The in-memory computing paradigm aims at overcoming the intrinsic inefficiencies of Von-Neumann computers by reducing the data-transport per arithmetic operation. Crossbar arrays of multilevel memristive devices enable efficient calculations of matrix-vector-multiplications, an operation extensively called on in artificial intelligence (AI) tasks. Resistive random-access memories (ReRAMs) are promising candidate devices for such applications. However, they generally exhibit large stochasticity and device-to-device variability. The integration of a sub-stoichiometric metal-oxide within the ReRAM stack can improve the resistive switching graduality and stochasticity. To this purpose, a conductive TaO_x layer is developed and stacked on HfO₂ between TiN electrodes, to create a complementary metal-oxide-semiconductor-compatible ReRAM structure. This device shows accumulative conductance updates in both directions, as required for training neural networks. Moreover, by reducing the TaO_x thickness and by increasing its resistivity, the device resistive states increase, as required for reduced power consumption. An electric field-driven TaO_x oxidation/reduction is responsible for the ReRAM switching. To demonstrate the potential of the optimized TaO_x/HfO₂ devices, the training of a fully-connected neural network on the Modified National Institute of Standards and Technology database dataset is simulated and benchmarked against a full precision digital implementation.

中文翻译：

---

用于神经网络训练的灯丝 TaOx/HfO2 ReRAM 设备，带有模拟内存计算

内存计算范式旨在通过减少每个算术运算的数据传输来克服冯诺依曼计算机固有的低效率。多级忆阻器件的交叉开关阵列能够有效计算矩阵向量乘法，这是人工智能 (AI) 任务中广泛调用的操作。电阻式随机存取存储器 (ReRAM) 是此类应用的有希望的候选设备。然而，它们通常表现出很大的随机性和设备到设备的可变性。在 ReRAM 堆栈中集成亚化学计量的金属氧化物可以提高电阻切换的渐变性和随机性。为此，开发了导电的 TaO _x层并将其堆叠在 HfO _2上在 TiN 电极之间，以创建互补的金属氧化物半导体兼容的 ReRAM 结构。根据训练神经网络的要求，该设备在两个方向上显示累积的电导更新。此外，通过减小TaO _x厚度并增加其电阻率，器件电阻状态增加，这是降低功耗的要求。电场驱动的 TaO _x氧化/还原负责 ReRAM 切换。为了展示优化后的 TaO _x /HfO ₂设备的潜力，我们在修改后的美国国家标准与技术研究院数据库数据集上对全连接神经网络的训练进行了模拟，并针对全精度数字实施进行了基准测试。