Université de Strasbourg, CNRS, Institut de Physique et Chimie des Matériaux de Strasbourg (IPCMS), UMR 7504, 23 rue du Loess, Strasbourg 67034, France.
CNRS, Centre de Nanosciences et de Nanotechnologies, Université Paris-Saclay, 91120 Palaiseau, France.
ACS Appl Mater Interfaces. 2023 Mar 29;15(12):15732-15744. doi: 10.1021/acsami.3c00092. Epub 2023 Mar 15.
Interface-driven effects in ferroelectric van der Waals (vdW) heterostructures provide fresh opportunities in the search for alternative device architectures toward overcoming the von Neumann bottleneck. However, their implementation is still in its infancy, mostly by electrical control. It is of utmost interest to develop strategies for additional optical and multistate control in the quest for novel neuromorphic architectures. Here, we demonstrate the electrical and optical control of the ferroelectric polarization states of ferroelectric field effect transistors (FeFET). The FeFETs, fully made of ReS/hBN/CuInPS vdW materials, achieve an on/off ratio exceeding 10, a hysteresis memory window up to 7 V wide, and multiple remanent states with a lifetime exceeding 10 s. Moreover, the ferroelectric polarization of the CuInPS (CIPS) layer can be controlled by photoexciting the vdW heterostructure. We perform wavelength-dependent studies, which allow for identifying two mechanisms at play in the optical control of the polarization: band-to-band photocarrier generation into the 2D semiconductor ReS and photovoltaic voltage into the 2D ferroelectric CIPS. Finally, heterosynaptic plasticity is demonstrated by operating our FeFET in three different synaptic modes: electrically stimulated, optically stimulated, and optically assisted synapse. Key synaptic functionalities are emulated including electrical long-term plasticity, optoelectrical plasticity, optical potentiation, and spike rate-dependent plasticity. The simulated artificial neural networks demonstrate an excellent accuracy level of 91% close to ideal-model synapses. These results provide a fresh background for future research on photoferroelectric vdW systems and put ferroelectric vdW heterostructures on the roadmap for the next neuromorphic computing architectures.
界面驱动效应对铁电范德华(vdW)异质结构中的铁电性能有着重要影响,为寻找替代器件结构以克服冯·诺依曼瓶颈提供了新的机会。然而,其应用仍处于起步阶段,主要通过电控制来实现。开发额外的光控制和多态控制策略,以用于新型神经形态架构的研究,这一点至关重要。在这里,我们演示了铁电范德瓦尔斯异质结构中,铁电晶体管(FeFET)的铁电极化状态的电和光控制。FeFET 完全由 ReS/hBN/CuInPS vdW 材料制成,实现了超过 10 的导通/关断比、高达 7 V 的滞后记忆窗口以及具有超过 10 s 寿命的多个剩余状态。此外,通过光激发 vdW 异质结构,可以控制 CuInPS(CIPS)层的铁电极化。我们进行了依赖波长的研究,这使得我们能够确定在极化的光控中起作用的两种机制:进入 2D 半导体 ReS 的带带光生载流子产生和进入 2D 铁电 CIPS 的光伏电压。最后,我们通过将 FeFET 操作在三种不同的突触模式下:电刺激、光刺激和光辅助突触,演示了异突触可塑性。模拟了包括电长期可塑性、光电可塑性、光增强和脉冲率相关可塑性在内的关键突触功能。模拟的人工神经网络显示出接近理想模型突触的 91%的出色准确性水平。这些结果为未来对光铁电 vdW 系统的研究提供了新的背景,并将铁电 vdW 异质结构置于下一代神经形态计算架构的道路上。