In general, frequency of characteristic x-ray emanating from any certain element rises parabolically with Atomic Number, as per Moseley's law on X-ray. Lighter elements emit lower frequency higher wavelength X-rays, so WDS analyzer crystal must have lattice parameter large enough to diffract that large wavelength.
As i come across this reference (Yang Lang, Materials Characterization, 2e, Wiley-VCH) Page 201, Some large crystal that can detect characteristic x-ray from lighter elements and also supposed to acts just fine to detect x-ray from heavier elements fail to do so, far example in the attached image, PE (above K), TAP (above Na) and LSM (above F) . Why does that happen?
Dear Sumit Bhowmick ,
from the Bragg equation (either written in x-ray wavelength lambda or in photon energy E)*) itself there should be no limit to have small wavelengths lambda (or high photon energies E) even for large 2d interplanar spacings of the analyzing crystals. One has only to go to quite small thetas in order to address high atomic elements.
But when doing this, one has to use a very small diameter primary x-ray beam in order to cover the sample, without illuminating the sample beyond its rims. However even in the case when the beam forming slit close to the sample is nearly zero, one will have a finite angular divergence of the net primary beam, which is not close to zero.
And this divergence in combination with the divergence of the slit system at the detector side is responsible for some kind of an uncertainty delta_theta of the x-ray optical system.
This delta_theta affects the wavelength (or the energy) resolution of the system:
When simply differentiating the Bragg equation one will have:
delta_lambda/lambda = cot(theta)*delta_theta
or
delta_E/E = -cot(theta)*delta_theta.
We see here, that even in the case when delta_theta is very small, at small theta the relative spectral uncertainty delta_lambda/lamda (or delta_E/E) will increase due to the cot(theta) function when further decreasing the theta.
In other words; the spectral resolution is going worse when you have too small theta, even when you nearly close all slits at the cost of a very lot of photon flux. Thus for the large 2d spacings there is an upper limit for reliably detecting high atomic order elements due to a) worse spectral resolution combined with b) low photon flux.
To my opinion the limits are not so strong as they are given for the case of K(19) and Na(11) in the table above, but they might be close to them.
*) lambda=2d*sin(theta) or E/(h*c) = 2d*sin(theta)
Final remark: the Fe(9) in the TAP line in your list above should read F(9); fluorine and not iron.
These suggested limits for what elements you can analyse with which analyser crystal look to me rather like a mixture of what can be done, confused with what should be done. Take potassium as an example. OK, you can probably measure it using LiF (200) but the peak would be rather broad (approx. 137 °). Pentaerythritol would be better at about 51 °. You *can* use pentaerythritol for much light elements than potassium, but you'll get better intensity with LiF (200). There are several other possible analyser crystals with d-spacings between those of LiF(200) and pentaerythritol (we've got graphite and germanium - Ge gives narrow peaks, zero intensity for 2nd-order reflections, but less intensity than graphite).
Try out a few samples with different analyser crystals and see what happens - look for interferences, peak width, peak intensity. We've got a circular device for showing what lines come where - produced many years ago by UKAEA - impossible to describe succinctly! - but invaluable for picking the best line.
- Badges
- Science method