinsic band atoms. Throughout this study, the xy component of the DL structure modification . spin Hall conductivity with the spin axis along the z- This study extends this Ta-based theoretical investi- direction was considered. For the alloys, the spin Hall gation. In addition, we expect V to play a similar role in conductivity at concentration x was calculated for all A15-structured β-W because V has the same number of possible configurations and was thermodynamically valence electrons as Ta. Thus, we explore Ta and V averaged using the Boltzmann factor by considering the substitutions in β-W for a wide range of alloy con- relative energies. centrations. Based on our first-principles band calcula- tions, an alloy layer, W X (where X is either Ta or Sample preparation 100−x x V), was introduced between the β-W and CoFeB layers. Samples were sputtered onto 1.25 × 1.25 cm thermally −9 We fabricated β-W/W X /CoFeB/MgO/Ta hetero- oxidized Si wafers under a base pressure below 5 × 10 100−x x structure film stacks on Si wafers coated with thermal Torr. The thickness of the Si-oxide layer was 300 nm. The oxides for various alloy compositions, 0 < x < 100 at%. stacking structure of each sample was W (4)/W X 100−x x Due to the presence of bottom β-W layers, the β-phase (2)/CoFeB (0.9)/MgO (1)/Ta (2) (the number in par- in the W X alloy layers was expected to be main- entheses represents the thickness in nanometers), where 100−x x tained to the maximum extent. Over most of the com- X is either Ta or V. A 4-nm-thick β-W layer was position range, PMA was exhibited. Harmonic Hall employed to maintain the β-phase in the 2-nm-thick W measurements confirmed that ξ increased to −0.49, X alloy layer. The alloy layer composition was varied in DL −x x corresponding to an approximately 40% enhancement 10 at% steps by changing the sputtering power densities of compared to that of β-W, which agreed well with the the W and Ta (or V) targets during co-deposition. The calculated results. This large SOT was further validated composition of the CoFeB target was Co Fe B in at%. 40 40 20 using other measurement methods, such as domain wall The metallic layers and the MgO interlayer were pro- depinning and the propagation model .Thiscombined duced by DC and radio frequency magnetron sputtering, theoretical and experimental study provides insight into respectively. All samples were post-annealed at 300 °C for engineering opportunities for developing materials 1 h under a magnetic field of 6 kOe applied perpendicular −6 exhibiting high SOT efficiency. to the film plane under a base pressure of 10 Torr. Each film was patterned into a 5-μm-wide and 35-μm-long Hall Materials and methods cross-device by photolithography (Karl Suss MA6) and Ar Theoretical calculations ion milling. An electrical contact pad of a Ti/Au bilayer First-principles calculations were performed using the was fabricated using an e-beam evaporator followed by a 23 2 Vienna Ab Initio Simulation Package with the projector lift-off process. We also constructed 4 × 4 μm -sized fer- augmented wave basis . An energy cutoff of 500 eV was romagnetic islands at the center of the Hall bar devices for selected, with a 16 × 16 × 16 k-point mesh for summation current-induced SOT switching, in which all parts in the Brillouin zone. The generalized gradient approx- excluding the islands were W/W–Ta or W/W–V layers. imation was employed for the exchange-correlation potential as parametrized by Perdew, Burke, and Ernzer- Measurements hof . The spin Hall conductivity (σ ) was calculated Hysteresis loops were measured using a vibrating sam- SH using the Kubo formula in the linear response theory as ple magnetometer (Microsense EV9). To evaluate the Kim et al. NPG Asia Materials (2021) 13:60 Page 3 of 9 60 SOT efficiency, the harmonic Hall response method was obtained for both W–Ta and W–V alloys at x = 25 at% employed (Supplementary Note 1, Figs. S1a, b). Two lock- (see Fig. 1c, d). The two alloys exhibit quite different in amplifiers were also utilized to simultaneously access behaviors based on their composition. σ for W–Ta SH the first- and second-harmonic responses of the magne- decreases monotonically with x, whereas that of the W–V tization. An alternating current (AC) of frequency 13.7 Hz alloy is enhanced as x increases from 12.5 to 50 at%. The and amplitude 1 mA was applied during the measurement W–Ta alloy favors the cc’ configuration with a 96% while changing the external magnetic field from −18,000 probability; hence, other configurations are not con- to 18,000 Oe with different azimuthal angles φ. The polar sidered. In comparison, for the W–V alloy, the bb’ con- angle of the sample was tilted (~5°) to prevent the for- figuration has the highest probability (60.5%), and the bc mation of multiple domains. configuration has a lower probability (35%). The prob- The current-induced SOT switching was measured by abilities are estimated using Boltzmann factors, con- the anomalous Hall voltage of the samples. The Hall sidering the relative total energies at room temperature voltage was detected by a 100-μA direct current (DC) (Supplementary Fig. S2e). after applying each current pulse is 0.5-mA steps and a The spin Hall conductivity of the bc configuration 10-μs width from −15 to 15 mA under a constant external reaches as high as −1.98 × 10 S/cm, equivalent to 141% field along the x-direction. The resistivity of the alloyed enhancement relative to that of β-W (−0.82 × 10 S/cm) layer was evaluated using a constant DC supply. Both (Supplementary Fig. S2f). This enhancement is attributed switching and resistivity were examined using a four- to the occurrence of strong symmetry breaking when V point electrical property measurement station (MSTECH simultaneously adopts bcc and chain sites. Specifically, the M7VC). The crystal structures of the W/W–Ta, and W/ Berry curvatures with opposite signs, a characteristic of W–V layers were characterized by grazing incidence (GI) β-W with a resonant double degenerate state, no longer X-ray diffraction (XRD, Rigaku ATX-G) with incident cancel out (Supplementary Note 3 and Fig. S3a–d). angles ranging from 30° to 80° in 0.02° steps. The atomic Moreover, the increased degeneracies associated with the distribution after annealing at 300 °C was evaluated by lowered symmetry increase the spin Hall conductivity, scanning transmission electron microscopy (STEM, FEI whose effect is most drastic when x = 25 at% in the W–V Double Cs Corrected Titan3 G2 60-300) with energy- alloy. Based on the calculations, a W–V alloy with a dispersive X-ray spectroscopy (EDS). For the STEM composition range of 12.5–50 at% is explored experi- sampling, a focused ion beam (FIB, FEI Quanta3D) system mentally, as discussed below. was used. Secondary ion mass spectrometry (SIMS, ION- TOF TOF, SIMS 5) was also used to examine the atomic SOT efficiency estimation employing the harmonic Hall distribution. method Based on first-principles calculations, various alloy lay- Results and discussion ers were introduced between the β-W and CoFeB layers in Materials screening by ab initio calculations a heterostructure consisting of β-W/CoFeB/MgO/Ta. The A previous investigation on x = 12.5 at% Ta alloying employment of the alloy layer was motivated by our was extended in this study to other concentrations of Ta previous study , which showed that a W layer could not and V. The crystal structure of A15 β-W comprises two maintain the β-W phase when it was alloyed with 10 at% symmetrically inequivalent sites: body-centered cubic Ta. A film stack of β-W (4)/W X (2)/CoFeB (0.9)/ 100−x x (bcc) and chain (c) sites. Due to the presence of these two MgO (1)/Ta (2) (thickness in nm) was fabricated with inequivalent sites, several configurations are possible for alloy X (Ta or V) to prevent the phase transition of W each x, where the acceptance of a configuration is deter- from β to α. Note that the β-phase is a fundamental mined by energetics. The spin Hall conductivity is esti- assumption in the theoretical calculations. mated by thermodynamic averaging (Supplementary Note The alloyed layer composition ranged from 0 to 100 at% 2). In particular, for an alloy with a solute