Strain-Engineered Adaptive 2D Photodetectors: A New Approach to Miniaturized Reconstructive Spectrometry

We present adaptive, strain-engineered photodetectors based on 2D semiconductors. Our devices present a footprint of only about 40 × 20 μm², and they are fabricated on transparent polypropylene substrates and integrate lithographically designed microheaters that induce controllable biaxial strain via thermal actuation, thus leading to a red shift in their spectral response. This strain-induced tunability is combined with a machine learning-based spectral reconstruction approach, where a neural network is trained on an extensive data set of synthetic spectra and corresponding photocurrents generated from measured device responsivities under varying strain conditions. Although the system demonstrates spectral reconstruction over a relatively narrow range, it provides a compelling proof-of-concept for a miniaturized adaptive spectrometer.