In situ copper isotope analysis by femtosecond laser ablation multicollector inductively coupled plasma mass spectrometry (fs-LA-MC-ICP-MS) on historical gold coins

Abstract

This study investigates the potential of femtosecond laser ablation coupled with multicollector inductively coupled plasma mass spectrometry (fs-LA-MC-ICP-MS) for copper isotopic analysis in gold matrices applied to cultural heritage. Elemental analyses, which have commonly been used so far, provide information on the circulation of metal stocks based on elemental signatures but fail to pinpoint the precise source of gold. In contrast, isotopic analyses can offer a more accurate means of identifying the source of the metal, yet their application to gold matrices remains a challenge. For the first time, we successfully determined copper isotope ratio in gold matrices and achieved repeatabilities of 0.12‰ to 0.26‰ (2SD) for δ⁶⁵Cu analyses carried out over up to 8 days, demonstrating the feasibility of copper isotopic analyses in gold coins at the micron-scale. This work was conducted using isotopically characterised in-house matrix-matched gold standard with copper concentrations varying from 4.5 wt% to 9.6 wt%. Our results open new avenues of research for provenance studies of precious museum artefacts and archaeological finds, with potential applications in authentication analyses on similar gold materials. The micro-sampling performed by femtosecond laser ablation minimises the damages on such ancient artifacts. However, the Cu concentrations had to be of at least 4 wt% with our analytical set up and a special care must be taken on the laser beam focusing in order to obtain accurate δ⁶⁵Cu measurements in gold matrices.

---

1. Introduction

The analyses carried out on ancient gold artefacts, mainly coins (e.g.^1,2) or jewellery (e.g.^3,4), have mostly concerned so far elemental compositions. The use of laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS) for elemental analyses is especially developed in cultural heritage thanks to its only micro-invasive nature, virtually imperceptible to the naked eye (e.g.^5–9). The aim of these elemental analyses was to provide information on the manufacturing techniques (e.g.^10–12) or on the characterisation, circulation and provenance of gold (e.g.^13–19). Most of the time, the results of these studies provide information that enable objects to be grouped together on the basis of the similarity of their chemical signatures, or even on the existence of metal stocks on a larger scale. However, it is very difficult to establish the geographical/geological source of gold on the basis of elemental analyses alone and isotopic analyses can help determining a much more accurate origin for the metal used, by providing information on the type or age of mineralisation, for example. Such analyses for various ancient metals (gold, silver, bronze, iron, among others) have been carried out using different isotopic systems. The longest-established approach of this kind used Pb isotopes (e.g.^20–32), but stable isotopes of Cu (e.g.^21,23–25 and ^33–38), Fe,^39–42 Ag (e.g.^{21,23,25,43,44}) and also Sn ^45,46 or Os ^47,48 have been developed and some have proved their worthiness.

In the case of gold, the use of isotopes for provenance studies began with the studies by Bendall⁴⁹ and Bendall et al.⁵⁰ for Celtic gold coins. They used Pb isotopes via in situ sampling with nanosecond laser ablation coupled to multicollector ICP-MS (ns-LA-MC-ICP-MS). However, as the Pb content was low (0.5 to 50 ppm), only ²⁰⁶Pb, ²⁰⁷Pb and ²⁰⁸Pb were determined, although ²⁰⁴Pb is the most relevant isotope for provenance studies as it serves as a reference since it is the only non-radiogenic isotope. The same conclusions were drawn by Hauptmann and Klein.⁵¹ Even if Standish et al.³⁰ showed that only a factor of 4 worse in reproducibility could be obtained relative to wet chemistry for isotopic ratio involving ²⁰⁴Pb in gold artefacts, it was still not measured with satisfactory precision by ns-LA-MC-ICP-MS for provenance interpretations. Whereas successful in situ Cu stable isotope analysis by ns-LA-MC-ICP-MS in copper ores has been reported two decades ago,⁵² its preliminary application in gold matrices yielded inaccurate and poorly reproducible results.⁴⁹ The use of femtosecond laser ablation (fs-LA) instead of ns-LA greatly improved results with matrix-matched and non-matrix-matched standards as demonstrated in copper-rich minerals.^53–57 As reviewed in Poitrasson and d'Abzac,⁵⁸ one of the advantages of fs-LA compared to ns-LA is the so-called a-thermal ablation: ablation pulses are so short – in the order of hundred femtoseconds – that they drastically limit heating of the sample during the laser ablation. This is thus of great interest for isotope analyses since thermal effects can fractionate isotopes, notably during non-matrix-matched calibration (i.e. use of a standard with a different mineralogical structure or composition relative to the samples). Overall, fs-LA compared to ns-LA improves the sensitivity of the LA-MC-ICP-MS thanks to the higher ablation yields. Furthermore, compared to ns-LA, it produces smaller particles that are more effectively atomised in the plasma torch, reducing the risk of isotopic fractionation caused by unevenly sized particles.

Here, this article reports for the first time copper isotope analyses performed on gold coins using a fs-LA to assess the relevance of this analytical approach for cultural heritage studies in gold.

2. Materials and methods

2.1. Corpus

Three gold coins from the 19th century were analysed in this study. They consist in an 1813 coin of 20 francs-or from the reign of Napoleon I, an 1863 coin of 10 lire from the Kingdom of Italy and an 1878 coin of 20 francs-or from the reign of Napoleon III. Moreover, some preliminary tests were undertaken on a dinar (medieval Islamic coin) minted during the Ayyubid dynasty in Egypt by al-Adil I (1199-1218).

As there are no referenced standards for Cu isotope measurement in a gold matrix, we employ one of our in-house prepared gold standard Au3, selected among three in-house standards (i.e., Au1, Au2 and Au3) for its high copper content (8.2 wt%), comparable with the three 19th c. coins (with respectively 4.5, 8.1 and 9.6 wt% Cu for 1813, 1863 and 1878). For a detailed description of the choice of the gold coins corpus, the in-house gold standards and for their full chemical (major, minor and trace elements) and isotopic characterisation (Pb, Fe, Cu), see de Palaminy et al.⁵⁹

2.2. Methods

The isotopic measurements were carried out by MC-ICP-MS (Nu Plasma HR, Nu Instruments, Wrexham, UK) at the IPREM laboratory. The mass spectrometer was coupled with a fs-LA system (Lambda 3 Nexeya, France). This laser is fitted with a diode-pumped KGW-Yb crystal source delivering 360 femtosecond pulses at 257 nm. Laser ablation and MC-ICP-MS parameters are listed in Tables 1 and 2.

