https://orcid.org/0000-0002-4199-0253
Figures
Abstract
A selection of metallurgical slag artifacts from the Early Bronze Age site of Tepe Hissar, Iran, were characterized using X-ray computed tomography (XCT), X-ray fluorescence (XRF), X-ray diffraction (XRD), and optical and scanning electron microscopy with energy-dispersive spectrometry (SEM-EDS). The CT scans were used to identify regions of interest and internal features for sectioning, including pores, high-density inclusions, and differences in slag density and porosity distribution. The scans revealed internal features and patterns; however, contiguous metal-rich regions and thick surface minerals masked some features and misrepresented others. This study demonstrates how XCT enables the identification and collection of salient diagnostic information from slag artifacts before sectioning, providing a prescience of internal features and a volumetric 3D record of each artifact. After studying the 3D images, the artifacts were precisely sectioned and examined using optical microscopy, SEM-EDS, and XRD, revealing primary pyrometallurgical phases and secondary mineralizations, stratified slag layers, leaching channels, and internal microenvironments. Secondary precipitates and mineral parageneses within these environments are discussed, referencing the archaeological record, soil salinity, and Pourbaix diagrams. An explanation for the presence of speiss in some slag artifacts from Hissar is also detailed.
Citation: Sabatini B, Allanore A (2025) A novel application of X-ray computed tomography towards the characterization and interpretation of phase formations, mineral parageneses, and internal features in ancient copper slag from Tepe Hissar, Iran. PLoS One 20(11): e0336603. https://doi.org/10.1371/journal.pone.0336603
Editor: Hayrunnisa Nadaroglu, Ataturk University: Ataturk Universitesi, TÜRKIYE
Received: August 28, 2025; Accepted: October 28, 2025; Published: November 11, 2025
Copyright: © 2025 Sabatini, Allanore. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: Reduced chemical data tables are given within the manuscript. The entire dataset is included in the attached zip file “Appendices.zip”. All chemical and thermodynamic data are included in Excel format in the the zip file. All code and libraries used to produce the diagrams are given in the Markdown file in the zip file.
Funding: The author(s) received no specific funding for this work.
Competing interests: The authors have declared that no competing interests exist.
Introduction
X-ray computed tomography, or, more colloquially, CT scanning, in archaeology and related fields, has dramatically increased in recent years since its first application to an Egyptian mummy in 1977 [1,2]. Since then, along with more nuanced internal scanning methods such as neutron diffraction [3–5](e.g., Cantini et al., 2024; Fedrigo et al., 2023; Festa et al., 2020), XCT has increased in frequency and application despite being logistically challenging to facilitate and cost-prohibitive [1,6–11]. In the last few years, newly imagined applications of CT, well beyond tangentially related medical purposes, have increased in number and expanded in scope to the study of ceramics [6], ancient glass [12], metal and metal statuary, mainly consisting of copper- and iron-based alloys [6,8,10,13–15], various human and animal remains, including but not limited to mummies [1,16,17], and prolifically, tree rings using high-resolution µ-CT to determine age and species [18–24].
The technique’s increased application and diverse use in the humanities and social sciences in recent years is undoubtedly due to its ability to impart internal structural information non-destructively [1,6,11,14,17]. This aspect of XCT is well-suited for studying irreplaceable cultural objects, spurring the development of cost-reducing, custom-made XCT apparatuses, software, and portable scanners designed for on-site analyses. A handful of such devices have been developed to overcome size constraints, for instance, when scanning a large, dense cast-iron engine block, or, conversely, nanoscale samples of wood [8,24], all while offsetting the costs of medical-grade CT scanners and the impracticality of transporting invaluable and delicate artifacts to medical facilities [7,9,10,13,25]. These recent trends indicate a growing desire to utilize CT across various fields, and with the advent of industrial scanners, it has become possible to produce higher-quality scans at lower cost with greater versatility and adaptability.
Contrary to its hallmark, this study employs XCT as a preliminary step in the destructive analysis of slag artifacts. Three-dimensional volumetric images may provide invaluable information on internal diagnostic pyrotechnological features, such as charcoal, prills, and porosity, as well as their relationship to external phenomena, including corrosion and cracks. This method may also enable the documentation of artifacts and the precise targeting of features for sectioning. Without these data, metallurgical slags are typically cut, and portions are pulverized after surface observations and chemical analyses have been performed with optical microscopy and X-ray Fluorescence (XRF) [26–28], respectively. Following preliminary characterization, sets or typologies and representative specimens are selected based on inferential sectioning processes, where superficially observed and sometimes chemically analyzed features, such as color, texture, geometry, and corrosion patches, serve as indicators of internal structures and chemical phenomena. After sectioning, fortuitously discovered pores, inclusions, minerals, and metal phases are then documented and studied [29–32].
Archaeological slag, however, is a highly inhomogeneous and altered product, containing glassy, crystalline, and fused phases, unreacted ore minerals, varied textures [26,29,32–37], and a host of secondarily precipitated minerals and compounds [34,38–42]; their mineralogical and pyrometallurgical descriptions are, therefore, incredibly challenging. And while the inferential technique described above has been adequate, as this study will show, the lack of prescience regarding the specimens’ internal structure inherently means that even well-informed inferential sectioning will miss diagnostic features such as metal droplets, overlaid surfaces, and trapped inclusions. The implications are that processual and procedural information may be unaccounted for, or multiple sections might be needed to achieve the same level of knowledge imparted by structurally detailed 3D volumetric images. In short, destructive analyses of slags that rely on traditional investigative techniques may not capture the most salient aspects of the pyrotechnological processes and conditions that produced them. Indeed, the surface of slags, having been exposed to various environmental conditions over several centuries or even millennia, alters, resulting in external crusts and coatings that may be unreliable indicators of internal structures and compound and mineral paragenesis [34,39].
The slags analyzed in this study originate from Tepe Hissar, Iran, and had been studied, as part of a larger corpus, by Thorton (2009), who determined their date by radiocarbon to periods IIB–IIIB (c. 3350–3000 BCE) and the late Chalcolithic to Early Bronze Age (Table 1) [43–47]. These dates were further refined based on pottery typologies to ca. 3100–2900 BCE [48]. Several of the slags were found to contain small speiss prills [43,49], which, by modern-day definition, are an undesirable byproduct; however, archaeologically, they are a significant outcome [50–53]. Understanding the processual circumstances leading to their formation, presence, and origins was the impetus for revisiting the slag artifacts; however, the focus of this article is primarily on the application of XCT. Nevertheless, speiss is discussed in a limited capacity because it has been a point of contention in the field, particularly concerning these specific artifacts. Speiss in archaeometallurgical contexts is defined as a di-iron arsenide (Fe2As), iron arsenide (FeAs), or iron di-arsenide (FeAs2) that lacks base metal and is theorized to have been intentionally produced, traded, and used as an additive to make arsenical copper as early as the 4th millennium BCE [44,47,49,52,54–58]. Understanding its formation in ancient smelting contexts is therefore of the utmost importance for deciphering prehistoric metallurgical practices; however, this question will only be briefly addressed in this work [59–61].
A small selection of slags from Tepe Hissar, Iran, dating to the 4th millennium BCE, was loaned to Professor Allanore at MIT in 2022 by the University of Pennsylvania’s Penn Museum [LOAN AGREEMENT LO-2022–7 ST-2022-14], located in Philadelphia, PA. The Museum later granted permission to section three of the artifacts in 2023, corresponding to two types: H76-S37A and H76-S45B, which are assuredly rich in copper and covered in green corrosion, and H76-S39, which potentially has no copper but may contain speiss [43,44,49,62]. No permits were required for the described study, which complied with all relevant regulations. They originate from a collection of metallurgical materials excavated in 1976 with support from the Iranian Center for Archaeological Research in Tehran by the Penn Museum and the University of Turin under the direction of Robert H. Dyson and Maurizio Tosi [63,64]. From these materials, slags H76-S37A and H76-S45B also have embedded charcoal and large exposed voids and pores. Comparatively, slag H76-S39, a possible tap slag with a flowed and layered appearance, has a smooth upper surface and a rough underside with far smaller pores [27,30,65]. These slags were selected for XCT and sectioning. Their findspot details and dates are given in Table 1.
Materials and methods
Before CT scanning, following typical study practices, the slags were described by texture, color, shape, mass, size, and relative magnetism [29,30,32]. After these basic characterizations, their surfaces were chemically analyzed using X-ray fluorescence (XRF), followed by XCT to identify internal features and determine their relationship to external phenomena, such as pores and corrosion products. The scans showed clear density differences and internal patterning, which guided later sectioning. Following sectioning, the exposed surfaces of the slags were ground and polished, followed by the imaging and chemical analysis of select features using optical, X-ray diffraction (XRD), and scanning electron microscopy with energy-dispersive X-ray spectrometry (SEM-EDS). In brief, XCT was integrated into a metallurgical slag characterization and sampling workflow, enabling precise selection and sectioning of internal features. Specific features identified by XCT were studied in greater detail after the specimen was sectioned.
Surface elemental analysis
A Thermo Scientific Niton XL3t GOLDD+ portable XRF (pXRF) was used to acquire surface chemistry data. The device was set to “AllGeo” mode, and the data were exported using the provided NDT software. The acquired elemental data were later converted to oxides typically found in copper smelting slags [32,40,66]. The artifacts were analyzed on exposed flat surfaces, rather than over crevices or angular areas, avoiding obstructions where the analyzer’s objective window could sit flush. The XRF data compiled in this work are given in Table 2. They show noticeable variation between analyses, reflecting the typical compositional inhomogeneity of ancient slag and slag surfaces [30,65,67], and to some degree, the absence of several elements. Of note, slag H76-S39 had been analyzed previously using a Spectrolab 2000 ED-XRF and found to contain Na2O, TiO2, SrO, and BaO at 1.22, 0.33, 0.11, and 0.15, respectively [43], which were not detected in the current XRF analyses. This data is also shown in Table 2.
X-ray computed tomography
Each slag was packed into secure craft foam holders and imaged using a Lumafield Neptune CT scanner. Slags H76-S37A and -S45B were scanned at 120 kV for 151 min using a 1 mm Cu filter, having less dense, non-glassy outward appearances, with an exposure time of 8.2 s; H76-S39, being far larger with fewer pores, thus being more dense and possibly glassy, was scanned at 190 kV for 339 min with a 6 mm Cu filter and exposure time of 20.2 s. The acquired data consists of a series of two-dimensional (2D) radiograph projections arranged in sequence as the artifacts were rotated on top of the stage within the sample chamber [1,11,68]. The radiographs are then synthesized to create 3D representations, incorporating the relative density data of the scanned object in voxels, or 3D pixels. Since each voxel has an attenuation value, expressed in Hounsfield units (HU) within the total volume, they can be selectively isolated based on these values and viewed separately in Voyager, the accompanying browser-based software [69–71]. The contrasting voxels created a 3D internal map of the slags, showing the precise locations of internal features. These locations guided the sectioning of the slags.
Denser materials, such as metals, strongly attenuate X-rays and consequently exhibit higher HU values [4,14], while less dense materials, like voids or organic matter, are lower [3,10,72,73]. Based on their attenuation, several features of interest were chosen for sectioning, including those expected to be metal droplets, which show the highest relative density in the slags, and less dense regions thought to be charcoal, technical ceramics, or other low-density materials. Since CT scans only display differences in density, suppositions of the internal features were made based on prior studies of slag from the region [26,28,44,54,56,59,60,74,75], protruding features such as charcoal, and the 3D-rendered shapes and distribution of these features. The slags were sectioned with a diamond-tipped blade and water lubricant along precise X, Y, and Z coordinates. After sectioning, the surfaces were polished using kerosene down to 1 µm using diamond paste in preparation for imaging and chemical analysis.
Optical and EDS characterization
Optical images of the exposed features were captured in both light- and dark-field using a Leica DM/LM reflected light microscope (type 020–520.008), an attached Axiocam 305 color camera (5 megapixels), and ZEN Core 3.6 software at magnifications ranging from 50 to 500 µm. Following optical microscopy, additional images were collected, and chemical analyses were performed using a JEOL JSM-6610LV SEM operated at 10–20 keV, equipped with a Silicon Drift EDS detector. Chemical quantifications were performed using Iridium Ultra software.
Mineral and phase identification
After chemical analysis, the powder from each precision-cut sample was filtered and dried for XRD analysis using a PANalytical Empyrean diffractometer equipped with a GaliPIX3D detector. The device was equipped with a molybdenum MoKα radiation source (λ = 0.7107 Å) operating at 60 kV and 40 mA, scanning from 5 to 60° 2θ with a step size of 0.014 ° 2θ at a rate of 150 s/step for 22 minutes. Mineral and phase identification were performed using HighScore Plus (v. 5.1.0) with the ICDD PDF5 + Converted database, utilizing Rietveld refinement based on the major phases identified in each slag, given the proportion of amorphous phase in each powder sample. Refinement can significantly alter XRD spectra, potentially misrepresenting phases and minerals, necessitating corroborating chemical and structural analyses.
Phase and mineral alteration Eh-pH (Pourbaix) diagrams
The stability of pyrometallurgical phases and relic minerals in the slags, considering regional and localized environmental and macroscopic conditions, was thermodynamically calculated and presented in Eh-pH (Pourbaix) diagrams. The influence of soil salinity on secondary precipitate formation, based on the phases and minerals found in the slags, is highlighted in respective diagrams, as it dramatically affects paragenesis. Soil conductivity surveys conducted at Dāmghān in the Semnan province, Iran [76,77], provided an estimate for chloride activity (γ) and the soil solution’s ionic strength (IS), accounting for non-ideal behavior at higher concentrations. The salinity, measured at 4830 μS/cm, was converted to Total Dissolved Solids (TDS) by multiplying it by an empirical conversion factor of 0.65. Assuming the majority of dissolved solids were NaCl, IS is equal to its molar concentration (ci) at ~ 0.054 mol/L, according to Equation 1. The extended Debye-Hückel equation was then used to determine γ, with parameters A ~ 0.51 and B ~ 0.33, ion size parameter a0 = 3.5 Å, the net charge of Cl- (zi), and IS at 273.15 K and 1.4 bar, approximating the pressure at 3 m below the surface (Equation 2). The Cl- activity was then calculated using IS and the logarithm (base 10) of γ, resulting in a value of −1.3626, which was used whenever chlorine-containing species, phases, and minerals were present.