composition x X in 10 at% steps. After annealing at 300 °C for 1 h in a = 25 at%, there are four possible configurations: cc, bc, cc’, vacuum, most film structures present PMA. However, and bb’, where b and c denote either a bcc site or a chain when the V composition exceeds 80 at%, a PMA to in- site. Specifically, the cc and cc’ configurations exist when plane magnetic anisotropy transition occurs. At this the solute atoms, either Ta or V, occupy the same and composition, the effective magnetic anisotropy energy eff different chain sites, respectively. Similarly, bb’ denotes (K ) decreases. (Supplementary Note 4, Fig. S4a–f, and the case when two different bcc sites are replaced, and bc Table S1). At x = 100 at%, the magnetic hysteresis loop is eff represents the case when each of the bcc and chain sites is not observed; hence, the sign of K cannot be deter- substituted (Supplementary Fig. S2a–d). mined. The degradation of the magnetic properties when The σ values were calculated as a function of x (see x > 80 at% is attributed to the formation of a magnetic SH 30,31 Fig. 1a, b), and the k-resolved Berry curvatures were dead layer between V and CoFe . Thus, in the Kim et al. NPG Asia Materials (2021) 13:60 Page 4 of 9 60 Fig. 1 Theoretical spin Hall conductivity calculation. Spin Hall conductivities of a W Ta and b W V and k-resolved Berry curvature of c 100−x x 100−x x W Ta and d W V alloy, where x = 25 at%, with two Ta or V adopting cc’ and bc configurations, respectively. 100−x x 100−x x Fig. 2 SOT measurement. a Schematic description of the device structure and stack of films. The coordinate axis is shown above the device structure. b DL and c FL-SOT efficiencies. d Experimentally estimated spin Hall conductivity. Error bars in figures represent standard errors. following, the SOT efficiency is discussed for the com- illustrates the harmonic Hall measurement, where an AC position range exhibiting PMA. is applied along the x-direction, following which the Hall Harmonic Hall measurements were performed to voltage (V ) along the y-direction is measured. During assess the SOT efficiencies. Figure 2a schematically the V measurement, sweeping the external in-plane H Kim et al. NPG Asia Materials (2021) 13:60 Page 5 of 9 60 magnetic field parallel (perpendicular) to the current In accordance with the ab initio calculations, the spin yields traces of DL-SOT (FL-SOT). Fig. 2b, c shows the Hall conductivity was estimated by employing the rela- DL-SOT (ξ ) and FL-SOT (ξ )efficiencies per unit tionship between the resistivity and the spin Hall angle, DL FL current density of the W/W Ta /CoFeB/MgO/Ta and θ ¼ σ ρ þ b , where σ comprises the intrinsic 100−x x SH SH SH xx W/W V /CoFeB/MgO/Ta heterostructures versus and side-jump contributions, ρ is the longitudinal 100−x x xx alloy composition x (in at%), respectively. The changes in resistivity, and b represents the skew-scattering con- ξ and ξ with respect to x are quite different. As the Ta tribution. Presumably, parameter b is negligible because DL FL content increases, ξ gradually decreases from −0.35 ± of the high impurity content in the studied system, and it DL 0.002 for a 6-nm-thick W to −0.06 ± 0.06 for the W/ is assumed that T is 1. Hence, the spin Hall conductivity int W Ta bilayer. Notably, the ξ of the W/W Ta can be expressed by the ratio of θ to ρ . The ρ value 0 100 DL 0 100 SH xx xx bilayer is similar to that of the highly resistive material of each device was determined by employing the parallel β-Ta . This suggests that the spin current generated in resistance model assuming ρ = 170 μΩ cm for CoFeB xx W inefficiently penetrates the W–Ta alloy layer and fur- (Supplementary Note 5, Fig. S5a, b). The results are dis- ther intensifies with increasing Ta composition. Conse- played in Fig. 2d, and the maximum SHC value is quently, in the extreme case of a bilayer NM structure, (−2.77 ± 0.31) × 10 S/cm at 20 at% V. This is reasonably most of the DL torque occurs due to the spin current consistent with the first-principles calculation results of generated in Ta. Furthermore, this is consistent with this study, which yield a maximum value of −1.98 × 10 S/ previous studies on the spin diffusion length of Ta, which cm at 25 at% V. Consequently, by using the W−V alloy 34,35 is 2–3nm . However, the ξ values of the W/W layer, the band structure modification predicted by the DL 100 V /CoFeB/MgO/Ta heterostructure exhibit interesting calculations is successfully captured. −x x composition-dependent behavior. At V = 20 at%, the ξ DL value reaches a maximum of −0.49 ± 0.05. This large Assessment of SOT switching value is in agreement with that of a W-based SOT tunnel Current-induced SOT switching in patterned devices 19,20 junction . was examined. A current sweep was applied from –15 to The FL-SOT (ξ )efficiencies do not present any dis- 15 mA. The current pulse width was 10 μs and was FL tinct compositional dependence, as depicted in Fig. 2c. In changed in 0.5-mA steps while the external magnetic field both the Ta and V cases, ξ  −0.10 to −0.20, suggesting strength was varied from −100 to 100 Oe in the x-direc- FL that the potential gradient at the interface (i.e., the Rashba tion. An additional fabrication process was utilized to effect) has a slight correlation with the alloy layer. In form an island pattern on the devices to test the current- addition, in the thin FM layer, in the layered structure induced SOT switching (Methods). with PMA, ξ depends on the magnetic anisotropy Figure 3a shows the switching loops for a hetero- FL energy and temperature . Moreover, the PMA in the structure with a 20 at% V alloy, which shows the highest NM/CoFeB/MgO structure mainly originates from the ξ . The sample clearly exhibits external field depen- DL CoFeB/MgO interfacial anisotropy energy, and the dence. When the polarity of the external field is varied, portion of NM/CoFeB is relatively small. Thus, the the switching polarity also changes. A strong applied insensitive behavior of ξ in this study is in line with a magnetic field implies a small observed switching current. FL previous report because only the W/CoFeB interface is Furthermore, magnetization switching was examined in modified. all patterned devices used in the harmonic measurements. Fig. 3 In-plane pulsed current-induced SOT switching. a SOT switching curves for different external field strengths for 20 at% V-incorporated devices. b SOT switching current density versus alloy composition for entire PMA samples. Under a constant 100 Oe external magnetic field along the x-direction. Kim et al. NPG Asia Materials (2021) 13:60 Page 6 of 9 60 As depicted in Fig. 3b, the magnetization direction was The DL effective field saturates at approximately 24.7 ± successfully manipulated by applying an in-plane current 2.33 Oe cm /MA when the external field exceeds 120 Oe. pulse and a constant external field, 100 Oe. The switching This is consistent with the effective DL field obtained current density (J ) and ξ are inversely proportional in using the harmonic Hall method, as indicated by the red SW DL single-domain devices as follows : dotted line in Fig. 4d. Because the two measurement schemes yield similar magnitudes for the DL-SOT effec- eff 2e H H tive field, it is concluded that the enhancement in the ext pffiffiffi ð3Þ J ¼ M t SW S FM SOT originates from the 20 at% V incorporation in the hξ 2 DL film structure. where t represents the thickness of the FM layer and Furthermore, we examine the 1/cos θ dependence of all FM eff H and H are the anisotropy field and the external devices and confirm that SOT switching occurs in a ext field, respectively. However, based on Figs. 2b and 3b, ξ domain wall-mediated process. From this perspective, the DL and J do not show inverse proportionality. A 4 μm× SOT switching efficiency (η), considering the domain wall SW 4 μm-sized island pattern was used to test the SOT depinning field, is investigated (Supplementary Note 7 switching, and this size of the island was extremely large and Fig. S7a). The enhancement ratio of the SOT to maintain a single-domain state. Therefore, this switching efficiency between pristine W and the 20 at% deviation is attributed to the