Table 1 Laser ablation parameters

LA parameters		Pre-cleaning	Ablation
Laser		Lambda 3 (Nexeya)
Wavelength (nm)		257
Carrier gas (He)		0.5 L min^−1
Pulse duration (fs)		360
Energy (μJ per pulse)		2	2
Diaphragm aperture (mm)		2	2
Spot diameter (μm)		20	20
Frequency (Hz)		3300	30 or 50
Scan speed (mm s^−1)		300	3
Ablation pattern	Square size (μm)	150 × 150	80 × 80
	Interval between lines (μm)	10	10
Pattern repetition		3000	1700
Resulting depth (μm)		∼65	∼130

---

Table 2 MC-ICP-MS parameters (Nu plasma HR)

Instrument parameters
RF power	1300 W
Instrument resolution	LR
Sample gaz (mix)	0.6 L min^−1 Ar (+0.5 L min^−1 He from LA)
Coolant gas (Ar)	13 L min^−1
Auxiliary gas (Ar)	1 L min^−1
Faraday cup set-up	H7	H2
	^65Cu	^63Cu

---

This laser has a narrow spot size (<20 μm) with a Gaussian beam shape. Thanks to the optical properties of the laser, and using a 2D-galvanometric scanner that allows to move the laser beam extremely rapidly (up to 2 m s⁻¹) and with great accuracy (±1 μm), it is possible to draw very precise patterns and complex trajectories specific to the study samples and issues, especially when combined with high repetition rate.^60–64 It was thus possible to design an ablation strategy specific to our needs, described below.

It is known that metal artefacts such as ancient coins, after repeated use and ageing, or through intentional modification, can undergo some surface enrichment or alloying element depletion. For gold, the most common modification is a fraudulent enrichment in gold of the surface, accompanied by depletion in silver and/or copper, in order to mislead the user on the real value of the coin.^65,66 Severe alteration can also cause silver and/or copper leaching from the coin surface, resulting in a gold-enriched surface. This process has been well studied for gold nuggets in alluvial contexts (e.g.^67,68) has been observed in gold coins surfaces, down to ∼40 μm deep from the modified surface (e.g.⁵⁰). As a consequence, in elemental analyses with laser ablation ICP-MS, the beginning of the time-resolved ablation signal, that corresponds to this altered surface, is not taken into account.⁵ The surface enrichment/depletion is the result of chemical reactions that may involve redox processes likely to generate isotopic fractionation of Cu. Therefore, this enriched/depleted surface layer of metal should not be integrated for the isotopic analysis as well.

It was thus decided to carry out a pre-ablation, or “pre-cleaning” (Fig. 1a), before the analysis itself, in order to remove the first ∼50 μm of the coins and a comparable approach was adopted for the calibration standard Au3. Because of the Gaussian shape of the laser beam, a continuous ablation would result in a conical-shaped crater with the produced aerosol being a mixture of the material ablated from the more internal portions with those of the surface of the sample. To avoid such focusing issue, the pre-cleaning crater had to be almost twice as wide (∼150 × 150 μm) as the second crater deemed for the analysis (∼80 × 80 μm), deeper in the sample (Fig. 1b–d). It was also important to consider the micro-destructive nature of the sampling, as the ablation should not be easily visible to the naked eye, as is mandatory for cultural heritage artefacts. The pre-cleaning (Fig. 1a) involved scanning a square of 150 × 150 μm repeated 3000 times at a speed of 300 mm s⁻¹ and a frequency of 3300 Hz, with a 10 μm gap between each scanning lines. For each sample, several pre-cleanings were carried out one after the other using an automatic program (Fig. 1b). Under our operating conditions, the pre-cleaning depth was measured at 65 μm ± 5 μm (see below in Section 4.3), giving a large amount of ablated gold (∼25 μg) in approximately 96 s. In order to avoid contamination of the LA-MC-ICP-MS transfer line by particle deposits, plasma overloading, material deposits on the cones and, in general, contamination that could affect MC-ICP-MS performance, the ablation cell used routinely has been slightly adapted for these pre-cleanings. The purge-exhaust of the ablation cell was diverted through a 6 mm i. d. (Internal Diameter) tube (much larger than the one connected to the plasma – 1.5 mm, i. d.) to the gas exhaust pipe of the MC-ICP-MS. In these conditions, contamination of the LA-MC-ICP-MS during pre-cleaning (via the helium flow carrying the ablated material) was avoided while maintaining safe working conditions for the operators.

Fig. 1 SEM – SE (Secondary Electron) images of craters made with the fs-LA in the 1863 gold coin. (a) Pre-cleaning crater. (b) Two sets of three analysis with the isotopic analysis crater inside the pre-cleaning one – note that the isotopic analysis crater is not always centered in the pre-cleaning one. (c) Detail of the two craters – note the redeposition of ablated particles in the pre-cleaned crater. (d) Crater size (μm) – note that the shape of the craters is slightly rectangular.

The second crater integrated for the actual analysis – the inner square within the pre-cleaning (Fig. 1b–d) – was scanned 1700 times in a ∼80 × 80 μm square with a laser beam speed of 3 mm s⁻¹ and a frequency of 30 Hz or 50 Hz. This latter frequency for the analysed squares had to be adapted in order to have the same overall Cu isotope intensity (V) on the MC-ICP-MS cups between the coins and Au3 in-house standard, for a smooth sample-standard bracketing. For coins 1863 and 1878, having a similar Cu content as Au3 (respectively 8.1, 9.6 and 8.2 wt% Cu), frequency was set at 30 Hz; whereas for coin 1813 with less copper (4.5 wt%) it was increased to 50 Hz.

The laser energy for pre-cleaning and actual analysis craters was set at 2 μJ per pulse.

In order to characterise the morphology and shape of ablation craters, surface observations of the coins were performed optically (OM) using a stereomicroscope ZEISS AxioZoom V16 and by scanning electron microscopy (SEM). The SEM apparatus used was a ZEISS EVO25 equipped with a LaB₆ emission gun, working at 20 kV acceleration voltage and 250 nA under high vacuum. Scanning electron microscope observations were performed without any conductive coating given the gold matrix.

