Impact of X-ray Spectra Modeling on Gain and Noise Determination in High-dynamic Range Detection Systems

The accuracy in the determination of gain and noise from X-ray spectra becomes critical when the data binning is larger with respect to noise, situation often encountered in high-dynamic range systems. In such conditions, the spectral components in the peak shape of the X-ray line to be fitted, if not correctly modelled, cause significant distortions in the fit results.We tested fitting models with increasing levels of accuracy, in order to study the impact on gain and noise determination from X-ray spectra. The main fitting options (e.g. low-energy shoulder related to charge collection inefficiency, bremsstrahlung, beam filtration, bin integration) have been tested on experimental spectra acquired by DEPFET pixels at low bin resolution. A dedicated analysis has been carried out to compare noise estimation by spectra fitting with and without bin integration on simulated data sets and on experimental data.