Modelling delayed luminescence decay kinetics in plant leaves:Balancing accuracy and robustness under experimental variability.

Charge recombination within the photosynthetic apparatus produces ultra-weak photon emission known as delayed luminescence (DL). Its complex, non-exponential behaviors make it difficult to accurately model its decay kinetics, despite its great potential as a non-invasive diagnostic tool. We benchmark a low-cost, economical manual DL detection system based on a single-photon avalanche diode (SPAD). This study focuses on the systematic comparison between four commonly used models: mono-exponential, bi-exponential, hyperbolic, and a combined power-law with exponential function. Using the Akaike Information Criterion (AIC) we assess statistical robustness in comparing the four mathematical decay models via curve fitting. After analysing the curve fits and decay parameters, the simpler mono-exponential model offers the required robustness and reproducibility against manual timing errors. The reproducibility makes it highly suitable for basic experimental and educational setups. However, the hyperbolic model consistently yields the most accurate global fit for the decay curve as seen in previous studies as well.