The vaporised matter was carried with a He flux towards a double path cyclonic spray chamber (Twister, Glass Expansion) that helps smoothing the MC-ICP-MS signal arising from the variations in the quantity of material introduced into the plasma due to laser shots at these frequencies. The nebuliser inlet of the spray chamber was used for argon addition (auxiliary gas) in order to achieve the optimum He–Ar mixture flow rate in the injector resulting in the best signal sensitivity and stability (Tables 1 and 2).

The sample-standard bracketing method was employed to acquire δ⁶⁵Cu. The in-house standard Au3 (STD) was measured 4 times and the sample (SMP) thrice following the scheme STD-SMP-STD-SMP-STD-SMP-STD, resulting in a ∼7 min measurement overall. Each sample was measured as triplicate, four to five times over the two weeks duration of the coin's analyses, resulting in 12 to 15 individual analyses per coin. As we conducted time-resolved analyses, the beginning and end limits of each integration limits were set manually in order to consider the acquisition interval with the most stable signals for both isotopes ⁶⁵Cu and ⁶³Cu. The ⁶⁵Cu/⁶³Cu ratios were calculated for each signal acquisition point of 2 s after blank subtraction with the Point By Point method (PBP),⁶⁹ the average ratio and 2SD being calculated from all the points within the selected stable measurement interval. In agreement with the previous study of García-Poyo et al.,⁶² the PBP method was chosen preferentially over the Linear Regression Slope (LRS) method as the analytical precision was 2 to 3 times worse for the LRS method (in 2SD, without outlier data rejection outside 2SD, for both methods). δ⁶⁵Cu (‰) of the sample (SMP) was calculated against the in-house standard Au3 following the equation:

δ ⁶⁵Cu was then recalculated against the international standard reference material SRM-976 adding +0.25‰, the δ⁶⁵Cu value measured for our in-house Au3 standard (see de Palaminy et al.;⁵⁹Table 3), and errors were propagated to the final δ⁶⁵Cu results.

Table 3 Copper isotopic results for coins 1813, 1863 and 1878. Samples were measured as sample standard bracketing against the Au3 in-house standard (8.2 wt% Cu) and δ⁶⁵Cu values have been recalculated against the international standard SRM-976, with Au3 δ⁶⁵Cu value (0.25 ± 0.07‰) obtained in solution MC-ICP-MS analysis after purification chemistry (see de Palaminy et al.⁵⁹)

1813					1863					1878
Cu (wt%)	4.5				Cu (wt%)	8.1				Cu (wt%)	9.6
Measure		δ ^65Cu‰ (SRM 976)	2SD	2SE	Measure		δ ^65Cu ‰ (SRM 976)	2SD	2SE	Measure		δ ^65Cu‰ (SRM 976)	2SD	2SE
2SD: 2 × standard deviation. 2SE: Standard deviation × t/√N. t/√N constant value at 95% probability level as a function of N-1 degree of freedom (Platzner, 1997).^73 /: Not relevant. °: See de Palaminy et al. (2024).^59
03-03-2022	1	0.01			07-03-2022	1	−0.26			04-03-2022	1	0.23
	2	0.24				2	0.42				2	0.15
	3	0.30				3	0.36				3	0.46
	Mean	0.18	/	0.38		Mean	0.17	/	0.94		Mean	0.28	/	0.42
04-03-2022	1	−0.08			07-03-2022 bis	1	0.28			07-03-2022	1	0.12
	2	−0.21				2	0.16				2	0.42
	3	0.15				3	0.25				3	0.19
	Mean	−0.05	/	0.46		Mean	0.23	/	0.19		Mean	0.25	/	0.40
04-03-2022 bis	1	0.23			08-03-2022	1	0.14			07-03-2022 bis	1	0.16
	2	0.22				2	0.14				2	0.07
	3	0.27				3	0.12				3	0.23
	Mean	0.24	/	0.11		Mean	0.14	/	0.10		Mean	0.15	/	0.23
11-03-2022	1	−0.05			11-03-2022	1	0.31			08-03-2022	1	0.23
	2	0.21				2	0.40				2	0.22
	3	0.36				3	0.09				3	0.32
	Mean	0.17	/	0.53		Mean	0.27	/	0.42		Mean	0.26	/	0.16
					11-03-2022bis	1	0.20			11-03-2022	1	0.47
						2	0.09				2	0.37
						3	0.33				3	0.47
						Mean	0.21	/	0.30		Mean	0.44	/	0.16
Mean (single analyses)		0.14	0.36	/	Mean (single analyses)		0.20	0.35	/	Mean (single analyses)		0.28	0.28	/
Mean (pooled analyses)		0.14	0.26	/	Mean (pooled analyses)		0.20	0.12	/	Mean (pooled analyses)		0.28	0.22	/
Solution MC-ICP-MS°		0.18	0.03	/	Solution MC-ICP-MS°		0.35	0.22	/	Solution MC-ICP-MS°		0.20	0.06	/

---

3. Results

We obtained average δ⁶⁵Cu values of 0.14 ± 0.26‰ (2SD) for coin 1813, 0.20 ± 0.12‰ (2SD) for coin 1863 and 0.28 ± 0.22‰ (2SD) for coin 1878. These values are accurate, within uncertainties, with respect to those obtained after chemical purification of Cu from gold chips drilled from these coins, and measured as solutions with MC-ICP-MS in “wet” plasma mode (Table 3 and Fig. 2).⁵⁹ Regarding our three samples, the wet method uncertainties (repeatability ∼0.1‰, 2SD, n = 10), obtained after sampling, acid digestion, ion chromatography extraction and dilution are often more precise compared to those from the present in situ fs-LA-MC-ICP-MS method using the same pooling approach (repeatability ∼0.2‰, 2SD). However, the 2SD of our individual δ⁶⁵Cu analyses are rather ∼0.3‰, and thus even less precise.

Fig. 2 Copper isotopic results for coins 1813, 1863 and 1878 measured against Au3 in-house standard and recalculated relative to SRM-976. Error bars are the 2SE for each of the triplicated sample-standard bracketing analysis and shaded areas represent expected values ± 2SD (from previous wet MC-ICP-MS analysis after wet chemistry Cu purification⁵⁹).

