Programmable photovoltaic devices are promising for self-powered sensing, visual perception, and neuromorphic computing, but existing systems often require complex architectures or pixel-level wiring, limiting high-density integration. Recently, a research team led by Dr. Xujie Lü from the Center for High Pressure Science and Technology Advanced Research (HPSTAR), Prof. Fuqiang Huang from Shanghai Jiao Tong University, and Prof. Xing Zhou from Huazhong University of Science and Technology proposed a light-electric dual-field programming strategy to realize reconfigurable and non-volatile anomalous photovoltaic responses in the two-dimensional van der Waals semiconductor ZnIn2S4. Their study, “Programmable Anomalous Photovoltaics Enabled by Light-electric Dual-Field Control”, is published in the latest issue of the Journal of the American Chemical Society.
Programmable photovoltaic functionalities are emerging as a platform for integrating energy conversion with sensing, memory, and information processing. Existing strategies, relying on either heterostructures or gate-free architectures based on ferroelectricity and ion migration, suffer from complex device architectures inherent to single-field modulation, where electrical bias simultaneously accommodates signal addressing and state switching. This coupling imposes an intrinsic limitation on spatial selectivity, architectural simplicity, and operational efficiency, reflecting a fundamental restriction in the available control degrees of freedom.
The research team overcomes this limitation by establishing a light-electric dual-field programming strategy in which light acts as an independent, spatially resolved addressing degree of freedom, rather than merely an excitation source. Spatially localized illumination defines where programming occurs, while a global electric field switches the state, thereby decoupling addressing from switching within a simple two-terminal architecture. This enables selector-free operation without per-pixel wiring or complex circuit control, and inherently suppresses crosstalk in integrated arrays.
Using ZnIn2S4 as a model system, the research team realizes reconfigurable multilevel anomalous photovoltaic states with retention exceeding 100 days. They identify [ZnS4] tetrahedral distortion as the microscopic origin, and demonstrate that such distortion can be further tuned by pressure to enhance the photoresponse by about 20-fold while reducing the programming voltage to 0.5 V. Leveraging this architecture, they demonstrate self-powered visual information processing with 87.9% accuracy under noisy conditions.
Caption: Schematic illustration of light-electric dual-field programming in the ZnIn2S4 device for generating spontaneous photovoltaic responses and enabling visual information processing.
可编程光伏器件在自供能传感、视觉感知和神经形态计算等领域具有重要应用前景,现有体系通常需要复杂器件结构或逐像素独立布线,严重制约了器件集成度和运行效率。为了解决这一问题,北京高压科学研究中心(HPSTAR)吕旭杰团队联合上海交通大学黄富强团队与华中科技大学翟天佑、周兴团队等,提出了一种光-电双场协同调控策略,在二维范德华半导体ZnIn2S4中实现了可重构、非易失的反常光伏效应,并展示了其在多态存储、噪声抑制和视觉信息处理中的应用潜力。研究团队通过高压调控揭示了[ZnS₄]多面体结构畸变是该反常光伏效应的微观起源,并借此将器件性能提升约20倍,同时将编程电压降低至0.5 V。北京高压科学研究中心的付同欢博士、卜克军博士和华中科技大学的博士生苟根畅为该论文共同第一作者。该工作还得到了复旦大学、上海科技大学以及中山大学等单位研究人员的帮助与支持。该研究获得国家自然科学基金、国家重点研发计划、上海市科委等项目资助,并得到上海同步辐射光源BL15U1线站支持。相关成果以“Programmable Anomalous Photovoltaics Enabled by Light-electric Dual-Field Control”为题发表于《美国化学会志》(Journal of the American Chemical Society)上。