Furthermore, for each Pourbaix diagram, several adjustments were made to the activity of contributing species to account for the in situ conditions of the slags, given their geographical location just south of the Zagros Mountains and proximity to the Caspian Sea [77–79]. The activity of carbonates in soil air was set to −3 to meet the minimum partial pressure required for the formation of siderite [80,81], as identified by chemical and XRD analysis. In some instances, the presence of minerals and phases dictated adjustments to the activities of species; however, unity was assumed for arsenic and copper, given their ubiquity in some slags and the presence of copper-based sulfides in others, and unity activity for H2O.
The predominance diagrams were calculated with published datasets containing Gibbs energy of formation parameters for the discovered minerals and compounds. These data were used to calculate the standard molal thermodynamic properties and chemical affinities of reactions using the open-source CHNOSZ R package to create Eh-pH diagrams [82]. The data used for these calculations [83–85], which consist of calculated parameters based on the revised Helgeson-Kirkham-Flowers (HKF) equation for aqueous species, were selected for their suitability to the slag burial environment and the rigor of their data evaluation. Several supplementary data sources were also added to account for As-containing minerals and aqueous species, and updated thermodynamic parameters [58,86–97]. The scripts and data used to produce the diagrams are available in the S3 and S4 Appendices.
Results
Macroscopic and archaeological interpretation
Visual inspection of the ‘green slags’ H76-37A and S45B, in addition to the subsequent XRF analyses, correlated with prior scholarship [56,98,99], affirming they are products of copper smelting, showing substantial green corrosion and charcoal among highly porous, grey-textured surfaces [29,32]. Slag H76-S39, however, is noticeably darker, with reddish-brown corrosion along its top surface; nevertheless, it is believed to have derived from the same high-temperature smelting process. Although not studied in this current work, slag H76-S30, a tap slag containing small amounts of iron (di)arsenide speiss (FeAs and FeAs2), is chemically similar and stated to have come from the same smelting operation [43]. Slag H76-S39 was therefore labeled a ‘speiss slag’ thought to have been adjoined to H76-S30, but, having cooled slowly, conditions did not prevail to form speiss [43,49].
Slag H76-S39 is dark in color and weakly magnetic, with a smooth top surface, a flow-shaped midsection, a discernible crease along the midsection, and a rough underside, where it likely settled on soil. Iron corrosion products cover portions of the top surface, along with more extensive localized surface disruptions. Notably, this slag had been previously studied and sectioned; its bulk, having been exposed, appears very dense, with most pores being less than 1 mm in diameter, a few ranging from 1 to 5 mm, and a ~ 1 cm unreacted charcoal inclusion (Fig 1).
Based on the outer texture of the slag, the left side represents the bottom, and the opposing side represents the weathered or altered top. X-ray fluorescence analyses revealed that the top layer exhibits increased concentrations of arsenic, likely resulting from the release of arsenic from the slag matrix and pyrometallurgical phases during alteration. Additional images are provided in S1 Appendix.
Slags H76-37A and H76-45B can be broadly attributed to copper smelting. Copious copper oxide and chloride corrosion cover much of their surfaces where sediments have not adhered, suggesting appreciable amounts of copper metal or phases. Slag H76-S37A is moderately magnetic and triangular prism-shaped with large inclusions, such as charcoal, and significantly large and angularly or teardrop-shaped pores on every side that, in some instances, extend beyond ~ 1 cm at their widest (Fig 2). Its bulk is grey and far less dense than slag S39, considering the high proportion of pores, likely having formed in a different part of a smelting apparatus [29,100–103]. This slag and H76-S45B have been weathered extensively through various pores and cracks. They also contain unreacted inclusions of charcoal, gangue, quartz, ores, and associated minerals [29,36,102]. Slag H76-S45B is weakly magnetic but appears denser despite having many relatively large pores. It also has a smooth-flowing red surface, likely cuprite (Cu2O), and is covered in clay and sediment. Interestingly, the slag contains embedded pieces that appear to be from a furnace lining or other pyrotechnic ceramic (Fig 3). Its shape also suggests it may have been wrapped around a cylindrical or round object, possibly a tuyѐre or stone.
Additional images are provided in S1 Appendix.
This slag is theorized to have been wrapped around a cylindrical object, possibly a tuyère. Additional images are provided in S1 Appendix.
These slags have all been associated with related copper smelting operations, despite being found in disturbed levels of the site [62,104]. Their outward and inward appearances, in the case of H76-S39, inform about their role and how they were coerced into existence from compositionally complex copper ore smelting operations conducted in the early 4th millennium, Iran [29,46,59,74,105–108].
X-ray fluorescence surface analyses
Slag H76-S39 showed no visible copper or copper corrosion and is likely a tap or hearth slag, suggesting that high temperatures and low viscosity were achieved during smelting [59,66,109–111]. The other slags, H76-S37A and H76-S45 B, are likely the result of incomplete copper/slag separation, as evidenced by their large pores and significant copper patina among unreacted inclusions [30,105,112]. X-ray fluorescence analyses confirmed high concentrations of Cu, Al, Si, S, Ca, and Fe oxides in both, agreeing, in complexity, with prior studies of slags from the site [43,49]. The Cl concentration on the surface of H76-S45B is of particular importance, as it is likely due to high-salinity weathering and the deposition of secondary corrosion products [30,32]. For instance, secondary surface enrichment of atacamite or paratacamite (Cu2Cl(OH)3) can occur due to coprecipitation and adsorption near saltwater bodies, whereby the precipitation of chlorides sequesters released copper [32,40]. Ordinarily, some Cl is present in copper smelting slag, especially if polymetallic ores, atacamite, and gangue minerals had been smelted [30,100,113], but its concentration would be mere fractions of a percent compared to those shown in the surface analyses. Analysis of slag H76-S37A revealed significantly less copper than iron oxide, at 5.47 and 25.92 wt.%, respectively. High Fe concentrations are not without precedent for slags at the site [43,49], and are likely due to higher concentrations of fayalite (Fe2SiO4), magnetite, and hematite (Fe2O3) at the analysis location [66,114–118], and an overabundance of Fe in the Cu ores [26,66,67].
The largest slag in the set, H76-S39, also showed high CaO, and, along the top, in the region visually identified as iron corrosion, were significant concentrations of arsenic trioxide (As2O3) at up to 6.13 wt.%. While this percentage of As is not within the range of an iron arsenide speiss, it suggests that substantial amounts of As, and perhaps speiss, resided just below the surface, or that the slag was in contact with an external source of As. Given the disrupted appearance of this region, however, the high As concentrations are likely due to weathering, leaching, and reprecipitation [39,119,120]. The high concentration of CaO in the corresponding analysis may have also contributed to the increase in As, as it has been shown that higher slag alkalinities promote the formation of stable calcium arsenates, such as Ca3(AsO4)2, increasing As content by as much as 20–50% in modern-day copper smelting operations. Similarly, an increased Fe[O]:SiO2 ratio, which is the highest in this analysis, can lead to a reaction between Fe3O4, or Fe2O3, under oxidizing conditions with As2O5 to form ferric arsenate (FeAsO4) [121,122]. These reactions and others like them are responsible for increasing the concentration of As in the slags.
X-ray computed tomography
X-ray Computed tomography confirmed the internal porosity first observed along the previously sectioned surface of slag H76-S39 (Fig 4). Furthermore, additional internal features and patterns were revealed, showing the precise location of diagnostic features despite the slag’s high-density matrix and brown exterior corrosion, which attenuate X-rays. Given that this is a tap slag, the scan showed regions of high porosity along its top and bottom, which appeared as small vacuoles or pores where the slag flowed onto the soil, stone, or perhaps other slag or technical ceramic. These pores can form when the molten slag meets a cooler material that contains moisture [30], which rapidly vaporizes and cools the slag, causing localized porosity [34,40,66,123]. Far smaller pores also appear on the surface of the slag, varying from 0.00241 to 0.289 mm³; however, these are likely the result of weathering [32,40].
The crosshairs shown in RGB correspond to X, Y, and Z coordinates and intersect precisely with features selected for further analysis and characterization. (A) Transparency of the slag delineating internal pore distribution, including one large, irregularly shaped void that was sectioned along the X-axis. (B) A 2D image perpendicular to the X-axis, where the slag was later sectioned. (C) A three-dimensional image with slight transparency shows the location of small and medium-sized pores that formed between successively deposited layers of slag. (D) A two-dimensional image perpendicular to the Z-axis showing density differences and spaces between layers, associated porosity, and where the layers and cracks intersect.
Slightly larger median-sized pores, between 0.289 and 0.574 mm³, are also located on the surface and form a semi-contiguous band across the midsection of the slag. Larger still, pores greater than 0.569 mm³ appear internally near the midsection. The clustering of median-sized pores along the midsection and uneven stratification are discernible (Fig 4C and 4D), suggesting that the slag formed from two layers, where a first, cooled and hardened layer could have met a consecutive one or folded onto itself in a continuous molten pour. Prior visual observations support the likelihood that this sequence of events occurred. In addition, the exposed charcoal inclusion and the congregation of tiny pores around it, further support the supposition that the slag settled in layers. The charcoal sits encased in the second, or top layer; once encased, as the slag cooled, decreasing temperatures would have prevented it from reacting further [110,114,118,124,125]. The largest feature identified by CT was, at its widest, 3.49 mm (Fig 4A and 4B). It appeared slightly denser than the pores, hinting that it contained X-ray attenuating materials. Slag H76-S39 was sectioned along the coordinates of this feature, which also bisected the external As-rich surface identified by XRF analysis.
The CT scans of slags H76-37A and H76-45B showed a striking difference in their internal pore distributions compared to H76-S39, in addition to several high-density internal features and surface corrosion products (Figs 5 and 6). Slag H76-S37A exhibited surface corrosion with a similar relative density to the presumed internal copper-containing droplets that had congregated together (Figs 4A–4D). Since the surface corrosion had previously been identified as atacamite/paratacamite, the internal droplets were thus likely to contain chlorides or other impurities. One of the largest of these high-density droplets, a nearly perfect sphere ~ 1.5 mm in diameter, was later targeted for sectioning. Based on the distribution of the larger droplets, although size and shape are not necessarily determinative of composition or process [102,109], its location suggests that it resides near the bottom of the slag. Due to gravity, larger copper droplets tend to settle, and while viscosity, surface tension, and unreacted inclusions may impede their descent [109,126], this droplet, along with similarly dense droplets nearby, appeared to have migrated downward. Slag H76-S45B contains similarly spherical droplets, in addition to ovoid and elongated ones, which appear as semi-circular cutouts in 3D. However, in this instance, they are possibly composed of copper, copper sulfides, or matte, based on the varied amounts of S and Fe shown by the XRF analyses (Fig 5A and 5B. The corroded surfaces and internal features are similarly dense and thus challenging to differentiate in the CT scan; however, the droplets are slightly more attenuated. Assuming the internal high-density features are copper droplets, high-density sulfides, or matte, their size and distribution suggest the green corrosion-rich surface is the top of the slag. The droplets would have settled, being denser [127], on the upward-facing inner border of the semicircle, which was theorized to have been formed by a tuyѐre. However, unlike H76-S37A, this slag has randomly distributed pores of various sizes, ranging from 0.00249 to 31.1 mm³, suggesting that the voids were produced under conditions that were efficient at evolving gases, such as SO2, rather than from contact with a lower-temperature material. Since this slag contains a multitude of similar high-density droplets, only one was targeted for sectioning.
(A) A high-density metal droplet from slag H76-S37A, ~ 1 mm in diameter and chosen for sectioning, is shown in 3D near the exterior of the slag. (B) The image and coordinates of a high-density droplet pre-section are shown perpendicular to the X-axis. (C) Location of a high-density droplet in slag H76-S45B and the overall 3D image. The droplet resides near the exterior of the slag, where it settled above a location presumed to be a cylindrical object or tuyère. (D) Image of the pre-sectioned location of a high-density droplet perpendicular to the X-axis.
(A, B) Cross-section of the large covellite droplet showing variably dense but distinguishable phases. In contrast, note that the smaller, unaltered droplet immediately next to it, which has no connecting conduits, is contiguous and shows no signs of leaching or corrosion. (C, D) CT scan aligning with the sectioned droplet. Note the three external pores and the high-density internal pore of the droplet. The internal pore has a surface layer of cuprite, which attenuates the X-rays from the CT scan, giving it a more intense red color.
Microscopy and X-ray diffraction
The largest pore in H76-S39 was initially presumed to be charcoal or technical ceramic, having a slightly higher density than the externally exposed inclusion; however, after sectioning, it was revealed to be a void lined with dense iron-rich magnetite, hematite, or oxyhydroxide phases and minerals (Table 3), within which were opaque and prismatic crystals. These secondary calcite and gypsum mineralizations, although prominent in the optical images, were not visible in the CT scans due to their small size and attenuation caused by the iron oxide-rich lining [128,129]. Optical microscopy revealed opaque rhombohedral and colorless crystals, with the latter showing multi-colored hues [130,131], suggesting calcite (CaCO3) and gypsum (CaSO4·2H2O), respectively (Fig 7A and 7C. Energy-dispersive spectroscopy corroborated calcite and gypsum stoichiometrically, along with minor concentrations of Na, Al, Si, P, K, and Sc, which, in combination, could have instilled variegated blue, green, orange, yellow, and purple hues (Table 3). The detection of Fe was associated with interspersed hematite and mixed weather-derived secondary precipitates, with the former appearing ruby-red under optical microscopy (Fig 7B and 7D). These minerals were also confirmed by targeted EDS, which revealed Si, Ca, and Cu impurities, as well as in the bulk via powder XRD (Fig 8).
