Tritium (T) behavior in a breeder blanket is a key design issue because of its impact on safety and fuel-cycle best performance. Considering the difficulty in handling T, the validation of an alternative method is an issue. In an attempt to simulate T residence time (or T diffusion) in a DEMO-like ionizing environment, gas sorption-desorption experiments of deuterium (D) during ionizing radiation in breeder ceramics were carried out to elucidate the involved D2 or T trapping centers. A deconvolution method is here proposed for the identification of the release processes occurring at the different temperatures, therefore providing the basic information for identifying the nature of they-associated defects acting as D or T-trapping centers. The deuterium release in the EU-BB solid candidate material is studied in Li4 Si04 ceramics with variable LizTi03 content (pebble samples fabricated by KIT). The samples, previously dehydrated, are exposed to a deuterium atmosphere inside a pressurized steel capsule (I bar) during 25 days and immersed in the 6°Co Nayade facility located at CIEMAT. Deuterium desorption curves are then registered by ... mehr means of Thermal Induced Desorption (TID) technique. A low temperature process of low activation energy occurs below 300°C suggesting that surface defects are acting as trapping centers for D2. D2 is also trapped in deeper centers being released at temperatures above 600°C. Selected optical measurements (cathodoluminescence, thermoluminescence) were also carried out in irradiated materials, the resultant curves contributing to the explanation of the radiation-induced surface defects involved in D trapping. A deconvolution method for a better interpretation of the obtained enveloping curves was implemented in Matlab© and tested in this set of measurements. It is based on a Levenberg Marquardt iterative fitting routine that minimizes the difference between an experimental TID spectrum and a model spectmm given by the sum of Gaussian bands. The methodology used to solve non-linear least squares problems, allows a complete deconvolution of the overlapping peaks with good precision.