We published a paper entitled “Antiferromagnetic ordering enhanced magnetic damping in Mn₂Au/CoFeB bilayers” in [Applied Physics Letters 129, 022401 (2026)]. Antiferromagnets hold significant potential for spintronic devices owing to their insensitivity to external magnetic fields and the absence of stray fields. In this work, we investigated the magnetic dynamics of Mn₂Au/CoFeB bilayers via ferromagnetic resonance. We found that the Néel temperature of 2-nm-thick Mn₂Au is as low as 40 K, in sharp contrast to that of bulk Mn₂Au which exceeds 1000 K. As the temperature decreases from 160 to 10 K, the magnetic damping α of the CoFeB layer increases from 0.013 to 0.047, accompanied by a synchronous increase of the exchange coupling field. Such an enhancement of damping is attributed to the spin angular momentum transfer from CoFeB to Mn₂Au, mediated by the AFM–FM exchange coupling which is strengthened by the antiferromagnetic ordering of Mn₂Au at low temperatures. Our study provides deeper insights into AFM/FM dynamics and spintronic storage technology. Congratulations to Donghang Xie and co-workers!

We published a paper entitled “Copper-doped sputtered BiSb topological insulators for field-free perpendicular magnetization switching via out-of-plane spin polarization” in [Physical Review Materials 10, 064201 (2026)]. Achieving high SOT efficiency together with deterministic field-free switching is a key challenge for perpendicular-magnetization SOT-MRAM. In this work, we propose an effective strategy to realize high-efficiency field-free switching with an ultralow critical current density by doping copper into the topological insulator Bi_0.9Sb_0.1 (BiSb). The cosputtered Cu-BiSb alloy film achieves a high damping-like SOT efficiency (ξ_y,DL = 0.53) and, in contrast to undoped BiSb, a significant field-like torque induced by spin currents with out-of-plane spin polarization (ξ_z,FL = 0.55). This unconventional out-of-plane spin polarization enables deterministic field-free switching of perpendicular magnetization with a critical current density as low as ~4 MA/cm² in Cu-BiSb/Pt/Co/Pt heterostructures. Our work provides critical support for the development of TI-based, high-performance and integrable SOT-MRAM devices. Congratulations to Haotian Duan and co-workers!

We published a paper entitled “Achieving high spin–orbit torque efficiency and robust thermal stability in the sputter-grown topological bilayer BiSb/Pt” in [Applied Physics Letters 128, 212409 (2026)]. Topological insulators (TIs) are promising candidates for spin–orbit torque (SOT) devices owing to their high charge-to-spin conversion efficiency, but the SOT efficiency of sputtered TI films is usually compromised by the elevated thermal budgets required by CMOS back-end-of-line processing. In this work, we examined the thermal stability of sputter-grown Bi_0.9Sb_0.1/Py films. While the system exhibits a high SOT efficiency (ξ_FMR = 0.83) after annealing at 230 °C, the efficiency decreases by ~65% when the annealing temperature is increased to 400 °C. By inserting a 1.5 nm Pt layer at the Bi_0.9Sb_0.1/Py interface, interfacial intermixing is effectively suppressed, and the stack maintains a strong SOT efficiency of 0.85 even after annealing at 400 °C. These results pave the way for BiSb-based SOT-MRAM and its integration with CMOS technologies. Congratulations to Zui Tao, Zeyi Zhu (equal contribution) ,  and co-workers!!

We published a paper entitled “Phase transition to failure: Quantifying critical thresholds of gradient conflict in PINN for LLG dynamics” (Chinese title: 从弱非线性可解到强非线性失效：LLG方程中梯度冲突诱导的PINN失效边界) in [ Progress in Physics 62(2), 309-322 (2026)]. In this work, we systematically investigated the capability of physics-informed neural networks (PINNs) for solving the Landau–Lifshitz–Gilbert (LLG) equation, the core governing equation of micromagnetics. By varying the magnetocrystalline anisotropy constant K_u and the demagnetization factor N to tune the strength of nonlinearity, we found that PINNs can solve the LLG equation only under weakly nonlinear conditions. In strongly nonlinear regimes, gradient conflicts during the training iterations lead to divergence or a catastrophic loss of accuracy, which defines a quantitative failure boundary for machine-learning solvers of nonlinear magnetization dynamics. This work provides a rational understanding of the performance boundaries of PINNs and helps to correct the overly optimistic assessment of ML-based differential-equation solvers. Congratulations to Ding Ma and Co-workers!