(A) Opaque and nested calcite with interspersed hematite (Hem) crystals. The black outlined boxes in (A) and (C) correspond to the SEM image and analysis location in (B) and (D), respectively. (B) Secondary electron image and broad EDS analysis corroborating calcite mineralizations. (C) Prismatic and multi-hued gypsum crystals. (D) Secondary electron image and broad EDS analysis confirming gypsum. Of note is that calcite or gypsum precipitated in the pore but never in the same location. Mineral abbreviations are from Warr (2021) [132].
Several detected phases and minerals, identified via optical microscopy and EDS, were used for Rietveld refinement of the spectra. Aluminum oxide, calcium arsenate, cristobalite, iron arsenate, iron carbide, löllingite, and siderite were not seen in significant quantities by analysis, suggesting they may exist but in limited proportions throughout the slag. Due to the large percentage of amorphous phase in the slag, many mineral-defining peaks may have also been obfuscated.
Calcite and gypsum frequently form as secondary precipitates in slag voids and pores through the weathering of pre-existing calcareous and sulfidic materials [32,34,40,133]. Metallurgical copper slags are rich in Ca-bearing minerals such as pyroxenes (hedenbergite (CaFeSi2O6), in addition to primary calcite, gypsum, dolomite (CaMg(CO3)2), and gangue and added fluxes [26,40,56,66,102,106,109]. Similarly, copper slags contain sulfate ions from sulfur-bearing ores such as chalcopyrite (CuFeS2), bornite (Cu5FeS4), chalcocite (Cu2S), pyrrhotite (Fe1-xS), and pyrite (FeS2), in addition to entrapped matte droplets [29,32,109]. These minerals and phases would have contributed to the reprecipitation of calcite and gypsum within pores and voids. As discussed in greater detail elsewhere [32,39,113,134], dissolved Ca2+ ions in solution, from CO2 in soil and the atmosphere, which dissolve in water, form carbonic acid (H2CO3), bicarbonate (), and carbonate ions
, leading to neutral–alkaline conditions, with the latter reacting with the former to form calcite (Equation 3). These same Ca2+ ions, when saturated in solution with SO42- ions from oxidized and weathered sulfide minerals, lead to secondary gypsum precipitates (Equation 4).
These minerals commonly nucleate in Ca-containing copper slags, where an increase in surface area creates favorable microenvironments for crystal growth. The voids, which act as conduits for the flow and transport of water, provide these microenvironments [32,34,39,40,119,133,135]. X-ray diffraction also identified iron and calcium iron silicates such as fayalite and hedenbergite (CaFeSi2O6), oxides such as quartz and possibly cristobalite (SiO2), hematite and magnetite and iron oxide hydroxides, iron arsenate hydrates such as scorodite/parascorodite (FeAsO4·2H2O), iron sulfides like pyrrhotite (Fe1-xS) and pyrite (FeS2), with some of the latter being As-bearing, in addition to arsenopyrite (FeAsS), and löellingite (FeAs2). These minerals are typical of archaeological slags and those derived from modern flash smelters [26,29,32,40,56,100,102,109].
In addition, optical microscopy and EDS analyses of the ‘top’ and ‘bottom’ of pore lining showed a black and ruby-red border of stoichiometric magnetite and hematite, long dark skeletal lathes of fayalite, often arranged in parallel, and colorless calcite (Fig 9A and 9C). These contain significant concentrations of As, Ca, and Al, along with a contingent of minor impurities (Table 4). The outliers in this group are analyses 2 and 11 (Fig 9B and 9D), which correspond to regions with high Al content in the slag. The accompanying Ca likely corresponds to small feather-like mineralizations of hedenbergite (CaFeSi2O6) that formed due to rapid cooling [29,38,46,105,118,136,137]. Of note, analysis 9 is likely scorodite, but with noticeably low As, which could have crystallized from weathered As-bearing minerals such as arsenopyrite [119,138,139], dissolved pyrite [140], relic sulfide inclusions from partially reacted As-bearing sulfides, or dissolved As in the slag [32,102]. Corrosion of these minerals, as shown in the Eh-pH diagram (Fig 10), can create localized high acidity, allowing for the spontaneous formation of scorodite from Fe3+ and at < 4 pH or stepwise from hydrated Fe3+ and As5+ ions, such as
–the octahedral form in aqueous solution– and
at pH ranging from 2–7 (Equations 5), and the oxidation of arsenopyrite (Equation 6), which, along with magnetite and other iron oxides, release Fe3+ ions into solution when weathered (Equation 6) [39,138–142]. Given the stability of slag and its relatively neutral pH, the simplified complexation reaction shown in Equation 7 represents a paragenetic sequence that leads to the formation of scorodite. Siderite, another secondary precipitate, is known to form between pH 6.5 and 7 in anoxic, carbonated environments, in association with chukanovite, a metastable iron hydroxycarbonate (Fe2(OH)2CO3) [81,96,97,143–145], which, together, are associated with magnetite, goethite, scorodite, and iron sulfides such as pyrite [80,143,144,146]. These minerals are also shown in the diagram, and their proportions or presence/absence in the slag would depend on localized conditions. The association of these minerals is not unexpected, as Fe oxides and hydroxides, sulfates, carbonates, pyroxenes, and arsenate minerals are known to precipitate together in the pores of weathered copper slags [32,39,119,138,139,147–149]. Accordingly, these minerals, in addition to aluminum oxide (Al2O3), iron carbide (Fe5C2), löllingite (FeAs2), iron and calcium arsenate (FeAsO4), and parascorodite, were identified in the XRD spectra. Goethite was not present in the spectra, either due to its low concentration in the slag or because its peaks had been obscured by amorphous material.
Numerous fayalite (Fa) lathes are also visible, in addition to isotropic magnetite (Mag) between them, interspersed within the pore lining with hematite (Hem). The white and black outlined boxes in (A) and (C) correspond to the SEM image and analysis locations for (B) and (D), respectively. (B) Backscatter image of the pore lining corresponding to the ‘Top’. Imaging and EDS analyses of the pore revealed the presence of calcite (Cal), fayalite (Fa) lathes in parallel, magnetite (Mag), hematite (Hem), scorodite (Scd), and hedenbergite (Hd). The outlined boxes represent analysis regions rather than points. Also shown is a crack/conduit filled with calcite and hematite precipitates that penetrated the pore’s Fe-oxide lining. (D) Backscatter image and EDS analyses of the ‘Bottom’ lining of the pore, showing hematite, magnetite, geothite (Gth), calcite, scorodite, siderite (Sd), fayalite, and quartz (Qz). The lining of the pore was intact, except for cracks where concentrations of calcite and hematite were found. Mineral abbreviations are from Warr (2021) [132].
Mineral and aqueous species stability regions are outlined in black and blue, respectively, with corresponding text color, and red dashed lines and text delineate arsenic. Soluble sulfur species are not shown, but are necessary for the calculation. Related stable minerals that may form at specific Eh-pH in each instance include hematite (Hem), pyrite (Py), magnetite (Mag), arsenopyrite (Apy), and löllingite (Lo). According to these calculations, scorodite precipitates at < 2 to ~ 6.5 pH, and ~ 1 to 0.74 Eh (volt) in Fe3+(aq) and . Activity for each component was set as follows: S = −7, Fe = −2, As = 0, Si = −2, and C = −3. The activity of Cl was calculated to be −1.26985 based on the soil’s salinity and ionic strength (IS), 0.05372177 (see S3 Appendix).
Finally, the high concentrations of As, especially for analyses 9 ‘Top’ and 1, 4, 5, and 11 near the surface of the slag, require explanation (Fig 11A). Table 5 presents analyses of the area immediately below the observed surface corrosion, identified by XRF as As-rich. Stoichiometrically, analyses 1, 5, and 11 are closest to scorodite, with 4 having As percentages consistent with parascorodite (Fig 11B). Both are secondary precipitates formed from weathered As-bearing minerals at ambient temperatures, pressures, and < 3.5 pH [138,139,150]. Furthermore, given the availability of As and the multitude of impurities in the analyses, much of the discovered scorodite may be amorphous ferric arsenates (AFA), also referred to as hydrous ferric arsenates (HFA), which form as an intermediary step to crystalline scorodite or parascorodite, but only at pH < 3 [119,138,139]. These arsenates quickly transition to thermodynamically stable scorodite or parascorodite [138], but can also slowly grow in neutral to alkaline pH environments [119,139,151].
A crack or conduit connects the surface to the inner pores. The box outlined in white corresponds to the SEM image and analysis. (B) Backscatter image of secondary scorodite (Scd), parascorodite (Pscd), and numerous small quartz (Qz) crystal precipitates. Mineral abbreviations are from Warr (2021) [132].
In addition to ferric arsenates and scorodite minerals, what is evident from these analyses is the appreciable amounts of As, regardless of mineralogy. The EDS data, in agreement with XRF, showed a more significant proportion of As-bearing phases residing just below the surface of the slag in the altered regions. And while XRD confirmed the presence of As-bearing secondary precipitates, it also showed löellengite or FeAs2. Iron diarsenide can form naturally as a mineral through hydrothermal reactions in geological settings at temperatures ranging from 400 to 500°C in pyrrhotite and serpentinite veins [152], or during high-temperature pyrometallurgical processing between 900 and 1200°C under reducing conditions, followed by the thermal decomposition of arsenopyrite and other iron arsenides, such as FeAs [58]. Processually, where high temperatures and chemistries persisted to form Fe2As and FeAs during smelting, the latter likely resulted in FeAs2 upon cooling at temperatures between 510–690°C, as solid FeAs reacted with highly active As from the melt [58,86,153]. These anthropogenically created iron arsenides could then, according to the Eh-pH diagram, oxidize to scorodite [119,138,139,141], following Equation 8 at low pH. Notably, at pH above ~ 4.7, arsenic mobilization would be limited.
Understanding the role of As during smelting and its omnipresence in the slag is of the utmost importance to archaeometallurgists and modern material scientists alike. What is clear from the studied slag is that while the proportions of As necessary to form iron arsenides existed, the conditions were insufficient to form large homogeneous speiss phases. Prior studies have, however, identified small amounts of noncontiguous speiss prills [49,52,54,154].
The likely explanation for the observed high As in these slags is that, during cooling, it was incorporated into the fayalite glassy silicate matrix and, in some instances, became encapsulated, forming intermetallic phases with Fe [138]. Arsenic can also be distributed to the fayalite phase and glassy silicate matrix through the substitution of for
tetrahedral groups during the smelting of ore-containing olivines in an oxygen-rich atmosphere [138,155,156]. Notably, olivines are prevalent in Iranian copper ores [26,102,157,158]. As for any serendipitously discovered speiss prills, these explanations would account for their scarcity and location in the slag (see S5 Appendix for a more detailed explanation). Tangentially related to speiss, arsenic could have also been distributed through the alteration of primary arsenic minerals, phases, and As-bearing fluids, resulting in ferric arsenates, hydrous ferric oxide (HFO), and minerals like scorodite [119,138,139].
After sectioning slag H76-S37A at the prescribed coordinates, a large metallic teal droplet and multiple smaller Cu-colored droplets were exposed; the smaller ones were unresolvable due to their small size. Optical imaging of the large droplet showed distinct teal and white crosshatching in light-field and a metallic luster in dark-field (Fig 12A and 12C), which we could not attribute to any known slag minerals, phases, or precipitates [32,159–164]. Encircling and between the grains are emerald-green colored phases identified as atacamite and later confirmed by EDS and XRD. The droplet exhibited a Widmanstätten-like structure. In this instance, given the extensive intergranular and transgranular corrosion throughout (Fig 12B and 12D), the pattern is undoubtedly the result of leaching and the formation of secondary precipitates along slip lines and planes [161,165,166]. Energy-dispersive X-ray spectroscopy of the intergranular, transgranular, and slip line alterations consistently showed atacamite or paratacamite (Table 6), with the latter being the more thermodynamically stable polymorph at high chloride concentrations [84,167,168]. The grains themselves were also consistently identified as copper sulfides by EDS with varying ranges of Cu, which, having been leached, predominantly resulted in ions in solution while enriching the overall proportion of S [159,169–172]. Small copper droplets, magnetite, atacamite or paratacamite, and fayalite with possible Al and Ca substitutions were also found adjacent to the larger droplet in the slag’s matrix.
The white outlined boxes in (A) and (C) correspond to the SEM image and analysis locations for (B) and (D), respectively. Also shown are the extensive intergranular and transgranular atacamite (Ata) or paratacamite (Pata) secondary precipitates surrounding and pervading the droplet. Numerous intact Cu droplets are also visible, having not been connected to the surface through cracks or conduits. The cracks or conduits extend from the surface to the droplet. (D) Backscatter image, with location shown in (A) of variously oriented leached covellite grains with intergranular, transgranular, and Widmanstätten-like corrosion patterns. The outlined boxes represent analysis regions rather than points in (B) and (D). (D) Backscatter image of the slag, delineated in (D), showing smaller copper (Cu) droplets, magnetite (Mag), and feather-like hedenbergite (Hed) within the fayalite (Fa) matrix. Mineral abbreviations are from Warr (2021) [132].
