Antimony and Nickel Impurities in Blue and Green Copper Pigments

: Impurities in paint layers executed with green and blue copper pigments, although relatively common, have been studied only little to date. Yet, their proper identiﬁcation is a powerful tool for classiﬁcation of paintings, and, potentially, for future provenance studies. In this paper, we present analyses of copper pigments layers from wall paintings situated in the vicinity of copper ore deposits (the palace in Kielce, the palace in Ciechanowice, and the parish church in Chotk ó w) located within the contemporary borders of Poland. We compare the results with the analyses of copper minerals from three deposits, two local, and one historically important for the supply of copper in Europe, i.e., Miedzianka in the Holy Cross Mountains, Miedzianka in the Sudetes, and, as a reference, Špania Dolina in the Slovakian Low Tatra. Optical (OM) and electron microscopy (SEM-EDS), Raman spectroscopy, and laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS) have been used for a detailed investigation of the minute grains. Special attention has been devoted to antimony and nickel phases, as more unusual than the commonly described iron oxides. Analyses of minerals from the deposits helped to interpret the results obtained from the paint samples. For the ﬁrst time, quantitative analyses of copper pigments’ impurities have been described. microscopy with energy-dispersive spectrometry; Backscatter Diffraction; (p)XRF—(portable) ﬂuorescence; MAXRF—X-ray ﬂuorescence radiation; XRD—X-ray powder diffraction; MAXRD—X-ray powder diffraction FTIR—micro-Fourier rFTIR—external reﬂection range reﬂectance FORS—ﬁber optic reﬂectance PIXE—particle induced X-ray emission; elements in bold—antimony and nickel impurities have been reported only rarely.

The polymetallic ore deposit in Miedzianka in the Sudetes is situated in the eastern part of the metamorphic envelope of the Karkonosze granitoid massif [44]. Ore mineralization mainly occurs in amphibolites, amphibolite, or quartz-sericite schists. The main ore minerals include: chalcopyrite, bornite, chalcocite, and minerals from the tetrahedrite group [45]. There are many secondary copper minerals in the weathering zone of the Miedzianka deposit, mainly arsenates and phosphates [46]. Malachite and azurite are rare. Malachite forms thin crusts on the surfaces of rocks containing weathered copper sulfides.
The Špania Dolina deposit is situated in the Starohorské Mountains (central Slovakia). Primary copper mineralization occurs in quartz-siderite-sulfide veinlets and as impregnations. Mineralized zones are present in the Permian sandstones and conglomerates [47]. The main copper ore mineral is the variety of tetrahedrite. This mineral is accompanied by chalcopyrite and variable amounts of other ore minerals [48]. There are many secondary minerals in the oxidation zone-mainly copper arsenates and sulfates [49]. The weathering zone of the Špania Dolina deposit is famous for large accumulations of copper carbonates. Malachite occurs as globular and reniform aggregates, up to several cms in diameter. This type of malachite was used to produce a high-quality green pigment. Azurite forms tiny crystals that form crusts on rock surfaces. Sometimes one may come across larger, cryptocrystalline azurite aggregates, with a diameter of up to several cms. Noteworthy is the high activity of antimony in the weathering zone of the Špania Dolina deposit [49]. This element is released during weathering of the tetrahedrite series and can form its own secondary minerals. It can also be incorporated into the crystal lattices of other minerals.
Analysis of trace minerals present in paint samples is a challenging task, mainly due to the small amounts of material (limited number of samples, small samples, thin paint layers, only few grains of impurities present, particularly minute grains of impurities-in this case, the grains ranged from 1 to 7 µm). Heterogenous composition of paint layers (mixed organic and inorganic components, variable elemental composition, variable size of the measured grains) is another complication that needs to be overcome, and is frequently mentioned by researchers [36]. Moreover, as the objects of art were exposed to adverse environmental conditions over a prolonged period, they were affected by degradation processes. Especially pigments in wall paintings are exposed to considerable humidity variations, as well as salts and microorganisms; together, these lead to alterations of minerals [12,50]. Hence, identification of the mineral present in the paint layer may be difficult, and identification of the primary mineral used by the artist is even more complicated. All of the studied samples were collected from wall decorations, and some of them suffered visible color alterations.
Unfortunately, minute size of the impurities made it impossible to examine them through an optical microscope, thus making it impossible to draw conclusions about their color and influence on the final hue of the paint layer. Based on energy dispersive spectrometry, their elemental composition was determined quantitively, but due to the limited space resolution of Raman spectroscopy coupled with optical microscopy, the Minerals 2021, 11, 1236 4 of 22 molecular structure still needs to be addressed. Nevertheless, we consider presenting reliable data on elemental composition of trace elements and their compounds present in copper-rich paint layers as a valuable contribution to technical art history.

Materials and Methods
Micro-samples of green and blue paint layers and samples of minerals for stratigraphic studies were embedded in epoxy resin, Epofix (Struers), and polished on a mechanical grinder-polisher Labo-Pol 2 (Struers ApS, Ballerup, Denmark) using silicon carbide waterproof abrasive foils at successive grits of 180, 320, 500, 1000, 2000, and 4000. For the final diamond polishing, suspensions of 3 and 1 µm were used.
Cross-sections were further studied in visible light using an Olympus CX41 microscope with an Olympus UC30 camera (Olympus Corporation, Tokyo, Japan).