existence of domains and V-incorporated samples is qualitatively consistent with domain walls . The next section describes tests on the enhancement ratio of ξ (Supplementary Fig. S7b). DL whether SOT switching entails domain wall motion and Therefore, increasing the switching current per increase double-checks of the high SOT efficiency. To this end, a in the V content, as shown in Fig. 3b, does not consider domain wall depinning and propagation model for the the effect of the domain wall depinning field. When the 20 at% V-incorporated devices was employed. domain wall depinning field during magnetization switching is considered, the W–V alloy achieves a higher Domain wall depinning and propagation model SOT switching efficiency than that in the single W layer. To examine the high SOT efficiency in the 20 at% V Note that if domain wall depinning and propagation alloy sample, the domain wall depinning and propagation occur during SOT switching, J follows Eq. (4) . sw model was applied. First, we determined whether SOT switching occurs through domain wall propagation or 4eμ M t H s FM c single domain reversal under the current application. J ¼ ð4Þ SW πhξ DL H (θ)is defined as the external switching field at polar angle θ. Figure 4a shows that the H (θ)/H (0) ratio p p accurately presents a 1/cos θ dependence, which confirms where e, μ , M ,t , and H represent the elementary 0 s FM c that in the studied Hall cross-device, magnetization charge of an electron, vacuum permeability, saturation switching involves a domain wall-mediated process. magnetization, thickness of the FM layer, and coercivity of Based on the SOT evaluation process, the measuring the ferromagnetic layer, respectively. The W–V alloyed time (t ) dependence of H was also examined, as shown structures reasonably follow the coercivity of CoFeB with m p in Fig. 4b. We find that the intrinsic coercive field (H )is regard to V composition (Supplementary Note 8 and Fig. c0 393.03 Oe and that the coercivity of the device is S8a, b). 274.82 Oe (Supplementary Fig. S6a). In addition, the Joule heating coefficient (κ) was determined utilizing two dif- Improvement in the DL-SOT efficiency of W–V alloys ferent channel resistance (R ) measurements: the tem- To elucidate the increase in the ξ in a W–Valloy,the xx DL perature dependence of R was determined by heating microstructure of the films was analyzed. GI-XRD was xx the device externally (Supplementary Fig. S6b), and the used to examine the phases of the W/W–Ta and W/ current dependence of R was determined from the W–V layers, and the results are shown in Fig. 5aand b. xx relationship T(I) ≈T + κI , where T is the room tem- Because the layers are ultrathin, the XRD spectra show 0 0 perature and I is the amplitude of the input current no peaks corresponding to CoFeB, MgO, or Ta. The (Supplementary Fig. S6c). W–Ta alloy retains the β-W phase up to 80 at% Ta When a current of I = 4.22 mA is applied, the tem- content. For the W–Valloy,asillustrated in Fig. 5b, the perature of the device increases by 157 K relative to room characteristic peak for α-W appears when 100 at% V is temperature, where κ is determined to be 0.02485 K/mA . incorporated. β-W has the A15 structure, which is an Using these parameters, the switching state diagram (Fig. energetically unstable phase. In addition, the atomic radii 4c) is converted into the DL effective field normalized by of W, Ta, and V are 135, 146, and 135 pm, respectively; the input current density (Fig. 4d). The detailed conver- therefore, the residual stress in the W–Ta alloy promotes sion process is described in Supplementary Note 6. the phase transition of β-W. Consequently, the phase Kim et al. NPG Asia Materials (2021) 13:60 Page 7 of 9 60 Fig. 4 Domain wall depinning and propagation switching model in the 20 at% V alloy-inserted heterostructure. a Polar angle dependence of the switching field. The green dotted line represents the fitted line for 1/cos θ. b Measuring time (t )-dependent behavior of the switching field. Error bars represent the standard deviation from 35 independent measurements. c Switching state diagram as a function of the external field in the x- direction. d Effective DL field evaluated from the switching state diagram. The red dotted line represents the effective DL field estimated from the harmonic measurement. The error bars in the figures represent the standard error. Fig. 5 Microstructural analysis of the alloy inserted heterostructure. a XRD patterns for W/W Ta /CoFeB/MgO/Ta films and b W/W V / 100−x x 100−x x CoFeB/MgO/Ta films after annealing at 300 °C for 1 h. Reference peak positions for the α (ICDD No. 00-004-0806) and β (ICDD No. 00-047-1319) phases of W are denoted below. stability of β-W in the W–V alloy is one of the reasons Schematics of the layered structures are placed on the for the enhancement in ξ . right side of the figures. The interfaces between each layer DL Because V is a lighter element than W, we examined become clear and sharp after the films undergo 300 °C atomic interdiffusion after annealing at 300 °C. To test annealings. EDS was conducted to locate the atomic whether ξ was enhanced due to its atomic interdiffusion positions in the film structure. To obtain the EDS inten- DL of V, atomic distribution profiles employing STEM were sity, the sample prepared for high-resolution (HR) TEM obtained. Figure 6a, b clearly shows the layered structures. measurement was extremely thin (20–30 nm), and thus, a Kim et al. NPG Asia Materials (2021) 13:60 Page 8 of 9 60 Fig. 6 Scanning transmission electron microscopy (STEM) data for the W/W–V/CoFeB/MgO/Ta heterostructure. a, b HR-TEM images of the as-deposited and 300 °C annealed samples. c, d HAADF-STEM image and element line profiles of the as-deposited and 300 °C annealed samples. The composition of the alloy layer in the atomic ratio is W V . 80 20 relatively thick specimen (80–100 nm) was prepared using composition-dependent behavior. Specifically, for the a FIB system. The atomic positions in the film structure 20 at% V alloy layer, ξ became −0.49 ± 0.05, an DL were assigned by EDS, as presented in Fig. 6c, d. Note that enhancement of 40% over that of pristine β-W. This the high-angle annular dark-field images appear blurred significant enhancement in ξ was confirmed by two DL because of the TEM sampling thickness. different measurement schemes: harmonic measure- We also conducted EDS on the samples used in the HR- ments and a domain wall depinning and propagation TEM measurement (Supplementary Fig. S9a, b). Based on model. The domain wall depinning and propagation Fig. 6c and d, V atoms are located under the CoFe layer, model presented a reasonable value for the DL-SOT regardless of heat treatment. The elemental profile of V effective field 24.7 ± 2.33 Oe cm /MA. observed in the SiO /W and Ta layers is confirmed to be The SOT switching efficiency (η) also improved, similar an EDS measurement artifact (Supplementary Note 9 and to the enhancement ratio of ξ . The XRD and TEM DL Fig. S10a–f). Furthermore, SIMS was performed, showing analyses confirmed that the high ξ was attributed to the DL that the atomic interdiffusion of V is negligibly small phase stability of β-W achieved by employing the W–V (Supplementary Note 10 and Fig. S11a–d). Therefore, we alloy layer. We expect that the optimized composition of can exclude the notion of atomic interdiffusion in the the W V alloy layer could become an energy-efficient 80 20 heterostructures. spin current source layer because of its high ξ and good DL process compatibility for semiconductor manufacturing. Conclusion Acknowledgements A combined theoretical and experimental study was The authors thank Yun Jung Jang of the Korea Institute of Science and presented for W/W−Ta/CoFeB/MgO/Ta and W/W–V/ Technology for helping us with the ToF-SIMS experiments. This research was CoFeB/MgO/Ta heterostructures. A series of first- supported by the National Research Foundation (NRF) of Korea (2015M3D1A1070465 and 2020M3F3A2A01082591). Y. K. Kim was supported principles calculations were conducted to investigate by Samsung Electronics Co., Ltd. (IO201211-08104-01). O. Lee was supported the W–Ta and W–V alloy compositions leading to high by the NRF of Korea (2020M3F3A2A01081635). The study at the University of spin Hall conductivity. Subsequently, fabrication and Ulsan was supported by the Basic Research Lab Program through the NRF of Korea (2018R1A4A1020696). measurement were performed by experimentally com- piling the composition dependencies of the SOT effi- Author details ciency. The W–Ta alloypresented agradually Department of Materials Science and Engineering, Korea University, Seoul decreasing ξ as the Ta content increased, and the ξ 2 DL DL 02841, Korea. Department of Physics, University of Ulsan, Ulsan 44610, Korea. of the W–V alloy demonstrated interesting Department of Physics and Computer Science, Vietnam National University, Kim et al. NPG Asia Materials (2021) 13:60 Page 9 of 9 60 Ho Chi Minh City 700000, Vietnam. Center for Spintronics, Korea Institute of 15. Mangin, S. et al. Current-induced magnetization reversal in nanopillars with Science and Technology, Seoul 02792, Korea. Seoul Center, Korea Basic perpendicular anisotropy. Nat. Mater. 