Metal-bearing sulfides are particularly susceptible to weathering under oxidizing conditions in both air and aqueous environments, releasing metallic elements into solution [32,40,173], as shown in the Eh-pH diagram (Fig 13). Slag H76-S37A is a unique example of this phenomenon, as Cu ions were slowly leached from the covellite droplet [169,172], resulting in surrounding and infiltrating atacamite or paratacamite precipitates, even along slip planes (Table 6 ‘Inner’). Based on these findings, a likely scenario for its structure and composition is that the copper sulfides formed during smelting, under low oxygen fugacity, or during cooling [26,30,37,40,59,102,109–111], were subsequently leached and corroded through two successive reactions. Beginning with chalcocite, or, more precisely, a comparable Cu2S phase formed during smelting [109,174], the sulfide would have been rapidly oxidized by Fe3+ ions at ambient temperatures, resulting in Cu-deficient “blue-remaining” covellite with hexagonal structure (Equation 9) [170,175–181]. Following this transformation, if the droplet were not already monoclinic Cu2S or white metal, leaching would produce Cu²⁺ ions and elemental S0 [159,169–172]. Small copper droplets, magnetite, atacamite or paratacamite, and fayalite with possible Al and Ca substitutions were also found adjacent to the larger droplet in the slag’s matrix.
Mineral and aqueous species stability regions are outlined in black and blue, respectively, with corresponding text color. Soluble sulfur species are not shown, but are necessary for the calculation. Related stable minerals that may form at specific Eh-pH include antlerite (Atl), chalcanthite (Cct), tenorite (Tnr), cuprite (Cpr), nantokite (Nik), djurleite (Dju), covellite (Cv), and chalcopyrite (Ccp). According to these calculations, (par)atacamite precipitates at ~ 4 to 6.4 pH and ~ 0.3 to 1.2 Eh (volt) in the presence of and oxygen, resulting in the formation of Cu+(aq). Activity for each component was set as follows: S = −2, Cu = 0, Fe = −15. Furthermore, the activity of Cl was calculated to be −1.26985 based on the soil’s salinity and ionic strength (IS), 0.05372177 (see S3 Appendix).
The first oxidation transformation (Equation 9) is controlled by the diffusion of the oxidant on the mineral surface, and the second (Equation 10), by chemical and/or electrochemical reactions [159,169,170,172,175,182–184]. During leaching, Fe3+ ions, released from efflorescent salts and weathered minerals such as magnetite and goethite [34,80], react, releasing Cu2+ ions and shifting the redox potential towards more reducing conditions [83,175,185]. If oxygen remained, Fe2+ could then oxidize to Fe3+ [30,40,186,187], contributing to the continued dissolution of chalcocite and covellite. Availability of Fe3+ could have also come from the oxidation of fayalite to hematite and silica [67,188], a reaction between chalcocite and fayalite [189], or by arsenate oxidants and selective sulfide-reduction-volatilization [190].
In these reactions, Cl⁻ ions, from salts such as NaCl in solution, enhance the corrosion kinetics such that [182], if it is absent, a layer of dense amorphous elemental S can form on the mineral surface, hindering diffusion and slowing the reaction [170,175,182]; however, if present, they promote the formation of a porous crystalline sulfur layer, allowing the leachant to penetrate and react with the underlying copper sulfide phases [159,170,171,175,182,191]. Chloride ions also stabilize Cu⁺ cations by forming , facilitating the dissolution of Cu2+ ions into solution while avoiding the intermediary step of oxidizing Cu+ [172,175]. Further accelerated leaching occurs from increased proportions of chlorides (Equations 11 and 12) [167,175,192], which, in this slag, resulted in CuS and soluble
, and potentially elemental S and
[159,169,175,182,191]. Eventually, the remaining Cu would be leached from the covellite for as long as chlorides and water were available for the reaction, leaving elemental S behind.
These combined reactions likely explain the droplet’s slow inwardly progressing transformation through narrow surface-connected cracks/conduits, as well as intergranular and transgranular leaching across the slight potential differences between grains, grain boundaries, and slip planes (Fig 12B) [80,193–196]. Our extensive SEM-EDS investigation furthermore revealed the presence of several cracks/conduits throughout the slag, which often led to leached copper sulfide droplets, as shown in Fig 12A and 12C. Where infiltration channels exist, the leaching of Cu2+ ions was favored, and at increasing Cl- concentrations and in the presence of S, intergranular leaching along high-angle grain boundaries was more probable [171,172,175,183,191,193]. The reactions are facilitated through the formation of large S crystals, which create porous, penetrable layers for the reactions to continue unabated [169,182]; however, when formed, basic salts such as atacamite and paratacamite would have slowed or halted the reaction [169], effectively freezing further leaching/corrosion and passivating the system.
Having leached Cu2+ and stabilized , the formation of atacamite could proceed according to Equation 13, as shown in the Pourbaix diagram (Fig 13). Notably, this is not the only pathway leading to copper oxychloride precipitates [167,192,197], but it is the most likely one, given the association of copper sulfides with high Cl- concentrations, the latter being common in the form of sodium chloride (NaCl) in modern-day Semnan [76,198].
In addition, of note, leached Cu2+ leads to a series of non-stochiometric copper sulfides, or polysulfides or intermediaries, such as djurleite, digenite, anilite (Cu₁.₇₅S), geerite (Cu₁.₆S), spionkopite (Cu₁.₄S), and yarrowite (Cu₁.₁S) [159,181]. Of these minerals, only covellite and digenite were confirmed via XRD, suggesting the discussed transformation of chalcocite into non-stochiometric copper sulfides between end-members Cu2S and Cu2S2 had occurred [181,199]. Aluminum oxide, hedenbergite, arsenolite (As2O3), augite (CaMg0.75Fe0.25Si2O6), quartz, and, possibly, sodium sulfide (Na2S) were also identified in the XRD spectra (Fig 14).
Several of the detected phases and minerals were previously identified using optical microscopy and EDS, except for aluminum oxide, arsenic oxide, hedenbergite, digenite, augite, quartz, and sodium sulfide. The presence of digenite is likely evidence of the gradual leaching of Cu2S phases. Similar to the other studied slags, a large percentage consists of an amorphous phase, which obscures many of the mineral-defining peaks. Rietveld refinement was employed to determine the mineral composition of the slag based on the phases and minerals identified by optical microscopy and SEM-EDS.
A large droplet in slag H76-S45B was identified using XCT, sectioned, and subsequently studied using optical and scanning electron microscopy, as well as X-ray diffraction (XRD). Due to the droplet’s size and the fact that it was sectioned along the same axis as its scanned orientation, the droplet, its internal spherical void, and the surrounding pores aligned precisely with the layout observed in the CT scan. When polished, the droplet appeared silvery with Cu and quartz distributed near it within the slag matrix, and its void contained an internal lining of red mineralizations (Fig 15A and 15C). The void minerals exhibited intense X-ray attenuation compared to the droplet, suggesting that they contained metal-rich secondary precipitates. Optical microscopy and EDS analyses showed the droplet to be a silvery-white metal with appreciable concentrations of Se, As, low Fe (Table 7 ‘Inner’), and intergranular cuprite (Fig 15B and 15D). The void mineralizations stoichiometrically conformed to cuprite. X-ray diffraction confirmed the presence of Cu2S, in addition to: aluminum oxide, copper-iron-arsenide (Cu0.73Fe0.99As), tennantite-(Cu) (Cu12As4S13), copper-selenide-sulfide (Cu2S0.06Se0.94), clinoatacamite (Cu2(OH)3Cl), atacamite, magnetite, maghemite (γ-Fe2O3), and fayalite (Fig 16). Several of these minerals appear in the Pourbaix diagram (Fig 13); however, since thermodynamic data is lacking for Cu-, Se-, and Fe-containing species, more representative stability regions could not be calculated for this slag.
The droplet is metallurgical Cu2S with fine Cu droplets and quartz (Qz) surrounding it. Notably, the large silver droplet also contains As, which contributes to its silvery appearance. The black and white outlined boxes in the optical images correspond to SEM-EDS images (B) and (D). Within the larger droplet is a void containing cuprite (Cpr), as shown in (A), (D), and between grains in (B). Also shown in the optical images are regions E and F, corresponding to the SEM-EDS images and analysis locations. Mineral abbreviations are from Warr (2021) [132].
Several of the detected phases and minerals were identified via optical microscopy and EDS, except for clinoatacamite, copper-iron arsenide, tennantite, magnetite, and hematite. Magnetite and hematite were prominently observed via optical microscopy elsewhere in the slag. Comparatively, cuprite was not detected, despite its assured existence as oxidation products within the Cu2S droplet. As before, the large percentage of amorphous phase obscured many of the peaks, requiring Rietveld refinement based on known phases and minerals.
Notably, the analyses of the droplet and surrounding phases reveal significant concentrations of As, Se, or both, which, together, could have imparted a silvery coloration to the already white-metal-colored droplet [200,201]. The droplet is a curiosity, given its similarity to modern-day copper matte white metal in copper converting, which contains ~ 75% Cu, with Se tending to distribute to it under reducing conditions [107,202–204]. Arsenic, on the other hand, tends to be distributed to Cu [205,206], although it can also be found in the white metal with increasing concentrations of As from primary ores [200,207,208]. Furthermore, archaeometallurgical studies of As-containing Cu have shown that silvery surface coloration, resulting from inverse segregation under non-equilibrium conditions, can occur through the formation of a γ-phase (Cu3As) layer [209–211]; however, even incomplete layers and low As concentrations can contribute to silvered appearances. In brief, while chalcocite is typically dark blue or black [30,32,105], pyrometallurgical Cu2S is white, and when alloyed with As and Se, it can result in a silvery color [59,102,212]. Abutting the droplet immediately beside the mid-sized pore is a minor atacamite/paratacamite secondary mineralization, and within it, shown in Table 7 ‘Outer’, analyses 7 and 12, are copper-iron-arsenates or -arsenites (Fig 17E and 17F).
Analyses (E) confirm that the droplet, as previously observed, comprises metallurgical Cu2S or chalcocite (Cc) and cuprite (Cpr) with high concentrations of Se and As. Immediately next to the droplet, between the slag matrix and the droplet, lies an atacamite/paratacamite (P/Ata) precipitate, within which are two copper-iron-arsenides (CuFeAs). Also shown are a Cu droplet, hedenbergite (Hed), and fayalite (Fa). SEM-EDS image (F) shows several tetragonal copper-iron-arsenide crystals. These reside within a crevice beside the large droplet and the ‘Mid-size’ pore. This is a unique microenvironment where these crystals and pure Cu had precipitated. Mineral abbreviations are from Warr (2021) [132].
Within the largest of the three pores surrounding the droplet are cuprite and calcite mineralizations, which contain copper-carbon-oxides or possibly Cu carbonates. These minerals can sequester Cu and inhibit further dissolution of metal-bearing phases, forming minerals like malachite and azurite [40,213,214]. The difference between the cuprite in the ‘Inner’ and ‘Large’ void and pore, respectively, seems to be in the contiguousness of the precipitate, with the void mineralization being an uninterrupted layer capable of attenuating X-rays. In contrast, those in the pore are sporadically distributed among less dense minerals such as calcite. Furthermore, as seen in Fig 15A and 15C, the pore resides between the droplet and the slag matrix; however, the void is entirely enclosed within the droplet, thus creating a microenvironment with limited ion migration.
As for the copper-iron-arsenide, SEM-EDS clearly showed several truncated tetragonal (di)pyramidal crystals in the ‘Mid-sized’ pore and adjacent crevice, which are not pyrometallurgical in origin; rather, they are secondary precipitates borne of a unique microenvironment, underscoring the fact that weathering and the dissolution and recrystallization of phases and minerals is commonplace in ancient, historical, leading to minerals typically uncharacterized in modern slags [39–41,133,149,215–217]. X-ray diffraction identified tetragonal copper-iron-arsenide, which may be related to those found in the crevice; however, this is uncertain, as they were not targeted separately. The high concentration of Se is also of interest, as it is prevalent in almost every EDS analysis (Table 8), likely due to its known distribution to white metal during the smelting process.
Discussion
Through the application of industrial XCT, this study identified several diagnostic features before the destructive analysis of archaeological slag. Included features were the position and distributions of metal and alloy droplets, pores and voids containing microenvironments, the conditions governing precipitates, and the orientation of the slags as they formed during the smelting process. Regarding the interpretation of the CT scans, in addition to what has already been covered elsewhere in this paper, the magnetite and hematite lining, as well as the pores in slag H76-S39, with densities of 5.11 and 5.14 g/cm³, respectively, sufficiently attenuated the X-rays to mask the inner materials (Fig 4A and 4B).
The presence of these minerals explains why the space within the pore appeared slightly denser than the surrounding air, since, when combined with the predominant fayalite phase in the slag, at 4.47 g/cm3, the less dense calcite and gypsum minerals lining the pore, at 2.71 and 2.3 g/cm3, respectively, were hidden. In brief, higher-density minerals and phases can obscure or mask lesser ones when uninterrupted high-density mineralizations encircle them [4,9]. This effect, however, can be mitigated by increasing the peak tube voltage [8–10,13,218]. Furthermore, it was shown that for the CuS grains and the intermingled atacamite/paratacamite phases in the droplet in slag H76-S37A, at densities of ~ 4.78 and 3.7 g/cm³, respectively, some indications of different phases and minerals could be seen; however, due to their diminutive size and interspersing, visual separation of them in the CT scan was not possible (Figs 6A, 6B, 12A and 12C). While one cannot identify the intergranular divisions between the covellite grains, discrepancies are evident, shown as non-contiguous internal density differences. In this instance, working inward, the surrounding material is the slag matrix, rich in silicates, followed by the atacamite/paratacamite outer rim and intergranular separations. Internally, CuS and possibly instances of intact Cu2S are present, depending on the progression of leaching. For comparison, a small, unaltered, high-density Cu2S (~5.65 g/cm³) feature is shown alongside the leached droplet (Fig 6A and 6B), with similar color intensity. In this instance, CT revealed, prior to sectioning, the state of preservation of the leached droplet. For H76-S45B, the picture was far clearer, as the adjacent pores shown around the droplet in the CT scan exactly matched the slag after it was sectioned (Figs 6C, 6D, 15A and 15C). In this instance, the benefits of CT before sectioning are readily apparent. Of note, the internal density of the H76-S45B droplet is more uniform than that of H76-37A, suggesting it was unaltered.
