Please suggest methods to calculate bar value from EDS analysis.
Your question becomes more specific when you let us know exactly what you are analyzing...answers can be more specific then.
EDS usually gives relative quantification ....from atomic concentrations....so the error should also be relative. However, major statistical error contribution is from the peak-fitting. Systematic errors depends on what you analyze.
Thank you sir
I am calculating the composition variation along the length of the crystal grown by
Bridgman method. the composition of Ag, Ga, In and Se are varies from bottom to top, but i don't know how to calculate the error values.
Bruker SEM instrument gives the error value, but FEI Quanta FEG 200 instrument does not give the error values.
though i search google, i got one reference "Energy Dispersive Spectroscopy on the SEM: A Primer" by Bob Hafner from university of minnesota, in that material he told that Accuracy (the closeness of agreement between anobserved value and an accepted reference value): (95% analysis).
±1% for polished bulk target, pure standards on site
±2% for polished bulk target, standards collectedon another SEM and then corrected for the
geometry and settings of the present microscope (NSS terms this “without standards”).
±5% for particles and rough surfaces “without standards”
our sample has polished bulk sample, shall i take this common material for all SEM analysis sir?
Please give me your valuable suggestion sir.
I am not into the type of research you are doing...but still from the SEM/EDX point of view I will make a few comments...hope it helps. You can use your own common sense and logic:
(1) Your crystal does not seem to be homogeneous, how are you going to correlate with the performance on standard perfectly polished homogeneous crystals?
(2) It's good that you are getting point to point info so I think its better not to report a bulk data (averaged). It's better to compare with a standard (if available) representing your sample closely or at least technically.
(3) What acc. voltage are you using?...what is the probe size or spot size of electron-beam on sample when taking EDX data?...this should directly relate to your sample homogeneity.
(4) Have you seen the EDX spectra?...and are you measuring the L-lines of the mentioned elements?...Are they distinctly identifiable?
(5) If you can extract the EDX spectra in ASCII form..then its better to fit the spectra and analyze separately (using some software..like CASINO for EPMA or EGS4, etc) so that you can use the peak fitting error to get the contribution for your statistical part.
(6) You are the best person to be able to gauge your systematic errors..and hence combined errors (Statistical + Systematic) ...you can refer standard text (e.g. G.F. Knoll).
(7) Generally as analytical chemists we would rather go for (a) say five repeat measurements of the same sample to get the standard deviation (Statistical SD) and (b) 5 repeat measurements of 5 replicate samples..i.e if you are able to grow five such crystals (1 spectra for each)...to get the systematic error or SD..or you can opt for only (b). I guess this is more practical and realistic than a single measurement....In this you don't have to use special softwares also..the instrument outputs should suffice.
Sorry in point # 7 the repeat measurement of 1 sample does not only account for the statistical but instrumental error also...but its best to go for 5 replicate crystals. Your no of replicates may vary according to convenience...and the desired confidence level you want to get.
- Badges
- Science topic
- Similar topics
- Engineering
- Mechanical Engineering