When I measure F content in HA, the Line of F overlaps with that of P. Because I use Rh source, I found line Fka=90.68 degree, RhLa-4=91.15, RhLa2-4=91.36, Pksat-3=90.90, Pka-3=91.69.
How can I fit the peak profile with all these five lines? For simplification, I chose only three lines for fitting, the result is good as shown. But I wonder if this is accurate enough? The intensity of RhLa-4 is small in this process. Is this result reasonable?
First of all this broad peak doesn't seem to be an XRF peak, but a continuum. The X-axis should be in Å, (WDXRF) or keV (EDXRF), why is it in deg? Is it a diffraction peak (XRD)? From the shape of this hump (rather than peak) it's completely meaningless to analyze leave alone deconvolution of different elemental signatures. First of all please make it clear whether it's XRF or XRD data.
The simplification seems to work and results seem to be reasonable, but it should be validated with known materials with lower and higher P contents.
A general remark: keep in mind that XRF based fluorine analysis is a surface analysis. The relevant depth is approx. 1 µm.
We use Fka with a collimator width 0.46 and a XS 55 crystal at 2Th = 37.947 . Variation of parameters may help to improve resolution.
I completely agree with Dhrubajyoti Gupta... I am not getting how you have fitted that continuum into three peaks,, That looks like a single peak...
Ludwig Gruber: Thanks for your answer. The HA samples were synthesized by Sol-gel mehtod. We changed the F addition in this procedure. So the basic compositions of mateials were well known in theoretically. We just want to verify if the F content is same as our expectation. The XRF samples were impacted from HA blended powder, So the surface effect can be ommited in such case. My question is how to ascertain the intensity of RhLa-4 is proper or not.
Energy of Ka line of F is 0.68 keV, La1,2 of Rh is 2.70 keV, Ka of P is 2.01 keV (satellites are slightly more energetic, but even Ka hypersatellite line of P is less energetic than La of Rh). I can't understand why you can put Ka of F and Ka of P in this same energy range.
Please keep in mind that the intensity of the scattered Rh peaks will vary as a function of your samples bulk mass attenuation coefficient (as the composition and hence matrix edfects change).
I like the comment about F analyses being only a surface measurement.
Excuse me my ignorance: As far as I know fluorescence (XRF) radiations are known to be isotropic...I wonder how they can be resolved into degrees (angular signature and that too for satellite lines)! X-ray scattered radiations are known to be anisotropic.
I would like to request Dr. Qun-Li Rao and Dr. Ludwig Gruber to enlighten us in this aspect. It will be very kind of you sirs if you could let us know the source of the literature i.e. Fka=90.68 degree, RhLa-4=91.15, RhLa2-4=91.36, Pksat-3=90.90, Pka-3=91.6
Dhrubajyoti Gupta:when WDXRF used, either degree or Å can be uesd. They can be exchanged each other with the wavelength of source.
Dr.Ludwig Gruber and Dr.Maciej Śliwiński,what should I notice when varying parameters?
Thank you, if I understand you right: You have the angle set by the goniometer for WD detection of the elements taken on the x-axis instead of Å.
But coming back to your experiment...is this the best way to resolve the elements, why Å is not used?...Is it the same either way? It seems the continuum from the source Rh-L X-rays is broad enough to eat up P and F peaks. There are no edges on the peak which gives a clear signal .How much is the Chi-square? The undulations on the peak (hump) suggests that there is high-noise in this region...so one has much less confidence while de-convoluting the peaks...however good the chi-square may be..there's a doubt.
If you have access to one you can try "Grazing-exit X-ray fluorescence" for detection of a light element like F .You can have better F signal/noise..and since F signals are only from the surface...it can be helpful.
Assuming You use a classical wavelength-dispersive XRF there is a number of options to vary: at least You can change the collimator width, choose several analyzer crystals and vary the excitation energy. I am not able to predict the results...
Filtering is not possible for F for obvious reasons. I don't know what instrument You use but it is also possible to optimize the parameters of the detector. Do You use a proportional counter?
What kind of spectrometer is it (who made it) and what software package did it come with? Just curious.
Is it possible for you to vary the excitation energy? Say to set the keV and nA such that the maximum of the continuum is below the P absorption edge energy?
The P La - F Ka overlap is well know in Electron microprobe world. Im under the impression that its commonly dealt with through pulse height analysis (detector settings).
Oops except that varying the excitation energy to below the P abs. Edge would still leave it above the P La edges... Please disregard that thought
Than your suggestions. The spectrometer is XRF 1800 made by Shimadzu Co. The data were proceed with the supplied software of of the machine. The detector is FPC(a kind of gas detector).
Dear Sir,
Sorry for this late reply. Though it is a commercial XRF, if you can change to a multilayer crystal (Johansson type) you may have better resolution. You will have to check with Shimadzu may be or other bent-crystal vendors. Also as Dr.Ludwig Gruber suggested you may if possible play with the soller collimators to further optimize.
Although I still feel that your spectral ROI (region of interest) is dominated by the X-ray tube Bremsstrahlung or continuum from the look of it. Hope you have already got some answer. Would be interested to know.
- Badges
- Science topic
- Similar topics
- Physical Sciences
- Physical Phenomena
- Luminescence
- Fluorescence