Regarding the location of the droplets relative to the slag orientation, as shown in the CT scans, they were, at least in part, influenced by the cooling rate and viscosity. In general, where cooling occurs rapidly, smaller droplets become entrapped within the silicate glass matrix; conversely, slower cooling, which occurred in this slag, permitted the formation of larger droplets [59,109]. These droplets, influenced by gravity, would then have settled [126], proceeding downward just above the area theorized to have been resting atop a tuyѐre or cylindrical object in slag H76-S45B and the bottom of H76-S37A (see S1 Appendix). In these instances, CT non-destructively and non-invasively aided in interpreting the orientation of the slag, while imparting valuable processual and post-depositional information, especially regarding the preservation state and influencing factors that contributed to corrosion, leaching, and secondary precipitates. An obvious example was the discovery of surface-linking channels or cracks in slag H76-S37A, which facilitated the leaching of Cu2+ from the copper sulfide droplet and the subsequent formation of atacamite, as illustrated in the Pourbaix diagram.
Computed tomography limitations
X-ray computed tomography offered several advantages when studying slags; however, caution is necessary when interpreting the results. Relative density in CT refers to the comparison of a material’s density against a reference standard, typically water, which is assigned an HU value of 0 [72], such that an uncorroded metal particle can have a HU value of 4500–10950 while air, at the bottom of the scale, registers as −1000 [10,14,71,73,219]. The relative density of a material dictates the brightness of the corresponding voxels in the scan, and their size, in turn, determines the level of detail and resolution attainable in the reconstructed image [9,72,73,129,218]. Despite fixing the HU values for water and air, voxel intensity cannot currently be used to determine metals, alloy phases, minerals, or necessarily void space, as was encountered in slag H76-S39. A related issue is that of attenuation and the misrepresentation of density, which can be explained by the emission of polychromatic beams and the Beer-Lambert law of exponential attenuation, which defines the relationship between attenuation and the thickness of a homogeneous material where µ represents the attenuation coefficient, I the intensity of the beam after passing through the object, I0 the initial beam intensity, and D the thickness of the object (Equation 14) [220]. The X-rays emitted from the source pass through the sample and are attenuated proportionally to the features and their thicknesses, but, importantly, since only atomic mass and physical density are considered, similar intensities, that is, the energy of the photons, result for different materials [71,72,221]. Attenuated X-rays may also under- or overrepresent density, as they encompass a range of energies. This is due to beam hardening, which causes a proportionally greater number of higher- or lower-energy photons, and by passing through dense materials that hinder the visualization of lower-density materials within [4,9,68,71].
High attenuation can hinder the penetration depth of the X-ray beam, making it challenging to image features located deep within objects [9,129]. This limitation is particularly pronounced in dense materials, such as slags and metals, necessitating high-energy X-ray sources for sufficient penetration. Another issue is that while XCT can provide high-resolution images, it may not always capture fine details of artifacts with complex internal structures, such as those with overlapping or superimposed elements. Additionally, the convex surfaces of some artifacts can complicate the scanning process, as the curvature can result in some information being cut off in the image slices, or the streaks and Moiré patterns [6,7,11,71].
Slag alternation and speiss formation
Returning now to the alteration of the slags, a plethora of secondary precipitates from the smelted anthropogenic phases, as well as the distribution of elements, inhomogeneity, and morphological differences, such as voids and surface-connected conduits, resulted in paragenetic mineralizations in distinct microenvironments that require explanation. As prior scholarship has shown, the slag and glassy silicate matrices are susceptible to weathering [38,40,222]. Sulfides are particularly deleterious, as they generate acidity (Equation 15), which can lead to continued alteration of the slag, exposing more sulfides and dissolving sulfates and metals into solution [40,223,224]. Through weathering, anions such as ,
,
,
, and
and metal cations are released, which in solution favor the precipitation of secondary mineralizations.
These secondary minerals are typically oxides and sulfides, but they rarely include iron arsenides such as Fe2As, FeAs, and FeAs2, which are the speiss phases archaeologists refer to as having formed from pyrometallurgical processes. The first forms in high-temperature melts through a eutectic reaction between the liquid and bcc Fe solid solution at 768–833°C and ~ 24.4 at.% As. The second, from the thermal decomposition of FeAs2 between 450–800°C or the eutectoid decomposition of an epsilon phase (~ Fe3As2) at 824°C and 40 at.% As, and as naturally occurring westerveldite. The last corresponds to the mineral löellingite, and like westerveldite, is stable at ambient temperatures and pressures [58,86,152,153]. Therefore, in a prehistoric copper smelt, Fe2As could have formed in the absence of oxygen at high temperatures from a mixture of oxides and sulfides, and remained stable upon cooling [56,58,86], as was likely the case at Shahr-i-Sokhta and Arisman [52,59,209,225–227]. Meanwhile, westerveldite and löellingite, if present, would have likely derived from relics in the ore charge, or they were the result of rapid cooling or quenching, which would have been unlikely given recent reevaluations of copper smelting in Iran [56], meaning that only Fe2As can be directly attributed to pyrometallurgy. If quenching had occurred, however, an indicator would be the presence of the NiAs-type polymorph, rather than the MnP-type, of FeAs, which forms at temperatures above 800°C [58]. These facts, along with the previously discussed mechanisms by which As can disseminate within copper smelting slag, raise questions regarding the interpretation of speiss from prehistoric smelting operations. For this research, neither the presence of As and Fe, nor their coexistence in higher concentrations, necessarily indicates that the intended outcome was speiss. Instead, some iron arsenides in the slag are relics, and others likely resulted from mixed ore smelting and processual happenstance. The overall distribution of arsenic in the slag is simply the result of smelting arsenic-rich ores. These ores were further altered through complex corrosion processes, leading to the precipitation of minerals like scorodite from catalysts such as hematite, which were abundant [35,40,213,222,228,229].
Conclusion
This study incorporated XCT into an archaeological slag study workflow, successfully extracting more information than would have been possible with prescribed analytical approaches and tools alone. Several unique features were discovered, and in-depth interpretations were made possible by interactive 3D images, which enabled the visualization of the overall internal structure of the slags, the distribution of pores, phases, cracks, crevices, and metal droplets, as well as the paragenesis of secondary mineralizations from primary phases and environmental factors in microenvironments. Moreover, while traditional slag analytical methods were employed, the use of XCT proved most fruitful, unmasking regions of interest that were later sectioned, dramatically augmenting not only the pyrometallurgical interpretive outcome but the relationship between the slag’s state of preservation and current chemistry and related mineralizations.
Furthermore, since slags are diagnostic of pyrometallurgical processes and site-specific conditions, it has been shown that comprehensive characterization and interpretation approaches combining metallurgy, geochemistry, corrosion science, and thermodynamics, among other fields, are necessary to draw substantive conclusions. This paper marks the beginning of that endeavor, but more is required to fully elucidate the processes of metal and slag production at Hissar, as well as the relationship between corrosion, leaching, and the post-smelting state of slag artifacts. These slags in particular are the basis of much scholarship in archaeometallurgy, particularly concerning speiss or iron arsenide production, which, here, has only been partially addressed. This work contributes to the corpus of ongoing scholarship on arsenic in prehistoric smelting, reinvigorating the field and offering new, comprehensive perspectives. However, despite the novelty of applied XRT, we must emphasize that traditional analytical techniques are irreplaceable.
This work relied on an array of analytical techniques, which, when combined, confirmed the chemical and mineral phases derived from the source materials, smelting processes, and subsequent interment, leaching, and corrosion activities. Specific ions, such as iron, silicate, and arsenate tetrahedra, along with chlorides and contributions from oxides including arsenic, calcium, and sodium, played distinct roles in elucidating the complexities of the slags and their current state of preservation. Phases formed during smelting, such as magnetite, fayalite, and copper sulfide droplets, along with their high-temperature reactions that sequestered arsenic in calcium arsenates and within the fayalite slag matrix, helped explain prior scholarship on the presence of iron arsenides in ancient Iranian slag, and perhaps more, which is a function of phase stabilities, thermodynamics, and mass transport. The presence of secondary precipitates, such as calcite, atacamite, and scorodite, in the context of thermodynamics and considering environmental conditions in Semnan, Iran, helped determine the types of secondary precipitates identified. Despite the success of XCT in this study, it cannot replace destructive analysis, as it cannot be used to assess phase or composition. Nevertheless, it has now been demonstrated to be indispensable in planning slag analyses and interpreting the significance behind traditional analytical results. In addition to offering a non-destructive and non-invasive method for analyzing the internal structure of slags, the technique further provides digital copies of the artifacts, which are particularly important given their limited number and availability for destructive analyses.
Supporting information
S4 Appendix. Thermodynamic data for pourbaix diagrams.
https://doi.org/10.1371/journal.pone.0336603.s004
(XLSX)
S5 Appendix. The distribution of arsenic and formation of iron arsenides.
https://doi.org/10.1371/journal.pone.0336603.s005
(DOCX)
Acknowledgments
Firstly, we thank Vincent Pigott and Christoper Thornton, with whom we discussed speiss, Hissar, and archaeometallurgy. We would also like to thank the staff and our colleagues at the Penn Museum for facilitating the slag loan agreement and permitting their sectioning. This work was furthermore supported by William Gilstrap and Jenny Meanwell at the Center for Materials Research in Archaeology and Ethnology (CMRAE), who instructed on how to section, polish, and prepare the slag for analysis. Lastly, we thank Kevin Cedrone at Lumafield for his consultation regarding use of the Neptune CT scanner. Part of this work was made possible thanks to the support of the MIT Human Insight Collaborative (MITHIC).
References
- 1. Hughes S. CT scanning in archaeology. In: Computed tomography - special applications. InTech; 2011.
- 2. Harwood-Nash DC. Computed tomography of ancient Egyptian mummies. J Comput Assist Tomogr. 1979;3(6):768–73. pmid:389964
- 3. Cantini F, Creange S, Li Y, van Eijck L, Kardjilov N, Kabra S, et al. Morphological and microstructural characterization of an ancient Chola bronze statuette by neutron-based non-invasive techniques. Archaeol Anthropol Sci. 2024;16(3):45.
- 4. Fedrigo A, Grazzi F, O’Flynn D, Kockelmann W, Cantini F, Scherillo A. How can neutron imaging contribute to heritage science? An overview at the ISIS neutron and muon source. J Phys: Conf Ser. 2023;2605(1):012019.
- 5. Festa G, Romanelli G, Senesi R, Arcidiacono L, Scatigno C, Parker SF, et al. Neutrons for cultural heritage-techniques, sensors, and detection. Sensors (Basel). 2020;20(2):502. pmid:31963186
- 6. Applbaum N, Applbaum YH. The use of medical computed tomography (CT) imaging in the study of ceramic and clay archaeological artifacts from the ancient near east. In: X-rays for archaeology. Berlin/Heidelberg: Springer-Verlag; 2005. p. 231–45.
- 7. Bossema FG, Palenstijn WJ, Heginbotham A, Corona M, van Leeuwen T, van Liere R, et al. Enabling 3D CT-scanning of cultural heritage objects using only in-house 2D X-ray equipment in museums. Nat Commun. 2024;15(1):3939. pmid:38744870
- 8. Böhnel M, Reims N, Prjamkov D, Salamon M. High Energy CT applications for cultural heritage. eJNDT. 2023;28(3):2022–3.
- 9. Sun W, Symes DR, Brenner CM, Böhnel M, Brown S, Mavrogordato MN, et al. Review of high energy x-ray computed tomography for non-destructive dimensional metrology of large metallic advanced manufactured components. Rep Prog Phys. 2022;85(1). pmid:35138267
- 10. de Lapérouse J-F, Eppihimer M, Flisch A, Zboray R. Revealing ancient technology: a high-energy X-ray computed tomography examination of a Mesopotamian copper alloy head. Herit Sci. 2024;12(1):307.
- 11. Uda M, Demortier G, Nakai I, editors. X-rays for archaeology. Berlin/Heidelberg: Springer-Verlag; 2005.
- 12. Zhang X, Lei Y, Cheng Q, Zhou G. Application of Computed Tomography in the analysis of the manufacture of eye beads technique. Microchem J. 2020;156:104798.
- 13. Bettuzzi M, Casali F, Morigi MP, Brancaccio R, Carson D, Chiari G, et al. Computed tomography of a medium size Roman bronze statue of Cupid. Appl Phys A. 2015;118(4):1161–9.
- 14. Nicolas T, Gaugne R, Tavernier C, Gouranton V, Arnaldi B. Preservative approach to study encased archaeological artefacts. In: Lecture notes in computer science. Springer International Publishing; 2014. p. 332–41.
- 15.
Scott DA, Podany J, Considine BB, Museum JPG, Institute GC. Ancient & historic metals: conservation and scientific research: proceedings of a symposium organized by the J. Paul Getty Museum and the Getty Conservation Institute; 1994.
- 16. Hawass Z, Saleem SN. Mummified daughters of King Tutankhamun: archeologic and CT studies. AJR Am J Roentgenol. 2011;197(5):W829-36. pmid:22021529
- 17. Czeibert K, Nagy G, Csörgő T, Donkó T, Petneházy Ö, Csóka Á, et al. High-resolution computed tomographic (HRCT) image series from 413 canid and 18 felid skulls. Sci Data. 2024;11(1):753. pmid:39013883
- 18. Geijer H, Ndongozi F, Edvardsson J. Dendrochronology with a medical X-ray photon counting computed tomography scanner. Dendrochronologia. 2024;86:126233.
- 19. Stelzner J, Million S, Stelzner I, Martinez-Garcia J, Gwerder D, Nelle O, et al. Non-destructive dendrochronology with X-ray computed tomography: the influence of different conservation methods for waterlogged archaeological wood. Dendrochronologia. 2023;78:126065.
- 20. Martinez-Garcia J, Stelzner I, Stelzner J, Gwerder D, Million S, Nelle O, et al. Curvilinear-tree-ring measurements in archaeological wood samples from X-ray computed tomography. Dendrochronologia. 2022;76:126002.