Elemental composition measurements and initial mineral identification were performed using a Jeol scanning electron microscope (SEM) JSM-6510LV (JEOL Ltd., Tokyo, Japan) equipped with an energy dispersive spectrometer (EDS) by Oxford Instruments (Abingdon, UK), with a silicon drift detector (X-act, Oxford Instruments, Abingdon, UK). The first EDS semi-quantitative measurements were carried out with no standardization in low vacuum mode under the pressure of 50 Pa and accelerating voltage of 20 kV, working distance of 10 mm, and acquisition time from 35 to 100 s.
After this preliminary analysis, samples of paint layers with more unusual impurities were selected for further studies, including molecular identification of the main copper pigment, and of bigger grains of impurities in the layers, by means of Raman spectroscopy. µRaman measurements were performed with a Nicolet Almega Dispersive Raman Spectrometer (XR, Thermo Fisher Scientific, Waltham, MA, USA), equipped with an Olympus confocal microscope and motorized stage. Depending on the properties of the sample, a 780 or 532 nm excitation line was used. Power of the laser was reduced to 25-50% to avoid sample overheating. Typical exposure time was 15-45 s. Each spectrum was collected from two scans. Spectral resolution was approximately 2 cm −1 .
Small grains of impurities in copper-based paint layers were subsequently analyzed with a ZEISS SIGMA VP field emission scanning electron microscope (Carl Zeiss Microscopy GmbH, Oberkochen, Germany) equipped with two energy-dispersive X-ray spectrometers Bruker XFlash 6|10 (SEM-EDS) (Bruker, Karlsruhe, Germany). The analyses were conducted under a 10 kV accelerating voltage, 0.77 nA beam current, and 7.7 mm working distance. The counting times applied were 300 s for elemental mappings, and 60 s for point analyses. The most interesting impurities were mapped. Samples were then repolished and carbon-coated, with coating thickness of 20 nm, to ensure good electrical conductivity. EDS was recalculated to chemical composition using NIST DTSA-II public domain software by the National Institute of Standards and Technology (NIST). For standard-based quantification of the analyzed material, sets of metal (SPI 44 metals) and mineral standards (SP088-Cameca geological set) were used. NIST DTSA-II allows two kind of standards, for composition, and for peak-reference. Metal standards for elements were used as composition standard only if no mineral standard for a given element was in the set, otherwise, it was used only as peak-shape standard. Composition standards were as follows: Ca-diopside, Ni from metal Ni, Cu from metal Cu, As from ZnAs 2 , Si from albite, Pb from PbTe, In from metal In, Sb from metal Sb, Sn from SnO, P from apatite. For the K52 sample, we present semi-quantitative data without normalization to 100%, from Bruker-Esprit software; the results are expressed as oxides wt%. This sample contains Br, and the available standard of Br-Tl(Br,I) is not stable under the electron beam and might contaminate the vacuum chamber.
Samples of minerals were further analyzed with Laser Ablation Inductively Coupled Mass Spectrometry (LA-ICP-MS) to detect trace elements. An NexION 300D Inductively Coupled Plasma Mass Spectrometer (PerkinElmer, Waltham, MA, USA) equipped with LSX-213 laser ablation system (CETAC, Omaha, NE, USA) was used during the measurements, with Argon (Ar) as carrier gas. Three ablation lines were selected on the surface of the sample (E = 5.0 mJ; φ = 100 µm; SR = 50 µm/s; 20Hz). Transient signals were recorded and evaluated for subsequent elemental quantification for the selected isotopes ( 23 S, 35 Cl, 39 K, 43 Ca, 45 Sc, 49 Ti, 51 V, 53 Cr, 55 Mn, 57 Fe, 59 Co, 61 Ni, 65 Cu, 66 Zn, 71 Pt,197 Au, 208 Pb, 209 Bi, 232 Th, 238 U). The LA-ICP-MS signals were background-corrected and integrated using Microsoft Excel. Standard glass NIST SRM 610 was used as the external standard. A rough estimation of elemental samples inhomogeneity was approximated by 2D map reconstruction based on 3-line ablation.
The unpolarized µRaman spectra of minerals were collected in back-scattered geometry with a Horiba Labram HR (Horiba, Ltd., Kyoto, Japan) spectrometer integrated with an Olympus BX 40 confocal microscope. The parameters of Raman measurements were as follows: estimated analytical spot size of~1-2 µm, 532 nm line of a solid-state Nd-YAG laser (10 mW), 1800 g/mm grating, microscope objective 100×, acquisition time of 20, and accumulation of 2 scans. The spectra were recorded in the range of 4000-50 cm −1 on a randomly oriented surface of crystals mounted in epoxy resin in 0.7-inch plate that was used also for SEM-EDS studies.