5,210–215 (2006). Science Institute, Seoul 02841, Korea 16. Ikeda, S. et al. A perpendicular-anisotropy CoFeB-MgO magnetic tunnel junction. Nat. Mater. 9,721–724 (2010). 17. Pai, C.-F. et al. Spin transfer torque device utilizing the giant spin Hall effect of Author contributions tungsten. Appl. Phys. Lett. 101, 122404 (2012). G.W.K. and Y.K.K. conceived and designed the experiments. D.D.C., S.C.H., and S. 18. Hao,Q.&Xiao,G.Giant Spin Hall Effect andswitching inducedby spin-transfer H.R. performed the first-principles calculations. G.W.K., T.K., and M.H.L. torque in a W/Co Fe B /MgO structure with perpendicular magnetic ani- fabricated the film structures and devices. G.W.K., Y.J.K., and I.H.C. conducted 40 40 20 sotropy. Phys. Rev. Appl. 3, 034009 (2015). the SOT efficiency evaluation and analyzed the microstructural transition of β.- 19. Demasius, K.-U. et al. Enhanced spin-orbit torques by oxygen incorporation in W. G.W.K. conducted the domain wall propagation experiments with the help tungsten films. Nat. Commun. 7, 10644 (2016). of O.J.L. H.B. performed the TEM experiment and analyzed the atomic 20. Chen, W., Xiao, G., Zhang, Q. & Zhang, X. Temperature study of the giant spin distribution. G.W.K. and D.D.C. wrote the paper after discussion with all the Hall effect in the bulk limit of β-W. Phys. Rev. B 98, 134411 (2018). authors. Y.K.K. supervised the entire project. 21. Sui,X.etal. Giant enhancement of the intrinsic spin Hall conductivity in β-tungsten via substitutional doping. Phys. Rev. B 96, 241105(R) (2017). Competing interests 22. Lee, O. J. et al. Central role of domain wall depinning for perpendicular The authors declare no competing interests. magnetization switching driven by spin torque from the spin Hall effect. Phys. Rev. B 89, 024418 (2014). 23. Kresse, G. & Furthmüller, J. Efficient iterative schemes for ab initio total-energy Publisher’s note calculations using a plane-wave basis set. Phys.Rev.B 54, 11169 (1996). Springer Nature remains neutral with regard to jurisdictional claims in 24. Blöchl, P. Projector augmented-wave method. Phys. Rev. B 50, 17953 (1994). published maps and institutional affiliations. 25. Perdew, J. P., Burke, K. & Ernzerhof, M. Generalized gradient approximation made simple. Phys.Rev.Lett. 77, 3865 (1996). Supplementary information The online version contains supplementary 26. Guo, G. Y., Murakami, S., Chen, T.-W. & Nagaosa, N. Intrinsic spin Hall material available at https://doi.org/10.1038/s41427-021-00326-8. effect in platinum: first principles calculations. Phys. Rev. Lett. 100, 096401 (2008). 27. Souza, I., Marzari, N. & Vanderbilt, D. Maximally localized Wannier functions for Received: 13 March 2021 Revised: 30 June 2021 Accepted: 19 July 2021 entangled energy bands. Phys. Rev. B 65, 035109 (2001). 28. Mostofi, A. A. et al. wannier90: a tool for obtaining maximally-localised Wannier functions. Comput. Phys. Commun. 178, 685 (2008). 29. Cha,I.H., Kim, T.,Kim,Y.J., Kim, G. W. &Kim,Y. K.Thickness andcomposition- References dependent spin-orbit torque behaviors in perpendicular magnetized Ta/W (t)/ 1. Miron, I. M. et al. Perpendicular switching of a single ferromagnetic layer CoFeB and Ta W /CoFeB junction structures. J. Alloy. Compd. 823, 153744 1-x x induced by in-plane current injection. Nature 476, 189–193 (2011). (2020). 2. Liu, L. et al. Spin torque switching with the giant spin Hall effect of tantalum. 30. Devolder, T.,Tahmasebi, T., Eimer, S.,Hauet,T.&Andrieu, S. Compositional Science 336, 555–558 (2012). dependence of the magnetic properties of epitaxial FeV/MgO thin films. Appl. 3. Slonczewski, J. C. Current-driven excitation of magnetic multilayers. J. Magn. Phys. Lett. 103, 242410 (2013). Magn. Mater. 159,L1–L7 (1996). 31. Fackler, S. W. et al. Combinatorial study of Fe-Co-V hard magnetic thin films. 4. Berger,L.Emissionofspinwaves by a magnetic multilayer traversed by a Sci. Tech. Adv. Mater. 18,231 (2017). current. Phys. Rev. B 54, 9353 (1996). 32. Avci, C. O. et al. Interplay of spin-orbit torque and thermoelectric effects in 5. Fukami,S., Anekawa, T., Zhang,C.& Ohno,H. A spin-orbit torque switching ferromagnet/normal-metal bilayers. Phys. Rev. B 90, 224427 (2014). scheme with collinear magnetic easy axis and current configuration. Nat. 33. Yu, R. et al. Determination of spin Hall angle and spin diffusion length in Nanotechnol. 11,621–625 (2016). β-phase dominated tantalum. Phys. Rev. Mater. 2, 074406 (2018). 6. Baek, S. C. et al. Complementary logic operation based on electric-field 34. Niimi, Y. et al. Extrinsic spin Hall effects measured with lateral spin valve controlled spin-orbit torques. Nat. Electron. 1,398–403 (2018). structures. Phys. Rev. B 89, 054401 (2014). 7. Zhang, J. et al. Role of micromagnetic states on spin-orbit torques-switching 35. Kim, D.-J., Kim, S.-I., Park, S.-Y., Lee, K.-D. & Park, B.-G. Ferromagnetic resonance schemes. Nano Lett. 18, 4074–4080 (2018). spin pumping in CoFeB with highly resistive nonmagnetic electrodes. Curr. 8. Song, C. et al. Spin-orbit torques: materials, mechanisms, performances, and Appl. Phys. 14,1344–1348 (2014). potential applications. Prog. Mater. Sci. 118, 100761 (2021). 36. Ou, Y.,Pai, C.-F., Shi,S., Ralph, D. C. &Buhrman,R.A.Origin of field-like spin- 9. Sinova,J., Valenzuela,S.O., Wunderlich, J.,Back, C. H. &Jungwirth,T.SpinHall orbit torques in heavy metal/ferromagnet/oxide thin film heterostructures. effects. Rev. Mod. Phys. 87,1213–1259 (2015). Phys. Rev. B 94, 140414(R) (2016). 10. Manchon, A. & Zhang, S. Theory of nonequilibrium intrinsic spin torque in a 37. Fritz, K., Wimmer, S., Ebert, H. & Menert, M. Large spin Hall effect in an single nanomagnet. Phys. Rev. B 78, 212405 (2008). amorphous binary alloy. Phys.Rev.B 98, 094433 (2018). 11. Wang, W. et al. Anomalous spin-orbit torques in magnetic single-layer films. 38. Lee,K.-S., Lee,S.-W., Min,B.-C. &Lee, K.-J.Thermally activatedswitching of Nat. Nanotech. 14,819–824 (2019). perpendicular magnet by spin-orbit spin torque. Appl. Phys. Lett. 104, 072413 12. Fan, Y. et al. Magnetization switching through giant spin-orbit torque in a (2014). magnetically doped topological insulator heterostructure. Nat. Mater. 13, 39. Emori,S., Bauer, U.,Ahn,S.-M.,Martinez, E. &Beach,G.S.D.Current-driven 699–704 (2014). dynamics of chiral ferromagnetic domain walls. Nat. Mater. 12,611–616 13. Wang, Y. et al. Room temperature magnetization switching in topological (2013). insulator-ferromagnet heterostructures by spin-orbit torques. Nat. Commun. 8, 40. Zhu, L., Ralph, D. C. & Buhrman,R.A.Lack ofsimple correlation between 1364 (2017). switching current density and spin-orbit torque efficiency of perpendicularly 14. Li, P. et al. Spin-momentum locking and spin-orbit torques in magnetic nano- magnetized spin-current-generator-ferromagnet heterostructures. Phys. Rev. heterojunctions composed of Weyl semimetal WTe . Nat. Commun. 9,3990 Appl. 15, 024059 (2021). (2018). http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png NPG Asia Materials Springer Journals

Spin–orbit torque engineering in β-W/CoFeB heterostructures with W–Ta or W–V alloy layers between β-W and CoFeB

Loading next page...
