Magnetoresistive random access memory(MRAM)is a promising non-volatile memory technology that can be utilized as an energy and space-efficient storage and computing solution,particularly in cache functions within circ...Magnetoresistive random access memory(MRAM)is a promising non-volatile memory technology that can be utilized as an energy and space-efficient storage and computing solution,particularly in cache functions within circuits.Although MRAM has achieved mass production,its manufacturing process still remains challenging,resulting in only a few semiconductor companies dominating its production.In this review,we delve into the materials,processes,and devices used in MRAM,focusing on both the widely adopted spin transfer torque MRAM and the next-generation spin-orbit torque MRAM.We provide an overview of their operational mechanisms and manufacturing technologies.Furthermore,we outline the major hurdles faced in MRAM manufacturing and propose potential solutions in detail.Then,the applications of MRAM in artificial intelligent hardware are introduced.Finally,we present an outlook on the future development and applications of MRAM.展开更多
基于自旋电子学的磁性随机存储器(Magnetic Random Access Memory,MRAM)具有非易失性、可无限擦写、低功耗和快速写入等优点,从而有望成为下一代通用存储器.在MRAM家族中,根据磁写入方式和磁媒介的不同,人们提出了自旋转移力矩(Spin Tra...基于自旋电子学的磁性随机存储器(Magnetic Random Access Memory,MRAM)具有非易失性、可无限擦写、低功耗和快速写入等优点,从而有望成为下一代通用存储器.在MRAM家族中,根据磁写入方式和磁媒介的不同,人们提出了自旋转移力矩(Spin Transfer Torque,STT)、自旋轨道力矩(Spin Orbit Torque,SOT)、电压控制型(如电控磁各向异性(Voltage Controlled Magnetic Anisotropy,VCMA)型和电控SOT(Voltage Gated-SOT,VG-SOT)型)、磁畴壁(Domain Wall,DW)和磁斯格明子(Magnetic Skyrmion)型等不同类型的各具特色的MRAM,共同推动着磁存储技术的多元化发展.近年来STT-MRAM商用芯片的成功问世进一步推动了MRAM器件的研究与应用.本文首先简要介绍了存储器技术的历史,然后介绍了MRAM的基本工作原理、从MRAM中读取和写入信息背后的技术、材料和不同的物理机制以及潜在的挑战问题.接下来介绍了近些年发展的新型的磁写入机制的进展.本文末尾讨论了一些可能有助于行业超越传统MRAM的技术,最后是总结和展望.展开更多
Recent progresses in magnetic tunnel junctions with perpendicular magnetic anisotropy （PMA） are reviewed and summarized. At first, the concept and source of perpendicular magnetic anisotropy （PMA） are introduced. ...Recent progresses in magnetic tunnel junctions with perpendicular magnetic anisotropy （PMA） are reviewed and summarized. At first, the concept and source of perpendicular magnetic anisotropy （PMA） are introduced. Next, a historical overview of PMA materials as magnetic electrodes, such as the RE-TM alloys TbFeCo and GdFeCo, novel tetragonal manganese alloys Mn-Ga, L10-ordered （Co, Fe）/Pt alloy, multilayer film [Co, Fe, CoFe/Pt, Pd, Ni, AU]N, and ultra-thin magnetic metal/oxidized barrier is offered. The other part of the article focuses on the optimization and fabrication of CoFeB/MgO/CoFeB p-MTJs, which is thought to have high potential to meet the main demands for non-volatile magnetic random access memory.展开更多
This work presents a new approach for simulating the random waves in viscous fluids and the associated bottom shear stresses. By generating the incident random waves in a numerical wave flume and solving the unsteady ...This work presents a new approach for simulating the random waves in viscous fluids and the associated bottom shear stresses. By generating the incident random waves in a numerical wave flume and solving the unsteady two-dimensional Navier-Stokes equations and the fully nonlinear free surface boundaiy conditions for the fluid