Copper Pigments Paint Layers
During the first stage, samples of paint cross-sections were studied with optical and electron microscopy. Based on the qualitative and semi-quantitative elemental composition obtained by SEM-EDS, further analyses were planned to identify the molecular structure of the main components and impurities. In the second stage, X-ray intensity maps were collected from the small antimony and nickel-rich grains, to establish elemental distribution within the grains. Finally, quantitative EDS measurements of the grains were performed, and the results are summarized in Tables 1 and 2 Contrary to initial expectations, only one sample with azurite was identified in the room. In sample K52 (Figure 1), azurite (399/400, 1093 cm −1 ) with individual grains of malachite (270, 431 cm −1 ), hematite (~292, 224 cm −1 ), and calcite (1086 cm −1 ) were detected by means of Raman spectroscopy. In all the other blue areas, smalt was present in the layer dated to the first decoration, ca 1640. In the subsequent overpaints, a mixture of lead white, charcoal black, Prussian blue, and artificial ultramarine were identified by EDS, µRaman spectroscopy, and optical microscopy. Copper pigments were detected in the green areas; in most of the samples, malachite (178, 217, 268, 431, 534, 1058, 1098 cm −1 ) with, probably, cuprite (150, 210 cm −1 ) and azurite (398, 1094 cm −1 ) as impurities. Copper sulphates-posnjakite (444, 505, 974 cm −1 ) were detected, mixed with malachite (177, 269, 431 cm −1 ) and impurities of hematite (220, 291, 406 cm −1 ). Thus, posnjakite should rather be interpreted as an alteration product of malachite, and not the initially used mineral [12,30]. • Piano nobile of the Ciechanowice Palace (before 1945, Rudelstadt) near the polymetallic deposit in the Rudawy Janowickie (the Sudetes).
The 16th century decoration of the piano nobile is rich in copper pigments. In the green areas of the oldest painted decoration, malachite with typical impurities of iron oxides was detected. High amounts of silica were found during EDS analyses; some clusters of silica and copper minerals were detected as well. In the blue areas, azurite (153, 170, 248, 399, 1094 cm −1 ) with impurities of rutile (447, 607 cm −1 ), goethite (246, 300, 388, 500 cm −1 ), and calcite (1086 cm −1 ) were identified. Azurite paint layers were particularly rich in titanium and iron oxides with manganese and phosphorous impurities. Less common grains of barium sulphate were also found. In one case, EDS measurements revealed zinc arsenate or copper-zinc arsenate. In samples G5 and G4 (Figure 2), collected from the bluegreen areas of a layer belonging to a later decoration, Raman spectroscopy measurements did not identify the copper component detected with EDS-probably due to pigment degradation [51]; only calcite (1085 cm −1 ) was detectable.  Malachite (179, 215, 268, 432, 1101 cm −1 ) and tenorite (294, 603 cm −1 ) were identified in the green and black leaves of the vegetal decoration of the church's vaults. In this case, tenorite is an alteration product of malachite, resulting from the action of oxalic acid as metabolic product of microorganisms, or the influence of alkaline environment on finely ground malachite grains [27,50,52,53], because, most probably, the painting was executed The 16th century decoration of the piano nobile is rich in copper pigments. In the green areas of the oldest painted decoration, malachite with typical impurities of iron oxides was detected. High amounts of silica were found during EDS analyses; some clusters of silica and copper minerals were detected as well. In the blue areas, azurite (153, 170, 248, 399, 1094 cm −1 ) with impurities of rutile (447, 607 cm −1 ), goethite (246, 300, 388, 500 cm −1 ), and calcite (1086 cm −1 ) were identified. Azurite paint layers were particularly rich in titanium and iron oxides with manganese and phosphorous impurities. Less common grains of barium sulphate were also found. In one case, EDS measurements revealed zinc arsenate or copper-zinc arsenate. In samples G5 and G4 (Figure 2), collected from the blue-green areas of a layer belonging to a later decoration, Raman spectroscopy measurements did not identify the copper component detected with EDS-probably due to pigment degradation [51]; only calcite (1085 cm −1 ) was detectable. • Piano nobile of the Ciechanowice Palace (before 1945, Rudelstadt) near the polymetallic deposit in the Rudawy Janowickie (the Sudetes).
The 16th century decoration of the piano nobile is rich in copper pigments. In the green areas of the oldest painted decoration, malachite with typical impurities of iron oxides was detected. High amounts of silica were found during EDS analyses; some clusters of silica and copper minerals were detected as well. In the blue areas, azurite (153, 170, 248, 399, 1094 cm −1 ) with impurities of rutile (447, 607 cm −1 ), goethite (246, 300, 388, 500 cm −1 ), and calcite (1086 cm −1 ) were identified. Azurite paint layers were particularly rich in titanium and iron oxides with manganese and phosphorous impurities. Less common grains of barium sulphate were also found. In one case, EDS measurements revealed zinc arsenate or copper-zinc arsenate. In samples G5 and G4 (Figure 2), collected from the bluegreen areas of a layer belonging to a later decoration, Raman spectroscopy measurements did not identify the copper component detected with EDS-probably due to pigment degradation [51]; only calcite (1085 cm −1 ) was detectable. • Floral decoration of the vaults in the Chotków church (before 1945, Hertwigswaldau), near the Legnica-Głogów copper deposit.
Malachite (179, 215, 268, 432, 1101 cm −1 ) and tenorite (294, 603 cm −1 ) were identified in the green and black leaves of the vegetal decoration of the church's vaults. In this case, tenorite is an alteration product of malachite, resulting from the action of oxalic acid as metabolic product of microorganisms, or the influence of alkaline environment on finely ground malachite grains [27,50,52,53], because, most probably, the painting was executed ground malachite grains [27,50,52,53], because, most probably, the painting was executed in fresco technique. This degradation is clearly visible in the samples where the green layer meets the black layer (see Figure 3b). Chemical analyses of samples Ch 33 and Ch 35 detected calcium oxalate-weddellite (1475, 909 cm −1 ), which would support the biodegradation hypothesis. Influence of high temperatures is also reported to be the cause of degradation of azurite to tenorite [26], but no evidence of fire in the church was found, so this cause may be ruled out. Calcite (1131, 1085, 492 cm −1 ) and gypsum (1007 cm −1 ) were also detected in the copper pigments layer.