4. Discussion

4.1. Comparison with previous works

Previous studies involving the in situ determination of δ⁶⁵Cu were focused on copper-rich materials such as metals and ores. The greater the concentration of copper in a material, the easier it is to obtain precise and accurate isotopic measurements because matrix effects are minimised for major elements.⁵⁸ With a UV nanosecond LA-MC-ICP-MS on chalcopyrite and with a matrix-matched standard, Graham et al.⁵² obtained a long term reproducibility for δ⁶⁵Cu of 0.12‰ (2SD) and Molnár et al.⁷⁰ had a precision between 0.4‰ and 1.2‰ (2SD). For the latter, the precision on NIST-978 was ± 0.24‰ (2SD, n = 14) against ± 0.06‰ for wet analyses. Ikehata et al.⁵⁵ were the first to experiment δ⁶⁵Cu analyses with a fs-LA-MS-ICP-MS (Near Infra Red/NIR LA) on sulfide minerals. Their reproducibility measured on an external standard (JMC Cu) yielded 0.05‰ (2SD) that was equivalent to that for the wet analyses they performed, whereas their precision on all their copper-rich material samples reached 0.14‰ (2SD). Ikehata et al.⁵⁴ and Ikehata and Hirata⁵⁶ stated that matrix-matched analysis was essential for copper-rich material analysed by both NIR and UV fs-LA-MC-ICP-MS, and that best precision of ∼± 0.09‰ (2SD) could be reached by UV fs-LA-MC-ICP-MS compared to NIR. Also with an in-house standard matrix-matched method and a UV laser, Bao et al.⁵³ obtained an excellent precision of ± 0.05‰ (2SD) for chalcopyrites, to be compared to a precision of ± 0.02‰ (2SD) with wet analysis. Lazarov and Horn⁵⁷ showed that UV laser long-term reproducibility of ± 0.08‰ (2SD) could be reached on SRM-976 as well as on copper-rich samples. But they also underlined that an uncertainty of ± 0.1‰ (2SD) could be reached without matrix-matched standard, in contrast to Ikehata et al.⁵⁴ and Ikehata and Hirata⁵⁶ conclusions (see also synthesis in Poitrasson et d'Abzac⁵⁸). Lazarov and Horn⁵⁷ showed that this can be achieved provided the energy density (fluence) of the laser is controlled and not too high. Accordingly, Lv et al.⁷¹ also reached similar uncertainties (∼0.10‰, 2SD) with a UV laser without matrix-matched standard for copper-rich minerals.

Only few studies focused on other materials than copper-rich ones, and even fewer with fs-LA, such as García-Poyo et al.⁶² on dried blood droplets for biomedical purposes. Regarding gold, Bendall⁴⁹ was the only one to conduct preliminary LA-MC-ICP-MS copper isotope analyses within gold matrices, specifically focusing on Celtic gold coins. However, it was rather a testing approach, with probably a ns-LA at the time (not specified in the text), yielding δ⁶⁵Cu uncertainties of up to 2‰ in a single sample, too large to be exploited in view of the smaller range of variations usually encountered in geological materials (e.g. ³⁷).

We thus conclude that the uncertainties obtained in this study, from 0.12‰ to 0.26‰ (2SD), for historical gold coin samples represent a significant improvement. Moreover, while dealing with material containing minor amounts of Cu, these uncertainties are also comparable to measurements of copper-rich matrices (e.g., copper-based metals or Cu-ore sulfides like chalcopyrite) using UV LA and matrix-matched method. As in the recent study of García-Poyo et al.⁶² on δ⁶⁵Cu, our fs-LA-MC-ICP-MS methodology has been pushed to the limits in order to obtain the best δ⁶⁵Cu results for such gold matrices. Comparing the performances of these LA-MC-ICP-MS measurements is however not always relevant because of the different matrices and Cu contents.

4.2. Signal behaviour during femtosecond laser ablation

We generally observed that the Cu signal intensity varies at the beginning of each ablation run (e.g. signal of ⁶³Cu for Au3 standard, Fig. 3a), but the Cu isotopic raw ratio remains rapidly stable after the very beginning of the ablation signal. We could explain the low Cu signal at the beginning of the ablation by the presence of a mixing spray chamber at the entry of the MC-ICP-MS torch, which dead volume and associated turbulence impose an equilibration (or mixing) time of about 30 s. or sometimes less. This is longer than the signal integration time (2 s), hence the signal start observed (Fig. 3). In rare cases the copper signal intensity was gradually decreasing over time after rapidly reaching a plateau (Fig. 3b). This was attributed to a problem of excessive defocusing of the laser beam, due to the manipulator having most likely kept the focus on the surface of the coin, forgetting to refocus the laser at the bottom of the pre-cleaning crater. Nevertheless, whatever the evolution of the signal intensity evolution, the raw ⁶⁵Cu/⁶³Cu ratio remained stable, thereby demonstrating the robustness of our fs-LA-MC-ICP-MS approach. This has been also observed by Graham et al.⁵² for chalcopyrite analysis. However, it is worth noticing that in the case of ⁶³Cu/⁶⁵Cu isotope ratio measurements in gold, the space charge effect on copper isotopes is expected to be more pronounced due to the high mass of gold (m/z = 197), compared to more copper-rich samples like chalcopyrite. Due to the increased isotopic fractionation, the analysis of copper isotopes in Au matrices is less ideal. Nevertheless, this study demonstrates that our method can adequately correct for this isotopic bias given the accuracy and precision reached with our measurements (Table 3 and Fig. 2).

Fig. 3 (a) Typical time-resolved signals of ⁶³Cu(V) and the raw ratio of ⁶³Cu/⁶⁵Cu for gold samples (here Au3). (b) An example of signal profile observed when the laser beam was probably focused above the bottom of the pre-cleaning crater (here on 1813 coin).

4.3. Depth and morphology of the laser ablation craters

For a given sample, the uncertainties of triplicated measurements can vary greatly depending on the analyses (Fig. 2 and Table 3). For instance, for the 1813 coin, the uncertainty over the four analyses (of three measurements each) varies between 0.11‰ (2SE) and 0.53‰ (2SE), while for the 1863 coin, over five analyses carried out, the uncertainty ranges between 0.10‰ and 0.94‰ (2SE). Having a closer look to the craters made by the fs-LA, it may be possible to understand these differences in 2SE for a given sample. As seen in Fig. 1b and c, the second crater made after the pre-cleaning is not always exactly centred. The particles sent to the ICP torch may in the latter case sometimes originate from the pre-cleaning crater slopes and not of the pre-cleaning bottom (Fig. 1c), in which case the particles analysed would come from the first 50 μm of the coin and might have a different isotopic signature due to possible surface alteration of the coin as discussed above. This could be caused by chemical surface processes (coin wearing or intentional gilding to hide debasement) that may translate into fractionation of the copper isotopic signature. This could otherwise be caused by the process described by d'Abzac et al.⁷² for iron isotopes, where the biggest particles can have a heavier isotopic signature due to preferential condensation – and those big particles would be the ones preferably deposited on the crater slopes. This means that if the analysis crater was better centred in the pre-cleaning crater, more homogeneous material would be ablated and the measurement 2SE would be smaller. Another way to simplify this centring would be to widen the pre-cleaning crater. However, these analyses are still bound to their micro-invasive nature and they should be as minimally visible as possible to the naked eye. Widening the pre-cleaning would thus be in contradiction with this precept.

