Don't Always Trust XRF Results

[Cmarett]

Good morning everyone,

Just a quick story to share on something that recently occurred that could possibly help someone if they are experiancing purity issues when having their samples tested via XRF.

I started refining silver from sterling a few months back as a hobby. Over these past few months I've been trying to work on the various techniques of the refining process with the goal of achieving at least 99.9 silver and pouring my own bars. Through the process of this, I would take my various samples to a local refinery that has an XRF machine to have the purity tested. The results were not great; 96% was the best I could get so I decided to build a silver cell as my finishing step. The silver came out beautiful and I was confident I had achieved my goal. Unfortunately the XRF only showed 98% purity. Frustrated, I decided to run the silver crystal back through the silver cell with electrolyte made from dissolving pure silver minted bars, ultra clean. I had the sample retested and it was the same purity of 98%. I then decided to melt a pure silver eagle stamped 99.9 as a test and use a new crucible with no possibility of contamination. I also remelted my crystal that had been ran through the silver cell twice, and took them both to be tested. The results clearly showed what I had began to expect; the XRF machine was not giving accurate results.

In the attached photos you can see the melted silver eagle coin which is a known 99.9 from the US Mint. It tested just over 96% and had several others trace metals contained. Obviously this shows the results of the machine are not accurate. The other picture shows a bar I poured from the refined silver crystal which tested higher than the silver eagle (YES) but still only shows a purity of just over 97% but with no other trace metals.

Conclusion is that I believe their machine is either mis-calibrated or malfunctioning, but as long as I have a known sample of 99.9 that I am testing against on the same machine, I can still judge the purity of my product.

Help this story helps anyone who may be experiencing the same issues.

 

[anachronism]

Hi there

Thanks for the post. Can I ask how old that machine is and when was it last calibrated by the manufacturer?

 

[4metals]

Your logic and your method to determine how accurate their instrument is are very good. Sad to say that, while XRF has made refining easier in many ways, the fast read has adversely effected the people selling scrap. There is no way a well calibrated instrument should be off by 2%. But on the other hand, the buyer made an extra 2%.

 

[Cmarett]

Hi there

Thanks for the post. Can I ask how old that machine is and when was it last calibrated by the manufacturer?

To be honest, I am not sure on either question. They do the testing for free, but don't seem to be overjoyed to do it, so I never wanted to push for more information or suggest that I thought their machine was inaccurate.

A part of the story I omitted to keep it short was that prior to the last visit which we got the above pictures, my wife took the same twice refined bar, and the silver eagle coin (before we melted it) and they would not test the coin as they said its obviously pure silver. That, and the continual low test results on my samples are what made me suspect their gun machine was off. Did they know? Not sure but would imagine they probably did, but I did not want to be the jerk that used their free service and then complained about it lol. That is when we decided to just melt the coin so they would test both.

 

[Cmarett]

Your logic and your method to determine how accurate their instrument is are very good. Sad to say that, while XRF has made refining easier in many ways, the fast read has adversely effected the people selling scrap. There is no way a well calibrated instrument should be off by 2%. But on the other hand, the buyer made an extra 2%.

Thank you. I feel like I should have mentioned it to them, but they do the service for free so I didn't want come across accusatory or as if I were complaining.

 

[Shark]

I’ve had the same thing happen here. Having a bit better relationship with the buyer. We discussed it, after discussing it on GRF. He admitted that when he got his XRF it wasn’t calibrated for silver as he didn’t deal in silver until much later, and then only bought marked sterling or bars, and how he still didn’t like to because of the low profit margin for him. In my case the gun consistently showed the contamination was 3+ percent rhodium.

 

[goldandsilver123]

Only time this happened to me there was a little miniscule copper shaving in the xray window.

Spent many minutes rechecking everything, when I cleaned the window, the copper came on the cloth and everything went back to normal

 

[galenrog]

Only time this happened to me there was a little miniscule copper shaving in the xray window.

Spent many minutes rechecking everything, when I cleaned the window, the copper came on the cloth and everything went back to normal

That happens frequently in the scrap business. I watched an experienced operator at major scrap buyer do exactly that for exotic alloys that had significant value. Cleaned window, shot, cleaned window, shot again. Method worked very well to get repeatable results.

 

[anachronism]

To be honest, I am not sure on either question. They do the testing for free, but don't seem to be overjoyed to do it, so I never wanted to push for more information or suggest that I thought their machine was inaccurate.

A part of the story I omitted to keep it short was that prior to the last visit which we got the above pictures, my wife took the same twice refined bar, and the silver eagle coin (before we melted it) and they would not test the coin as they said its obviously pure silver. That, and the continual low test results on my samples are what made me suspect their gun machine was off. Did they know? Not sure but would imagine they probably did, but I did not want to be the jerk that used their free service and then complained about it lol. That is when we decided to just melt the coin so they would test both.

Thanks for the detailed reply. For me, the situation in the highlighted text would have set all my alarm bells clanging. For the sake of 20 seconds to zap the coin it would have removed the main variable in this situation and would have meant there were no doubts as to the gun....

So why did they not check the coin?

 

[Lou]

The main issue is that the people using these handheld XRF instruments don't know the context in which to move to a different assay method. They also don't know any better. That's why I see samples with 90% iridium 10% gold (news flash, something like that just doesn't exist!) or silver heavily contaminated with rhodium (amazing! they mixed oil with water). Metallurgical impossibilities. But for John Q. Scrapbuyer, XRF is a magic instrument that gives them a number and so it must be right but in reality, barring a matrix-matched standard, anything coming from it should be taken it with a large halite crystal! Critical reasoning skills aside, any instrument producing a result algorithmically must be scrutinized. This is why using a trusted calibration standard, like what you did for an eagle, quickly tells the tale.

Knowing when the XRF reading is horsepoo is a skill I've had to develop over the years. Often times it's when I'm seeing things that don't exist, or the look isn't right appropriate to the reading. I have more than a handful of handhelds and one desktop. They all lie. Even ICP will lie. At least a properly done fire assay puts metal in the hand.

With careful titration work, you can very accurately and very precisely determine the purity of your silver and the quantity. Sure, it won't tell you what else might be in there like XRF or as fast, but I'd take a Volhard titration for settlement over any other method.

 

[anachronism]

Exactly. It takes time to get to grips with the idiosyncrasies of handhelds. A compnent that has a whole host of PGMs all under 1% is a sure red herring for a start. They've developed software packs for ewaste that distinguish Pd far better instead of presenting it as Ir, alongside various other fixes but even so, unless you've got a handle on what your gun is likely to interpret and why it's very easy to be misled.

In this case I'm really quite suspicious as to why they didn't gun the coin Lou.

 

[Martijn]

Thank you. I feel like I should have mentioned it to them, but they do the service for free so I didn't want come across accusatory or as if I were complaining.

Do they also buy pm's based on their 'free' test?
Tolerances on measuring devices are a great way to make money.
That's why most 'we buy gold' shops can test up to 22K with acids and put the difference in their deep pockets.

Edited for spelling.

 

[Cmarett]

Thanks for the detailed reply. For me, the situation in the highlighted text would have set all my alarm bells clanging. For the sake of 20 seconds to zap the coin it would have removed the main variable in this situation and would have meant there were no doubts as to the gun....

So why did they not check the coin?

I am speculating, but my guess would be that they probably knew their gun was off. That’s when I really became suspicious.