- 21. Stelzner J, Million S. X-ray Computed Tomography for the anatomical and dendrochronological analysis of archaeological wood. J Archaeol Sci. 2015;55:188–96.
- 22. Stelzner J, Million S, Stelzner I, Nelle O, Banck-Burgess J. Micro-computed tomography for the identification and characterization of archaeological lime bark. Sci Rep. 2023;13(1):6458. pmid:37081053
- 23. Martinez-Garcia J, Stelzner I, Stelzner J, Gwerder D, Schuetz P. Automated 3D tree-ring detection and measurement from X-ray computed tomography. Dendrochronologia. 2021;69:125877.
- 24. Dierickx S, Genbrugge S, Beeckman H, Hubau W, Kibleur P, Van den Bulcke J. Non-destructive wood identification using X-ray µCT scanning: which resolution do we need? Plant Methods. 2024;20(1):98. pmid:38915095
- 25. Vigorelli L, Re A, Guidorzi L, Brancaccio R, Bortolin C, Grassi N, et al. The study of ancient archaeological finds through X-ray tomography: the case of the “Tintinnabulum” from the Museum of Anthropology and Ethnography of Torino. J Phys: Conf Ser. 2022;2204(1):012034.
- 26. Emami M, Kowald T, Trettin R. Preliminary archaeometallurgical studies on copper extraction from polymetallic ore sources in Meymand, south central Iranian desert. Archaeol Anthropol Sci. 2016;9(7):1515–28.
- 27. Hall S, Miller D, Anderson M, Boeyens J. An exploratory study of copper and iron production at Marothodi, an early 19th century Tswana town, Rustenburg District, South Africa. JAA. 2006;4(1):3–35.
- 28. Sodaei B, Kashani P. Analytical assessment of chaltasian slag: evidence of early copper production in the Central Plateau of Iran. IANSA. 2017;VIII(2/2017):137–44.
- 29. Bourgarit D. Mineralogy of slags: a key approach for our understanding of ancient copper smelting processes. In: The contribution of mineralogy to cultural heritage. Mineralogical Society of Great Britain and Ireland; 2019. p. 203–32.
- 30. Hauptmann A. Archaeometallurgical slags and other debris. Archaeometallurgy - materials science aspects; 2020. p. 199–293.
- 31. Hauptmann A. The investigation of archaeometallurgical slag. In: Roberts BW, Thornton CP, editors. Archaeometallurgy in global perspective. New York (NY): Springer New York; 2014. p. 91–105.
- 32.
Bachmann HG. The identification of slags from archaeological sites. London: Institute of Archaeology; 1982.
- 33.
Haubner R, Strobl S, Trebsche P. Metallographic analyses from the late Urnfield period copper mining settlement at Prigglitz-Gasteil in Lower Austria. In: Alpine Copper II - Alpenkupfer II - Rame delle Alpi II - Cuivre des Alpes II, new results and perspectives on prehistoric copper production. Bochum: VML Verlag Marie Leidorf GmbH; 2019. p. 323–32.
- 34. Lottermoser BG. Evaporative mineral precipitates from a historical smelting slag dump, Rio Tinto, Spain. NJMA. 2005;181(2):183–90.
- 35. Sueoka Y, Sakakibara M. Primary phases and natural weathering of smelting slag at an abandoned mine site in Southwest Japan. Minerals. 2013;3(4):412–26.
- 36. Kupczak K, Warchulski R, Gawęda A, Ślęzak M, Migas P. The use of predominance area diagrams (PAD) to determine the oxygen and sulfur fugacities prevailing during historical metallurgical processes: the case of fifteenth to seventeenth century copper slags from Polichno (Old Polish industrial district). Herit Sci. 2024;12(1):49.
- 37. Jalkanen H, Vehviläinen J, Poijärvi J. Copper in solidified copper smelter slags. Scand J Metall. 2003;32(2):65–70.
- 38. Kierczak J, Pietranik A. Mineralogy and composition of historical cu slags from the rudawy janowickie mountains, Southwestern Poland. Can Mineral. 2011;49(5):1281–96.
- 39. Kierczak J, Pietranik A, Piatak NM. CHAPTER 4. Weathering of slags. In: Weathering of slags. Royal Society of Chemistry; 2021. p. 125–50.
- 40. Potysz A, van Hullebusch ED, Kierczak J, Grybos M, Lens PNL, Guibaud G. Copper metallurgical slags – current knowledge and fate: a review. Crit Rev Environ Sci Technol. 2015;45(22):2424–88.
- 41. Fernández-Caliani JC, Pérez-Macías JA. Tracing soil contamination from pre-roman slags at the monte romero archaeological site, Southwest Spain. Soil Syst. 2024;8(3):78.
- 42. Ettler V, Johan Z, Kříbek B, Šebek O, Mihaljevič M. Mineralogy and environmental stability of slags from the Tsumeb smelter, Namibia. Appl Geochem. 2009;24(1):1–15.
- 43. Thornton CP. The chalcolithic and Early Bronze Age metallurgy of Tepe Hissar, Northeast Iran: a challenge to the levantine paradigm. Anthropology. 2009.
- 44. Thornton CP. The emergence of complex metallurgy on the Iranian Plateau. In: Archaeometallurgy in global perspective. New York (NY): Springer New York; 2014. p. 665–96.
- 45. Gürsan-Salzmann A. Appendix 4. Radiocarbon dates from the 1976 restudy project. In: The new chronology of the Bronze Age settlement of Tepe Hissar, Iran. University of Pennsylvania Press; 2016. p. 371–4.
- 46. Pigott VC, Howard SM, Epstein SM. Pyrotechnology and culture change at Bronze Age Tepe Hissar (Iran). Early Pyrotechnology: The Evolution of the First Fire-Using Industries: Papers Presented at a Seminar on early Pyrotechnology. Washington (DC): Smithsonian Institution, Gaithersburg (MD): National Bureau of Standards; 1982. p. 215–36.
- 47.
Helwing B. Early metallurgy in Iran: an innovative region as seen from the inside. In: Metal matters: innovative technologies and social change in prehistory and antiquity; 2013. p. 105–35.
- 48.
Gürsan-Salzmann A. Analysis of ceramic complexes of the main mound and the north flat. In: The new chronology of the Bronze Age settlement of Tepe Hissar, Iran. University of Pennsylvania Museum of Archaeology and Anthropology; 2016.
- 49. Thornton CP, Rehren Th, Pigott VC. The production of speiss (iron arsenide) during the Early Bronze Age in Iran. J Archaeol Sci. 2009;36(2):308–16.
- 50. Mendoza D, Hino M, Itagaki K. Distribution equilibria between Cu-Fe-As ternary speiss and slag phases. Shigen-to-Sozai. 2002;118(3/4):197–201.
- 51.
Mendoza DG. Thermodynamic study of arsenic and antimony in molten copper matte, speiss and slag. 2002:114–20.
- 52. Rademakers FW, Vandenbeusch M, Vassilieva E, Vanhaecke F, Degryse P. Speiss at Amarna (Egypt, c. 1353–1336 BCE) – Exotic anachronism or cherished commodity? J Archaeol Sci. 2024;170:106043.
- 53.
Weeks L. Iran in the 4th millennium: metallurgy. 2009:52–6.
- 54. Rehren T, Boscher L, Pernicka E. Large scale smelting of speiss and arsenical copper at Early Bronze Age Arisman, Iran. J Archaeol Sci. 2012;39(6):1717–27.
- 55. Weeks L. Iranian metallurgy of the fourth millennium BC in its wider technological and cultural contexts. In: Petrie CA, editor. Ancient Iran and its neighbours. Oxbow Books; 2019. p. 277–92.
- 56. Nezafati N, Pernicka E, Helwing B, Kirchner D. Reassessment of the prehistoric metallurgy at Arisman, Central Iran. On salt, copper and gold. In: MOM Éditions; 2021. p. 231–60.
- 57. Thornton CP, Lamberg-Karlovsky CC. A new look at the prehistoric metallurgy of Southeastern Iran. Iran. 2004;42:47.
- 58. Seitkan A, Lampronti GI, Widmer RN, Casati NPM, Redfern SAT. Thermal behavior of iron arsenides under non-oxidizing conditions. ACS Omega. 2020;5(12):6423–8. pmid:32258877
- 59. Hauptmann A, Rehren T, Schmitt-Strecker S. Early Bronze Age copper metallurgy at Shahr-i-Sokhta (Iran), reconsidered. Man Min. 2003:197–213.
- 60. Emami M. Toroud, the late motion for As-Sb bearing Cu production from 2nd millennium BC in Iran: an archaeometallurgical approach. MAA. 2014;14:185–204. Available from: https://www.researchgate.net/publication/263505740
- 61. De Ryck I, Adriaens A, Adams F. An overview of Mesopotamian bronze metallurgy during the 3rd millennium BC. J Cult Herit. 2005;6(3):261–8.
- 62.
Dyson RH, Howard SM, editor. Tappeh Hesar: reports of the restudy project, 1976/ ed. by Robert H. Dyson and Susan M. Howard. Tappeh Hesar: reports of the restudy project, 1976. Firenze: Casa editrice Le Lettere; 1989.
- 63.
Gürsan-Salzmann A. Tepe Hissar, an introduction. In: The new chronology of the Bronze Age settlement of Tepe Hissar, Iran. University of Pennsylvania Press; 2016. p. 1–12. doi:10https://doi.org/10.9783/9781934536841-006
- 64.
Gürsan-Salzmann A. Concluding remarks. The new chronology of the Bronze Age settlement of Tepe Hissar, Iran. University of Pennsylvania Press; 2016. p. 267–98. Available from: http://www.jstor.org/stable/j.ctt1c0gmq8.14
- 65. Addis A, Angelini I, Nimis P, Artioli G. Late Bronze Age copper smelting slags from Luserna (Trentino, Italy): interpretation of the metallurgical process. Archaeometry. 2015;58(1):96–114.
- 66.
Piatak NM, Ettler V, Hoppe D. CHAPTER 3. Geochemistry and mineralogy of slags. In: Geochemistry and mineralogy of slags. The Royal Society of Chemistry; 2021. p. 59–124. doi:10https://doi.org/10.1039/9781839164576-00059
- 67. Berger D, Brügmann G, Friedrich R, Lutz J, Meyer H-P, Pernicka E. Shiny bronze in glassy matter: an inconspicuous piece of slag from the Bronze Age mining site of Mušiston (Tajikistan) and its significance for the development of tin metallurgy in Central Asia. Archaeol Anthropol Sci. 2022;14(8):150.
- 68. Bossema FG, Coban SB, Kostenko A, van Duin P, Dorscheid J, Garachon I, et al. Integrating expert feedback on the spot in a time-efficient explorative CT scanning workflow for cultural heritage objects. J Cult Herit. 2021;49:38–47.
- 69. Lanska DJ. Hounsfield, Godfrey N. In: Encyclopedia of the neurological sciences. Elsevier; 2014. p. 616–7.
- 70. Modelling M, Science E. Medical imaging 2 2.1; 2023.
- 71.
Kak AC. Principles of tomographic imaging. Society for Industrial and Applied Mathematics; 2001.
- 72. Kong Y, Kato M, Kurumisawa K. Recent advances in X-ray computed tomography for alkali-activated materials: a review. ACT. 2023;21(8):573–95.
- 73. Bernabale M, Cognigni F, Mancini C, Proietti A, Mura F, Montanari D, et al. 3D fractures analysis and conservation assessment of wrought iron javelin through advanced non-invasive techniques. Sci Rep. 2023;13(1):10142. pmid:37349367
- 74. Oudbashi O, Mishmastnehi M. Archaeometallurgy of copper in the Middle Elamite period of southwestern Iran: analytical investigation of various parts of the copper production in Haft Tappeh. J Archaeol Sci Rep. 2020;30:102216.
- 75. Baqershahi M, Samet HF. Revision of the history of smelting and mining of copper in the center of the Iranian plateau based on the evidence and petrographic tests of the slag of Gazanak Zarandieh. AIS. 2023;2:35–47.
- 76. Mohammadi Arasteh S, Shoaei SM. A comprehensive review of the salinity assessment in groundwater resources of Iran. Acta Geophys. 2023;72(1):385–403.
- 77. Ghorbani M. The economic geology of Iran. In: The economic geology of Iran: mineral deposits and natural resources. Dordrecht: Springer Netherlands; 2013.
- 78. The Caspian Sea treaty. Strat Comments. 2018;24(9):i–ii.
- 79. Kulekeyev ZhA, Nurtayeva GKh, Mustafin ES, Pudov AM, Zharikessov G, Taylor P, et al. Studies in support of the regulation of dispersant use in the Kazakhstan Sector of the Caspian Sea (KSCS). Int Oil Spill Conf Proc. 2014;2014(1):463–75.
- 80. Dillmann P, Béranger G, Piccardo P, Matthiesen H. Corrosion of metallic heritage artefacts. In: Dillmann P, Béranger G, Piccardo P, Matthiesen H, editors. Corrosion of metallic heritage artefacts: investigation, conservation and prediction of long term behaviour. New York: Woodhead Publishing Limited; 2007.
- 81. Saheb M, Neff D, Bellot‐Gurlet L, Dillmann P. Raman study of a deuterated iron hydroxycarbonate to assess long‐term corrosion mechanisms in anoxic soils. J Raman Spectroscopy. 2010;42(5):1100–8.
- 82. Dick JM. CHNOSZ: thermodynamic calculations and diagrams for geochemistry. Front Earth Sci. 2019;7:1–18.
- 83.
King F, Ahonen L, Taxén C, Vuorinen U. Copper corrosion under expected conditions in a deep geologic repository. Technical Report TR-01-23; 2001.
- 84.
King F, Lilja C, Pedersen K, Pitkänen P, Vähänen M, Oy P. An update of the state-of-the-art report on the corrosion of copper under expected conditions in a deep geologic repository. Technical Report TR-10-67; 2010. Available from: www.skb.se
- 85.