Minerals 2021, 11, x FOR PEER REVIEW 7 of 23 in fresco technique. This degradation is clearly visible in the samples where the green layer meets the black layer (see Figure 3b). Chemical analyses of samples Ch 33 and Ch 35 detected calcium oxalate-weddellite (1475, 909 cm −1 ), which would support the biodegradation hypothesis. Influence of high temperatures is also reported to be the cause of degradation of azurite to tenorite [26], but no evidence of fire in the church was found, so this cause may be ruled out. Calcite (1131, 1085, 492 cm −1 ) and gypsum (1007 cm −1 ) were also detected in the copper pigments layer.

Antimony and Nickel Impurities Analysis
For the second phase of the research, samples containing unusual antimony and nickel impurities were selected. By means of electron microscopy, X-ray intensity mapping of individual grains and semi-quantitative (sample K52) and quantitative (samples G4, Ch33, Ch35) elemental point measurements were performed (see Tables 1 and 2). EDS measurements revealed the presence of areas of lead concentration in azurite crystals-probably inclusions of cerussite, Pb(CO3), a mineral associated with azurite in copper deposits [54]. In some cases, the lead-rich areas also contained antimony ( Figure  4). Along with relatively small amounts of antimony (~3 wt% Sb2O3), elevated concentrations of lead (~20-25 wt% PbO), arsenic (~20 wt% As2O5), and bromine (~2-4 wt% Br2O) were detected. Other elements detected in the antimony-bearing grains were silicon, potassium, and, probably, phosphorus. The elemental composition may be indicative of a mixture. A possible mineral could be murdochite PbCu6O8-x(Cl,Br)2x-a common associate of malachite. Presence of cerussite, bindheimite (Pb2Sb2O6O), and an arsenate is not ruled out. Mindat.org lists iron and phosphorous among common impurities of olivenite, a copper arsenate, while for bindheimite, it lists arsenic, bismuth, calcium, iron, and sodium [54]. Moreover, bindheimite is an associate of azurite, malachite, and cerussite [54]. Substitution of Pb by Sb in bindheimite, as well as bindheimite hosting copper and arsenic has already been reported [55]. This mixture of minerals is, however, a vague hypothesis and would need to be confirmed by supplementary studies. Presence of copper sulphosalts would be another possibility, yet no sulphur was found in the EDS analysis.

Antimony and Nickel Impurities Analysis
For the second phase of the research, samples containing unusual antimony and nickel impurities were selected. By means of electron microscopy, X-ray intensity mapping of individual grains and semi-quantitative (sample K52) and quantitative (samples G4, Ch33, Ch35) elemental point measurements were performed (see Tables 1 and 2). EDS measurements revealed the presence of areas of lead concentration in azurite crystals-probably inclusions of cerussite, Pb(CO 3 ), a mineral associated with azurite in copper deposits [54]. In some cases, the lead-rich areas also contained antimony ( Figure 4). Along with relatively small amounts of antimony (~3 wt% Sb 2 O 3 ), elevated concentrations of lead (~20-25 wt% PbO), arsenic (~20 wt% As 2 O 5 ), and bromine (~2-4 wt% Br 2 O) were detected. Other elements detected in the antimony-bearing grains were silicon, potassium, and, probably, phosphorus. The elemental composition may be indicative of a mixture. A possible mineral could be murdochite PbCu 6 O 8-x (Cl,Br) 2x -a common associate of malachite. Presence of cerussite, bindheimite (Pb 2 Sb 2 O 6 O), and an arsenate is not ruled out. Mindat.org lists iron and phosphorous among common impurities of olivenite, a copper arsenate, while for bindheimite, it lists arsenic, bismuth, calcium, iron, and sodium [54]. Moreover, bindheimite is an associate of azurite, malachite, and cerussite [54]. Substitution of Pb by Sb in bindheimite, as well as bindheimite hosting copper and arsenic has already been reported [55]. This mixture of minerals is, however, a vague hypothesis and would    In sample G4, EDS measurements revealed the presence of an antimony-rich grain (G4/4) with over 20 wt% Sb 2 O 3 content, and grains (G4/1, G4/2) with antimony as trace element. The antimony-rich grain (G4/4) also contained a high percentage of nickel and copper. Moreover, some lead, tin, and arsenic were detected. The elemental composition may be indicative of an oxidized mineral zlatogorite (CuNiSb 2 ), or a mixture with breithauptite (NiSb), with lead and tin. Tin was the main component of one grain with antimony as trace element (G4/2); this could be cassiterite (SnO 2 ), and as for the other (G4/1) a lead-tin oxide [56] or a mixture of minerals containing lead, tin, and copper could be expected.

•
Samples Ch33 and Ch35 from the church in Chotków.
