An XPS surface analysis is not like slicing off a few nanometers of the sample surface and counting the atoms in the slice. The Beer-Lambert law
tells us that the signal intensity from a layer of atoms buried a distance z beneath the surface is less than the signal intensity one would obtain from that layer had it been at the surface by a factor
where λ is the photoelectron attenuation length and θ the photoemission angle.
" 95% of the information obtained by XPS comes from within three attenuation lengths of the surface "
Pictorially, this means that we obtain information from the sample as presented in the figure below, in which the vertical width of the orange "trombone" figure represents the preponderance of information as a function of depth. We can say that 95% of the information obtained by XPS comes from within three attenuation lengths of the surface, a distance sometimes referred to as the "sampling depth", but we must remember that the information is not obtained in a uniform manner within that depth. Because the photoelectron attenuation length is typically on the order of 2-5 nm, XPS is a very surface-sensitive technique.
We can also note at this point that at higher photoemission angles, the information is obtained from shallower depths than at lower photoemission angles (compare the 0° degree and 75° figures below.) By varying the photoemission angle, therefore, we vary the effective "sampling depth", because we oblige the emerging photoelectrons to travel through more material at the higher photoemission angles, so that only those photoelectrons originating from closer to the surface tend to escape with no energy loss. Whatever the photoemission angle, however, most of the information comes from the very surface of the sample, and so we must expect the fidelity of a reconstruction of the depth profile to degrade as a function of depth.