flows in the flume, the viscous flows and laminar bottom shear stresses induced by random waves axe determined. The deterministic spectral amplitude method implemented by use of the fast Fourier transform algorithm was adopted to generate the incident random waves. The accuracy of the numerical scheme is confirmed by comparing the predicted wave spectrum with the target spectrum and by comparing the nanlerical transfer function between the shear stress and the surface elevation with the theoretical transfer function. The maximum bottom shear stress caused by random waves, computed by this wave model, is compared with that obtained by Myrhaug＇ s model （1995）. The transfer function method is also employed to determine the maximum shear stress, and is proved accurate.展开更多
地质灾害易发性评价对防灾减灾至关重要,但镇域常因样本稀缺导致评价因子权重难以确定。以陕西省安康市镇坪县城关镇(目标域)为例,提出基于迁移学习的地质灾害易发性预测模型。采用TrAdaBoost算法,将镇坪县全域数据(源域)与城关镇数据结...地质灾害易发性评价对防灾减灾至关重要,但镇域常因样本稀缺导致评价因子权重难以确定。以陕西省安康市镇坪县城关镇(目标域)为例,提出基于迁移学习的地质灾害易发性预测模型。采用TrAdaBoost算法,将镇坪县全域数据(源域)与城关镇数据结合,以CatBoost为基准学习器构建TrAdaBoost-CatBoost模型,实现知识迁移。结果表明:迁移策略显著提升了目标域性能,TrAdaBoost-CatBoost模型受试者工作特征曲线(receiver operating characteristic curve,ROC)下面积(area under the curve,AUC)达0.96,较仅用城关镇数据的CatBoost模型ROC曲线AUC(0.94)提升了0.02;与传统模型对比,TrAdaBoost-CatBoost模型ROC曲线AUC显著优于支持向量机(support vector machine,SVM)模型ROC曲线AUC(0.92)和随机森林(random forest,RF)模型ROC曲线AUC(0.93),分别高出0.04和0.03;迁移框架具普适性,TrAdaBoost-SVM模型ROC曲线AUC为0.94(较SVM模型ROC曲线AUC提升了0.02),TrAdaBoost-RF模型ROC曲线AUC为0.95(较RF模型的AUC提升了0.02),两者性能均得到提升,但TrAdaBoost-CatBoost模型(AUC=0.96)仍保持最优。该模型为小样本区域地质灾害评价提供了高精度解决方案,验证了迁移学习在数据稀缺场景的有效性,对类似区域灾害风险防控具有实际参考意义。展开更多
基金supported in part by the Youth Innovation Promotion Association of Chinese Academy of Sciences(CAS)under Grant 2020118Beijing Nova Program under Grant 20230484358Beijing Superstring Academy of Memory Technology:under Grant No.E2DF06X003。
文摘Magnetoresistive random access memory(MRAM)is a promising non-volatile memory technology that can be utilized as an energy and space-efficient storage and computing solution,particularly in cache functions within circuits.Although MRAM has achieved mass production,its manufacturing process still remains challenging,resulting in only a few semiconductor companies dominating its production.In this review,we delve into the materials,processes,and devices used in MRAM,focusing on both the widely adopted spin transfer torque MRAM and the next-generation spin-orbit torque MRAM.We provide an overview of their operational mechanisms and manufacturing technologies.Furthermore,we outline the major hurdles faced in MRAM manufacturing and propose potential solutions in detail.Then,the applications of MRAM in artificial intelligent hardware are introduced.Finally,we present an outlook on the future development and applications of MRAM.
文摘基于自旋电子学的磁性随机存储器(Magnetic Random Access Memory,MRAM)具有非易失性、可无限擦写、低功耗和快速写入等优点,从而有望成为下一代通用存储器.在MRAM家族中,根据磁写入方式和磁媒介的不同,人们提出了自旋转移力矩(Spin Transfer Torque,STT)、自旋轨道力矩(Spin Orbit Torque,SOT)、电压控制型(如电控磁各向异性(Voltage Controlled Magnetic Anisotropy,VCMA)型和电控SOT(Voltage Gated-SOT,VG-SOT)型)、磁畴壁(Domain Wall,DW)和磁斯格明子(Magnetic Skyrmion)型等不同类型的各具特色的MRAM,共同推动着磁存储技术的多元化发展.近年来STT-MRAM商用芯片的成功问世进一步推动了MRAM器件的研究与应用.本文首先简要介绍了存储器技术的历史,然后介绍了MRAM的基本工作原理、从MRAM中读取和写入信息背后的技术、材料和不同的物理机制以及潜在的挑战问题.接下来介绍了近些年发展的新型的磁写入机制的进展.本文末尾讨论了一些可能有助于行业超越传统MRAM的技术,最后是总结和展望.
