Original title: Machine learning-assisted design and explainable optimization of CdSnP2-based integrated solar-photodetector devices
Md. Alamin Hossain Pappu, Kazi Abrar Shafin, Mainul Hossain, Jaker Hossain · September 2026
Devices that act as both solar cells to make electricity and photodetectors to sense light can be slow and hard to design by hand. Researchers tackled this by combining a computer simulation tool called SCAPS-1D with machine learning models, which learn patterns from simulated data. They tested different layer thicknesses, doping concentrations, which add specific impurities to guide electric charges, and defect amounts. The models found that stacking n-CdS, p-CdSnP2, and a p+-CuGaSe2 back surface field layer works best out of thirteen designs. Adding a 200 nm CuGaSe2 layer boosts energy efficiency from 20.67% to 32.69%. It also improves light-sensing responsivity from 0.53 AW-1 to 0.72 AW-1 and raises detectivity from 2.51x1014 Jones to 1.78x1016 Jones.
Imagine organizing a bookshelf where each book must fit without wasting space. The hard part is deciding which books to place where to maximize access and stability. In this paper, researchers are arranging layers of materials in a solar cell, optimizing their thickness and properties to maximize energy conversion and light detection, using machine learning to find the optimal arrangement quickly.
This research could lead to better solar panels and light sensors, reducing costs for clean energy and improving everyday electronics like cameras and solar-powered devices.
The abstract doesn't specify if results are validated experimentally, as it relies on simulations, and the study focuses on specific materials and configurations.
Ask anything about this paper. Answers come from its abstract, and say so when the abstract doesn't cover your question.
Machine learning-assisted design and explainable optimization of CdSnP2-based integrated solar-photodetector devices — Md. Alamin Hossain Pappu, Kazi Abrar Shafin, Mainul Hossain et al.
Abstract licence: CC0
Explanation generated by Synthesis from the abstract — it is not the original paper. This is a preprint, so it hasn't necessarily been peer reviewed.