pXRF of coins? What does that mean?
Portable X-ray fluorescence (pXRF) is an analytical technique using an instrument that can detect and identify most elements within a material. This makes it useful in instances where an investigator is trying to determine the material makeup of an analyte (solid, liquid and/or gas). In archaeology and conservation the analyte will most commonly be an object of some sort.
How does the instrument work? Well, it is complicated but here a very basic explanation. The pXRF sends out photons (packets of energy with no mass) from a source material inside the instrument (in our case, rhodium). These photons can eject electrons within the object causing a separate photon to be ejected from the object. These photons will have a specific energy related directly to the element from which they originated. If the generated photon strikes the detector inside the pXRF it will register at that certain energy level (kilo-electron volts (keV)) and the identity of the element from which it was generated will be known. The instrument readout is usually in the form of a spectrum where the individually identified elements will be represented by a peak at specific energies. For instance, X-ray fluorescence of a solid gold object will return a peak at the gold energy levels. An object composed of a copper alloy with silver will return a spectrum with peaks at copper, silver, tin, antimony and zinc depending on the type of copper alloy.
The setup parameters of the pXRF are important for analyses. Voltage, for example, can have a major impact on the elements the pXRF will detect and their depth of detection (object density and other parameters will also affect depth of detection and X-ray penetration). Different voltages are optimal for different elements. For the life of the instrument we restrict our measurements between 1 and 40 kilovolts (kV). A selection of 40 kV will reveal a great number of elements and will detect them from deeper (microns to millimeters) within the object but is not really optimized for any specific group of elements. A selection of 8 kV will detect very few elements at much shallower depths (microns) and is optimal for much lighter (lower molecular weight) elements. A typical full analysis of an object will be conducted at two or three voltages to produce optimal results for a wide variety of elements expected within an object.
The pXRF is a wonderful tool but it is only going to record that at which it is directed. This places a lot of responsibility in the hands of the investigator. Can pXRF be used to analyze an object just excavated from an archaeological site? The simple answer is ‘yes’ but unless the object is cleaned the soil will show up in the spectrum as well. The depth of penetration depends on the setup parameters of the instrument but also the density of the object being analyzed. So, if we have a suspect gold coin that has a surface obscuring concretion (for whatever reason), we can still analyze the object. The return of a gold peak (amongst the peaks from the concretion elements) in the spectrum may warrant further investigation of this coin. However, if it is a thick or relatively dense concretion we may have no return for gold leaving us to conduct investigative cleaning on the object until we can make a presence/absence confirmation.
With this background information, it should be obvious how pXRF can help in the identification of archaeological coins which can be composed mostly of copper, gold, silver, lead and even iron. But the makeup of coins, even of the same base metal, can change overtime with the ebbs and flows of the economy of a civilization; times of prosperity may be reflected in more precious metals in a coin whereas a decline may reveal less.
Now for some Hands-on experience
Something atypical about this typical Cardiff Castle Roman coin
The reader will see spectra of 40 kV measurements of the coins in this blog post. We, however, used both 40 kV (to identify most elements) and 15 kV (optimal for lighter elements) measurements with associated change to other setup parameters which are provided on the figures but will not address here for simplification. We have selected a single coin to illustrate these voltages.
Coin SF077 is a typical Roman copper alloy coin recovered from Cardiff Castle (…or is it?). Two or more measurements are necessary when analyzing objects. Measurements at 40 kV (Spectrum S1 (logarithmic scale to show more clarity in the comparison)) reveal a great number of elements present in the coin. The 15 kV measurement are incapable of returning peaks much greater than 15 kilo-electron volts (keV – the X axis) but do provide greater clarity of the spectra below ~ 10 keV.
Spectrum S2a and S2b are the same spectrum measured using 15 kV (only difference are the associated labels with the peaks in S2b). For simplicity we will look at the elements above 5 keV. We can see that there is a high peak for copper (Cu) and small peaks for iron (Fe) and bromide (Br). The presence of bromide provides me with a clue as to what to expect with a higher measurement. Read on…
Armed with the information provided above, we know that a single measurement at 15 kV is not enough to determine all the elements within this coin. Therefore we conducted another measurement at 40 kV (Spectrum S3a and S3b). Now we see the addition of another element – silver (Ag). This was suspected due to the presence of bromide which is a constituent of silver bromide, a silver corrosion product. Bromide is usually not found on copper alloy coins containing no silver. The rhodium (Rh) and palladium (Pd) peaks are associated with the instrument and will almost always be present in the spectra.
So, perhaps this coin is not so typical of the Roman copper alloy coins found on the site. There is not enough silver to call this a silver coin. However, there may be enough to have given the coin a silver finish. As a comparison, here are the spectra (S4a and S4b) of a silver coin where we can see a much higher ratio of silver (Ag) to copper (Cu).
The exercise above was just to illustrate a point; that one measurement is usually not enough for a fair assessment of the elemental makeup of the coin. In all fairness, we would be very interested in the elements below 5 keV in the 15 kV measurements. Additionally, we can expand the spectrum to see additional elements (Spectrum S5a and S5b) as can be seen in the logarithmically scaled spectrum S1.
A sign of declining times?
Four coins (SF0333, SF0438, SF2019 and SF2021) recovered from the Cardiff Castle excavations are yellowish brown in colour. Three of these were identified as 4th century coins. The forth was not assigned to any century. The colour of these coins is different from the green, silver and black of most of the coins recovered from the excavations. They were analyzed and compared to four green copper alloy coins (SF0142, SF0376, SF0339 and SF0496) identified as the 3rd century or before to determine the reason for the colour differentiation. The spectra for the green copper alloy coins (Spectra S6a and S6b) exhibit a high peak for copper (Cu), medium peak for iron (Fe) and low peak for lead (Pb). The spectra for the yellowish brown coins were adjusted to represent the same scale (y-axis) as the green copper alloy coin spectra. These spectra (S7a and S7b) exhibited low peaks for copper (Cu) and iron (Fe), and medium peaks for lead (Pb). The spectra for two of the comparative coins are presented to show the difference (Spectra S8a and S8b).
So, the yellow coins appear to have a greater concentration of lead and a much lower peak for copper whereas the green coins have high peaks for copper and very low peaks for lead. Why is there such a difference in the composition of the coins? Are they lead tokens, fakes, forgeries or true coins with a clue to the economic times in which they were produced? They probably represent the latter. The fourth century, and especially the late forth century, was a time of upheaval. Roman Britain had to contend with invasions from Ireland and Scotland while with reduced resources due to campaigns and insurrections on the continent. Greater quantities of lead were introduced to some coins to extend general copper supplies.
In conclusion, we have seen how pXRF can identify the various elements within an object and how this can be used to inform on the greater economic situation at the time in this scenario. PXRF will be used again on some other objects recovered from Cardiff Castle.
We will now close this chapter on coins. The next blog post will be on the conservation of the general copper alloy objects.
Note: The coin captions will change as I get post-conservation identifications from a GGAT coin expert. All the coin posts will be amended to reflect those changes.