The 3D Monte-Carlo code ERO, which calculates erosion processes, impurity transport and deposition, has been coupled to the Monte-Carlo code SDTrimSP to simulate material mixing processes in wall components more precisely. SDTrimSP calculates the transport of ions in solids by means of the binary collision approximation. It keeps track of the depth dependent material concentration caused by implantation of projectiles in the solid. Modelling with the coupled code ERO-SDTrimSP is compared with dedicated TEXTOR experiments, in which the formation of mixed surface layers has been studied. In these experiments, methane 13CH4 was injected through graphite and tungsten spherical limiters during plasma exposure and the local redeposition probability was measured post mortem by surface analysis. A significant difference in the carbon 13C deposition efficiency, i.e. the ratio of the locally deposited to the injected amount of 13C, between graphite and tungsten was found, 4% for graphite and 0.3% for tungsten. Modelling of these experiments with ERO-SDTrimSP reproduces the clear substrate dependence with about 2% deposition efficiency on graphite and less than 0.5% on tungsten in good agreement with the experiment. The reason for the substrate dependence is partly explained by the higher physical sputtering yield of a thin carbon film on top of a tungsten substrate compared with a graphite substrate. Surface roughness of the materials has been identified to be another important parameter for the interpretation of the results.