In the samples from Chotków, minute antimony-bearing grains were quite numerous. All the measured particles contained, apart from often-elevated concentrations of antimony (two grains with the content of 25.83 and 29.71 wt% Sb 2 O 3 ), different amounts of lead (ranging from 0.39 to 49.73 wt% PbO). In all measurements, nickel was detected, in variable amounts (0.18-46.27 wt% NiO). Again, as in the case of the Ciechanowice Palace, a Cu-Sb-Ni grain was detected (compare Ch35/1 and G4/4 in Table 2). Although most of the studied grains had irregular shapes, the grain measured as Ch35/1 ( Figure 5) was characterized by angular, rod-like shape. In one measurement (Ch35/2), antimony was associated with lead and arsenic. Elemental composition of the grain in sample Ch33 and grain Ch35/5 is similar to bindheimite, with arsenic, copper, and silicon as trace elements. Grain Ch35/4 consisted mainly of nickel and copper, and lower amounts of silicon and calcium.

Trace Elements and Impurities in Minerals from Selected Copper Deposits
Concurrently with the examination of paint layers, samples of green and blue copper minerals from two possible local sources (Piotr mine shaft in Miedzianka in the HCM, and Segen Gotttes mine in Miedzianka in the Rudawy Janowickie Mts.-the Sudetes), and a third, more distant, but historically significant source, Piesky mine in Špania Dolina in the Slovakian Low Tatra (formerly Hungary), were examined. The mineralogical research was far from exhaustive but provided preliminary referential data for comparative analyses.

Trace Elements and Impurities in Minerals from Selected Copper Deposits
Concurrently with the examination of paint layers, samples of green and blue copper minerals from two possible local sources (Piotr mine shaft in Miedzianka in the HCM, and Segen Gotttes mine in Miedzianka in the Rudawy Janowickie Mts.-the Sudetes), and a third, more distant, but historically significant source, Piesky mine in Špania Dolina in the Slovakian Low Tatra (formerly Hungary), were examined. The mineralogical research was far from exhaustive but provided preliminary referential data for comparative analyses.
In the first step, samples of minerals were measured as fragments, without grinding, with EDS and LA-ICP-MS, and during the next step, cross-sections from the fragments were prepared, mapped, and measured with EDS. Finally, selected cross-sections, with significant concentrations of antimony detected with EDS were studied with point µRaman spectroscopy to characterize antimony-bearing phases. Unfortunately, due to the minute size of the particles, only the matrix in which they were embedded was identified.
Slovakian copper ore deposits, as well as Hungarian and Romanian, are reported to be dominated by Cu-Sb-As sulphosalts [57]. Therefore, we expected a high content of antimony and arsenic in the samples from Špania Dolina (samples 13-15, Table 4). However, preliminary studies showed inclusions of antimony in samples from Miedzianka in the Sudetes and Miedzianka in the HCM, but no antimony in the samples from Špania Dolina.
In the antimony-bearing grains of sample No. 3 (Miedzianka in the Sudetes), EDS measurements revealed, apart from antimony, high concentrations of arsenic, copper, iron, and small amounts of zinc, calcium, silicon, and aluminum ( Figure 6). In the matrix, mainly copper was detected, with arsenic and zinc, suggesting a Cu-Zn arsenate with traces of lead, calcium, and silicon. In sample No. 9 (Miedzianka in the HCM), the predominant elements of the antimony-bearing grains were, again, arsenic, copper, lead, iron, while calcium, zinc, and silicon were detected in low concentrations. The matrix contained mainly copper with arsenic, and calcium, a small amount of zinc, and traces of lead and silicon, indicative of a Cu-Ca-Zn arsenate. In both cases, lead was in the structure of antimony-bearing grains. Nickel was found as trace element (≥1 wt%) in samples No. 5, 6, 7, and 11-all from the HCM deposit.
traces of lead, calcium, and silicon. In sample No. 9 (Miedzianka in the HCM), the predominant elements of the antimony-bearing grains were, again, arsenic, copper, lead, iron, while calcium, zinc, and silicon were detected in low concentrations. The matrix contained mainly copper with arsenic, and calcium, a small amount of zinc, and traces of lead and silicon, indicative of a Cu-Ca-Zn arsenate. In both cases, lead was in the structure of antimony-bearing grains. Nickel was found as trace element (≥1 wt%) in samples No. 5, 6, 7, and 11-all from the HCM deposit.  (Tables 3  and 4). The results were consistent with the EDS results; indeed, highest values of antimony content were observed in sample 3 (three areas of the sample were measured, and labeled as 3a and 3b (green area), 3n (blue area)), and 9. Yet in samples 5 and 10, some antimony concentrations were also recorded. Maps calculated based on LA-ICP-MS results showed considerable heterogeneity of the mineral samples ( Figure 7). Thus, it is probably more representative to consider the maximal values than the mean. Antimony was usually, but not always, correlated with arsenic. The calculated correlation coefficient values of antimony were quite high for Sb-Pb (0.60-0.86) and Sb-As (0.68-0.82), but lower for Sb-Ni (0.37-0.45). Correlation coefficient of Sb-Cu could be high: up to 0.86 but could also be negative. Table 3. Sb and Ni content in copper-based minerals coming from three different deposits, measured by LA-ICP-MS expressed in wt%, normalized to 100%.  (Tables 3 and 4). The results were consistent with the EDS results; indeed, highest values of antimony content were observed in sample 3 (three areas of the sample were measured, and labeled as 3a and 3b (green area), 3n (blue area)), and 9. Yet in samples 5 and 10, some antimony concentrations were also recorded. Maps calculated based on LA-ICP-MS results showed considerable heterogeneity of the mineral samples ( Figure 7). Thus, it is probably more representative to consider the maximal values than the mean. Antimony was usually, but not always, correlated with arsenic. The calculated correlation coefficient values of antimony were quite high for Sb-Pb (0.60-0.86) and Sb-As (0.68-0.82), but lower for Sb-Ni (0.37-0.45). Correlation coefficient of Sb-Cu could be high: up to 0.86 but could also be negative.     Samples 3 and 9 were measured with Raman spectroscopy in order to identify the matrix in which antimony-bearing crystals were present-the spectra are shown in Figures 8-10. Results obtained from the studied samples were compared with the data from the literature. Two Cu-arsenates-olivenite and conichalcite-were identified in both localities (Miedzianka in the Sudetes, and the HMC).

