Everything (hopefully) you ever wanted to know about XRF

Shop deals on ebay
xelexo
Advertise with us
shop deals on amazon
Gold Refining Forum

Help Support Gold Refining Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
4metals

4metals

Moderator
Staff member
Supporting Member
Moderator
Joined
Apr 5, 2009
Messages
6,239
Location
northeast USA
I have been in this refining game for a long time, way before the luxury of XRF. I have found it to be a useful tool as well as a curse all based on one's understanding of the information the XRF spits out.

I started in this industry from the analytical side so fire assays and fire assays with instrumental finish was how I got started. When the XRF came along it was great to use it as a guide to determine approximate concentrations to cut down on the quantity of cupellations required or the number of dilutions for analysis. But I never used XRF as an end all definitive tool.

There are lots of reasons for this such as the form of the sample (powder or melted sample) or the amount of surface oxides throwing off a reading. And I am sure limitations I've never considered. (Like picking up a rock with a vein of something in it, shooting it with an XRF, and thinking I'm a millionaire)

If you were to go through postings from many members there are some amazing expectations made based on XRF results which really shouldn't be relied on. These are the kind of things I'd like to see this thread clear up.

What I would like to see develop here is a definitive thread about XRF with posts from members with hands on working experience with the instrumentation. There is data which is displayed with a result like sigma values which indicate just how much you can rely on a result. These things should be explained, and as per good old forum style, debated and discussed here so everyone who ever has the need to rely on an XRF result can be better informed.

So I'll step aside and let those members with more XRF experience than I have develop this thread.
 
Well, I have no personal experience with XRF, but I have a project I've been working on for some time that includes a brief description.

How it Works - XRF works by analyzing the secondary fluorescent energy emitted when a substance is excited by a primary X-ray source. The device directs an X-ray at the sample, measures the fluorescent energy given off by the sample, and uses specialized software to determine the composition of the sample.

There are two main types of XRF technology, Energy Dispersive (EDX) and Wavelength Dispersive (WDX). EDX technology is commonly used in the portable, handheld “gun” type analyzers and some benchtop units, while WDX is more common in larger units. EDX is convenient and less expensive than WDX, but WDX is more sensitive than EDX.

Problems with XRF - While XRF is a fast, non-destructive testing method, there are potential problems. The results can be affected by the operator’s technique, the programming and calibration of the unit, the metals in the sample, sample preparation, etc.

Lou gave an excellent summary of the state of the technology in 2017:
An average energy dispersive bench top XRF cannot determine impurities and distinguish readily between 99.5 and 99.95 purity. If you wish to truly know with XRF, you need to have the surface machined perfectly flat, run it on wavelength dispersive, under vac or helium, and wait about 30 minutes while it raster scans the whole of the surface at a powerlevel 100X that of a usual 40W EDX system (or 1000X in the case of a handheld, which are 4 W). Even then, that instrument, very expensive as it is, is subject to certain limitations for certain elements and require different crystal/filters.
Click to expand...

If you sell to a buyer who bases their payment on XRF, it is a good idea to keep a known sample to check the calibration of the XRF. A number of mints have produced coins of 999 or 9999 gold. The U.S. Mint introduced the 9999 fine American Buffalo in 2006. Canadian Maple Leafs from 1979 to 1982 were 999 fine. There are many other coins, commemoratives, and ingots. If you feel your buyer’s reading is questionable, have them shoot your known sample and check the reading.

Dave
 
Excellent idea. In one period of my refining "career", I was using XRF on daily basis to determine progress of the precipitations or dissolutions of the PGMs. And I can definitely say, it is very convenient and straightforward. On the other hand, there are situations, when the gun decide to tell you something like "you have whole periodic table inside". And you know there are three elements and bit of water.

