This study evaluates a model combining Monte Carlo radiation transport and spatially dependent charge collection to characterize the BPW34 photodiode for diagnostic X-ray dosimetry. Geant4 simulations were coupled to a depletion-and-diffusion model with an effective diffusion length constrained by published experimental sensitivities. Three collection scenarios—complete collection, depletion-only collection, and depletion plus diffusion—were compared for N40, N80, and N120 beams. Reverse-bias effects were assessed between 0 and 25 V. A calibration relationship linking the charge ratio without and with a 0.1 mm Cu filter to charge-to-air-kerma sensitivity was fitted at five tube potentials between 40 and 120 kVp and evaluated at four intermediate potentials excluded from fitting. Including diffusion reduced the maximum absolute sensitivity deviation from 52.8% under complete collection to 11.3%, with deviations of +11.3%, −4.71%, and −8.24% for N40, N80, and N120, respectively. Relative Monte Carlo statistical uncertainties in charge and air kerma ranged from 0.1% to 0.4%. At 10 V, sensitivity reached approximately 97% of its value at 25 V. The calibration fit achieved R2 = 0.988, and air-kerma estimates had a mean absolute percentage error of 0.52% relative to simulated reference values. By distinguishing generated from collected charge and connecting spectral response to a measurable charge ratio, the model supports operating-voltage selection and energy-dependent calibration of photodiode dosimeters. Experimental validation of a complete detector assembly remains necessary.