Sample No 121 Sb 61 Ni
In the spectra (Figure 8), we observed one strong band at~851-850 cm −1 attributed to AsO 4 symmetric stretching vibrations (v1), the bands at~525 and 495 cm −1 related to AsO 4 symmetric bending vibrations (v4), and bands at 278 and 213 cm −1 attributed to Me-O stretching vibration [58]. The band of the hydroxyl-stretching region of olivenite is visible at 3479 cm −1 (Figure 9). Figure 8 shows the spectra of olivenite and adamite from the RRUFF database (olivenite R050532, adamite R060593), two arsenates forming solid solutions. The maximum position of the bands can be indicative of substitutions in the chemical composition. The band at 3550 cm −1 is characteristic of pure adamite, and 3440 cm −1 of pure olivenite. In the studied case, it can be concluded that olivenite has a substitution of zinc, considering the shift of the bands towards lower wavelengths. Minerals 2021, 11, x FOR PEER REVIEW 14 of 23     In the spectra (Figure 8), we observed one strong band at ~851-850 cm −1 attributed to AsO4 symmetric stretching vibrations (v1), the bands at ~525 and 495 cm −1 related to AsO4 symmetric bending vibrations (v4), and bands at 278 and 213 cm −1 attributed to Me-O stretching vibration [58]. The band of the hydroxyl-stretching region of olivenite is visible at 3479 cm −1 (Figure 9). Figure 8 shows the spectra of olivenite and adamite from the RRUFF database (olivenite R050532, adamite R060593), two arsenates forming solid solutions. The maximum position of the bands can be indicative of substitutions in the chemical composition. The band at 3550 cm −1 is characteristic of pure adamite, and 3440 cm −1 of pure olivenite. In the studied case, it can be concluded that olivenite has a substitution of zinc, considering the shift of the bands towards lower wavelengths. Figure 10 shows spectra of conichalcite from the two localities. They were also compared with the standard spectrum of conichalcite (RRUFF R040167). The very strong band at 831 cm −1 and weak band at 812 cm −1 (v1) are related to the substitution of Zn for Cu in the structure of the studied minerals (solid solution conichalcite-austinite (CaZn(AsO)4)(OH)). Even a small admixture of the elements in the structure can contribute to the change of the position of the band maximum.

Discussion
In this paper, results of rare copper pigments impurities containing antimony and nickel were reported. Grains of Pb-Al-Br-As-Cu-Sb, Pb-Sn-Cu (with traces of Ni and Sb), Pb-Sb-Cu-Ni, Pb-As-Sb, Sn-Cu (with traces of Sb and Ni), Cu-Ni-Sb (with variable content of Pb and Sn), and Ni-Cu-Ca-Pb-Si were identified in paint layers containing copper pigments. In the mineral specimens, grains containing lead and antimony were found in the olivenite-adamite and conichalcite-austinite matrix in samples from Miedzianka in the HCM and the Sudetes.
The deposit in Miedzianka-Ciechanowice, Rudawy Janowickie (samples 0-3, Table  4), had already been studied by Siuda et al. [38]. Among the phases which, because of the presence of antimony, nickel, or accessory elements such as arsenic, could be helpful for the interpretation of results of paint layers analyses, annabergite (Ni3(AsO4)2·8H2O), minerals from tetrahedrite group, and arsenates, such as bayldonite (PbCu3(AsO4)2(OH)2), cornwallite (Cu5(AsO4)2(OH4) ), duftite (PbCu(AsO4)OH), clinoclase (Cu3AsO4(OH)3), Figure 10. Reference spectra of conichalcite, with the spectra measured in samples 3 and 9. Figure 10 shows spectra of conichalcite from the two localities. They were also compared with the standard spectrum of conichalcite (RRUFF R040167). The very strong band at 831 cm −1 and weak band at 812 cm −1 (v1) are related to the substitution of Zn for Cu in the structure of the studied minerals (solid solution conichalcite-austinite (CaZn(AsO) 4 )(OH)). Even a small admixture of the elements in the structure can contribute to the change of the position of the band maximum.

Discussion
In this paper, results of rare copper pigments impurities containing antimony and nickel were reported. Grains of Pb-Al-Br-As-Cu-Sb, Pb-Sn-Cu (with traces of Ni and Sb), Pb-Sb-Cu-Ni, Pb-As-Sb, Sn-Cu (with traces of Sb and Ni), Cu-Ni-Sb (with variable content of Pb and Sn), and Ni-Cu-Ca-Pb-Si were identified in paint layers containing copper pigments. In the mineral specimens, grains containing lead and antimony were found in the olivenite-adamite and conichalcite-austinite matrix in samples from Miedzianka in the HCM and the Sudetes.