 
/lp/springer-journals/spin-orbit-torque-engineering-in-w-cofeb-heterostructures-with-w-ta-or-J39zVyhqGY

References (44)

Publisher
Springer Journals
Copyright
Copyright © The Author(s) 2021
ISSN
1884-4049
eISSN
1884-4057
DOI
10.1038/s41427-021-00326-8
Publisher site
See Article on Publisher Site

Abstract

The spin–orbit torque (SOT) resulting from a spin current generated in a nonmagnetic transition metal layer offers a promising magnetization switching mechanism for spintronic devices. To fully exploit this mechanism, in practice, materials with high SOT efficiencies are indispensable. Moreover, new materials need to be compatible with semiconductor processing. This study introduces W–Ta and W–V alloy layers between nonmagnetic β-W and ferromagnetic CoFeB layers in β-W/CoFeB/MgO/Ta heterostructures. We carry out first-principles band structure calculations for W–Ta and W–V alloy structures to estimate the spin Hall conductivity. While the predicted spin Hall conductivity values of W–Ta alloys decrease monotonically from −0.82 × 10 S/cm for W at% as the Ta concentration increases, those of W–V alloys increase to −1.98 × 10 S/cm for W V at% and then gradually decrease. 75 25 Subsequently, we measure the spin Hall conductivities of both alloys. Experimentally, when β-W is alloyed with 20 at% V, the absolute value of the spin Hall conductivity considerably increases by 36% compared to that of the pristine β-W. We confirm that the W–V alloy also improves the SOT switching efficiency by approximately 40% compared to that of pristine β-W. This study demonstrates a new material that can act as a spin current-generating layer, leading to energy- efficient spintronic devices. Introduction interface generate a SOT. The spin current, in turn, is In recent decades, there has been tremendous known to be generated by the spin–orbit interaction at 1,2 9,10 advancement in spintronics. Spin–orbit torque (SOT) , the NM or NM/FM interface . Recently, FMs (e.g., which is significantly more rapid and energy-efficient NiFe) have been shown to generate a spin current. 3,4 than spin-transfer torque (STT) ,has generated Other material classes, such as topological insulators 12–14 interest from technological and scientific perspectives. and Weyl semimetals, have also been studied . SOT is a critical part of magnetization switching, facil- However, their application in device manufacturing 5–8 itating devices with magnetic memory and logic .Ina appears to be limited due to the complexity of the typical nonmagnetic transition metal (NM)/ferromagnet growth method and low thermal stability. Therefore, it is (FM) heterostructure, polarized spin moments carried essential to design NMs with manufacturing process- by the spin current and accumulated at the NM and FM friendly materials, such as W, commonly used in the current semiconductor industry. It is also equally essential to identify material combinations that possess Correspondence: Sonny H. Rhim (sonny@ulsan.ac.kr) or Young perpendicular magnetic anisotropy (PMA), which is the Keun Kim (ykim97@korea.ac.kr) 1 15,16 Department of Materials Science and Engineering, Korea University, Seoul key to achieving high bit density . Two orthogonal 02841, Korea torques contribute to the SOT: damping-like (DL) and Department of Physics, University of Ulsan, Ulsan 44610, Korea field-like (FL) SOTs. Note that the DL-SOT efficiency Full list of author information is available at the end of the article These authors contributed equally: Gyu Won Kim, Do Duc Cuong © The Author(s) 2021 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to theCreativeCommons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/. 1234567890():,; 1234567890():,; 1234567890():,; 1234567890():,; Kim et al. NPG Asia Materials (2021) 13:60 Page 2 of 9 60 per unit current density (ξ ) is correlated to the spin follows DL Hall angle (θ ), which is the charge-to-spin conversion SH ratio, as follows: ξ = T θ ,where T is the inter- DL int SH int σ ¼ f Ω ðkÞ ð1Þ SH kn kn facial spin transparency. Unless we know the exact value of T ,the valueof θ is not equal to ξ . int SH DL Of the various NMs, β-W is a good candidate, showing where f is the Fermi–Dirac function for the nth band at kn PMA with CoFeB and a high DL-SOT efficiency ξ of k; therefore, the Berry curvature of the n-th band at k is DL 17,18 −0.33 to −0.40 . There have been several efforts expressed as made to improve the ξ of β-W derivatives, e.g., the DL 0 0 deposition of W by sputtering under an oxygen atmo- hj kn jji kn hj kn vbc kn x y Ω ðÞ k ¼ 2 Im 19  ð2Þ sphere and multistep W deposition to form a thick 0 e  e 0 n ≠n k;n k;n β-W layer of up to 16 nm . Both methods achieved ξ DL values of approximately −0.5. In the former study, the where j ¼ fg Σ; v is the spin current and Σ is the spin x x PMA strength weakened due to oxygen incorporation, operator in the full relativistic formalism. In this study, the 27,28 whereas the processing time increased in the latter. interpolation technique using Wannier90 was Moreover, Ta substitution into β-W was theoretically employed with s, p, and d orbitals of W, Ta, and V predicted to achieve a ξ of −0.5 due to intrinsic band atoms. Throughout this study, the xy component of the DL structure modification . spin Hall conductivity with the spin axis along the z- This study extends this Ta-based theoretical investi- direction was considered. For the alloys, the spin Hall gation. In addition, we expect V to play a similar role in conductivity at concentration x was calculated for all A15-structured β-W because V has the same number of possible configurations and was thermodynamically valence electrons as Ta. Thus, we explore Ta and V averaged using the Boltzmann factor by considering the substitutions in β-W for a wide range of alloy con- relative energies. centrations. Based on our first-principles band calcula- tions, an alloy layer, W X (where X is either Ta or Sample preparation 100−x x V), was introduced between the β-W and CoFeB layers. Samples were sputtered onto 1.25 × 1.25 cm thermally −9 We fabricated β-W/W X /CoFeB/MgO/Ta hetero- oxidized Si wafers under a base pressure below 5 × 10 100−x x structure film stacks on Si wafers coated with thermal Torr. The thickness of the Si-oxide layer was 300 nm. The oxides for various alloy compositions, 0 < x < 100 at%. stacking structure of each sample was W (4)/W X 100−x x Due to the presence of bottom β-W layers, the β-phase (2)/CoFeB (0.9)/MgO (1)/Ta (2) (the number in par- in the W X alloy layers was expected to be main- entheses represents the thickness in nanometers), where 100−x x tained to the maximum extent. Over most of the com- X is either Ta or V. A 4-nm-thick β-W layer was position range, PMA was exhibited. Harmonic Hall employed to maintain the β-phase in the 2-nm-thick W measurements confirmed that ξ increased to −0.49, X alloy layer. The alloy layer composition was varied in DL −x x corresponding to an approximately 40% enhancement 10 at% steps by changing the sputtering power densities of compared to that of β-W, which agreed well with the the W and Ta (or V) targets during co-deposition. The calculated results. This large SOT was further validated composition of the CoFeB target was Co Fe B in at%. 40 40 20 using other measurement methods, such as domain wall The metallic layers and the MgO interlayer were pro- depinning and the propagation model .Thiscombined duced by DC and radio frequency magnetron sputtering, theoretical and experimental study provides insight into respectively. All samples were post-annealed at 300 °C for engineering opportunities for developing materials 1 h under a magnetic field of 6 kOe applied perpendicular −6 exhibiting high SOT efficiency. to the film plane under a base pressure of 10 Torr. Each film was patterned into a 5-μm-wide and 35-μm-long Hall Materials and methods cross-device by photolithography (Karl Suss MA6) and Ar Theoretical calculations ion milling. An electrical contact pad of a Ti/Au bilayer First-principles calculations were performed using the was fabricated using an e-beam evaporator followed by a 23 2 Vienna Ab Initio Simulation Package with the projector lift-off process. We also constructed 4 × 4 μm -sized fer- augmented wave basis . An energy cutoff of 500 eV was romagnetic islands at the center of the Hall bar devices for selected, with a 16 × 16 × 16 k-point mesh for summation current-induced SOT switching, in which all parts in the Brillouin zone. The generalized gradient approx- excluding the islands were W/W–Ta or W/W–V layers. imation was employed for the exchange-correlation potential as parametrized by Perdew, Burke, and Ernzer- Measurements hof . The spin Hall conductivity (σ ) was calculated Hysteresis loops were measured using a vibrating sam- SH using the Kubo formula in the linear response theory as ple magnetometer (Microsense EV9). To evaluate the Kim et al. NPG Asia Materials (2021) 13:60 Page 3 of 9 60 SOT efficiency, the harmonic Hall response method was obtained for both W–Ta and W–V alloys at x = 25 at% employed (Supplementary Note 1, Figs. S1a, b). Two lock- (see Fig. 1c, d). The two alloys exhibit quite different in amplifiers were also utilized to simultaneously access behaviors based on their composition. σ for W–Ta SH the first- and second-harmonic responses of the magne- decreases monotonically with x, whereas that of the W–V tization. An alternating current (AC) of frequency 13.7 Hz alloy is enhanced