Crater depths (±5 μm) were obtained using SEM, based on the difference in WD (working distance) obtained after focusing the top and then the bottom of each crater. The depth measurements obtained are reproducible. The pre-cleaning depth measures ∼65 μm and the analysis crater ∼65 μm, so that the total ablation depth is ∼130 μm. The objective of avoiding the first 50 μm, that might contain fractionated copper isotope signatures, has therefore been achieved. Regarding the crater shape, the pre-cleaning crater measures in average (mean on a triplicate) 148 × 166 ± 3 μm and the inner crater 87 × 96 ± 3 μm (see Fig. 1d). The laser system was programmed to obtain a 150 × 150 μm square for the pre-cleaning and an 80 × 80 μm square for the analysis craters, with a 10 μm interval between ablation lines. There is a slight difference with a more rectangular shape obtained eventually, as the repeated ablation lines – well visible with the waves they generate on the edges (Fig. 1a) – are 16 of them, rather than the expected 15 (150 μm/10 μm) (see ESI† 1 for an image demonstration on a copper matrix). The inverted truncated pyramid shape of the crater, i.e. its slope sides are a result of both redeposition of the ablated particles (notably for pre-cleaning crater walls, Fig. 1c and d) and gradual defocusing of the laser beam through passes over time.

4.4. Limitations of the fs-LA-MC-ICP-MS method for δ⁶⁵Cu measurement in gold matrices

Despite the difficulty of measuring Cu isotopes in a gold matrix given the relatively huge quantities of gold sent into the plasma during the laser ablation, the accuracy and reproducibility obtained in this study remain suited for archaeological applications (see Section 4.5). Nevertheless, some limitations of the method may be drawn.

Matrix effects are inherent to LA-MC-ICP-MS analyses. Even though the first ionisation potential of Au is ∼1.5 eV higher than that of Cu,⁷³ it may induce a Cu isotopic bias effect due to the large amount of Ar plasma energy used by this Au matrix, along with space charge effects at the ICP interface. In brief, the space charge effect refers to the off-axis repulsion of isotopes in the ICP-MS interface due to their positive charge. This particularly affects light isotopes due to their lower kinetic energy, which then creates a mass bias that is incompletely compensated for by ionic lenses before the ions enter the mass spectrometer compartment.^74–7740Ar⁺ being the most abundant isotope in the plasma, it is mostly responsible for this mass bias. However, in our case, a large amount of gold is introduced into the plasma and ¹⁹⁷Au⁺ is likely to enhance the space charge effect with respect to ⁶³Cu and ⁶⁵Cu isotopes. It is therefore less favourable to measure copper isotopes in Au matrices due to increased mass bias. Moreover, this dominant Au matrix will use a large fraction of the Ar plasma energy which will reduce the overall instrument sensitivity for the analytes of interest such as Cu.

As a consequence, there is a limit to the minimum content of Cu present in the gold coins to achieve the uncertainties on δ⁶⁵Cu measurements reported in this study. For example, measurements were attempted on a Dinar coin with 0.1 wt% Cu and using Au1 standard (0.3 wt% Cu) as bracketing standard (ref. 59 for more info on these samples). But, while at least 2 V of signal are required on ⁶³Cu in order to achieve precise and accurate δ⁶⁵Cu measurements, it was not possible to obtain more than 1 V on the Dinar, despite all the laser parameters tested. In line with the average ⁶³Cu signals obtain for the more Cu-rich coins 1813, 1863 and 1878 of the study (with between 4.5 to 9.6 wt% Cu, see above), we can propose that the minimum Cu content in the gold to achieve δ⁶⁵Cu measurements at the precision reported should be at least 4 wt%. For Fe or Pb isotope analyses in gold matrices, minimum contents are also required. And as the Fe and Pb contents were even lower (<0.1 and <0.01 wt%) than for Cu in these gold coins, Fe and Pb isotopes could only be analysed using the wet method.⁵⁹

Comparison with ns-LA should be conducted to validate the true need for a fs-LA with our matrix-matched standards,⁷⁸ unfortunately, no ns-LA was available to us during the data acquisition.

Another concern is the size and shape of the craters. They are markedly smaller compared to the holes drilled with a 1.5 mm bit for wet analyses (Fig. 4b and 5). However, if one takes a closer look at them, their “squareness” may be less acceptable than a rough circle of the same size – like the crater from the ns-LA used for the elemental characterisation (Fig. 4a). This latter is more likely to be taken as a wear mark to the naked eye. The sharp shapes of the fs-LA could catch the eye more easily (Fig. 5). In fact, the size of the ns-LA crater is nearly the same as that the fs-LA inner analysis crater (Fig. 4). Whereas widening the pre-cleaning crater is good for a better reproducibility, it would lead the ablation to become too visible compared to nanosecond ablation crater. However, to overcome this visual problem, this squared crater shape could easily be turned into a disk given the goniometric system used in this work that allows to produce any desired shape.^61,62,64

Fig. 4 SEM – SE images of different sampling modes in the gold coin Dinar. (a) Crater made by a nanosecond excimer laser for elemental analyses. (b) Comparison between the hole drilled with a 1.5 mm drill bit performed for wet MC-ICP-MS analysis⁵⁹ and ablation tests made with the fs-LA. Note that the small craters (red arrow) are the same size as those in Fig. 1. The biggest craters (green arrow) are 300 μm wide tests. Also note that the nanosecond crater (a) is approximately the same size (∼80 μm) as the inner square craters used for fs-LA isotope analysis (b).

