Recent proposals by Wood and Wilson [C. E. C. Wood and R. A. Wilson, J. Vac. Sci. Technol. B 11, 1036 (1993)], to explain the formation of impurity spikes at substrate/epitaxial layer interfaces in GaAs prepared by molecular‐beam epitaxy (MBE), have been experimentally investigated. Their suggestion that the spikes form due to suboxide transport via reactions that involve the As2O3 released from the substrate during oxide desorption and hot Knudsen cells, is not supported by the experimental data. The same authors have also speculated that there may be significant flux leakage from nominally closed cells. For this to occur, reflection and scattering of flux by inadequately cooled cryoshroud baffle surfaces are necessary. Secondary ion mass spectrometry analyses of interfaces, at which the growth of GaAs and AlAs was interrupted for times up to 30 min, confirm that this takes place. However, flux leakage is only found to be significant for the high vapor pressure group III elements. For these elements, incorporation levels in the range 0.02%–0.1% are found under normal deposition conditions. These results suggest that careful attention should be given to increasing the internal MBE system baffling in order to eliminate cross contamination problems.