Beverskog B, Pettersson S-O. Pourbaix diagrams for copper in 5 m chloride solution. SKI Report; 2002.
- 86. Kidari O, Chartrand P. Thermodynamic evaluation and optimization of the As–Co, As–Fe and As–Fe–S systems. Calphad. 2023;82:102589.
- 87. Lu P, Zhu C. Arsenic Eh–pH diagrams at 25°C and 1 bar. Environ Earth Sci. 2011;62: 1673–83.
- 88. Liu Y, Ng WS, Chen M. Thermodynamic analysis of the immobilisation of arsenic during the pressure oxidation and curing processes. Miner Eng. 2022;185:107681.
- 89. Marini L, Accornero M. Prediction of the thermodynamic properties of metal–arsenate and metal–arsenite aqueous complexes to high temperatures and pressures and some geological consequences. Environ Geol. 2006;52(7):1343–63.
- 90. Nordstrom DK, Majzlan J, Konigsberger E. Thermodynamic properties for arsenic minerals and aqueous species. Rev Mineral Geochem. 2014;79(1):217–55.
- 91. Charykova MV, Krivovichev VG. Mineral systems and thermodynamic stability of arsenic minerals in the environment. In: Lecture notes in earth system sciences. Springer International Publishing; 2019. p. 259–76.
- 92. Mashadiyeva LF, Babanly DM, Hasanova ZT, Yusibov YA, Babanly MB. Phase relations in the Cu-As-S system and thermodynamic properties of copper-arsenic sulfides. J Phase Equilib Diffus. 2024;45(3):567–82.
- 93. Babanly MB, Gasanova ZT, Mashadieva LF, Zlomanov VP, Yusibov YuA. Thermodynamic study of the Cu-As-S system by EMF measurements with Cu4RbCl3I2 as a solid electrolyte. Inorg Mater. 2012;48(3):225–8.
- 94. Babanly MB, Muradova GV, Il’yasly TM, Babanly DM. Thermodynamic properties of arsenic sulfides studied by EMF measurements. Inorg Mater. 2011;47(3):227–30.
- 95. Robie R, Hemingway BS. Thermodynamic properties of minerals and related substances at 298.15K and 1 bar. US Geological Survey Bulletin; 1995. p. 1–461.
- 96. Lee S, Xu H. The crystal structure and Gibbs free energy of formation of chukanovite as an oxidation product of carbon steel in human liver. Chem Geol. 2018;488:180–8.
- 97. Nishimura T, Dong J. Corrosion behavior of carbon steel for overpack in groundwater containing bicarbonate ions. JPES. 2009;3(1):23–30.
- 98.
Boscher LC. Reconstructing the arsenical copper production process in early Bronze Age southwest Asia. University College London; 2016.
- 99. Thornton CP, Rehren T. A truly refractory crucible from fourth millennium Tepe Hissar, Northeast Iran. J Archaeol Sci. 2009;36(12):2700–12.
- 100. Nezafati N, Bochum DB, Pernicka E. Polymetallic ore and the simultaneous silver-copper production in the late chalcolithic-early Bronze Age of Central Iran. Metalla. 2023;12:67–8.
- 101. Taskinen P, Jokilaakso A. Reaction sequences in flash smelting and converting furnaces: an in-depth view. Metall Mater Trans B. 2021;52(5):3524–42.
- 102. Emami M, Dehno SE, Eskandari N, Pritzel C. Tracing the pyro-technological evidence during the 3rd millennium bc in “Shahdad” through archaeometallurgical remains. Archeometriai Műhely. 2024;21:253–64.
- 103. Kupczak K, Warchulski R, Gawęda A. Reconstruction of smelting conditions during 16th‐ to 18th‐century copper ore processing in the Kielce region (Old Polish Industrial District) based on slags from Miedziana Góra, Poland. Archaeometry. 2022;65(3):547–69.
- 104. Thornton CP, Gürsan-Salzmann A, Dyson RH. Tepe Hissar and the fourth millennium bc of North-Eastern Iran. In: Ancient Iran and its neighbours. Oxbow Books; 2019. p. 131–44.
- 105.
Bourgarit D, Mille B, Prange M, Ambert P, Hauptmann a. Chalcolithic fahlore smelting at Cabrières: reconstruction of smelting processes by archaeometallurgical finds. In: Archaeometallurgy in Europe; 2003. p. 431–40.
- 106. Greener A, Ben-Yosef E. Calcium content in metallurgical slag as a proxy for fuel efficiency of ancient copper smelting technologies. JLS. 2016;2:66–76.
- 107. Rostoker W, Pigott VC, Dvorak JR. Direct reduction to metal by oxide/sulphide mineral interaction. Archaeomaterials. 1989;3:69–87.
- 108. Plaza MT, Garrido F, Larreina-García D. A new piece of the puzzle: slag and ore analysis to reconstruct the prehispanic smelting technology at the Atacama Desert, Chile. Herit Sci. 2023;11(1):171.
- 109. Mohamadi Nasab S, Shafiei Bafti B, Yarahmadi MR, Mahmoudi Maymand M, Kamalabadi Khorasani J. Mineralogical properties of the copper slags from the SarCheshmeh smelter plant, Iran, in view of value recovery. Minerals. 2022;12(9):1153.
- 110. Gorai B, Jana RK, Premchand . Characteristics and utilisation of copper slag—a review. Resour Conserv Recycl. 2003;39(4):299–313.
- 111. Dzinomwa G, Mapani B, Nghipulile T, Maweja K, Kurasha JT, Amwaama M, et al. Mineralogical characterization of historic copper slag to guide the recovery of valuable metals: a namibian case study. Materials (Basel). 2023;16(18):6126. pmid:37763403
- 112. Shamsuddin M. Physical chemistry of metallurgical processes. Wiley; 2016.
- 113. Krismer M, Töchterle U, Goldenberg G, Tropper P, Vavtar F. Mineralogical and petrological investigations of early Bronze Age copper‐smelting remains from the kiechlberg (Tyrol, Austria). Archaeometry. 2012;55(5):923–45.
- 114. Ku J, Zhang L, Fu W, Wang S, Yin W, Chen H. Mechanistic study on calcium ion diffusion into fayalite: a step toward sustainable management of copper slag. J Hazard Mater. 2021;410:124630. pmid:33243648
- 115. Tian H, Guo Z, Pan J, Zhu D, Yang C, Xue Y, et al. Comprehensive review on metallurgical recycling and cleaning of copper slag. Resour Conserv Recycl. 2021;168:105366.
- 116. Zuo Z, Feng Y, Luo S, Dong X, Li X, Ren D, et al. Element distribution and migration behavior in the copper slag reduction and separation process. Front Energy Res. 2021;9:1–10.
- 117. Erdenebold U, Choi H-M, Wang J-P. Recovery of pig iron from copper smelting slag by reduction smelting. Arch Metall Mater. 2018;63:1793–8.
- 118. Guo Z, Zhu D, Pan J, Wu T, Zhang F. Improving beneficiation of copper and iron from copper slag by modifying the molten copper slag. Metals. 2016;6(4):86.
- 119. Majzlan J, Drahota P, Filippi M. Parageneses and crystal chemistry of arsenic minerals. Rev Mineral Geochem. 2014;79(1):17–184.
- 120. Drever JI, editor. The chemistry of weathering. Series C: mathematical and physical sciences, vol. 149. D. Reidel Publishing Company; 1985.
- 121. Jia W, Li B, Zhou S, Wei Y, Wang H. Distribution behavior and deportation of arsenic in copper top-blown smelting process. Metall Res Technol. 2023;120(5):507.
- 122. Zhao Z, Wang Z, Xu W, Qin W, Lei J, Dong Z, et al. Arsenic removal from copper slag matrix by high temperature sulfide-reduction-volatilization. J Hazard Mater. 2021;415:125642. pmid:34088174
- 123. Ettler V, Vítková M. CHAPTER 5. Slag leaching properties and release of contaminants. In: Slag leaching properties and release of contaminants. The Royal Society of Chemistry; 2021. p. 151–73.
- 124. Gupta CK. Pyrometallurgy. In: Chemical metallurgy. Wiley; 2003. p. 343–457.
- 125. Ren H, Min X, Ke Y, Wang Y, Peng C, Li Y, et al. Mechanism study of lime decomposing fayalite melt: thermodynamic basis for the sustainable and efficient recovery of iron resources from copper slag. J Clean Prod. 2024;462:142678.
- 126. Isaksson J, Andersson A, Vikström T, Lennartsson A, Samuelsson C. Improved settling mechanisms of an industrial copper smelting slag by CaO modification. J Sustain Metall. 2023;9(3):1378–89.
- 127. Zhang H, Li B, Wei Y, Wang H. The settling behavior of matte particles in copper slag and the new technology of copper slag cleaning. J Mater Res Technol. 2021;15:6216–30.
- 128. Rankin KE, Hazell ZJ, Middleton AM, Mavrogordato MN. Micro-focus X-ray CT scanning of two rare wooden objects from the wreck of the London, and its application in heritage science and conservation. J Archaeol Sci Rep. 2021;39:103158.
- 129. Lehmann EH, Deschler‐Erb E, Ford A. Neutron tomography as a valuable tool for the non‐destructive analysis of historical bronze sculptures. Archaeometry. 2010;52(2):272–85.
- 130. Verma PK. Optical mineralogy. Cham: Springer Nature Switzerland; 2023.
- 131. Braekmans D, Degryse P. Petrography: optical microscopy. In: The Oxford handbook of archaeological ceramic analysis; 2017.
- 132. Warr LN. IMA–CNMNC approved mineral symbols. MinMag. 2021;85(3):291–320.
- 133. Gilstrap WD, Meanwell JL, Paris EH, López Bravo R, Day PM. Post-depositional alteration of calcium carbonate phases in archaeological ceramics: depletion and redistribution effects. Minerals. 2021;11(7):749.
- 134. Kanth AP, Singh MR, Mani BR. Archaeometallurgical characterisation of ancient copper slags from pre-Harappan site, Kunal, India. Anal Sci Adv. 2022;3(7–8):226–34. pmid:38716081
- 135. Kasina M, Kowalski PR, Michalik M. Mineral carbonation of metallurgical slags. Mineralogia. 2015;45(1–2):27–45.
- 136.
Hauptmann A. The investigation of archaeometallurgical slag. In: Roberts BW, Thornton CP, editors. Archaeometallurgy in global perspective: methods and syntheses. New York (NY): Springer New York; 2014. p. 91–105.
- 137. Haubner R, Strobl S, Trebsche P. Materialographic investigations of plate slags from the Late Bronze Age copper production site of Prigglitz-Gasteil (Lower Austria). J Archaeol Sci Rep. 2023;48:103838.
- 138. Drahota P, Filippi M. Secondary arsenic minerals in the environment: a review. Environ Int. 2009;35(8):1243–55. pmid:19665230
- 139. Das B. Theoretical study of formation of secondary arsenic minerals: scorodite and pharmacosiderite. ACS Earth Space Chem. 2018;3(2):192–201.
- 140. Jelić I, Zavašnik J, Lazarov M, Zdravković A, Kovač S, Stojanović J, et al. Micro- to nanoscale textures of gold in arsenopyrite and scorodite from the As-Au-Bi assemblage of Drenjak locality (Serbia). Ore Geol Rev. 2023;163:105711.
- 141. Langmuir D, Mahoney J, Rowson J. Solubility products of amorphous ferric arsenate and crystalline scorodite (FeAsO4·2H2O) and their application to arsenic behavior in buried mine tailings. Geochim Cosmochim Acta. 2006;70(12):2942–56.
- 142. Tang Z, Tang X, Xiao Z, Liu H. Mechanism and thermodynamics of scorodite formation by oxidative precipitation from arsenic-bearing solution. Environ Res. 2024;250:118500. pmid:38387492
- 143. Saheb M, Neff D, Dillmann Ph, Matthiesen H, Foy E. Long-term corrosion behaviour of low-carbon steel in anoxic environment: characterisation of archaeological artefacts. J Nucl Mater. 2008;379(1–3):118–23.
- 144. Saheb M, Neff D, Dillmann P, Descostes M, Matthiesen H. Long-term anoxic corrosion of iron. In: Corrosion and conservation of cultural heritage metallic artefacts. Elsevier; 2013. p. 260–84.
- 145. Azoulay I, Rémazeilles C, Refait Ph. Determination of standard Gibbs free energy of formation of chukanovite and Pourbaix diagrams of iron in carbonated media. Corros Sci. 2012;58:229–36.
- 146. Saheb M, Descostes M, Neff D, Matthiesen H, Michelin A, Dillmann P. Iron corrosion in an anoxic soil: comparison between thermodynamic modelling and ferrous archaeological artefacts characterised along with the local in situ geochemical conditions. Appl Geochem. 2010;25(12):1937–48.
- 147. Toffolo L, Addis A, Martin S, Nimis P, Rottoli M, Godard G. The Misérègne slag deposit (Valle d’Aosta, Western Alps, Italy): insights into (pre-)Roman copper metallurgy. J Archaeol Sci: Rep. 2018;19:248–60.
- 148. Ettler V, Mihaljevič M, Jarošíková A, Culka A, Kříbek B, Majer V, et al. Vanadium-rich slags from the historical processing of Zn–Pb–V ores at Berg Aukas (Namibia): Mineralogy and environmental stability. Appl Geochem. 2020;114:104473.
- 149. Bril H, Zainoun K, Puziewicz J, Courtin-Nomade A, Vanaecker M, Bollinger J-C. Secondary phases from the alteration of a pile of zinc-smelting slag as indicators of environmental conditions: an example from swietochlowice, Upper Silesia, Poland. Can Mineral. 2008;46(5):1235–48.