Fig. 5 Optical microscope surface observation of the Dinar, displaying three sampling methods. Two fs-LA craters (box), one ns-LA crater and two holes drilled with a 1.5 mm drill bit.

4.5. Applicability of the fs-LA-MC-ICP-MS method for δ⁶⁵Cu measurement in gold for archaeological research

As copper isotopic composition may give information on the type of gold ore, it will reflect the types of mineralisation associated to a specific geodynamic context. They are yet very few studies for Cu isotopes in gold ores. While the Bulk Silicate Earth (BSE) δ⁶⁵Cu is measured at 0.07 ± 0.10‰ (2SD),⁷⁹ gold ores can range from −1‰ to +1‰ in diverse European gold ores.^21,49 Some gold objects have been analysed and their δ⁶⁵Cu ranged from −0.5‰ to +0.5‰ for gold artefacts from Mesopotamia and Georgia⁸⁰ and from 0.3‰ to 2.6‰ for medieval coins.⁵⁹ On the basis of these first results, we can consider that the 0.2‰ precision obtain with our analyses is sufficient for archaeological and historical gold studies. Moreover, some comparison can be made with the Cu isotope data existing on other metallic mineralisation (such as copper mineralisations): copper isotope signature in ores can range from around −15‰ to around +10‰.^37,81 Broadly speaking, positive values indicate oxide minerals, negative values indicate secondary sulfide minerals while values near zero indicate primary sulfide minerals.^82–86 Moreover, the insight provided by copper isotopes on the ore origin may be further enhanced when combined with other isotopic data, such as Pb isotopes. For instance, some gold ores may have the same model age calculated from Pb isotopes, but could be differentiated using contrasted Cu isotope signatures.^21,24,59

5. Conclusion

Copper isotope ratio measurements by LA-MC-ICP-MS had never been achieved in a gold matrix before with precisions and accuracies sufficient for historical and archaeological studies. Here we obtained with fs-LA precisions (repeatability of 4 to 5 triplicated sample-standard bracketing measured independently over a few days) ranging from 0.12‰ to 0.26‰ (2SD) for δ⁶⁵Cu analysis in gold coins. The use of a matrix-matched and isotopically characterised in-house gold standard allowed these precise and accurate Cu isotopic analyses with Cu concentrations ranging from 4.5 to 9.6 wt%. Future investigations will have to evaluate whether matrix-matched standards are necessary since non-matrix-matched standards have been shown to be effective using femtosecond laser for LA-MC-ICP-MS analysis for different matrices.

Our results on 19th c. gold coins pave the way for studies of older gold coins, along with other types of gold materials, and for authentication analyses. The micro-sampling performed by laser ablation leaves minimal damage on these precious ancient artefacts.

Data availability

The data is already in the manuscript as part of the discussion.

Conflicts of interest

There are no conflicts of interest to declare.

Acknowledgements

Authors would like to thank the Mission for Intersectional and Interdisciplinary Initiatives (MITI) from the CNRS (French National Center for Scientific Research) via “80|PRIME” and “Défi Isotopes” projects for funding Louise de Palaminy's PhD thesis. Analyses were conducted at the instrumental platforms PAMAL (Platform for Trace Metal Analysis by Laser Ablation) and ISOLOURD (isotopic measurements of “heavy” elements by MC-ICP-MS) from the Instrumental Service Centre UPPA-Tech (Pau, France). We also thank Maryse Blet-Lemarquand from IRAMAT-CEB (Orléans, France) for the coins and standard elemental analyses, Loïc Perrière from ICMPE (Thiais, France) for making the standard Au3 and Anne-Marie Cousin from GET. We are also grateful to Dr. Christopher Standish (University of Southampton) for his extensive revisions and comments on Louise de Palaminy’s PhD thesis chapter which led to this article. The two anonymous referees are also thanked for their reviews.

