## There are three main scattering mechanisms operating in these structures

Single scattering mechanism fitting is done first to extract the reasonable range of the parameters, the background impurity level NB, the activated dopants level ND, the alloy disorder potential Vad, the interface correlation length Λ and the average height of the surface Δ. Vad will be the same for each wafer since the alloy composition is identical among them. We assume that the dopant activation rate is the same across this AMG-47A set of wafers. According to [22], we assume that at the doping level in this paper, Si is fully activated with an activation rate of 1. Scattering from MD is not the limiting scattering mechanism since the mobility limited by MD is one magnitude larger (see Fig. 5). Therefore to simplify the fitting procedure the data using only BG and AD was carried out according to [1]. The extracted parameters are summarised in Table 2. By assuming the scattering from interface roughness is not significant, background impurity levels extracted from the fitting are very sensitive to changes in growth temperature and arsenic over pressure. One example fit is shown in Fig. 5 (right). As a comparison, the fitting that considers all four scattering mechanisms is presented in the same diagram. R-squared value has been calculated from 1.7×1011 cm−2 to 3.8×1011 cm−2 for these two fittings, the value from the four-mechanism fitting is 0.9573 and that from the two-mechanism fitting is 0.8386, therefore the four-mechanism fitting is more representative of the experimental data. Consequently, even though scattering from IR is not dominant, thigmotropism has a significant contribution to the total mobility.