From my experience, there are certain combinations of elements which have probably overlaying signals - and thus if you have them mutually in solution, results can often came out pretty skewed.
Also, with complicated mixtures of elements, readings can come out pretty strange.
We usually obtained more accurate readings on solid metallic materials than on powders. It came with how diffraction works and what could be the other factors like moisture, grain size, homogenity of the grain, distance from the gun etc.
Most skewed results came from measuring solutions. This is probably the one of the most notable features of XRF - that you can measure contents of the solution in seconds. Big drawback is - it is unreliable, if you do not calibrate it on standardized composition or check the known composition.

For silver and gold, I was very satisfied with results and outcomes. Refining karat using XRF and scales to measure the weight produced consistent +-1% results on the ammount of gold recovered.
With PGMs, it was a whole different story. Not recommended without special library or DIY calibration/check on known composition. On melted PGMs it worked relatively OK.

And yeah, we were using OLYMPUS Vanta machine. In warranty, 2 year old, callibrated.
 
I think the key to any XRF reading is to understand that your assay is only as good as your sampling. A pin has to be taken from a uniform melt, rolled, then polished, then checked in multiple locations.

The other key is the interpretation of the results. The best example I can remember is in the movie "The Hunt for Red October" when the sonar guy hears something and the computer calls it a whale noises. He makes the comment that he hears something mechanical, and that he believes the computer is "running home to momma". This is exactly the case with the analysis software. If it's set up for precious metals, it's not going to differentiate between tungsten and gold. If it's set for base metal alloys, it's likely to call gold tungsten.

I'll let someone else describe the spectroscopy.
 
One experience stands out over most for me. I took some silver that came from a nitric dissolution, ran through a cell twice and had the crystals shot with an XRF. For gold it had always shot good, and I have used this buyer for several years now. The crystal shot really low with several contaminants that should not have been here. Lou later explained about the rhodium content and how xrf may miss read it. Anyway, later I melted the same crystal and had the resulting 6 toz shot by the same xrf. As a bar it read at .9998. A 1toz bar I keep on hand marked as .999 has consistently read as .9993 even after carrying it for several years. I don’t sell silver as I still get very little if it, but the gold always reads good enough to please both myself and my buyer.

Edited for spelling
 
I get screen shots of XRF readings from clients all the time just like the one below. This lot in particular represented a bar which was about 270 ounces. Sure he wanted the Gold and Palladium but the Rhodium is what be wanted most. Even though the percentage is small, the $18000 an ounce made him drool for it.

I told him to aqua regia refine it and cement out the PM’s after the gold by Sulfur Dioxide and the Palladium by Ammonium chloride and sodium chlorate. Then, when he collects enough from the cementation tank to talk about that process. He really didn’t want to hear that it likely doesn’t have Rhodium anyway.

What do the sigma values represent to some of our XRF experienced members. What can they tell you about this bar?
 

Attachments

  • 9EA9CEE4-D01B-4EDD-9884-49590C2BC1CF.jpeg
    9EA9CEE4-D01B-4EDD-9884-49590C2BC1CF.jpeg
    1.1 MB · Views: 34
It's the standard deviation. How high that number indicates how spread out the curve is around the peak....inotherwords, the it is a measure of the precision of the numbers they are utilizing to determine that there is Rhodium there.
 
I don't do anything with XRF, but I often see screenshots here or on YouTube.

Maybe it's just me, but when I see a XRF showing a date way off, I suspect the owner probably doesn't care well about calibration also.

The last, I've seen was June, 28th 2030…
 
4metals said:
I get screen shots of XRF readings from clients all the time just like the one below. This lot in particular represented a bar which was about 270 ounces. Sure he wanted the Gold and Palladium but the Rhodium is what be wanted most. Even though the percentage is small, the $18000 an ounce made him drool for it.

I told him to aqua regia refine it and cement out the PM’s after the gold by Sulfur Dioxide and the Palladium by Ammonium chloride and sodium chlorate. Then, when he collects enough from the cementation tank to talk about that process. He really didn’t want to hear that it likely doesn’t have Rhodium anyway.

