XRF Results

Could somebody please explain these xrf results? The test indicates a negative value for Cu and over 100% for Au. This test is from the first button I melted solely out of ram fingers. Thanks

Would this be a good example of why not to trust an XRF reading? :shock:

Tom C.

Now thats a place to sell your gold :roll:
I'm sure that your gold is pretty pure but to get a real answer an assay is the only way.

I wont mention the name of the company but they are an extremely reputable swiss company that deals in bullion with an outlet in the Birmingham jewelery quarter.

I understand the XRF is a surface scan so no where near a proper assay, just wondered if anyone could make sense of the results

I never seen an XRF, and have no idea what the graph represents.

what strikes me funny is it has 100.4% gold, 0.07% silver, -0.49% copper.


How can it have a negative number percentage of copper, and over 100% gold?
also from what this looks like to me they are only showing gold, silver, copper, what about the many other metals that could be pssible contaminates?

vegaswinner said:

I wont mention the name of the company but they are an extremely reputable swiss company that deals in bullion with an outlet in the Birmingham jewelery quarter.

I understand the XRF is a surface scan so no where near a proper assay, just wondered if anyone could make sense of the results

Click to expand...

I know exactly who you mean and the result just proves exactly what I have always said they aren't the answer for all needs, they are useful but unless extremely high end units they are just a guide.
My advice is to get an assay done as your UK based,it's fast and cheap and accurate.
Try Guardian Labs they are good lads and it will cost £10 for a gold assay. Phone number 0121 359 8233 speak to Austin or Andy. If you want a full assay done it's not cheap but in honesty I can't see the need if your just selling your gold.

As with most measuring devices (mechanical and electronic), they tend to be most accurate in the middle 2/3 of their range. XRF has its place. It is a fast and "dirty" method of determining composition. Until we get the Star Trek (tm) transporter technology, the fire assay will be the standard.

Butcher, it was those very readings which struck me as strange. Had all the numbers added to 100% with a % for gold, silver, copper, other metals etc then it would probably appear to be a little more self explanatory.

Nickvc, the purpose of the XRF was not to obtain an assay. It was done more out of fascination. As I'm sure a lot of people here have experienced, when you tell someone your refining gold and they see you with a bucket and a load of computer bits the response is usually something like oh yeah haha so what are you really doing. As a friend was in a position to do the XRF and was also intrigued at the fact refining was indeed possible in the back yard I was happy to let him take my 1.4g button to work for a scan, even he couldn't explain the results though. Thanks for that contact number, I will definitely be getting an assay done when I finally pour my first oz, if anything for my own personal gratification and of course it would be great to know exactly what level my refined gold is coming out at.

Hello butcher,

You ask:

I never seen an XRF, and have no idea what the graph represents. what strikes me funny is it has 100.4% gold, 0.07% silver, -0.49% copper.
How can it have a negative number percentage of copper, and over 100% gold? also from what this looks like to me they are only showing gold, silver, copper, what about the many other metals that could be pssible contaminates?

The quality and significance of every single analysis result obtained by a spectroscopic measurement is dependent on the quality of the standard materials used for calibration. Principally every spectroscopy-based analysis-result relies actually on two measurements: one being done with the unknown sample, in this case, the gold-sample to be analyzed, and the second one, being done on a reference sample of predetermined, exactly known composition (purity), verified by an independent, different analysis-method. Thus it may be possible, that you obtain negative results (percentages) for individually measured items, in case the reference sample used for comparison contains more of the item concerned, than the sample to be analyzed.

Regards, freechemist

Thanks Freechemist,
I want to compliment you on the posts you make on the forum, I always find very helpful information in them, and enjoy reading what you have to say about these subjects.

Are you saying his gold being measured can be purer than the standard sample they use to get a base line reading?

If so why would they use a standard of that quality, or are they just not expecting to read a more pure gold?

Hello butcher,

You ask:
"I never seen an XRF, and have no idea what the graph represents. what strikes me funny is it has 100.4% gold, 0.07% silver, -0.49% copper.
How can it have a negative number percentage of copper, and over 100% gold? also from what this looks like to me they are only showing gold, silver, copper, what about the many other metals that could be pssible contaminates?"

XRF-measurements usually are done on solid samples, powder or very cleanly polished plane surfaces, like e.g. polished metal plates, and this holds true for test-samples and reference samples as well. Small differences measured, like the ones cited above, are determined accurately only difficultely, and thus very sensitive (and prone to false results) on physical properties, like eveneness and/or cleaneness of the surface of samples concerned, too. So, it may be possible, that the gold sample is indeed a little bit purer, than the reference, but it is as well possible, that these small differences are caused by not exactly equal preparation of reference and test-sample, small differences in their surface properties etc.

Regards, freechemist

Thanks freechemist, that explains a lot... Now I know what to ask I may be able to find out the purity/assay details of the calibration material.

It can also be explained by the standard used for calibration is of a lower carat alloy. It would explain the negative copper content too.

An XRF is only a tool and it has a built in measurement error as every instrument has. Via calibration and fancy computer calculations the errors are compensated. But so far noone has built an error free measuring device. The important question is how big are the errors? For example the result should be reported as Au 100.4% +/- 0.5% and Cu -0.47% +/- 0.5%. Then a button with 99.97% Au + 0.03% Cu could give that result without it being strange.

What affects the precision of a measurement is a lot of factors, for example a flat or curved sample, scratches on the surface, temperature, air pressure, which standards it's been calibrated against, how long it was since calibration, the operator, ... and so on.

