Nonlinear self-calibrated spectrometer with single GeSe-InSe heterojunction device.

Area of Science:

• Optoelectronics
• Spectroscopy
• Artificial Intelligence

Background:

• Computational spectrometry merges photodetection with algorithms for spectroscopic analysis.
• Compact photodetectors offer advantages over traditional spectrometers but often exhibit nonlinear responses.
• Nonlinearities complicate the extraction of accurate optical spectra.

Purpose of the Study:

• To train an artificial neural network (ANN) for recovering the nonlinear spectral photoresponse of a single GeSe-InSe p-n heterojunction device.
• To demonstrate a solution for metamerism in optical imaging using the developed computational spectrometry system.

Main Methods:

• Utilized a compact GeSe-InSe p-n heterojunction photodetector with a spectral range of 400-1100 nm.
• Trained an ANN to reconstruct the full nonlinear spectral photoresponse from the device's output.
• Quantified reconstruction accuracy with a mean error of 2 × 10^-4 for the power spectrum at 0.35 nm.

Main Results:

• Successfully recovered the nonlinear spectral photoresponse of the GeSe-InSe device using the trained ANN.
• Achieved high accuracy in spectral reconstruction, demonstrating the efficacy of the computational approach.
• Showcased the device's capability to address metamerism, a significant challenge in optical imaging.

Conclusions:

• Artificial neural networks can effectively interpret nonlinear photodetector responses for computational spectrometry.
• This compact, ANN-enhanced system offers a promising solution for accurate spectral measurements and color imaging applications.
• The developed technology advances the field of compact, high-performance spectroscopic instrumentation.