What Kinds Of Coins Can XRF Really Distinguish?

Let's add beam "strength" to the list of questions. What does that actually mean? Light is absorbed or reflected by various substances. To my understanding of X-rays (which is quite limited), if you blast 10x more of the same frequency at a surface, it will transmit or reflect the beam at the same rate. You would need to increase the frequency (and hence photon energy) to achieve a different result. But maybe that's wrong? My EM physics class spent most of its time on visual spectrum light because lasers are used in fiber optics.

The penetration depth of xrays is effectively inversely proportional to the heaviness of the atoms of the substance being penetrated. In effect, the further "down" the periodic table you go, the harder it is for xrays to penetrate.

This is why medical xrays work: they go straight through the "lightweight" elements such as carbon, nitrogen, oxygen and hydrogen that make up your skin and organs, but are stopped by the heavier elements such as calcium that are in your bones and teeth.

So, for coinage metals, copper nickel and iron are "easiest" for xrays to penetrate, silver and palladium are "medium", and gold and platinum are "hardest". What exactly this means in terms of actual penetration depth for a specific sample, depends on the power of the xray emitter used.

"Power" or "strength" is indeed equivalent to frequency. X-ray emitters emit a broad spectrum of frequencies. Each different kind of chemical element absorbs xrays at different frequencies, and then re-emits that absorbed energy at a different frequency - a process known as "flourescence", and fluorescence is the "F" in XRF. Fluorescent photons are emitted in random directions, rather than in the same direction as the beam, so "all you need to do" to turn an xray beam into an XRF is set up an xray detector somewhere other than directly in the path of the beam, measure how many xrays of different frequencies are been created within the sample while the xray beam passes through the sample, and thus calculate how much of each element is present in the sample.

The trick with using XRF on real world samples (such as coins) is proper calibration. Real-world coins aren't uniformly flat disks, they have lumps and bumps which causes the thickness to vary. This is the main cause of variation or error in reporting XRF results for coins.


XRF detects chemical elements, as individual xray photons interact with individual atoms, not with molecules or alloy structures. Thus, an XRF can tell you if something is made of iron, vanadium, and manganese, which you might presume to be stainless steel, but cannot tell whether that steel is ferromagnetic or paramagnetic, since that's a property of the alloy as a whole, not it's individual elemental components.

In terms of "which elements can XRF detect", the short answer is "in theory, all of them". Though some are better than others; as noted earlier, the lighter the element, the harder it is for an XRF to "see" it properly. Aluminium is the most lightweight element you're likely to encounter in numismatics, so smaller, thinner coins containing aluminium are most likely to return a "wrong" reading as it under-estimates the amount of aluminium present (and thus over-estimates everything else).

In practice, the limiting factor in the number of detectable elements any given XRF machine can detect is, well, the detector. The more frequencies you want to detect, the more complex and expensive the detector has to be. Thus, a "precious metals only" XRF will be cheaper than an "everything on the periodic table" XRF.


Platings and claddings do complicate things. XRF should in theory return "bulk results" for the sample. So aiming an XRF at a coin with a solid core and a thin plating (such as a zincoln), will return a result as if the entire coin were melted together into a single homogeneous lump of alloy (in the case of a zincoln, that would be 97.5% Zn, 2.5% Cu). Analysis of a zincoln should actually return such numbers. Likewise, the 40% silver coins will return "40% silver" and not notice or care that the silver actually comes in three distinct layers of different finenesses.

Note one caveat with this previous statement: if you aim the beam in such a way that it pass through only one layer (for example, by firing the beam sideways through a clad coin), then the results will be different. Exactly how different would be difficult to predict, but I suspect a beam aimed at the middle of the edge of a coin would probably return a result essentially equal to the composition of the core alone.

Penetration depth of the different metals can also affect the result. For example: imagine a coin made of solid gold, but given a thin silver plating; the xrays will penetrate the "top" layer of silver and into the gold, but won't make it all the way through several millimetres of gold to see the "bottom" layer of silver on the other side - thus giving you a lower-than-expected result for silver content.

In terms of the other way around - your gold-plated clad quarter, for example - you'd have to make sure the XRF was powerful enough to penetrate the plating, otherwise the answer returned will be "it's solid gold" because the beam wasn't strong enough to punch through the gold and see the underlying copper. Electroplating by these after-market coin marketers is usually done to be as cheap as possible, usually a very thin layer (less than 2 micrometres), and your typical handheld PM-specialist XRF rig can easily punch through 10 to 12 micrometres of pure gold, so there's no major issue there. It might not give you a "perfect" result in terms of chemical analysis, but it will definitely tell you "hey, that's not solid gold, it's mostly copper and nickel".

This is why you would need a very, very powerful XRF to spot the gold-foil-wrapped-tungsten fake bullion ingots, as the "gold foil" the counterfeiters use is quite thick (several thousand micrometres). Benchtop lab-grade XRF machines usually can't go through thicker than around 80 micrometres of gold, so XRF is essentially useless for these. They've developed ultrasound techniques which work better at spotting physical irregularities within a large block or piece of gold.

Sap, that is a remarkably detailed breakdown. Exactly what I was trying to understand! Thanks. =)

I am now left wondering if that 1983 transitional planchet certified by PCGS with a non-standard alloy was just a badly applied XRF reading.

Oddly, I can't find the thread where we were discussing that odd coin. It was eventually locked because the original poster kept arguing with AI, if memory serves correctly.

My limited experience comes from the scrap metal industry.
About 10 years ago we used XRF analyzers to identify metals. In theory they worked well, but in practice user scans often produced mixed results.
One of the most expensive mistakes I remember involved a catering company bringing in silver-plated serving dishes. The scanner identified them as silver, which resulted in a costly payout.
After incidents like that, many yards stopped relying on a single surface reading. Items were often cut and tested in multiple locations, with the results averaged before a final decision was made.
I'm sure the technology has improved significantly since then, but that was my experience using it in the field.
One of the early projects I worked on was a Bluetooth-connected Sigma-style device that paired with a phone. The theory was straightforward—measure conductivity and estimate silver content—but much of the challenge ended up being calibration and software rather than the hardware itself.
That experience is part of what pushed me toward software-assisted inspection. The sensor is only one piece of the puzzle; how the data is processed and presented to the user often determines whether the system is actually useful.