View Single Post
Old 8th November 2007, 07:10 PM   #95
The Almond
Graduate Poster
The Almond's Avatar
Join Date: Nov 2006
Posts: 1,015
Originally Posted by bofors View Post
The data itself is explictly quantitative, we are talking about numerical intensities of radiation at specific numerical wavelengths.
I would strongly dispute that statement. Intensity, measured as the integral of a region of interest, is no more quantitative (in the sense that it has no bearing on actual weight percent composition) than looking at your foot and trying to determine if it is "big."

Quantification of EDS spectra involves, in its simplest form, five steps:
1) Deconvolution of the region of interest to remove overlap and subtract the background (which often requires extensive modeling of the Bremstrahlung for lighter elements such as oxygen)
2) Integration of deconvolved peak
4) Determination of the k-ratio, which is the ratio of the peak integral of the experimental data to some known standard.
5) Correction of the k-ratios of all of the peaks present in a spectrum for the effects of atomic number, absorption and fluorescence. These corrections are collectively known as ZAF corrections, and they are frequently handled by software such as CITZAF, written by John T. Armstrong from the California Institute of Technology.

The point is that, while EDS data can be quantitative, it requires processing and comparison to known standards to be so.

Generally speaking, EDS data are best for qualitative determinations of the composition of the sample. It is possible to identify the elements present in a sample surface without knowing anything about the actual composition of the sample. This stands in stark contrast to a complimentary technique known as Wavelength Dispersive Spectroscopy. WDS is widely considered to be the better quantitative analysis technique. But due to limitations in Bragg's law, you really have to know something about the sample first in order to select the proper crystal to diffract the X-rays of interest.
However, interpreting analytical data can be more of a qualitative process. There are often mirror discrepancies and one needs to know what is important and what it not.
Again, it might be semantic, but I would contend that qualitative analysis aims to determine if an element is present or not. That being said, idiots, like Stephen Jones, misinterpret the data all the time.

Originally Posted by Mr. Skinny
Sorry, bofors, but that came across as a bit of word salad for me.

Does the process deliver results that say, for example: Pb = 11.2%, or does it just say: Pb, or perhaps Pb = High?
Referring to my comments above, Mr. Skinny, both are correct. It's just that the quantitative result takes a lot of work and isn't quite as accurate as wavelength dispersive spectroscopy. I would also add that making an analysis based entirely upon peak integral as to what is more or less is fraught with danger. Heavier elements, such as lead and bismuth, tend to absorb lighter elements, despite the fact that their overvoltage is not as great. That means that, while the lead peak might be smaller than the silicon peak, it does not mean there is more silicon in the sample than lead.

For now, erm...I'm just trying to learn if XEDS will give you results similar to an XRF.

Maybe I should just look it up myself.

Thanks anyhow, bofors.
X-ray fluorescence will yield a very similar result to electron probe microanalysis. In one instance, the rapid deceleration of an electron by a sample will cause the photoelectric absorption and emission of a characteristic X-ray. In the case of XRF, an incident beam of X-ray photons will undergo a process of photoelectric absorption and emission to produce a characteristic X-ray.

Fluoresced X-rays can be analyzed in two ways:
1) In an energy dispersive mode with a silicon lithium or silicon drift detector
2) In a wavelength dispersive mode with an appropriate crystal (TAP, LIF, LDEB1, LDEB2, etc.)

The principle advantage to XRF is that there is no bremstrahlung background, meaning that modeling and background correction are much simpler. This allows for much lower detection limits than are seen in traditional electron probe excited X-ray analysis.

The drawback to XRF is beam size and interaction volume. As a general rule, electrons and X-rays excited by electron photoelectric interaction, come from an area usually less than 10 microns in diameter, and less than 10 microns deep. XRF beams range anywhere from about 10 microns to 1 millimeter in diameter, with massive penetration into the sample. For that reason, XRF is traditionally viewed as a bulk analytical method.

If you're interested in both methods, I can suggest Scanning Electron Microscopy and X-ray Microanalysis by Goldstein, Newbury, Joy, et al. and X-ray Fluorescence by Ron Jenkins. Both books are extremely accessible, well referenced, and widely regarded in the microanalytical community.

This is an interesting side note: Today, November 8th, was the day that Roentgen discovered the X-ray.
"Perfection, even in stupidity, is difficult to achieve without a conscious effort."--pomeroo, JREF Forum Member
The Almond is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top