as x increases from 12.5 to 50 at%. The and amplitude 1 mA was applied during the measurement W–Ta alloy favors the cc’ configuration with a 96% while changing the external magnetic field from −18,000 probability; hence, other configurations are not con- to 18,000 Oe with different azimuthal angles φ. The polar sidered. In comparison, for the W–V alloy, the bb’ con- angle of the sample was tilted (~5°) to prevent the for- figuration has the highest probability (60.5%), and the bc mation of multiple domains. configuration has a lower probability (35%). The prob- The current-induced SOT switching was measured by abilities are estimated using Boltzmann factors, con- the anomalous Hall voltage of the samples. The Hall sidering the relative total energies at room temperature voltage was detected by a 100-μA direct current (DC) (Supplementary Fig. S2e). after applying each current pulse is 0.5-mA steps and a The spin Hall conductivity of the bc configuration 10-μs width from −15 to 15 mA under a constant external reaches as high as −1.98 × 10 S/cm, equivalent to 141% field along the x-direction. The resistivity of the alloyed enhancement relative to that of β-W (−0.82 × 10 S/cm) layer was evaluated using a constant DC supply. Both (Supplementary Fig. S2f). This enhancement is attributed switching and resistivity were examined using a four- to the occurrence of strong symmetry breaking when V point electrical property measurement station (MSTECH simultaneously adopts bcc and chain sites. Specifically, the M7VC). The crystal structures of the W/W–Ta, and W/ Berry curvatures with opposite signs, a characteristic of W–V layers were characterized by grazing incidence (GI) β-W with a resonant double degenerate state, no longer X-ray diffraction (XRD, Rigaku ATX-G) with incident cancel out (Supplementary Note 3 and Fig. S3a–d). angles ranging from 30° to 80° in 0.02° steps. The atomic Moreover, the increased degeneracies associated with the distribution after annealing at 300 °C was evaluated by lowered symmetry increase the spin Hall conductivity, scanning transmission electron microscopy (STEM, FEI whose effect is most drastic when x = 25 at% in the W–V Double Cs Corrected Titan3 G2 60-300) with energy- alloy. Based on the calculations, a W–V alloy with a dispersive X-ray spectroscopy (EDS). For the STEM composition range of 12.5–50 at% is explored experi- sampling, a focused ion beam (FIB, FEI Quanta3D) system mentally, as discussed below. was used. Secondary ion mass spectrometry (SIMS, ION- TOF TOF, SIMS 5) was also used to examine the atomic SOT efficiency estimation employing the harmonic Hall distribution. method Based on first-principles calculations, various alloy lay- Results and discussion ers were introduced between the β-W and CoFeB layers in Materials screening by ab initio calculations a heterostructure consisting of β-W/CoFeB/MgO/Ta. The A previous investigation on x = 12.5 at% Ta alloying employment of the alloy layer was motivated by our was extended in this study to other concentrations of Ta previous study , which showed that a W layer could not and V. The crystal structure of A15 β-W comprises two maintain the β-W phase when it was alloyed with 10 at% symmetrically inequivalent sites: body-centered cubic Ta. A film stack of β-W (4)/W X (2)/CoFeB (0.9)/ 100−x x (bcc) and chain (c) sites. Due to the presence of these two MgO (1)/Ta (2) (thickness in nm) was fabricated with inequivalent sites, several configurations are possible for alloy X (Ta or V) to prevent the phase transition of W each x, where the acceptance of a configuration is deter- from β to α. Note that the β-phase is a fundamental mined by energetics. The spin Hall conductivity is esti- assumption in the theoretical calculations. mated by thermodynamic averaging (Supplementary Note The alloyed layer composition ranged from 0 to 100 at% 2). In particular, for an alloy with a solute composition x X in 10 at% steps. After annealing at 300 °C for 1 h in a = 25 at%, there are four possible configurations: cc, bc, cc’, vacuum, most film structures present PMA. However, and bb’, where b and c denote either a bcc site or a chain when the V composition exceeds 80 at%, a PMA to in- site. Specifically, the cc and cc’ configurations exist when plane magnetic anisotropy transition occurs. At this the solute atoms, either Ta or V, occupy the same and composition, the effective magnetic anisotropy energy eff different chain sites, respectively. Similarly, bb’ denotes (K ) decreases. (Supplementary Note 4, Fig. S4a–f, and the case when two different bcc sites are replaced, and bc Table S1). At x = 100 at%, the magnetic hysteresis loop is eff represents the case when each of the bcc and chain sites is not observed; hence, the sign of K cannot be deter- substituted (Supplementary Fig. S2a–d). mined. The degradation of the magnetic properties when The σ values were calculated as a function of x (see x > 80 at% is attributed to the formation of a magnetic SH 30,31 Fig. 1a, b), and the k-resolved Berry curvatures were dead layer between V and CoFe . Thus, in the Kim et al. NPG Asia Materials (2021) 13:60 Page 4 of 9 60 Fig. 1 Theoretical spin Hall conductivity calculation. Spin Hall conductivities of a W Ta and b W V and k-resolved Berry curvature of c 100−x x 100−x x W Ta and d W V alloy, where x = 25 at%, with two Ta or V adopting cc’ and bc configurations, respectively. 100−x x 100−x x Fig. 2 SOT measurement. a Schematic description of the device structure and stack of films. The coordinate axis is shown above the device structure. b DL and c FL-SOT efficiencies. d Experimentally estimated spin Hall conductivity. Error bars in figures represent standard errors. following, the SOT efficiency is discussed for the com- illustrates the harmonic Hall measurement, where an AC position range exhibiting PMA. is applied along the x-direction, following which the Hall Harmonic Hall measurements were performed to voltage (V ) along the y-direction is measured. During assess the SOT efficiencies. Figure 2a schematically the V measurement, sweeping the external in-plane H Kim et al. NPG Asia Materials (2021) 13:60 Page 5 of 9 60 magnetic field parallel (perpendicular) to the current In accordance with the ab initio calculations, the spin yields traces of DL-SOT (FL-SOT). Fig. 2b, c shows the Hall conductivity was estimated by employing the rela- DL-SOT (ξ ) and FL-SOT (ξ )efficiencies per unit tionship between the resistivity and the spin Hall angle, DL FL current density of the W/W Ta /CoFeB/MgO/Ta and θ ¼ σ ρ þ b , where σ comprises the intrinsic 100−x x SH SH SH xx W/W V /CoFeB/MgO/Ta heterostructures versus and side-jump contributions, ρ is the longitudinal 100−x x xx alloy composition x (in at%), respectively. The changes in resistivity, and b represents the skew-scattering con- ξ and ξ with respect to x are quite different. As the Ta tribution. Presumably, parameter b is negligible because DL FL content increases, ξ gradually decreases from −0.35 ± of the high impurity content in the studied system, and it DL 0.002 for a 6-nm-thick W to −0.06 ± 0.06 for the W/ is assumed that T is 1. Hence, the spin Hall conductivity int W Ta bilayer. Notably, the ξ of the W/W Ta can be expressed by the ratio of θ to ρ . The ρ value 0 100 DL 0 100 SH xx xx bilayer is similar to that of the highly resistive material of each device was determined by employing the parallel β-Ta . This suggests that the spin current generated in resistance model assuming ρ = 170 μΩ cm for CoFeB xx W inefficiently penetrates the W–Ta alloy layer and fur- (Supplementary Note 5, Fig. S5a, b). The results are dis- ther intensifies with increasing Ta composition. Conse- played in Fig. 2d, and the maximum SHC value is quently, in the extreme case of a bilayer NM structure, (−2.77 ± 0.31) × 10 S/cm at 20 at% V. This is reasonably most of the DL torque occurs due to the spin current consistent with the first-principles calculation results of generated in Ta. Furthermore, this is consistent with this study, which yield a maximum value of −1.98 × 10 S/ previous studies on the spin diffusion length of Ta, which cm at 25 at% V. Consequently, by using the W−V alloy 34,35 is 2–3nm . However, the ξ values of the W/W layer, the band structure modification predicted by the DL 100 V /CoFeB/MgO/Ta heterostructure exhibit interesting calculations is successfully captured. −x x composition-dependent behavior. At V = 20 at%, the ξ DL value reaches a maximum of −0.49 ± 0.05. This large Assessment of SOT switching value is in agreement with that of a W-based SOT tunnel Current-induced SOT switching in patterned devices 19,20 junction . was examined. A current sweep was applied from –15 to The FL-SOT (ξ )efficiencies do not present any dis- 15 mA. The current pulse width was 10 μs and was FL tinct compositional dependence, as depicted in Fig. 2c. In changed in 0.5-mA steps while the external magnetic field both the Ta and V cases, ξ  −0.10 to −0.20, suggesting strength was varied from −100 to 100 Oe in the x-direc- FL that the potential gradient at the interface (i.e., the Rashba tion. An additional fabrication process was utilized to effect) has a slight correlation with the alloy layer. In form an island pattern on the devices to test the current- addition, in the thin FM layer, in the layered structure induced SOT switching (Methods). with PMA, ξ depends on the magnetic anisotropy Figure 3a shows the switching loops for a hetero- FL energy