I have a sample of aluminium which I'm interested in finding the rough composition of.
The sample is a part of a large number of samples which I sanded down and polished with a polishing table. I cleaned the samples with acetone, washed and rinsed in ultrapure water. Area spectrum SEM EDS analyses were then ran on all the samples.
Some impurities were left on the surface of the sample as expected. All other samples show oxygen, silicon and carbon impurities assumedly left from the sanding process (quartz), the carbon tape for the grounding and possibly carbon impurities in the SEM-chamber. The samples were left to dry overnight so assumedly the amounts of acetone and water on the surface are fairly small.
All the other samples show accuracies and composities that seem sensible to me. However, the aluminium sample that I have shows levels of oxygen much larger than I expect:
So, three measurements of four areal spectrums each. Carbon is left out. And a mean of 17.37-at% of oxygen?
I understand that aluminium has a passive oxide layer and that EDS is a relatively surface active technique. However this paper cites passive layer thicknesses in the order of nanometers, and I understand that the EDS has a penetration thickness in the order of micrometers. A comparable steel sample that I analysed has levels of oxygen that I expected (~~0.1%)
What's going on? The measurement device should be able to handle the relatively low Z aluminium according to the manufacturer and work I have seen previously. Is it simply not true in this case? Can oxides precipitate inside bulk aluminium during manufacture or aging?