Improvement of TimePix energy resolution correcting threshold variations

The Timepix detector is an important tool in spectroscopic imaging since it allows the determination of the energy of absorbed photons using its time over threshold capability. In this paper, however, we look at the energy resolution using the single photon counting mode combined with threshold scans. The pixel circuit of the TimePix detector involves a discriminator logic, which compares the collected amount of charge in each pixel with a threshold level (THL). The counter is incremented only if the induced signal surpasses this threshold. The THL is typically calibrated w.r.t. energy in the proximity of the noise edge, however, due to gain mismatch of the individual pixels, the threshold levels of pixels deviate if a DAC value far from the calibration point is set. This energy threshold mismatch far from the calibration point was corrected in this paper based on spectrum matching between individual pixels. The derived method is compared to straightforward calibration using multiple monochromatic peaks. To this end a threshold scan was performed using a TimePix Si-1mm detector (utilizing counting priciples) in presence of a Am-241 gamma source. The peak at energy 59.5 keV was fitted for each pixel to determine the actual THL values for all pixels at this energy. Then the full spectrums for all pixels were matched, using one fixed point. The matching process gave us two matrices of correction factors, the first matrix corrects the THL gain and the second its offset. In a second evaluation we measured and fitted the x-ray fluorescence of zirconium and copper during threshold scan in addition to the Am-241 peak. The gain and offset of each pixel were then determined by linear fit of the peak positions. The two calibration methods were evaluated by measuring the full width of half maximum FWHM of two other fluorescence peaks (23.17 keVand 26.2 keV peaks of cadmium). The evaluation shows that linear fit of multiple peaks results in a comparable FWHM (reduction of 21%) over spectrum matching the (reduction of 20%), while, opposite to spectrum matching, necessitating multiple measurement setps and THL scans.