The use of an internal O-18 implant to quantify the surface oxygen concentration, presented in Franzreb et al. [K. Franzreb, J. Lorincik, P. Williams, Surf. Sci. 573 (2) (2004) 291], is critically evaluated, in particular when applied for bombardment conditions leading to high depth resolution measurements. It is shown that this methodology gives no direct access to the surface composition, but only to the O/Si ratio in the total emission flux, phi(O16)/phi(Si), which reflects the total amount of oxygen that passes through the surface. When applied under conditions relevant for high depth resolution studies (oxygen bombardment at low energies and small angles), extremely large phi(O16/)phi(Si) -ratios are observed (> 50). These large values are interpreted in terms of the oxygen that leaves the surface by sputtering as well as by alternative processes such as by desorption. The latter is included in the total flux ratio phi(O16)/phi(Si), but has little interaction with the sample surface and the departing Si-atoms. Therefore the use of these phi(O16)/phi(Si) ratios in ionization mechanism studies becomes inappropriate unless the desorbed oxygen fraction phi(desorb)(O) can be calculated quantitatively or is negligible. Based on the angular dependence of the useful ion yield of oxygen, we make a rough estimate of the desorbed oxygen fraction phi(desorb)(O) and show that this becomes a dominant fraction (up to 70%) in the emitted flux at low bombardment energies. (c) 2006 Elsevier B.V. All rights reserved.