Anyone that wants to know how XRF works can read on, every one else just think of it as magic. 8)

The inner working of an XRF is quite a feat of engineering and physics. What you see on the printout above is the fingerprint of gold x-rays. Every metal has it's own unique fingerprint. To recognise a fingerprint is easy, the police is doing it all the time, but when you have two overlapping fingerprints it gets a lot more difficult.

Now, pretend that you have 118 fingerprints that you know about (number of known elements), some is so rare that you can just ignore them and some you couldn't even see (light elements). But in the end you have at least 30-50 different fingerprints that you need to recognise. Put them on top of each other along a line, with different strength, then try to sort out the mess you got (it's called a spectra). Not so easy, but there are ways to do it.

In a simple way you can do it like this...
Take the largest peaks and try to match up agains a known fingerprint. Okay, that's your first element. The height of the peak is proportional of how many percent of it you have. Substitute that from the mess. Now do it again and again for each peak in the spectra (the mess of overlapping fingerprints).

So, that didn't sound too bad, did it? To be frank, in the real life is a lot more difficult and messy. For example x-rays can reflect from other surfaces both outside of the XRF and inside the instrument. A paint flake that drops off can skew the results. Some elements have strong peaks that lies right on top of other, for example we have seen iridium appear in lead-copper slag. Some elements are good at shielding the x-rays coming from other elements. Lighter elements as beryllium is invisible to the XRF.
To help our poor computer with the task we can add spectras from standard alloys, so it can identify alloys immediately. Another trick is to let the computer calculate how the mix of elements will affect the x-rays in the case it first guessed at and compensate for it, getting closer and closer the more it calculates. But that is only for the best instruments as those calculations needs a lot of computer power.

Then we have the sensor part, it is quite delicate and to get a good signal we need to cool it down at the same time it works with a high voltage, just below the limit for arcing.
There have been some astounding developments in the sensor technology that let the makers create hand held devices. Before that the sensor was made from ultra pure silicone with a whiff of lithium diffused into the crystal structure to compensate for the last part of germanium that couldn't be refined off. Then cooled down with liquid nitrogen in vacuum, sitting behind a window of 0.001 mm thick beryllium film and driven with a kV in voltage. The sensor counted electrons given off the crystal when an x-ray hit. Every thaw or cooldown could break the sensor and the most dreaded sound was "pffffftt" when the beryllium window broke and got sucked into the vacuum flask.
The detector measures the energy from each x-ray in number of electrons that got knocked off the silicon atoms. An x-ray from the L-beta peak in gold has the energy of 11.44 keV, and knocks off 6,57 electrons from Si atoms, using up 1.74 keV of energy in each collision. Now, there are no part electrons so we can get at most 6 electrons per x-ray. Some energy could be lost on the way to the detector, some x-rays exits before they have used up all energy, some x-rays lost energy when it exited the tested sample. So in the end we got a whole lot of different number of electrons for the x-ray coming from a gold atom. By putting a high voltage (1 kV) the free electrons move to the surface where sensitive electronics detects them and counts how many electrons came and records the energy for that x-ray. This happens hundreds of thousands times a second. The electronic circuits also detects double strikes, when two x-rays hits too close to each other to discern beween them and discards this data.

Here is a good resource for looking at the spectras for the periodic system, and specifically the spectra of gold. Compare it to the printout.

But why does the gold give off x-rays? Normally it doesn't do it, but if you hit it with something small with high energy then it will give off x-rays. In an electron microscope (or x-ray tube) electrons hits a target and knocks out electrons from the shell of atoms. Depending on how high the energy is, electrons deeper and deeper in the atom can be knocked off. But nature hates when there are missing electrons in an atom and higher lying electrons drops down towards the center to fill the empty hole. To be able to do that it must get rid of the extra energy it has on the higher level and that exits as an x-ray. Since the electrons can only exist in a couple of fixed energy levels the difference is unique for each element. (Quantum physics, something for another topic)
The probability for an x-ray to belong to one or another peak in the x-ray spectra depends on how much energy the electron that hit the atom has. The higher the energy the bigger the probability that an x-ray from a higher energy peak are created. But even if the size of the peaks can change, the energy (eV, electron volt) of it never moves.

But why is it called XRF? X-Ray Flourescense?
Just as electrons can knock an electron out of an atom, the same thing happens if an x-ray hits an atom, an electron gets knocked off and then another falls in to fill the void and a new x-ray is emitted, but this time with a specific energy level less or equal to the one that knocked off the electron. That is the fluorescense part.

To summarize....
An x-ray is produced in the instrument and hits the unknown sample. The material emits new x-rays but with specific energy levels depending on the atoms in the sample. A detector measures the energy of every x-ray returned and sorts it into a spectra. The computer compares the spectra to known elements and alloys, trying to get the best mix of atoms to produce an equal spectra. When it is close enough it presents the analyse with elements and percentages.

So in the end, the fact that XRF can detect elements as good as it can is just at the edge of magic. even for me as a physicist.

Still here? Wow, I'm impressed. :lol:

Göran

Wow! Goran, that's all I can say, WOW!

That was a wealth of information. Thanks!

Glad you liked it, then it was worth writing it.

Göran

Very interesting, and informative, I liked your description of sorting out all of those fingerprints on top of each other.
Thank you for spending the time write this, it did help me to understand more of how this Xray machine works, and at the same time amazes me that the thing could even work at all.