Horowitz and collaborators have reported evidence on the structure of TLD-100 peak 5. A satellite peak, called 5a, has been singled out as arising from localised electron-hole recombination in a trap/luminescent centre, its emission mechanism would be geminate recombination and, therefore, its population would depend on incident radiation ionisation density. We report a phenomenological study of peak 4, 5a and 5 strengths for glow curves previously measured at UNAM for gammas, electrons and low-energy ions. The deconvolution procedure has followed strict rules to assure that the glow curve, where the presence of peak 5a is not visually noticeable, is decomposed in a consistent fashion, maintaining fixed widths and relative temperature difference between all the peaks. We find no improvement in the quality of the fit after inclusion of peak 5a. The relative contribution of peak 5a with respect to peak 5 does not seem to correlate with the radiation linear energy transfer.