and temperature . Moreover, the PMA in the structure with a 20 at% V alloy, which shows the highest NM/CoFeB/MgO structure mainly originates from the ξ . The sample clearly exhibits external field depen- DL CoFeB/MgO interfacial anisotropy energy, and the dence. When the polarity of the external field is varied, portion of NM/CoFeB is relatively small. Thus, the the switching polarity also changes. A strong applied insensitive behavior of ξ in this study is in line with a magnetic field implies a small observed switching current. FL previous report because only the W/CoFeB interface is Furthermore, magnetization switching was examined in modified. all patterned devices used in the harmonic measurements. Fig. 3 In-plane pulsed current-induced SOT switching. a SOT switching curves for different external field strengths for 20 at% V-incorporated devices. b SOT switching current density versus alloy composition for entire PMA samples. Under a constant 100 Oe external magnetic field along the x-direction. Kim et al. NPG Asia Materials (2021) 13:60 Page 6 of 9 60 As depicted in Fig. 3b, the magnetization direction was The DL effective field saturates at approximately 24.7 ± successfully manipulated by applying an in-plane current 2.33 Oe cm /MA when the external field exceeds 120 Oe. pulse and a constant external field, 100 Oe. The switching This is consistent with the effective DL field obtained current density (J ) and ξ are inversely proportional in using the harmonic Hall method, as indicated by the red SW DL single-domain devices as follows : dotted line in Fig. 4d. Because the two measurement schemes yield similar magnitudes for the DL-SOT effec- eff 2e H H tive field, it is concluded that the enhancement in the ext pffiffiffi ð3Þ J ¼ M t SW S FM SOT originates from the 20 at% V incorporation in the hξ 2 DL film structure. where t represents the thickness of the FM layer and Furthermore, we examine the 1/cos θ dependence of all FM eff H and H are the anisotropy field and the external devices and confirm that SOT switching occurs in a ext field, respectively. However, based on Figs. 2b and 3b, ξ domain wall-mediated process. From this perspective, the DL and J do not show inverse proportionality. A 4 μm× SOT switching efficiency (η), considering the domain wall SW 4 μm-sized island pattern was used to test the SOT depinning field, is investigated (Supplementary Note 7 switching, and this size of the island was extremely large and Fig. S7a). The enhancement ratio of the SOT to maintain a single-domain state. Therefore, this switching efficiency between pristine W and the 20 at% deviation is attributed to the existence of domains and V-incorporated samples is qualitatively consistent with domain walls . The next section describes tests on the enhancement ratio of ξ (Supplementary Fig. S7b). DL whether SOT switching entails domain wall motion and Therefore, increasing the switching current per increase double-checks of the high SOT efficiency. To this end, a in the V content, as shown in Fig. 3b, does not consider domain wall depinning and propagation model for the the effect of the domain wall depinning field. When the 20 at% V-incorporated devices was employed. domain wall depinning field during magnetization switching is considered, the W–V alloy achieves a higher Domain wall depinning and propagation model SOT switching efficiency than that in the single W layer. To examine the high SOT efficiency in the 20 at% V Note that if domain wall depinning and propagation alloy sample, the domain wall depinning and propagation occur during SOT switching, J follows Eq. (4) . sw model was applied. First, we determined whether SOT switching occurs through domain wall propagation or 4eμ M t H s FM c single domain reversal under the current application. J ¼ ð4Þ SW πhξ DL H (θ)is defined as the external switching field at polar angle θ. Figure 4a shows that the H (θ)/H (0) ratio p p accurately presents a 1/cos θ dependence, which confirms where e, μ , M ,t , and H represent the elementary 0 s FM c that in the studied Hall cross-device, magnetization charge of an electron, vacuum permeability, saturation switching involves a domain wall-mediated process. magnetization, thickness of the FM layer, and coercivity of Based on the SOT evaluation process, the measuring the ferromagnetic layer, respectively. The W–V alloyed time (t ) dependence of H was also examined, as shown structures reasonably follow the coercivity of CoFeB with m p in Fig. 4b. We find that the intrinsic coercive field (H )is regard to V composition (Supplementary Note 8 and Fig. c0 393.03 Oe and that the coercivity of the device is S8a, b). 274.82 Oe (Supplementary Fig. S6a). In addition, the Joule heating coefficient (κ) was determined utilizing two dif- Improvement in the DL-SOT efficiency of W–V alloys ferent channel resistance (R ) measurements: the tem- To elucidate the increase in the ξ in a W–Valloy,the xx DL perature dependence of R was determined by heating microstructure of the films was analyzed. GI-XRD was xx the device externally (Supplementary Fig. S6b), and the used to examine the phases of the W/W–Ta and W/ current dependence of R was determined from the W–V layers, and the results are shown in Fig. 5aand b. xx relationship T(I) ≈T + κI , where T is the room tem- Because the layers are ultrathin, the XRD spectra show 0 0 perature and I is the amplitude of the input current no peaks corresponding to CoFeB, MgO, or Ta. The (Supplementary Fig. S6c). W–Ta alloy retains the β-W phase up to 80 at% Ta When a current of I = 4.22 mA is applied, the tem- content. For the W–Valloy,asillustrated in Fig. 5b, the perature of the device increases by 157 K relative to room characteristic peak for α-W appears when 100 at% V is temperature, where κ is determined to be 0.02485 K/mA . incorporated. β-W has the A15 structure, which is an Using these parameters, the switching state diagram (Fig. energetically unstable phase. In addition, the atomic radii 4c) is converted into the DL effective field normalized by of W, Ta, and V are 135, 146, and 135 pm, respectively; the input current density (Fig. 4d). The detailed conver- therefore, the residual stress in the W–Ta alloy promotes sion process is described in Supplementary Note 6. the phase transition of β-W. Consequently, the phase Kim et al. NPG Asia Materials (2021) 13:60 Page 7 of 9 60 Fig. 4 Domain wall depinning and propagation switching model in the 20 at% V alloy-inserted heterostructure. a Polar angle dependence of the switching field. The green dotted line represents the fitted line for 1/cos θ. b Measuring time (t )-dependent behavior of the switching field. Error bars represent the standard deviation from 35 independent measurements. c Switching state diagram as a function of the external field in the x- direction. d Effective DL field evaluated from the switching state diagram. The red dotted line represents the effective DL field estimated from the harmonic measurement. The error bars in the figures represent the standard error. Fig. 5 Microstructural analysis of the alloy inserted heterostructure. a XRD patterns for W/W Ta /CoFeB/MgO/Ta films and b W/W V / 100−x x 100−x x CoFeB/MgO/Ta films after annealing at 300 °C for 1 h. Reference peak positions for the α (ICDD No. 00-004-0806) and β (ICDD No. 00-047-1319) phases of W are denoted below. stability of β-W in the W–V alloy is one of the reasons Schematics of the layered structures are placed on the for the enhancement in ξ . right side of the figures. The interfaces between each layer DL Because V is a lighter element than W, we examined become clear and sharp after the films undergo 300 °C atomic interdiffusion after annealing at 300 °C. To test annealings. EDS was conducted to locate the atomic whether ξ was enhanced due to its atomic interdiffusion positions in the film structure. To obtain the EDS inten- DL of V, atomic distribution profiles employing STEM were sity, the sample prepared for high-resolution (HR) TEM obtained. Figure 6a, b clearly shows the layered structures. measurement was extremely thin (20–30 nm), and thus, a Kim et al. NPG Asia Materials (2021) 13:60 Page 8 of 9 60 Fig. 6 Scanning transmission electron microscopy (STEM) data for the W/W–V/CoFeB/MgO/Ta heterostructure. a, b HR-TEM images of the as-deposited and 300 °C annealed samples. c, d HAADF-STEM image and element line profiles of the as-deposited and 300 °C annealed samples. The composition of the alloy layer in the atomic ratio is W V . 