XRF Gold Assay Experiment

[Sesselmann]

Some of you might know that I also have a jewellery business, and these days with most jewellery produced in China we want to verify if the gold quality holds the carat it should. Unlike in the UK where each piece of jewellery must pass through the government assay office, Australia has a self regulating system, so it's up to the manufacturer to make sure the gold holds the right carat.

Traditionally gold assays were done in a kiln where the pure gold was extracted from a small sample and weighed, but these days one can do a non destructive assay by XRF and get a very good result, the problem is the equipment costs $15,000 to $20,000 (Thermo Scientific).

Karl sent me a surplus 1.5" xrf crystal with a beryllium window, and I mounted it on a new ADIT PMT, so I thought I would give this a try.

I sandwiched a piece of copper and lead, and hammered it into a concave shape with a hole in the middle, I then attached three Am sources around the hole with a bit of glue.

front.jpg

back.jpg

As you can see, this was a very quick and crude experiment, proof of concept only ;)

ring.jpg

I suspended the gold ring above the hole with a copper wire, leaving about 5-10 mm between the Am sources and the ring.

au3.png

Using a GS-pro with PRA, I let the experiment run overnight, and recorded the spectrum above. I happen to know the exact composition of the 9 carat gold alloy in the ring as I am the manufacturer.

Cu 420/1000
Zn 70/1000
Ag 135/1000
Au 375/1000

The x rays from Cu and Zn are 8.04 and 8.63 keV respectively, and my NaI crystal is unable to resolve these peaks, so they show up as one big peak with the gross counts being around 48%, of the total counts in the three selected peaks (actual alloy abundance is 49% )

The next peak is from Ag and the ROI shows 17% of the total counts of selected regions vs actual 13.5%

The last peak which is Au is less defined and more difficult to establish, but I selected the ROI visually and it worked out to 35%.

As you can see from the spectrum above, some of the x-rays from my Am source made their way to the detector and gave me an unwanted peak at 59 keV, but this could be improved with a better shield. I already have an idea for how to improve this.

All up a better result than I was hoping for.

[pietkuip]

Knowing your setup, maybe I could fool you with gold-plated lead :)

Because I would guess that the gold K fluorescence is induced by alpha particles from the americium (PIXE). The alphas do not penetrate, so a thin surface layer gives the same intensity.

Also: the gold will scatter 60 keV from the source, you will not be able to get rid of that with shielding.

For x-ray spectra, I use the XR-100 from Amptek: [broken link removed - Steven]
Less expensive than the things that you mentioned. I assume that Si-detectors have come down in price since we bought our setup.
(I have not tried it yet with the sound card and PRA. I should test that.)

[Sesselmann]

Knowing your setup, maybe I could fool you with gold-plated lead :)

After 30 years casting jewellery and 10 years in the gold plating business, unlikely ;)

Because I would guess that the gold K fluorescence is induced by alpha particles from the americium (PIXE). The alphas do not penetrate, so a thin surface layer gives the same intensity.
Also: the gold will scatter 60 keV from the source, you will not be able to get rid of that with shielding.

Yes, I agree, the method has its shortcomings, but in conjunction with what is already known it can be and is used to verify gold content. I think the commercial XRF products use a small x-ray source to excite the fluorescence, but that comes with all the baggage of licensing and compliance. i wonder if a polonium source would work better?

[pietkuip]

I would not use an alpha emitter, unless it is the top few micrometers that one is interested in. Gamma (and beta) would make the setup sensitive to bulk composition.

Licensing is usually less of a problem than for radioisotopes. Everybody used to have x-ray sources: the CRT tv-sets and computer monitors that people spent hours in front of. Low-voltage sources (up to 40 kV) are usually exempt if power is low and shielding adequate. You could build something yourself out of old vacuum tubes and an electrostatic generator.

Amptek sells low-power miniature sources. They also have this pyroelectric generator: [broken link removed - Steven]

[Sesselmann]

Today I made myself a new shield for my new XRF detector. On the attached images you will se I used a small iron pot to melt some lead, which I cast into an old 65 mm aluminium detector housing. I them machined it to fit over my 1.5" XRF detector, drilled a 10 mm hole in the middle, and embedded three Am sources around the perimeter. A quick three hour test gave a reasonably good result. More sample excitation would be better, and as Pieter commented above, x-rays would be better than alpha.

pot.JPG

backpb.JPG

frontam.JPG

9 carat.png

Note: Looking at the spectrum above, the first peak looks as it it has been cropped on the left side, this might account for the relatively low copper high gold result.

[Sesselmann]

Here are a few more spectra...

This one is from an Australian Dollar coin. The alloy is Cu, Ni, Al

dollar.png

This is a spectrum of pure Indium...

Indium(2).png

This is a spectrum of a river pebble and I have no idea what it contains or why it emits these peaks.

pebbles.png

Drinking Glass

drinking glass.png

This is a simple drinking glass, and appears to have the same peaks as the pebbles, which makes me suspect there is a reason other than XRF causing these peaks.

I suspect Compton scattering, seeing as these objects are not metals?
The angle between the source and the detector is constant in both cases, and remarkably the two peaks visible above are 42 keV and 18 kev respectively, which sum to 60 KeV, very close to the 59.54 from the Am source.

[Geoff]

I have no clue what I'm talking about, so excuse me if I'm wrong here... Would it be possible that your rock has a high silica content, much like the glass? I don't know if there is a XRF peak for silicon, but that's the connection that jumped out at me while reading this.

[Sesselmann]

Geoff,

That was my first thought as well, but the x-rays from silica are nowhere near 45 keV.

As mentioned above the two peaks produced sum to approx. 60 keV which makes me suspect it is the 59 keV gamma from Am which is scattering at two different angles to produce these peaks,

Here is a good resource for looking up the x-ray energies..

http://nucleardata.nuclear.lu.se/toi/

[lodovico]

Hi to everybody,
I am interested in XRF too, I did a lot of tests with Am sources as excitation sources, a CsI detector and Theremino MCA.
I nearly tested all the elements, I miss few of them.
I attach a report with the results.
As far as I know with light elements compton scattering is dominant and low energy x rays (below 4-5 keV) are very difficult to see. My limit is chromium but with professional equipment you can go down till argon.
Lodovico

[lodovico]

The peaks on the glass XRF spectrum could be both compton scattering form light elements present in glass and XRF from heavy elements which are sometimes used as additives, such as barium (very common) or lead.
Lodovico