Computational design of magnon spintronic devices with multiscale approach by combining time-dependent quantum transport with classical micromagnetics

传统的自旋电子学利用电子的自旋电流来携带、传输和处理信息。然而，它们在短距离内衰减，同时也产生焦耳热损失。碰撞在磁绝缘体上的电子自旋电流被转换成由局域磁矩集体运动携带的自旋电流。这种自旋波或磁振子自旋电流可以长距离传播而没有焦耳热损失，因为电子不会穿过绝缘体。当自旋电流在具有易平面各向异性的室温磁绝缘体中通过进动磁矩的螺旋织构传输时，甚至可以实现更慢的空间衰减，在低温下具有相干和超流体传输的许多特征而不耗散。此外，自旋波可以干扰并表现出非线性波相互作用，这可以用于新型基于波的逻辑门，以及在同一芯片上混合逻辑和存储器，以避免所谓的存储器墙作为传统电子器件中远距离慢速存储器和快速逻辑门之间数据交换的瓶颈。超流体自旋输运可以用来制造超导约瑟夫森结的磁性类似物。该研究将采用新开发的计算工具来模拟电子自旋电流与自旋波或超流体自旋电流之间的相互转换，从而为器件制造提供精确的指导，以最优控制长距离和低耗散自旋编码信息流穿过磁性绝缘体及其在正常金属触点的体或界面附近的探测。拟议研究的更广泛影响将包括：设计具有超低功耗的新型计算技术的构建模块；高级量子器件建模和超级计算模拟的研究生培训；并创建新的公开可用的设备建模软件。磁振子自旋电子器件的可扩展性需要有效的方案来激发短波长的交换自旋波，这将通过模拟由两个磁畴壁湮灭或单个电流驱动的畴壁产生的纳米级波长的自旋波来研究。这些设置也将成为内存逻辑设备模拟的基础，其中域壁存储二进制信息，而自旋波通过遍历它们来执行逻辑操作。将对具有易轴向各向异性的铁磁和反铁磁绝缘体上调制自旋波电流的磁振子阀和晶体管进行仿真。将研究通过易平面各向异性磁绝缘体调制超流体自旋电流的方案，包括约瑟夫森结的磁性类似物。在所有这些装置中，磁绝缘体处理的信息读出需要最终将由局域磁矩动力学携带的自旋电流转换为传统的电子自旋和电荷电流。整个过程将通过利用和进一步推进最近开发的多尺度框架的理论和计算能力进行微观模拟，该框架将电子的时变非平衡格林函数算法与经典局域磁矩动力学的Landau-Lifshitz-Gilbert方程自洽地结合起来。这一框架使得首次以时间分辨的方式直接研究电子自旋电流（稳定或脉冲）的注入如何通过自旋扭矩激发自旋波或超流体自旋输运，以及它们如何通过自旋泵效应激发电子自旋电流成为可能。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

Conventional spintronics employs spin currents of electrons to carry, transport and process information. However, they decay over short distances while also generating Joule heat losses. Electronic spin current impinging onto a magnetic insulator is converted into spin current carried by collective motion of localized magnetic moments. Such spin wave or magnon spin currents can propagate over long distances without Joule heat losses since electrons do not move through an insulator. Even slower spatial decay of spin currents becomes possible when they are transported by spiraling textures of precessing magnetic moments within room temperature magnetic insulators with easy-plane anisotropy, sharing many features of coherent and superfluid transport without dissipation at cryogenic temperatures. In addition, the spin waves can interfere and exhibit nonlinear wave interaction which can be exploited for novel wave-based logic gates, as well as for mixing of logic and memory on the same chip to evade the so-called memory wall as a bottleneck in data exchange between distant slow memory and fast logic gates in conventional electronics. The superfluid spin transport can be exploited to create magnetic analogues of superconducting Josephson junctions. The proposed research will employ newly developed computational tools to simulate interconversion between electronic spin currents and spin wave or superfluid spin currents, thereby offering a precise guidance for device fabrication with optimal control of long-distance and low-dissipation spin-encoded information flow across magnetic insulators and its probing in the bulk or near interfaces with normal metals contacts. Broader impact of the proposed research will include: design of building blocks for novel computing technologies with ultralow power consumption; training for graduate students in advanced quantum device modeling and supercomputing simulations; and creation of new publicly available device modeling software.The scalability of magnon spintronic devices requires efficient schemes to excite exchange spin waves of short wavelength, which will be investigated by simulating spin waves of nanoscale wavelength generated by annihilation of two magnetic domain walls or by a single current-driven domain wall. These setups will also underlie simulations of logic-in-memory devices where domain walls store binary information while spin waves perform logic operations by traversing them. Magnon valves and transistors which modulate spin wave spin current across ferro- and antiferromagnetic insulators with easy-axis anisotropy will be simulated. Schemes to modulate superfluid spin currents across magnetic insulators with easy-plane anisotropy will be investigated, including magnetic analogues of Josephson junctions. In all of these devices, readout of information processed by magnetic insulator requires to eventually convert spin currents carried by the dynamics of localized magnetic moments into conventional electronic spin and charge currents. The whole process will be modeled microscopically by utilizing and further advancing theoretical and computational capabilities of recently developed multiscale framework which self-consistently combines time-dependent nonequilibrium Green function algorithms for electrons with the Landau-Lifshitz-Gilbert equation for classical dynamics of localized magnetic moments. This framework makes it possible for the first time to directly, and in time-resolved fashion, investigate how injection of electronic spin current (steady or pulsed) excites spin waves or superfluid spin transport by spin torque and, how they excite electronic spin currents via the spin pumping effect.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.