80 20 relatively thick specimen (80–100 nm) was prepared using composition-dependent behavior. Specifically, for the a FIB system. The atomic positions in the film structure 20 at% V alloy layer, ξ became −0.49 ± 0.05, an DL were assigned by EDS, as presented in Fig. 6c, d. Note that enhancement of 40% over that of pristine β-W. This the high-angle annular dark-field images appear blurred significant enhancement in ξ was confirmed by two DL because of the TEM sampling thickness. different measurement schemes: harmonic measure- We also conducted EDS on the samples used in the HR- ments and a domain wall depinning and propagation TEM measurement (Supplementary Fig. S9a, b). Based on model. The domain wall depinning and propagation Fig. 6c and d, V atoms are located under the CoFe layer, model presented a reasonable value for the DL-SOT regardless of heat treatment. The elemental profile of V effective field 24.7 ± 2.33 Oe cm /MA. observed in the SiO /W and Ta layers is confirmed to be The SOT switching efficiency (η) also improved, similar an EDS measurement artifact (Supplementary Note 9 and to the enhancement ratio of ξ . The XRD and TEM DL Fig. S10a–f). Furthermore, SIMS was performed, showing analyses confirmed that the high ξ was attributed to the DL that the atomic interdiffusion of V is negligibly small phase stability of β-W achieved by employing the W–V (Supplementary Note 10 and Fig. S11a–d). Therefore, we alloy layer. We expect that the optimized composition of can exclude the notion of atomic interdiffusion in the the W V alloy layer could become an energy-efficient 80 20 heterostructures. spin current source layer because of its high ξ and good DL process compatibility for semiconductor manufacturing. Conclusion Acknowledgements A combined theoretical and experimental study was The authors thank Yun Jung Jang of the Korea Institute of Science and presented for W/W−Ta/CoFeB/MgO/Ta and W/W–V/ Technology for helping us with the ToF-SIMS experiments. This research was CoFeB/MgO/Ta heterostructures. A series of first- supported by the National Research Foundation (NRF) of Korea (2015M3D1A1070465 and 2020M3F3A2A01082591). Y. K. Kim was supported principles calculations were conducted to investigate by Samsung Electronics Co., Ltd. (IO201211-08104-01). O. Lee was supported the W–Ta and W–V alloy compositions leading to high by the NRF of Korea (2020M3F3A2A01081635). The study at the University of spin Hall conductivity. Subsequently, fabrication and Ulsan was supported by the Basic Research Lab Program through the NRF of Korea (2018R1A4A1020696). measurement were performed by experimentally com- piling the composition dependencies of the SOT effi- Author details ciency. The W–Ta alloypresented agradually Department of Materials Science and Engineering, Korea University, Seoul decreasing ξ as the Ta content increased, and the ξ 2 DL DL 02841, Korea. Department of Physics, University of Ulsan, Ulsan 44610, Korea. of the W–V alloy demonstrated interesting Department of Physics and Computer Science, Vietnam National University, Kim et al. NPG Asia Materials (2021) 13:60 Page 9 of 9 60 Ho Chi Minh City 700000, Vietnam. Center for Spintronics, Korea Institute of 15. Mangin, S. et al. Current-induced magnetization reversal in nanopillars with Science and Technology, Seoul 02792, Korea. Seoul Center, Korea Basic perpendicular anisotropy. Nat. Mater. 5,210–215 (2006). Science Institute, Seoul 02841, Korea 16. Ikeda, S. et al. A perpendicular-anisotropy CoFeB-MgO magnetic tunnel junction. Nat. Mater. 9,721–724 (2010). 17. Pai, C.-F. et al. Spin transfer torque device utilizing the giant spin Hall effect of Author contributions tungsten. Appl. Phys. Lett. 101, 122404 (2012). G.W.K. and Y.K.K. conceived and designed the experiments. D.D.C., S.C.H., and S. 18. Hao,Q.&Xiao,G.Giant Spin Hall Effect andswitching inducedby spin-transfer H.R. performed the first-principles calculations. G.W.K., T.K., and M.H.L. torque in a W/Co Fe B /MgO structure with perpendicular magnetic ani- fabricated the film structures and devices. G.W.K., Y.J.K., and I.H.C. conducted 40 40 20 sotropy. Phys. Rev. Appl. 3, 034009 (2015). the SOT efficiency evaluation and analyzed the microstructural transition of β.- 19. Demasius, K.-U. et al. Enhanced spin-orbit torques by oxygen incorporation in W. G.W.K. conducted the domain wall propagation experiments with the help tungsten films. Nat. Commun. 7, 10644 (2016). of O.J.L. H.B. performed the TEM experiment and analyzed the atomic 20. Chen, W., Xiao, G., Zhang, Q. & Zhang, X. Temperature study of the giant spin distribution. G.W.K. and D.D.C. wrote the paper after discussion with all the Hall effect in the bulk limit of β-W. Phys. Rev. B 98, 134411 (2018). authors. Y.K.K. supervised the entire project. 21. Sui,X.etal. Giant enhancement of the intrinsic spin Hall conductivity in β-tungsten via substitutional doping. Phys. Rev. B 96, 241105(R) (2017). Competing interests 22. Lee, O. J. et al. Central role of domain wall depinning for perpendicular The authors declare no competing interests. magnetization switching driven by spin torque from the spin Hall effect. Phys. Rev. B 89, 024418 (2014). 23. Kresse, G. & Furthmüller, J. Efficient iterative schemes for ab initio total-energy Publisher’s note calculations using a plane-wave basis set. Phys.Rev.B 54, 11169 (1996). Springer Nature remains neutral with regard to jurisdictional claims in 24. Blöchl, P. Projector augmented-wave method. Phys. Rev. B 50, 17953 (1994). published maps and institutional affiliations. 25. Perdew, J. P., Burke, K. & Ernzerhof, M. Generalized gradient approximation made simple. Phys.Rev.Lett. 77, 3865 (1996). Supplementary information The online version contains supplementary 26. Guo, G. Y., Murakami, S., Chen, T.-W. & Nagaosa, N. Intrinsic spin Hall material available at https://doi.org/10.1038/s41427-021-00326-8. effect in platinum: first principles calculations. Phys. Rev. Lett. 100, 096401 (2008). 27. Souza, I., Marzari, N. & Vanderbilt, D. Maximally localized Wannier functions for Received: 13 March 2021 Revised: 30 June 2021 Accepted: 19 July 2021 entangled energy bands. Phys. Rev. B 65, 035109 (2001). 28. Mostofi, A. A. et al. wannier90: a tool for obtaining maximally-localised Wannier functions. Comput. Phys. Commun. 178, 685 (2008). 29. Cha,I.H., Kim, T.,Kim,Y.J., Kim, G. W. &Kim,Y. K.Thickness andcomposition- References dependent spin-orbit torque behaviors in perpendicular magnetized Ta/W (t)/ 1. Miron, I. M. et al. Perpendicular switching of a single ferromagnetic layer CoFeB and Ta W /CoFeB junction structures. J. Alloy. Compd. 823, 153744 1-x x induced by in-plane current injection. Nature 476, 189–193 (2011). (2020). 2. Liu, L. et al. Spin torque switching with the giant spin Hall effect of tantalum. 30. Devolder, T.,Tahmasebi, T., Eimer, S.,Hauet,T.&Andrieu, S. Compositional Science 336, 555–558 (2012). dependence of the magnetic properties of epitaxial FeV/MgO thin films. Appl. 3. Slonczewski, J. C. Current-driven excitation of magnetic multilayers. J. Magn. Phys. Lett. 103, 242410 (2013). Magn. Mater. 159,L1–L7 (1996). 31. Fackler, S. W. et al. Combinatorial study of Fe-Co-V hard magnetic thin films. 4. Berger,L.Emissionofspinwaves by a magnetic multilayer traversed by a Sci. Tech. Adv. Mater. 18,231 (2017). current. Phys. Rev. B 54, 9353 (1996). 32. Avci, C. O. et al. Interplay of spin-orbit torque and thermoelectric effects in 5. Fukami,S., Anekawa, T., Zhang,C.& Ohno,H. A spin-orbit torque switching ferromagnet/normal-metal bilayers. Phys. Rev. B 90, 224427 (2014). scheme with collinear magnetic easy axis and current configuration. Nat. 33. Yu, R. et al. Determination of spin Hall angle and spin diffusion length in Nanotechnol. 11,621–625 (2016). β-phase dominated tantalum. Phys. Rev. Mater. 2, 074406 (2018). 6. Baek, S. C. et al. Complementary logic operation based on electric-field 34. Niimi, Y. et al. Extrinsic spin Hall effects measured with lateral spin valve controlled spin-orbit torques. Nat. Electron. 1,398–403 (2018). structures. Phys. Rev. B 89, 054401 (2014). 7. Zhang, J. et al. Role of micromagnetic states on spin-orbit torques-switching 35. Kim, D.-J., Kim, S.-I., Park, S.-Y., Lee, K.-D. & Park, B.-G. Ferromagnetic resonance schemes. Nano Lett. 18, 4074–4080 (2018). spin pumping in CoFeB with highly resistive nonmagnetic electrodes. Curr. 8. Song, C. et al. Spin-orbit torques: materials, mechanisms, performances, and Appl. Phys. 14,1344–1348 (2014). potential applications. Prog. Mater. Sci. 118, 100761 (2021). 36. Ou, Y.,Pai, C.-F., Shi,S., Ralph, D. C. &Buhrman,R.A.Origin of field-like spin- 9. Sinova,J., Valenzuela,S.O., Wunderlich, J.,Back, C. H. &Jungwirth,T.SpinHall orbit torques in heavy metal/ferromagnet/oxide thin film heterostructures. effects. Rev. Mod. Phys. 87,1213–1259 (2015). Phys. Rev. B 94, 140414(R) (2016). 10. Manchon, A. & Zhang, S. Theory of nonequilibrium intrinsic spin torque in a 37. Fritz, K., Wimmer, S., Ebert, H. & Menert, M. Large spin Hall effect in an single nanomagnet. Phys. Rev. B 78, 212405 (2008). amorphous binary alloy. Phys.Rev.B 98, 094433 (2018). 11. Wang, W. et al. Anomalous spin-orbit torques in magnetic single-layer films. 38. Lee,K.-S., Lee,S.-W., Min,B.-C. &Lee, K.-J.Thermally activatedswitching of Nat. Nanotech. 14,819–824 (2019). perpendicular magnet by spin-orbit spin torque. Appl. Phys. Lett. 104, 072413 12. Fan, Y. et al. Magnetization switching through giant spin-orbit torque in a (2014). magnetically doped topological insulator heterostructure. Nat. Mater. 13, 39. Emori,S., Bauer, U.,Ahn,S.-M.,Martinez, E. &Beach,G.S.D.Current-driven 699–704 (2014). dynamics of chiral ferromagnetic domain walls. Nat. Mater. 12,611–616 13. Wang, Y. et al. Room temperature magnetization switching in topological (2013). insulator-ferromagnet heterostructures by spin-orbit torques. Nat. Commun. 8, 40. Zhu, L., Ralph, D. C. & Buhrman,R.A.Lack ofsimple correlation between 1364 (2017). switching current density and spin-orbit torque efficiency of perpendicularly 14. Li, P. et al. Spin-momentum locking and spin-orbit torques in magnetic nano- magnetized spin-current-generator-ferromagnet heterostructures. Phys. Rev. heterojunctions composed of Weyl semimetal WTe . Nat. Commun. 9,3990 Appl. 15, 024059 (2021). (2018).

Journal

NPG Asia MaterialsSpringer Journals

Published: Dec 1, 2021

There are no references for this article.