Through weight percentage you can compare initial weight values of added materials and final changes occurred in percentage variations in weight of respective elements in your prepared compound. And similarly, you can calculate how much changes occurred in your initial stoichiometrically added materials and final obtained compound using atomic percentage.
As the above attached image consists of only Al, O and Y without the mention of chemical formula, I am just taking Al5O12Y3 as an example.
From the above formula it can be said that total atom count is 5+12+3 = 20 and elemental presence is Al =25%, O = 60% and Y = 15% for 100% in the above stoichiometry . This is in terms of atomic percentage.
Thankyou.Your explanation was sufficient to clear my doubts.Is this necessary that results from EDS be accurate 100%.Suppose I want to publish above results in a research paper but I have taken EDS over an area as nanoparticles were not distinctly visible .As a result, some half nanoparticles must be included inside selected area and hence the results of atomic and weight %ages are different from what I expected, howerever I have already confirmed phase formed using XRD . In one sample it is near about ,but in other it shows too much difference in atomic %age. Can I still use them or will it cause me problem? Or can I use to them to confirm just presence of elements instead of their atomic %age ratios?
It is difficult to quantify O with EDS. Requires utilization of special standards and good command of the method. I am afraid you performed standardless analysis, which is mostly useless for quantifying oxides.
Usually we perform EDS analysis on multiple regions of the sample to confirm the presence and stoichiometry of the sample and later we average it. Usually, in a homogeneous sample the atomic % should not vary much from one region to another region.
I agree with @Vladimir Dusevich that its difficult to quantity O accurately with EDS. X-ray photoelectron spectroscopy (XPS) would be a grate tool for that I believe.