In the previous studies of Miedzianka deposit in the HCM (samples 5-11, Table 4), antimony was identified, along with sulphur, copper, and arsenic in the structure of miedziankite, a copper sulphosalt of the tetrahedrite and tennantite series solid solution [59]. Arsenic was identified also in arsenates (adamite, austinite, conichalcite, cornubite, olivenite, tyrolite, zincolivenite). Nickel was reported to be present in gersdorffite (NiAsS), and asbolane ((Ni,Co) 2-x Mn 4+ (O,OH) 4 nH 2 O) [54]. In the analysis of the soils of the former mining area, elevated concentrations of copper were strongly correlated with arsenic and nickel [60]. This is in accordance with our study, as nickel was detected as trace element by EDS, and LA-ICP-MS results confirmed higher values of 61 Ni in the samples from Miedzianka in the HCM compered to Miedzianka in the Sudetes and Špania Dolina.
As shown above, antimony may be found in all three copper deposits, mostly in tetrahedrite series. Our study confirmed the presence of antimony-rich grains in both Miedzianka (in the Sudetes and the HMC). In the case of the samples from Špania Dolina, antimony traces were detected by LA-ICP-MS. Nickel-bearing minerals were reported in the literature in both Miedzianka (in the Sudetes and the HMC), while, according to mindat.org [54], no nickel phases were identified in the Piesky mine in Špania Dolina. Nevertheless, in a nearby copper deposit in L'ubietová several nickel-bearing species were confirmed (annabergite, asbolane, gersdorffite, nickelskutterudite, siegenite). Thus, the sole presence of antimony or nickel phases cannot be regarded as indicative of the pigment's origin. However, it might help exclude some possibilities, such as the Piesky mine, when elevated concentrations of nickel occurs. The high correlation coefficient between antimony and lead (see Table 4) in the studied minerals might be indicative of bindheimite, while the high correlation coefficient between antimony and arsenic suggests a tetrahedrite group.
Antimony and nickel impurities have been reported only rarely (Table 5). In the case of the studied paint layers' samples (see Tables 1 and 2), antimony is almost always present in association with lead and, usually, arsenic. This is in accordance with the results of Mastrotheodoros et al. [68]. They reported, apart from Zn arsenates and barite, identification of antimony grains (Sb (Crete, 16th c), (unknown, 18th c); Sb-Pb (Crete 16th c); Sb-Zn-As, Sb-Cu-Pb (Crete, 17th c)) and grains containing nickel (Ni-Zn-As (Crete)) in azurite layers. In malachite layers, they reported finding, apart from Zn arsenates, barite, and dolomite, grains containing antimony (Zn-As-Sb-P (Ionian Islands, 17th c), Zn-As-Sb (Ionian Islands, 18th c)) and nickel (Zn-As-Ni (unknown, 18th c)). In Italian wall paintings, antimony and arsenic impurities were reported together with zinc [34], or barium [19]. In our study, zinc was not detected in the paint layers but was present in the matrix in mineral specimens in the form of olivenite-adamite and conichalcite-austinite. Smieska et al. [34] correlated antimony and barium with azurite from a French manuscript. Nickel was also reported by Vagnini et al. [66] but interpreted as impurity of iron oxides. In this paper, we present impurities containing both elements, nickel and antimony together.
The presence or absence of impurities is used as an argument when determining the natural or artificial origin of the pigment [12]. The presence of some elements, such as zinc for brass [15] or tin for bronze, might be indicative of the artificial origin of copper pigments produced from metal alloys. Concurrent presence of copper, nickel, and tin, observed in the case of samples from the Ciechanowice Palace, may be linked with the use of bronze as the source of copper in the production of the artificial green copper pigment, verdigris, or a kind of calcium copper acetate, like in the recipes reported by Scott [2] and Krekel et al. [73]. In the blue-green paint layer, copper and calcium were detected as the main elements, while Raman spectroscopy detected calcite only, and copper pigment remained undetermined. Small inclusions of Pb-Sb-Cu-O, Cu-As-O, Cu-Sb-O, Sb-P-Cu-O, Sb-Pb-Cu-As-O, or Pb-As-Sb in bronze objects were described, which suggested that As-rich copper ores, such as copper sulphosalts of the tetrahedrite and tennatite series or copper arsenates could have been used as the source of copper [57]. In the case of Ciechanowice, the artificial origin of the copper pigment is likely, but to confirm it studies of reconstructions of old recipes for copper pigments would be needed.
