What can be commented for the accuracy of quantitative phase analysis (QPA) by XRD Rietveld refinement for very fine micro-structures of multi-phase steels. If the grain size is ~10um. Does it negatively effect the accuracy? Can the microstrcutral observations done by SEM/optical microscopy have to be in strict agreement with the quantities of phases determined by the Rietveld QPA?
"Fine" is quite relative. 10µm are big for XRD and not small. In so far you might run into problems regarding the big grain size.
If you are talking about multi-phase steel you need to say which phase you want determine. As long as you only discriminate between ferrite and austenite, both materials have the "same" scattering power. In such case the error is commonly assumed in the scale of 2%. Phases with higher mean atomic scattering factor f can be distinguished even with smaller fraction (e.g. W carbide). The opposite happens for phases with lower f, e.g. Cr carbides or oxides. If you think you can define the phase fraction of martensite, I don't assume that this has a high reliability.
Compared to other techniques, XRD will be commonly better (because of statistical reasons). In order to get comparable data from SEM or TEM you have to collect many images and take them all into account. On the other hand, caused by sample preparation you possibly do not see at the surface what is going on in the depth below the surface (which is for a homogenized material visible for XRD as well).
BTW: I personally do not like the term Rietveld analysis for QPA since this code has been developed for the determination of the crystal structure of ONE phase. Many years later the so-called full-pattern refinement has been derived and so it is often also called (even as software name: Fullprof). The term Rietveld should be used exclusively for structure analysis.
@ Dr. Gert Nolze, Thanks for a comprehensive and enlightening response.
The phases concerned in our case are, ferrite/martensite/austenite with ~5% of W/Mo/V carbides. The fraction of carbides determined by full-pattern scan seems to be in good agreement with the microscopy observations, we did not try to determine the individual fractions of ferrite and martensite (which i believe is not very easy). However when we determine the austenite fraction, it seems to contradict with the optical microscopy (with selective etching) the fraction seems higher in optical case by more than 5%. its not clear if this discrepancy is normal or not?
Secondly, some researchers seem to term the fractions determined by full pattern scan as 'weight %age', is it the right term or the volume %age is more accurate?
The problem regarding residual austenite determination is often texture. in order to get as many as possible information such measurements are typically performed using W radiation. Otherwise, the texture is not very determined if you only have two or three peaks from one phase. In order to test texture, you can move omega without changing the detector (positioned at the strongest austenite peak (if there is no overlay with other peaks).
Regarding the minor phases...keep in mind that you need a certain amount of crystals which are required for all 2theta angles. Imgagine: if you only have one crystal, it is very likely that you do not get any reflection. If a crystal is in "reflection condition" you will not get any other reflection (except of higher orders from this lattice plane), i.e. you need to have that many grains as reflections. In order to exclude the size factor of the crystal you assume that they have all the same size which is fine for powders after sieving a certain particle size fraction. With bulk material everything is much more complex and uncertain.
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Techniques
- Structural Equation Modeling