In future fusion reactors the long-term tritium retention will be a critical issue. In ITER, which will be the first device to demonstrate a burning plasma, the in-vessel tritium inventory limit will be 350 g. The primary retention mechanism of tritium is co-deposition with eroded first wall material. The measured tritium build-up rates at JET and TFTR tokamaks, both of which were operating with carbon walls at the time of their tritium campaigns, are too high for ITER. For these reasons there is today a growing interest in carbon-free fusion machines:ASDEX Upgrade in Garching, Germany, has replaced, step-by-step, all carbon fibre composite structures by tungsten-coated ones.The tritium inventory from the fuel tritium is typically found on top of the material surfaces, where it is deposited as amorphous hydrogen-rich carbon layers. This is because the fuel tritons are thermal, i.e., have low energy (E 100 eV) and, thus, a penetration depth of at most a few nanometers. However, tritium is also formed in deuterium-deuterium (D-D) reactions. The quantity of tritium formed inside the plasma will be insignificant compared to the amount of fuel tritium: the D-D fusion reaction rate is about 1/200 of the deuterium-tritium reaction rate. But due to its very high energy, contribution of the fusion-born tritium to the long-term tritium inventory and material damages can still be significant. This difference becomes evident by comparing the mean range and backscattered fraction of tritons in carbon, tungsten and beryllium, shown in tables 1 and 2. Energy 100 eV 1 MeV tungsten (2.8 ± 1.5) nm (4.3 ± 0.1) µm carbon (2.8 ± 1.6) nm (7.7 ± 0.1) µm beryllium (3.3 ± 1.6) nm (10.2 ± 0.1) µm 2) % (0.11 ± 0.01) % carbon (9.9 ± 0.1) % (0.000 ± 0.001) % beryllium (6.2 ± 0.1) % (0.000 ± 0.001) %