Experiment and simulation in silicon PN-junction photodetectors: Insights into electrical and optical transport

Silicon PN junctions remain central to optoelectronic technologies due to their maturity and CMOS compatibility. We report the fabrication and comprehensive optoelectronic characterization of a silicon PN-junction photodiode demonstrating stable operation over a wide temperature range. The device exhibits excellent diode behavior, with a rectification ratio exceeding four orders of magnitude, an ideality factor close to unity above 0.3 V, and a series resistance below 100 Ω. Under white-light illumination, the photodiode shows a linear photocurrent response over broad optical power and temperature ranges, achieving an average responsivity of 0.3 A·W -1 . We implement a machine learning framework based on an Artificial Neural Network to perform global parameter estimation, demonstrating its effectiveness in generalizing across diverse experimental datasets. Moreover, we propose a comprehensive analytical model, validated by Atlas–Silvaco simulations, that successfully captures charge transport and photogeneration mechanisms. This integrated approach, combining experimental measurements, machine learning, numerical simulations and analytical modelling, provides a robust performance benchmark and deeper insights for optimizing silicon-based optoelectronic devices. • A silicon PN-junction photodiode was fabricated and thoroughly characterized, exhibiting excellent rectification (>10 4 ), low series resistance (<100 Ω), and near-ideal diode behavior. • Multiple charge-transport mechanisms were identified and modeled, with the full I–V characteristics accurately described using a double-diode phenomenological model. • The device shows highly linear photocurrent response over wide ranges of optical power and temperature, with a stable wavelength-averaged responsivity of ∼0.3 A W -1 under white-light illumination. • Temperature-dependent measurements reveal thermally activated leakage currents, with an activation energy close to half the silicon bandgap, limiting detectivity at elevated temperatures. • Excellent agreement between experiment, analytical theory, and Atlas–Silvaco simulations validates the proposed modeling framework and provides reliable insight into silicon PN photodiode physics.