What do the sigma values represent to some of our XRF experienced members. What can they tell you about this bar?
Click to expand...
Personally I would ignore the Pt and Rh readout as only a possibility and go for the gold foremost then cement out any PGMs , once you have the blacks dried shoot them and get excited if the Rh was now showing a good percentage.
 
snoman701 said:
It's the standard deviation. How high that number indicates how spread out the curve is around the peak....inotherwords, the it is a measure of the precision of the numbers they are utilizing to determine that there is Rhodium there.
Click to expand...
I would add to this information, that sigma value is computed by the machine. There can be a situation, where you have metal shown and good sigma value, but it is just not there. Pretty common with PGMs like Ru or Rh. Os is also sometimes shown. In our OLYMPUS Vanta, there was a special mode for catalysts, which was precisely calibrated for powder alumina/ceramic/zirconia matrixes containing PtPdRh. It gave relatively good numbers, also trying to compute absolute content of the PMs and light elements. But nothing reliable enough that I would buy a bag of spent catalyst based on this reading.

I liked a XRF a lot when working or buying the material. You can easily distinguish between base metals MLCCs and PdAg MLCCs right in the field after crushing one or two grams. Very convenient to zap old MLCCs to know if they are Pt or Pd ones (very important info). Also distinguish between grades of silver conveniently, if it is 800 or 925 or other fineness.
 
I am a certified XRF technician. I can do any XRF analysis for PMI. It’s a trick sciences based off radiology x-ray. When you get into the platinum groups especially extremely dense metals. X-ray analysis has the abilities to miss identify denser metals. For instances on the old version of Olympus guns. Gold registers as tungsten due to density’s not being to far off. All hand1held versions are way less accurate compared to traditional fire assaying. For instance I work in aerospace and defense work. We had almost a recall last year on 304ss that was testing as 303ss. Come to find out the Olympus couldn’t pick up on the extra silicone or phosphorus used in the manufacture process. Causing it to be miss identified.

Also most smelter when you read ASME specs on alloys you’ll get a allowable %. Smelters always uses the lowest amount of % to make alloys. There are some specialized smelters that sell high grade alloys with high % as well. Usually for R&D type work.

Also 95% of commercial metal stock is from recycling material. Very rarely is are mountains pulverized and milled into metal these days for environmental reasons.
 
Last edited:
Interesting perspective on the topic, but it is not the density that do the separation is it?
Except for the elements that is too light to reflect any of the X-rays.

It's the returned spectrum from the outgoing X-rays.
It is the X-rays that excite electrons in the outer shell of the elements and when they fall back to the original position they send out an photon with an energy in a certain frequency.
The receiver picks these up and estimates what is in the material being radiated based on libraries and algorithms.
Libraries to tell which spectrum belongs to which element and algorithms to estimate the amount of said element.
The operator have to make sure it is done in the proper way to get consistent reading and make sure the proper library are loaded for the test at work.
I'm sure there is a plethora of things I have not said and maybe some exaggerations.
But in short that is how I understand it.

Please do correct me if I'm wrong.
 
No, you are correct. The material is bombarded with radiation from an xray tube. The interaction between the material and the electrons releases energy called the Bremsstrahlung radiation. This radiation is then measured through a specialized sensor.
 
I'll ask a friend of mine who really, really knows XRF (Gabe!) to hop on this!

I'm sure the poor fellow will be flustered by the ads but hey, join the party!

We have half a dozen or so of the handhelds at work, and a desktop that we love. From 300 ppm to 300,000 ppm it is within about 1% of the ICP-OES instrument. So for general production stuff, it's awesome.
 
The XRF on the picture should be removed...But it dosent work properly...
The new XRF is much smaller and same power. If anyone have experience as technician with the one in the picture, he can PM me...
 

Attachments

  • 1.jpg
    1.jpg
    984.3 KB · Views: 16
  • 2.jpg
    2.jpg
    1.2 MB · Views: 16

Latest posts

Back
Top