References

1. M. Blet-Lemarquand, S. Nieto-Pelletier and F. Téreygeol, Bulletin de la Société Française de Numismatique, 2014, vol. 69.
2. M. F. Guerra, in Radiation in Art and Archaeometry, ed. D. C. Creagh and D. A. Bredley, Elsevier., 2000, pp. 378–416.
3. M. F. Guerra, T. Calligaro and A. Perea, Archaeometry, 2007, 49, 53–74.
4. M. Nordez, B. Armbruster, M. Blet-Lemarquand, S. Fürst, N. Lockhoff, P.-Y. Milcent, S. Nieto-Pelletier, L. Olivier, M. Schönfelder, R. Schwab and S. Sievers, in Matières premières en Europe au 1^er Millénaire av. n. è. Exploitation, transformation, diffusion. La Europa de las materias primas en el Ier milenio a.n.e. Explotación, transformación y difusión. Actes du 45^e colloque de l’AFEAF (Gijón, 13-15 mai 2021), ed. L. Valdès, V. Cicolani and E. Hiriart, 2023, pp. 23–39.
5. M. Blet-Lemarquand, S. Nieto-Pelletier and B. Gratuze, in Mines, Metals, and Money Ancient World Studies in Science, Archaeology and History, The Royal Numismatic Society Special Publication, 2020.
6. M. Blet-Lemarquand, A. Suspène and M. Amandry, in Archaeometallurgy in Europe III: Proceedings of the 3rd International Conference, Deutsches Bergbau-Museum Bochum, June 29–July 1, 2011, ed. A. Hauptmann and D. Modarressi-Tehrani, 2015, pp. 107–113.
7. L. B. Brostoff, J. J. González, P. Jett and R. E. Russo, J. Archaeol. Sci., 2009, 36, 461–466.
8. A. Gondonneau, M. F. Guerra and J.-N. Barrandon, ArcheoSciences, 1996, 20, 23–32.
9. R. Kovacs, S. Schlosser, S. P. Staub, A. Schmiderer, E. Pernicka and D. Günther, J. Anal. At. Spectrom., 2009, 24, 476.
10. M. F. Guerra, S. Fernández, I. Delaveris, J. O. Encuentra and J. Villanueva, J. Archaeol. Sci. Rep., 2019, 24, 967–977.
11. M. F. Guerra and I. Tissot, Nucl. Instrum. Methods Phys. Res., Sect. B, 2013, 306, 227–231.
12. L. G. Troalen, J. Tate and M. F. Guerra, J. Archaeol. Sci., 2014, 50, 219–226.
13. R. J. Chapman, R. C. Leake, R. A. Warner, M. C. Cahill, N. R. Moles, C. A. Shell and J. J. Taylor, Appl. Geochem., 2006, 21, 904–918.
14. A. Ehser, G. Borg and E. Pernicka, Eur. J. Mineral., 2011, 895–910.
15. A. Gondonneau and M. F. Guerra, ArcheoSciences, 2000, 24, 27–38.
16. A. Gondonneau and M. F. Guerra, Archaeometry, 2002, 44, 573–599.
17. S. Nieto and J.-N. Barrandon, Rev. Numis., 2002, 6, 37–91.
18. J. Ogden, Hist. Metall., 1977, 11, 53–70.
19. C. Sillon, L’or monnayé dans le Nord de la Gaule. Recherches sur les monnaies d’or frappées dans le Nord de la Gaule entre le IIIe et le Ier siècle avant notre ère, PhD thesis, Université d'Orléans, 2014.
20. S. Baron, C. G. Tămaş, B. Cauuet and M. Munoz, J. Archaeol. Sci., 2011, 38, 1090–1100.
21. S. Baron, C. G. Tămaş, M. Rivoal, B. Cauuet, P. Télouk and F. Albarède, Sci. Rep., 2019, 9, 17816.
22. S. Baron, C. Le-Carlier, J. Carignan and A. Ploquin, Appl. Geochem., 2009, 24, 2093–2101.
23. A.-M. Desaulty, P. Telouk, E. Albalat and F. Albarede, Proc. Natl. Acad. Sci. U. S. A., 2011, 108, 9002–9007.
24. M. Jansen, A. Hauptmann and S. Klein, Metalla, 2016, 22, 75–146.
25. M. Kałaska, R. Mathur, G. Kamenov, J. Chyla, P. Prządka-Giersz and M. Giersz, Archaeometry, 2022, 64, 1168–1186.
26. D. J. Killick, J. A. Stephens and T. R. Fenn, Archaeometry, 2020, 62, 86–105.
27. L. J. Molofsky, D. Killick, M. N. Ducea, M. Macovei, J. T. Chesley, J. Ruiz, A. Thibodeau and G. C. Popescu, J. Archaeol. Sci., 2014, 50, 440–450.
28. F. Nocete, R. Sáez, M. R. Bayona, J. M. Nieto, A. Peramo, P. López, J. I. Gil-Ibarguchi, N. Inácio, S. García and J. Rodríguez, J. Archaeol. Sci., 2014, 41, 691–704.
29. F. Nocete, R. Sáez, A. D. Navarro, C. San Martin and J. I. Gil-Ibarguchi, J. Archaeol. Sci., 2018, 92, 87–102.
30. C. Standish, B. Dhuime, R. Chapman, C. Coath, C. Hawkesworth and A. Pike, J. Anal. At. Spectrom., 2013, 28, 217–225.
31. C. D. Standish, B. Dhuime, C. J. Hawkesworth and A. W. G. Pike, Proc. Prehist. Soc., 2015, 81, 149–177.
32. Z. A. Stos-Gale, G. Maliotis, N. H. Gale and N. Annetts, Archaeometry, 1997, 39, 83–123.
33. E. Balliana, M. Aramendía, M. Resano, C. Barbante and F. Vanhaecke, Anal. Bioanal. Chem., 2013, 405, 2973–2986.
34. N. H. Gale, A. P. Woodhead, Z. A. Stos-Gale, A. Walder and I. Bowen, Int. J. Mass Spectrom., 1999, 184, 1–9.
35. S. Klein, G. P. Brey, S. Durali-Müller and Y. Lahaye, Archaeol. Anthropol. Sci., 2010, 2, 45–56.
36. S. Klein, Y. Lahaye, G. P. Brey and H.-M. Von Kaenel, Archaeometry, 2004, 46, 469–480.
37. R. Mathur, S. Titley, G. Hart, M. Wilson, M. Davignon and C. Zlatos, J. Archaeol. Sci., 2009, 36, 430–433.
38. D. Berger, G. Brügmann, J. Bunnefeld and E. Pernicka, Archaeometry, 2021, 64, 44–74.
39. J. Milot, M. Coustures, F. Poitrasson and S. Baron, Archaeometry, 2021, 64, 1–23.
40. J. Milot, S. Baron and F. Poitrasson, J. Archaeol. Sci., 2018, 98, 22–33.
41. J. Milot, F. Poitrasson, S. Baron and M.-P. Coustures, J. Archaeol. Sci., 2016, 76, 9–20.
42. T. Rose, P. Télouk, S. Klein and H. R. Marschall, J. Archaeol. Sci., 2019, 101, 52–62.
43. M. Vaxevanopoulos, G. Davis, J. Milot, J. Blichert-Toft, C. Malod-Dognin and F. Albarède, J. Archaeol. Sci., 2022, 145, 105645.
44. F. Albarède, J. Blichert-Toft, F. Callataÿ, G. Davis, P. Debernardi, L. Gentelli, H. Gitler, F. Kemmers, S. Klein, C. Malod-Dognin, J. Milot, P. Télouk, M. Vaxevanopoulos and K. Westner, Archaeometry, 2020, 63, 142–155.
