Application of the TSVD unfolding method for reconstruction of primary X-ray spectra using semiconductor detectors

A thorough knowledge of the primary spectrum is very important to perform a quality control (QC) of X-ray tubes. In previous works, a methodology to assess primary spectrum using a Compton spectrometer has been analyzed. In summary, this methodology consists of the use of a Monte Carlo model (normally the MCNP code is applied) to reproduce the physical phenomena involving the interaction of photons and electrons with the Compton spectrometer and with a high purity Germanium detector. By means of this Monte Carlo model, a response matrix can be built, relating the Pulse Height Distribution (PHD) registered in the detector, with the primary X-ray spectrum. Subsequently, an unfolding method based on the application of a Truncated Singular Value Decomposition (TSVD) is applied to the response matrix in order to assess the primary spectrum. Germanium detectors present an optimal Full Width at Half Maximum (FWHM) value for energies up to 150 keV covering mammographic and diagnostic energy range. However, the need of a liquid nitrogen cooling system introduces some troubles in the data acquisition process. In this work, two other detection systems are proposed, a Silicon and a Cadmium-Telluride detector (CdTe), which do not require liquid nitrogen cooling system, and consequently the acquisition process is simplified. The weak point of this kind of detectors is the loss of energy resolution. Despite their low resolution the Silicon and CdTe detector have been considered due to both their low cost and easy handling and portability. The main goal of this paper is to determine whether the TSVD unfolding method is adequate to provide an acceptable reproduction of characteristics lines despite the low resolution of the detectors considered. The MCNP5 code, based on the Monte Carlo method has been used to simulate the actual physic processes for spectra acquisition analysis with Germanium, Silicon and CdTe detectors. (author)