Accuracy of XRF scanners? 3

[RedVette]

Hey, guys.

Would any of you happen to know what the accuracy tolerance is with modern XRF alloy scanners? Does it vary by brand? I read a brochure from Niton outlying their advances in the past century--heightened sophistication allowing a wider range of data processing, but it doesn't talk about the accuracy range. Is the margin of error a constant percentage of the yield value, or is it a consistent +/- decimal of a percent?

Assuming that the surfaces are fully clean and prepped, what could be the expected error margin if a scan by a modern Niton yielded--for example--an iron content of 0.25% in an L605 alloy?

Thanks in advance.

 

[mrfailure]

Accuracy in part will vary by the element. Remember, XRF is only semiquantitative in nature - it works best to dentify alloys but is not such a good tool if you are trying to determine if an alloy is in compliance. Also remember that levels of detected elements may in fact be down in the noise region and are not a reflection of the true amount present if even really present.

 

[EdStainless]

Most of these devices have special double and quad precision modes.
The precision of and XRF device is a function of the number of counts for each element.
Lighter elements return fewer counts, as would lower concentrations. This makes small amounts of Al or Si very tricky.

Accuracy is a function of the reference standards. If you are not using type standards that are very well characterized and very similar to your unknown then you cannot expect accuracy. And your type standards should have the same surface finish as your unknown.

So in L605 the Fe usually runs about 1.5%, getting 0.25% should not be a problem. But you need to be aware of possible interference from other elements and the methods for using type standards.

= = = = = = = = = = = = = = = = = = = =
Plymouth Tube

 

[RedVette]

Thank you guys for clarifying on this matter.

Given the density factor, would I be correct to assume that, since carbon is even lighter than Al and Si, it is therefore practically impossible for an XRF to pinpoint its percentage to within any proportionate range of accuracy apart from luck chance?

In comparison, could "wet" chemical testing be expected to be the surefire method to determining the percentages of alloying elements to within a decimal percent range from the true value?

Thanks again for your collective input.

 

[mrfailure]

Quantitative techniques including wet chemical, ICP (inductive coupled plasma), OES (optical emission spectroscopy) and carbon/sulfur analyzers will give you that accuracy, yes.

 

[EdStainless]

You can't do C by XRF, not even with large fixed channel vacuum path machines. Mg is as light as I have ever seen done with XRF.
There are a few companies working on hand held, Ar purged, laser energized, OE devices. These should be able to do C and N and everything heavier. Probably only about $50k.

= = = = = = = = = = = = = = = = = = = =
Plymouth Tube

 

[dbooker630]

Last year I had our local NDT source do some alloy sorting. They used an Olympus Delta XRF (Innov-X) analyzer, which claims improvement as far as light elements (Mg, Al, Si, P, S) are concerned. I was more interested in heavy metals but was impressed with the overall package. The base version is around $25K but the version for light elements is more expensive.