基金supported by the State Key Project of Fundamental Research of Ministry of Science and Technology,China(Grant No.2010CB934400)the National Natural Science Foundation of China(Grant Nos.51229101 and 11374351)
文摘Recent progresses in magnetic tunnel junctions with perpendicular magnetic anisotropy （PMA） are reviewed and summarized. At first, the concept and source of perpendicular magnetic anisotropy （PMA） are introduced. Next, a historical overview of PMA materials as magnetic electrodes, such as the RE-TM alloys TbFeCo and GdFeCo, novel tetragonal manganese alloys Mn-Ga, L10-ordered （Co, Fe）/Pt alloy, multilayer film [Co, Fe, CoFe/Pt, Pd, Ni, AU]N, and ultra-thin magnetic metal/oxidized barrier is offered. The other part of the article focuses on the optimization and fabrication of CoFeB/MgO/CoFeB p-MTJs, which is thought to have high potential to meet the main demands for non-volatile magnetic random access memory.
基金the Science Council (Grant No. NSC95-2221-E-006-474)
文摘This work presents a new approach for simulating the random waves in viscous fluids and the associated bottom shear stresses. By generating the incident random waves in a numerical wave flume and solving the unsteady two-dimensional Navier-Stokes equations and the fully nonlinear free surface boundaiy conditions for the fluid flows in the flume, the viscous flows and laminar bottom shear stresses induced by random waves axe determined. The deterministic spectral amplitude method implemented by use of the fast Fourier transform algorithm was adopted to generate the incident random waves. The accuracy of the numerical scheme is confirmed by comparing the predicted wave spectrum with the target spectrum and by comparing the nanlerical transfer function between the shear stress and the surface elevation with the theoretical transfer function. The maximum bottom shear stress caused by random waves, computed by this wave model, is compared with that obtained by Myrhaug＇ s model （1995）. The transfer function method is also employed to determine the maximum shear stress, and is proved accurate.
文摘地质灾害易发性评价对防灾减灾至关重要,但镇域常因样本稀缺导致评价因子权重难以确定。以陕西省安康市镇坪县城关镇(目标域)为例,提出基于迁移学习的地质灾害易发性预测模型。采用TrAdaBoost算法,将镇坪县全域数据(源域)与城关镇数据结合,以CatBoost为基准学习器构建TrAdaBoost-CatBoost模型,实现知识迁移。结果表明:迁移策略显著提升了目标域性能,TrAdaBoost-CatBoost模型受试者工作特征曲线(receiver operating characteristic curve,ROC)下面积(area under the curve,AUC)达0.96,较仅用城关镇数据的CatBoost模型ROC曲线AUC(0.94)提升了0.02;与传统模型对比,TrAdaBoost-CatBoost模型ROC曲线AUC显著优于支持向量机(support vector machine,SVM)模型ROC曲线AUC(0.92)和随机森林(random forest,RF)模型ROC曲线AUC(0.93),分别高出0.04和0.03;迁移框架具普适性,TrAdaBoost-SVM模型ROC曲线AUC为0.94(较SVM模型ROC曲线AUC提升了0.02),TrAdaBoost-RF模型ROC曲线AUC为0.95(较RF模型的AUC提升了0.02),两者性能均得到提升,但TrAdaBoost-CatBoost模型(AUC=0.96)仍保持最优。该模型为小样本区域地质灾害评价提供了高精度解决方案,验证了迁移学习在数据稀缺场景的有效性,对类似区域灾害风险防控具有实际参考意义。