45. D. Berger, J. S. Soles, A. R. Giumlia-Mair, G. Brügmann, E. Galili, N. Lockhoff and E. Pernicka, PLoS One, 2019, 14, 1–46.
46. A. Mason, W. Powell, H. A. Bankoff, R. Mathur, M. Price, A. Bulatović and V. Filipović, J. Archaeol. Sci., 2020, 122, 105181.
47. M. Jansen, S. Aulbach, A. Hauptmann, H. E. Höfer, S. Klein, M. Krüger and R. L. Zettler, J. Archaeol. Sci., 2016, 68, 12–23.
48. S. A. Junk and E. Pernicka, Archaeometry, 2003, 45, 313–331.
49. C. Bendall, The Application of Trace Element and Isotopic Analyses to the Study of Celtic Gold Coins and their Metal Sources, Johann Wolfgang Goethe University-Frankfurt, PhD thesis, 2003.
50. C. Bendall, D. Wigg-Wolf, Y. Lahaye, H.-M. Von Kaenel and G. P. Brey, Archaeometry, 2009, 51, 598–625.
51. A. Hauptmann and S. Klein, ArcheoSciences, 2009, 75–82.
52. S. Graham, N. Pearson, S. Jackson, W. Griffin and S. Y. O'Reilly, Chem. Geol., 2004, 207, 147–169.
53. Z. Bao, N. Lv, K. Chen, Y. Luan, X. Sun, C. Zong and H. Yuan, Geostand. Geoanal. Res., 2021, 45, 401–418.
54. K. Ikehata, K. Notsu and T. Hirata, Econ. Geol., 2011, 106, 307–316.
55. K. Ikehata, K. Notsu and T. Hirata, J. Anal. At. Spectrom., 2008, 23, 1003.
56. K. Ikehata and T. Hirata, Anal. Sci., 2013, 29, 1213–1217.
57. M. Lazarov and I. Horn, Spectrochim. Acta B Atom Spectrosc., 2015, 111, 64–73.
58. F. Poitrasson and F.-X. d'Abzac, J. Anal. At. Spectrom., 2017, 32, 1075–1091.
59. L. de Palaminy, F. Poitrasson, S. Baron, M. Blet-Lemarquand and L. Perrière, J. Anal. At. Spectrom., 2024, 39, 1302–1321.
60. M. Aramendía, L. Rello, S. Bérail, A. Donnard, C. Pécheyran and M. Resano, J. Anal. At. Spectrom., 2014, 30, 296–309.
61. F. Claverie, Développement et applications d’un système laser femtoseconde infra-rouge basse énergie et haute cadence de tir pour l’analyse d’éléments trace dans les solides par couplage ablation laser/ICPMS, PhD thesis, Université de Pau et des Pays de l'Adour, 2009.
62. C. García-Poyo, S. Bérail, A.-L. Ronzani, L. Rello, E. García-González, B. Lelièvre, P. Cales, F. V. Nakadi, M. Aramendía, M. Resano and C. Pécheyran, J. Anal. At. Spectrom., 2021, 36, 968–980.
63. L. Martin, A. Galy, G. Barbotin, F. Claverie, E. Pons-Branchu, C. Tribolo, N. Mercier and C. Pécheyran, Anal. Chem., 2022, 94, 3046–3055.
64. M. Monperrus, C. Pécheyran and V. Bolliet, Appl. Sci., 2020, 10, 2463.
65. M. Blet-Lemarquand, M. Bompaire and C. Morrisson, Rev. Numis., 2010, 6, 175–198.
66. M. Blet-Lemarquand and F. Duyrat, in White Gold Studies in Early Electrum Coinage, The American Numismatic Society New York, The Israel Museum Jerusalem, ed. van Alfen P. and Wartenberg U., 2020, pp. 337–378.
67. R. J. Chapman, D. A. Banks, M. T. Styles, R. D. Walshaw, S. Piazolo, D. J. Morgan, M. R. Grimshaw, C. P. Spence-Jones, T. J. Matthews and O. Borovinskaya, Miner. Deposita, 2021, 56, 1563–1588.
68. E. B. Melchiorre and J. Henderson, Ore Geol. Rev., 2019, 109, 348–357.
69. F. Claverie, A. Hubert, S. Berail, A. Donard, F. Pointurier and C. Pécheyran, Anal. Chem., 2016, 88, 4375–4382.
70. F. Molnár, I. Mänttäri, H. O'Brien, Y. Lahaye, L. Pakkanen, B. Johanson, A. Käpyaho, P. Sorjonen-Ward, M. Whitehouse and G. Sakellaris, Ore Geol. Rev., 2016, 77, 133–162.
71. N. Lv, Z. Bao, K. Chen, K. Wu and H. Yuan, Sci. China Earth Sci., 2022, 65, 2005–2017.
72. F.-X. d'Abzac, B. L. Beard, A. D. Czaja, H. Konishi, J. J. Schauer and C. M. Johnson, Anal. Chem., 2013, 85, 11885–11892.
73. I. T. Platzner, Modern Isotope Ratio Mass Spectrometry, Wiley, Chichester, reprinted, 1997.
74. Y. Yu, E. Hathorne, C. Siebert, M. Gutjahr, J. Fietzke and M. Frank, Chem. Geol., 2024, 662, 122220.
75. S. D. Tanner, Spectrochim. Acta B Atom Spectrosc., 1992, 47, 809–823.
76. M. Rehkämper, M. Schönbächler and C. H. Stirling, Geostand. Newsl., 2001, 25, 23–40.
77. F. Albarède, P. Telouk, J. Blichert-Toft, M. Boyet, A. Agranier and B. Nelson, Geochim. Cosmochim. Acta, 2004, 68, 2725–2744.
78. K. P. Jochum, B. Stoll, U. Weis, D. E. Jacob, R. Mertz-Kraus and M. O. Andreae, Geostand. Geoanal. Res., 2014, 38, 265–292.
79. P. S. Savage, F. Moynier, H. Chen, G. Shofner, J. Siebert, J. Badro and I. S. Puchtel, Geochem. Persp. Let., 2015, 1, 53–64.
80. M. Jansen, Geochemie und Archäometallurgie des Goldes der Bronzezeit in Vorderasien, PhD thesis, Fakultät für Geowissenschaften der Ruhr-Universität Bochum, 2019.
81. F. Moynier, D. Vance, T. Fujii and P. Savage, Rev. Mineral. Geochem., 2017, 82, 543–600.
82. Y. Kim, I. Lee, S. Oyungerel, L. Jargal and T. Tsedenbal, Ore Geol. Rev., 2019, 104, 656–669.
83. G. Markl, Y. Lahaye and G. Schwinn, Geochim. Cosmochim. Acta, 2006, 70, 4215–4228.
84. W. Powell, R. Mathur, H. A. Bankoff, A. Mason, A. Bulatović, V. Filipović and L. Godfrey, J. Archaeol. Sci., 2017, 88, 37–46.
85. W. Powell, R. Mathur, A. H. Bankoff, J. John, O. Chvojka, M. Tisucká, A. Bulatović and V. Filipović, J. Archaeol. Sci., 2018, 93, 216–221.
86. M. Jansen, A. Hauptmann, S. Klein and H.-M. Seitz, Archaeol Anthropol Sci, 2017, 10, 1485–1502.

---

Footnotes

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d4ja00217b

‡ Now also at the Instituto de Geociencias, Universidade de Brasilia, Campus Darcy Ribeiro, 70910-900, Brasilia, Brazil.

§ Now at IPREM, UMR 5254, CNRS, Université de Pau et des Pays de l’Adour, Avenue de l’Université, BP 576 64012 Pau, France.

---