The CMMP EDS detector is what is known as a light element EDS detector. The sensitivity of the EDS detector is entirely determined by the thickness and composition of the detector window. The older "thick window" systems are generally sensitive to Al and higher Z elements. The light element systems are sensitive to C and above.
Even with a light element system one should be careful with the quantitation of light elements. One has no idea what the effective window thickness is and thus it is difficult to model and subsequently compensate for window absorption. The EDS detector is cooled to LN2 temperatures which causes the EDS detector window to be the coolest part of the SEM chamber. Diffusion pump fluids and volatiles evolving from samples in the SEM vacuum are cryopumped on the EDS detector window. This can result in a severe reduction in light element signal.
This EDS spectrum of quartz at 30 kV shows no sign of the O Ka peak. Quantitation without standards yields 99% silicon by weight. One could attempt to use standards to quantitate light elements such as C, N and O with a heavily contaminated EDS window like this-- but this leads to the problem of very high statistical uncertainty in both the standard and specimen data. Since the k-ratios used in quantitation are ratios of specimen and standard data, quantitative uncertainties are enormous.
This EDS spectrum of quartz is taken immediately after cleaning the light element window. The EDS detector was completely warmed, the electron trap removed, and the window cleaned by flowing a fluorohydrocarbon solvent across the window using a dropper. Standardless quant yielded 66% Si and 34% O by weight. This still underestimates the oxygen content which should be ~ 53% weight, but the O Ka peak is sufficiently strong to perform meaningful analysis. The use of standards would account for the remaining systematic error at low energies due to attenuation in the EDS detector window.