There are 2 aspects of microprobe analysis which demand the sample be flat, but they are both the result of a single inherent fact of detecting compositional information ... the fact both types of detectors are line-of-sight.
The first example is the compositional information acquired by detecting backscattered electrons. These trajectories of these high-energy electrons cannot be bent, and therefore cannot be attracted to a detector. The effects of sample flatness is evident in these 2 BSE images ...
Both of these backscattered electron images are natural samples of pure FeS (light) intergrown with pyrrhotite (dark). The image on the left was acquired after it was made flat but before polishing. The image on the right is the same sample, after it was polished . While BSE detectors are designed to minimize the effects from topography, they cannot be totally eliminated. The image on the left should be embarrassing to any micro-analyst!
Quantitative compositional analysis, which depends on x-ray measurement, is even more demanding of a sample polish. X-rays, after being generated by incident electrons within the sample, are absorbed before leaving the surface. The trajectory path to the detector is known as the takeoff angle, and if the sample is flat, the trajectory path within the sample is a constant. See the figure below for how surface defects can affect x-ray absorption.
As can be seen, only the first case (a) allows the absorption path to be a predictable length. This assumption allows us to acquire the emission, quantify the absorption, therefore the amount generated, and ultimately the weight fraction of the element.
This degree of sample preparation may seem extreme, but it really is not. It is normal practice to make rock samples flat for normal optical observation ... usually as a thin-section. It really is not too much more work to also polish the sample, and the microprobe's sample holder can readily accommodate typical thin-sections.
If your samples require thin-sectioning, or are not polished, you should contact Brian Loveridge, 864-8528 (tel) firstname.lastname@example.org
(email). The microprobe can accept a wide range of sample formats, but most geological work is performed on polished thin sections or solid mounts. Samples up to 4 x 7 x 2 cm can be accommodated easily; maximum dimensions are limited by the sample chamber size to 10 cm diameter, although surface topography is limited to only several mm. A highly polished surface is necessary for quantitative work, but semi-quantitative results can be achieved with unpolished surfaces.