Computational spectrometers with small footprints can be integrated with other devices for use in applications such as chemical analysis, medical diagnosis and environmental monitoring. However, their spectral resolution is limited because conventional photoelectric detectors only measure an amplitude-dependent response to incident light. Here we show that a deformable two-dimensional homojunction can be used to create a microspectrometer with dual-signal spectral reconstruction. The semifloating molybdenum disulfide homojunction exhibits a giant electrostriction effect through which the kinetics of photo-generated carriers can be manipulated via an in-plane electric field generated by gate voltage. By leveraging the tunability of both amplitude and relaxation time of the photoelectric response, a dual-signal response can be used with a deep neural network algorithm to reconstruct an incident spectrum. Our dual-signal microspectrometer has a footprint of 20 × 25 µm², offers a resolution of 1.2 nm and has a spectral waveband number of 380, which is comparable to benchtop spectrometers.

占地面积小的计算光谱仪可以与其他设备集成，用于化学分析、医疗诊断和环境监测等应用。然而，它们的光谱分辨率受到限制，因为传统的光电探测器仅测量对入射光的振幅相关响应。在这里，我们展示了可变形二维同质结可用于创建具有双信号光谱重建的微型光谱仪。半浮置二硫化钼同质结表现出巨大的电致伸缩效应，通过该效应，可以通过栅极电压产生的面内电场来操纵光生载流子的动力学。通过利用光电响应的幅度和弛豫时间的可调性，双信号响应可以与深度神经网络算法一起使用来重建入射光谱。我们的双信号微型光谱仪的占地面积为 20 × 25 µm ² ，分辨率为 1.2 nm，光谱波段数为 380，与台式光谱仪相当。