Identification of accompanying phases in copper-bearing paint layers is often subject to theoretical speculations about the provenance of the pigment. Scrivano et al. [66] described two different mineralogical associations of azurite in the studied polychromy of marble fragments, and, based on historical data and accompanying phases (lead minerals, monazite), identified the Calabona mine in Sardinia as a possible provenance for one type of azurite. Nevertheless, these are ubiquitous phases with over 800 localities in Europe [54]; thus, the hypothesis needs to be confirmed by additional research. Hradil et al. [18] identified grains of malachite, olivenite, and, possibly, adamite in a sample of azurite, and, based on olivenite's presence in the deposit, pointed to L'ubietová as a possible provenance. However, olivenite is a common mineral phase, with 370 localities in Europe listed by mindat.org [54]. Based on historical data, the Slovakian origin of the pigment is probable, but in this case the chemical analyses cannot be regarded as conclusive evidence. Nevertheless, they may certainly be considered as one of the arguments to support the hypothesis about the Slovakian origin. Buzgar et al. [13] identified conichalcite and pointed to German provenance of the pigment and dolerophanite indicating Vesuvius sources. As shown in this study, conichalcite is ubiquitous (with more than 150 localities in Europe [54]), and it cannot be indicative of a specific region. The dolerophanite is indeed a rare mineral, thus it could be indicative of a particular type of deposit. Finding an impurity or a set of impurities of a copper pigment rare enough to be specific for a particular ore deposit or particular mine is a challenge. Moreover, not all historic mines have been studied, as they are often inaccessible. They were not of interest to the scientists if they had not had any economic significance. Thus, we do not have a good comparative database for provenance studies. As presented in this paper, a random study of minerals is not conclusive.
Therefore, it seems that, at the current state of knowledge, it is more appropriate to use impurities to group paintings than to trace the provenance of the pigment. Buzgar et al. [13] discerned two different workshops on the basis of conichalcite and dolerophanite presence. Smieska et al. [34] correlated azurite with silver and high concentration of barium with a Spanish manuscript, and iron, antimony, barium without zinc, arsenic, zirconium, and bismuth with a French manuscript. Švarcová et al. [12] classified the wall paintings of the Sázava monastery into groups. Berrie et al. [19] cited analyses of paintings attributed to Giotto, in which arsenic, antimony and barium were detected in areas painted with copperbased pigments, and herself identified a copper-bismuth arsenate, mixite, in an azurite layer exceptionally free from other impurities. It shows that specific impurities can be a helpful tool in establishing the artist's choice of color, and, in this respect, the authorship, or simply in the studies of technical art history. It is important to remark that they provide hints not only about the source of the pigment, but also about the preparation, as all mineral pigments underwent purification. This should also be taken into consideration when grouping paintings.
In our study, the sample of copper pigment from Cracow Bishops' Palace in Kielce differs from the others as it contains bromine, aluminum and elevated concentrations of iron, and it does not contain tin nor nickel. Bromine has never been measured in the Miedzianka HMC deposit, but we cannot completely exclude its presence in low quantities. Among European localities with copper and bromine, Mindat.org lists the Ore Mountains in Czechia, Baden-Württemberg, Rhineland-Palatinate, and Saxony in Germany [54]. In this case, we may suggest with caution that the pigment was probably purchased from some more remote source than the local deposit. Our study showed some similarities between impurities of copper pigments in Ciechanowice and Chotków. In both places, tin was detected, and Cu-Ni-Sb phase was present (see G4/4, Ch35/1 in Table 2). Yet, there are also important differences between the two paintings, such as the type and morphology of the copper pigment. Furthermore, the concentrations of tin were much higher in samples from Ciechanowice Palace, most likely indicating the artificial origin of the pigment produced from bronze alloy. Some of the above mentioned impurities might be overlooked during research, e.g., nickel in XRF analyses, due to the presence of impurities in the X-ray tube [71]. In EDS analysis, bromine might be misinterpreted because of possible overlap between the most characteristic X-ray peaks of Br and Al in low energies: Al Kα1 (1.487 keV)-Br Lα1 (1.480 keV). Lead is a common component of various pigments and is usually attributed to lead white, a pigment often mixed with copper pigments to achieve a lighter hue [36]. Nevertheless, cerussite and hydrocerussite are natural associates of malachite and azurite [54]. In sample K52, from the former Cracow Bishops' Palace in Kielce, lead areas are embedded inside the crystals of azurite (Figure 1), while in sample Ch35 (Figure 3), in a Cu-Ni-Sb crystal, which is clear evidence of an impurity. Antimony is often interpreted as an artificially produced pigment-Naples yellow (the equivalent of naturally occurring mineral-bindheimite). Still, it does not necessarily have to be correlated with lead, although most often is, as shown in our study. Our research proved that antimony might be accompanied by arsenic (antimony grains were found in olivenite/adamite and conichalcite matrix in mineral specimens), but also by nickel (in Cu-Ni-Sb grains in copper pigments), and tin and copper (Sn-Cu grains with variable content of Sb).

Conclusions
Analyses presented in this paper show how challenging it is to identify the accessory phase in a thin layer of paint. The problems encountered are mainly as follows: minute samples, small number of impurities, and low spatial resolution of laboratory devices. In this respect, macro-scanning methods, such as MAXRF or MAXRD would definitely be more efficient than point analyses. However, since they are more precise, point analyses of micro-samples, provide useful information for the interpretation of macro-scanning methods. As demonstrated above, impurities might be used to group paintings, define the artist choice of colors, determine whether the pigment was of natural or artificial origin, and, to some extent, support the hypothesis of the pigment's origin but, at the current state of the art, are usually not conclusive about its provenance. Comparison of results obtained from paint layers and from minerals helped to better understand the composition of the impurities and the matrix in which they might be embedded. Studies of minerals from the known sources have once again shown the risks researchers might face when provenance studies are based on the presence of more or less ubiquitous impurities.