A new experimental approach to the study of collisions of hydroxyl radicals with liquid surfaces is described, incorporating a molecular–beam source of OH (or, in practice, OD, for technical reasons) radicals. This allowed the collision-energy dependence of the scattering to be examined. The incident and scattered OD molecules were detected by laser-induced fluorescence. The representative branched, long-chain alkane, squalane (2,6,10,15,19,23-hexamethyltetracosane), and its partially unsaturated analogue, squalene (2,6,10,15,19,23-hexamethyl-2,6,10,14,18,22-tetracosahexaene), were compared with perfluoropolyether as an inert reference liquid. Dynamical aspects of the scattering necessary to quantify the OD survival probability, and hence its complement, the reactive sticking coefficient, were determined. Results were obtained at average laboratory-frame kinetic energies of 7.2 and 29.5 kJ mol-1; they are compared with previous independent measurements using a photolytic source of OH with an average kinetic energy of 54 kJ mol-1. At lower collision energies, the survival probability is significantly lower on squalene than on squalane, but increases significantly with collision energy. This is consistent with a negatively-activated contribution to loss of hydroxyl through addition to double-bond sites at the squalene surface. In contrast, survival on squalane surface is found to be approximately independent of collision energy across the range examined. This is surprising, because it does not reflect the positively activated behavior typical of gas-phase OH + alkane reactions. We suggest that this may be explained by a higher probability of trapping dynamics at lower collision energies, enhancing the probability of reaction following migration to more reactive sites. The results have implications for the modelling of OH uptake on atmospheric aerosol surfaces as a function of chemical composition and temperature.