We published a paper entitled “Subterahertz collective spin-resonance modes and field-adaptive reservoir computing in the chiral helimagnet Cr_1/3TaS₂” in [Phys. Rev. Applied 24, 054022 (2025)]. Monoaxial chiral helimagnets (CHMs) host rich helical spin textures, including chiral soliton lattices (CSLs) with tunable periods. However, the spin resonance modes of existing CHMs lie in the gigahertz range, limiting their potential for high-speed signal processing. In this work, by combining ferromagnetic resonance, electron spin resonance, and magneto-Raman spectroscopy, we uncovered subterahertz CSL phonon modes in the CHM Cr_1/3TaS₂. Near the critical field, nontrivial CSL phonon modes reach 0.15 THz, while a uniform ferromagnetic resonance mode emerges at 0.375 THz in the forced ferromagnetic phase under 9 T. The CSL phonon frequency in Cr_1/3TaS₂ is five to six times higher than that of isostructural Cr_1/3NbS₂, owing to the larger spin–orbit-coupling-induced Dzyaloshinskii–Moriya interaction. Micromagnetic simulations further resolve the frequency spectrum and the spatial distributions of amplitudes, phases, and precession trajectories of each CSL resonance mode. Moreover, we demonstrate that physical reservoir computing exploiting the nonlinear collective spin dynamics and field-controlled hysteresis of these nontrivial spin textures achieves exceptional performance in time-series prediction tasks. Our findings pave the way for CHM materials toward subterahertz signal processing and neuromorphic computing applications. Congratulations to Zishuang Li, Shuai Zhang (equal contribution), and Co-workers!

We published an invited review entitled “自旋类脑神经形态计算” (Spin-based Brain-like Neuromorphic Computing) in [Journal of Sichuan Normal University (Natural Science) 48(2), 176-191 (2025)]. Brain-inspired neuromorphic computing aims to simulate and implement the information processing and learning capabilities of the biological brain, with one of its key ideas being to mimic the behavior of biological neurons and synapses to achieve information transmission, processing, and storage. Owing to their non-volatility, high speed, low power consumption, nearly infinite endurance, and inherent nonlinearity, spintronic devices have been widely explored and have shown remarkable performance in neuromorphic computing. In this invited review, we systematically introduce and summarize various spintronic effects, including magnetoresistance effects, spin-transfer torque and spin-orbit torque effects, voltage-controlled magnetic anisotropy, and nonlinear magnetization dynamics. Taking the applications of various spintronic devices in reservoir computing, Ising machines, spiking neural networks, and true random number generators as examples, we present an outlook on the prospects and trends of spin-based neuromorphic computing hardware for future artificial-intelligence chips. Congratulations to Shuai Zhang and Co-workers!

研究论文 | 基于随机自旋轨道矩器件构建人工脉冲神经网络硬件

《中国科学：物理学 力学 天文学》英文版(SCIENCE CHINA Physics, Mechanics & Astronomy, SCPMA)出版南京大学刘荣华教授团队成果，文章题为“Stochastic spin-orbit-torque device as the STDP synapse for spiking neural networks”，于2023年第66卷第5期刊出。

目前，人工智能技术已渗透到人们生活生产的各个方面，然而随着其高速发展，传统冯·诺依曼架构计算机已经难以满足其庞大的计算量要求。近期，受人脑学习方式的启发，基于各类物理器件构建的神经形态计算硬件在人工智能领域受到了广泛关注。在众多的人工神经网络模型中， 脉冲神经网络展现出低能耗和高并行率的特性。拥有短时记忆效应、亚纳秒尺度非线性磁动力学和随机翻转行为的纳米自旋电子器件，可以用来构建循环神经网络和脉冲神经网络硬件。相比于传统的自旋转移矩，近期发现的自旋轨道矩可以使自旋电子器件实现更长的使用寿命和更快的写入速度，因此，自旋轨道矩器件受到学术界和工业界青睐。

脉冲神经网络中的神经元之间是基于众多突触传递离散脉冲序列进行通信的，突触权重可以根据与突触连接的前后神经元的激发状态进行调整，即突触具有可塑性。本文介绍了在一类低功耗自旋轨道矩器件中，通过研究电流诱导的自旋轨道矩效应和焦耳热效应，发现器件存储层磁矩的随机翻转概率（对应于电导率）随输入脉冲电流及其时间间隔成指数衰减关系，即时间间隔越长，随机翻转概率越低。为了演示该类自旋轨道矩器件能模拟人脑神经突触可塑性，本文进一步基于该自旋器件构建了两种人工脉冲神经网络，分别进行了手写数字识别的无监督学习（准确率高达80%）和逻辑运算的学习。该工作为如何运用新型低功耗自旋电子器件实现神经形态计算硬件提供了新的思路和具体方案。

创新要点：

本文利用自旋轨道矩器件中电流驱动存储层磁矩随机翻转和其温度特性实现了人工突触。由于存储层磁矩的随机翻转概率随输入脉冲电流及其时间间隔变化，其构建的人工突触的可塑性可以由存储层磁矩方向对应的霍尔电阻变化实现。基于该类自旋人工突触，研究人员构建了两种脉冲神经网络硬件原型，实现了非监督的手写数字识别和逻辑运算学习。

原文信息：

H. Li, L. Li, K. Zhou, C. Yan, Z. Gao, Z. Li, and R. Liu, Stochastic spin-orbit-torque device as the STDP synapse for spiking neural networks, Sci. China-Phys. Mech. Astron. 66, 257512 (2023), https://doi.org/10.1007/s11433-022-2081-5