- 150. Villalobos M, García-Payne DG, López-Zepeda JL, Ceniceros-Gómez AE, Gutiérrez-Ruiz ME. Natural arsenic attenuation via metal arsenate precipitation in soils contaminated with metallurgical wastes: I. Wet chemical and thermodynamic evidences. Aquat Geochem. 2009;16(2):225–50.
- 151. Ghorbani M. The geology of Iran: tectonic, magmatism and metamorphism. Cham: Springer International Publishing; 2021.
- 152. Thalhammer O, Stumpfl EF, Panayiotou A. Postmagmatic, hydrothermal origin of sulfide and arsenide mineralizations at Limassol Forest, Cyprus. Miner Depos. 1986;21(2):95–105.
- 153. Okamoto H. The As-Fe (arsenic-iron) system. JPE. 1991;12(4):457–61.
- 154. Warchulski R, Kałaska M, Rizzuto BC, Sierpień P, Pisarek M, Kaproń G, et al. In‐depth study of a speiss/matte sample from Castillo de Huarmey, North Coast of Peru, and its implications for the pre‐Columbian production of arsenic bronze in the Central Andes. Archaeometry. 2024;67(1):112–31.
- 155. Zhang H, Wang Y, He Y, Xu S, Hu B, Cao H, et al. Efficient and safe disposition of arsenic by incorporation in smelting slag through copper flash smelting process. Miner Eng. 2021;160:106661.
- 156. Reddy RG, Font JM. Arsenate capacities of copper smelting slags. Metall Mater Trans B. 2003;34(5):565–71.
- 157. Movahednia M, Maghfouri S, Fazli N, Rastad E, Ghaderi M, González FJ. Metallogeny of Manto-type stratabound Cu-(Ag) mineralization in Iran: relationship with NeoTethyan evolution and implications for future exploration. Ore Geol Rev. 2022;149:105064.
- 158. Nezafati N, Stöllner T. Economic geology, mining archaeological and archaeometric investigations at the veshnaveh ancient copper mine, Central Iran. Metalla. 2018;23(2):67–90.
- 159. Miki H, Nicol M, Velásquez-Yévenes L. The kinetics of dissolution of synthetic covellite, chalcocite and digenite in dilute chloride solutions at ambient temperatures. Hydrometallurgy. 2011;105(3–4):321–7.
- 160. Rushworth DD, Christl I, Kumar N, Hoffmann K, Kretzschmar R, Lehmann MF, et al. Copper mobilisation from Cu sulphide minerals by methanobactin: effect of pH, oxygen and natural organic matter. Geobiology. 2022;20(5):690–706. pmid:35716154
- 161.
Grigoriev S. Metallurgical production in Northern Eurasia in the Bronze Age. 2015.
- 162.
Scott DA. Copper and bronze in art: corrosion, colorants, conservation. Getty Conservation Institute; 2002.
- 163. Kurmaev EZ, Ek J van, Ederer DL, Zhou L, Callcott TA, Perera RCC, et al. Experimental and theoretical investigation of the electronic structure of transition metal sulphides: CuS, and. J Phys: Condens Matter. 1998;10(7):1687–97.
- 164. Pronti L, Felici AC, Alesiani M, Tarquini O, Bracciale MP, Santarelli ML, et al. Characterisation of corrosion layers formed under burial environment of copper-based Greek and Roman coins from Pompeii. Appl Phys A. 2015;121(1):59–68.
- 165.
Scott DA, Schwab R. Metallography in archaeology and art. Cultural Heritage Science; 2019. Available from: http://link.springer.com/10.1007/978-3-030-11265-3%0Ahttp://files/468/ScottandSchwab-2019-MetallographyinArchaeologyandArt.pdf
- 166. Wanhill RJH, Oudbashi O. Fractography of ancient metallic artifacts, and restoration and conservation aspects. In: Fractography. ASM International; 2024. p. 49–63.
- 167. Hannington MD. The formation of atacamite during weathering of sulfides on the modern seafloor. Can Mineral. 1993;31:945–56.
- 168. Casaletto MP, De Caro T, Ingo GM, Riccucci C. Production of reference “ancient” Cu-based alloys and their accelerated degradation methods. Appl Phys A. 2006;83(4):617–22.
- 169. Cheng CY, Lawson F. The kinetics of leaching chalcocite in acidic oxygenated sulphate-chloride solutions. Hydrometallurgy. 1991;27(3):249–68.
- 170. Niu X, Ruan R, Tan Q, Jia Y, Sun H. Study on the second stage of chalcocite leaching in column with redox potential control and its implications. Hydrometallurgy. 2015;155:141–52.
- 171. Nicol M, Basson P. The anodic behaviour of covellite in chloride solutions. Hydrometallurgy. 2017;172:60–8.
- 172. Pérez K, Toro N, Saldaña M, Salinas-Rodríguez E, Robles P, Torres D, et al. Statistical study for leaching of covellite in a chloride media. Metals. 2020;10(4):477.
- 173. Todd EC, Sherman DM. Surface oxidation of chalcocite (Cu2S) under aqueous (pH = 2–11) and ambient atmospheric conditions: mineralogy from CuL- and OK-edge X-ray absorption spectroscopy. Am Miner. 2003;88(11–12):1652–6.
- 174. Ruiz MC, Abarzúa E, Padilla R. Oxygen pressure leaching of white metal. Hydrometallurgy. 2007;86(3–4):131–9.
- 175. Toro N, Moraga C, Torres D, Saldaña M, Pérez K, Gálvez E. Leaching chalcocite in chloride media—a review. Minerals. 2021;11(11):1197.
- 176. Ruiz MC, Gallardo E, Padilla R. Copper extraction from white metal by pressure leaching in H2SO4–FeSO4–O2. Hydrometallurgy. 2009;100(1–2):50–5.
- 177. Vračar RŽ, Cerović KP. Kinetics of oxidation of Fe(II) ions by gaseous oxygen at high temperatures in an autoclave. Hydrometallurgy. 1997;44(1–2):113–24.
- 178. Ruan R, Zou G, Zhong S, Wu Z, Chan B, Wang D. Why Zijinshan copper bioheapleaching plant works efficiently at low microbial activity – study on leaching kinetics of copper sulfides and its implications. Miner Eng. 2013;48:36–43.
- 179. Yu S, Yang B, Fang C, Zhang Y, Liu S, Zhang Y, et al. Dissolution mechanism of the oxidation process of covellite by ferric and ferrous ions. Hydrometallurgy. 2021;201:105585.
- 180. Fisher WW, Flores FA, Henderson JA. Comparison of chalcocite dissolution in the oxygenated, aqueous sulfate and chloride systems. Miner Eng. 1992;5(7):817–34.
- 181. Kumar P, Nagarajan R, Sarangi R. Quantitative X-ray absorption and emission spectroscopies: electronic structure elucidation of Cu2S and CuS. J Mater Chem C Mater. 2013;1(13):2448–54. pmid:23781327
- 182. Hashemzadeh M, Dixon DG, Liu W. Modelling the kinetics of chalcocite leaching in acidified cupric chloride media under fully controlled pH and potential. Hydrometallurgy. 2019;189:105114.
- 183. Herreros O, Quiroz R, Viñals J. Dissolution kinetics of copper, white metal and natural chalcocite in Cl2/Cl− media. Hydrometallurgy. 1999;51(3):345–57.
- 184. Lawson F, Cheng C-Y, Lee LSY. Leaching of copper sulphides and copper mattes in oxygenated chloride/sulphate leachants. Min Proc Ext Met Rev. 1992;8(1–4):183–203.
- 185. Crossland I. Long-term corrosion of iron and copper. Proceedings - 10th International Conference on Environmental Remediation and Radioactive Waste Management, ICEM’05; 2005. p. 1402–8.
- 186. Lu J, Dreisinger D. Copper chloride leaching from chalcopyrite and bornite concentrates containing high levels of impurities and minor elements. Hydrometallurgy. 2013;138:40–7.
- 187. Giacometti F, Rebay G, Riccardi MP, Tarantino SC, Tizzoni CC, Tizzoni M. Iron age silicate slags from Val Malenco (Italy): the role of textural and compositional studies in the reconstruction of smelting conditions. Period Mineral. 2014;83: 329–344. doi:10https://doi.org/10.2451/2014PM0018
- 188. Prince S, Young J, Ma G, Young C. Characterization and recovery of valuables from waste copper smelting slag. In: Advances in Molten Slags, Fluxes, and Salts: Proceedings of the 10th International Conference on Molten Slags, Fluxes and Salts 2016. Springer International Publishing; 2016. p. 889–98.
- 189. Wang Z, Xu W, Li Y, Zhao Z, Jie F, Zeng G, et al. Diffusion behaviors and mechanism of copper-containing sulfide in fayalite-type slag: a key step of achieving copper slag depletion. Colloids Surf A: Physicochem Eng Asp. 2022;638:128264.
- 190. Wang D, Zhao Z, Lin Z, Liang Y, Kang L, Peng B. Interaction mechanism between arsenate and fayalite-type copper slag at high temperatures. Trans Nonferrous Met Soc China. 2022;32(2):709–20.
- 191. Toro N, Briceño W, Pérez K, Cánovas M, Trigueros E, Sepúlveda R, et al. Leaching of pure chalcocite in a chloride media using sea water and waste water. Metals. 2019;9(7):780.
- 192. Sharkey JB, Lewin SZ. Conditions governing the formation of atacamite and paratacamite. Am Mineral. 1971;56:179–92. Available from: http://www.bcin.ca/Interface/openbcin.cgi?submit=submit&Chinkey=419471
- 193. Vargel C. Intergranular corrosion. In: Corrosion of Aluminium. Elsevier; 2020. p. 185–97.
- 194.
Carvalho M. Corrosion of copper alloys in natural seawater – effects of hydrodynamics and pH. 2014. https://doi.org/10.1109/TIM.2015.2477160
- 195. Marcus P. Corrosion mechanisms in theory and practice. CRC Press; 2002.
- 196. Cohen A. Corrosion of copper and copper alloys. In: Corrosion: Materials. ASM International; 2005. p. 125–63.
- 197. de Oliveira FJR, Lago DCB, Senna LF, de Miranda LRM, D’Elia E. Study of patina formation on bronze specimens. Mater Chem Phys. 2009;115(2–3):761–70.
- 198. Tale Fazel E, Mehrabi B, Tabbakh Shabani AA. Kuh-e Dom Fe–Cu–Au prospect, Anarak Metallogenic Complex, Central Iran: a geological, mineralogical and fluid inclusion study. Miner Petrol. 2014;109(1):115–41.
- 199. Goh SW, Buckley AN, Lamb RN. Copper(II) sulfide? Miner Eng. 2006;19(2):204–8.
- 200. Swinbourne DR, Yazawa A, Barbante GG. Thermodynamic modeling of selenide matte converting. Metall Mater Trans B. 1997;28(5):811–9.
- 201. Lossin A. Copper. In: Ullmann’s encyclopedia of industrial chemistry. Wiley; 2001.
- 202. Zhang J, Matsuura H, Tsukihashi F. Processes for recycling. In: Treatise on process metallurgy. Elsevier; 2014. p. 1507–61.
- 203. Morris AE, Wadsley M. Metal extraction: phase stability diagrams. In: Encyclopedia of materials: science and technology. Elsevier; 2001. p. 5362–77.
- 204. Yazawa A. Thermodynamic considerations of copper smelting. Can Metall Q. 1974;13(3):443–53.
- 205. Chaubal PC, Nagamori M. Thermodynamics for arsenic and antimony in copper matte converting—computer simulation. Metall Trans B. 1988;19(4):547–56.
- 206. Nagamori M, Mackey PJ, Tarassoff P. The distribution of As, Sb, Bi, Se, and Te between molten copper and white metal. Metall Trans B. 1975;6(1):197–8.
- 207. Voisin L, Itagaki K. Phase relations, activities and minor element distribution in Cu-Fe-S and Cu-Fe-S-As systems saturated with carbon at 1473 K. Mater Trans. 2006;47(12):2963–71.
- 208. Kashima M, Eguchi M, Yazawa A. Distribution of Impurities between Crude Copper, White Metal and Silica-Saturated Slag. Trans JIM. 1978;19(3):152–8.
- 209. Mödlinger M, Sabatini B. A Re-evaluation of inverse segregation in prehistoric As-Cu objects. J Archaeol Sci. 2016;74:60–74.
- 210. Haubner R, Strobl S. The copper-arsenic eutectic and the Cu3As phase. DDF. 2020;405:19–25.
- 211. Mödlinger M, Cziegler A, Macció D, Schnideritsch H, Sabatini B. Archaeological arsenical bronzes and equilibrium in the As-Cu system. Metall Mater Trans B. 2018;49(5):2505–13.
- 212. Lechtman H, Klein S. The production of copper–arsenic alloys (arsenic bronze) by cosmelting: modern experiment, ancient practice. J Archaeol Sci. 1999;26(5):497–526.
- 213. Lottermoser BG. Mobilization of heavy metals from historical smelting slag dumps, north Queensland, Australia. Mineral Mag. 2002;66(4):475–90.
- 214. Saez R, Nocete F, Nieto JM, Capitan MA, Rovira S. The extractive metallurgy of copper from Cabezo Jure, Huelva, Spain: chemical and mineralogical study of slags dated to the third millenium B.C. Can Mineral. 2003;41(3):627–38.
- 215. Alvarez-Valero AM, Perez-Lopez R, Nieto JM. Prediction of the environmental impact of modern slags: a petrological and chemical comparative study with Roman age slags. Am Mineral. 2009;94(10):1417–27.
- 216. Sobanska S, Deneele D, Barbillat J, Ledésert B. Natural weathering of slags from primary Pb–Zn smelting as evidenced by Raman microspectroscopy. Appl Geochem. 2016;64:107–17.
- 217. Navarro A, Cardellach E, Mendoza JL, Corbella M, Domènech LM. Metal mobilization from base-metal smelting slag dumps in Sierra Almagrera (Almería, Spain). Appl Geochem. 2008;23(4):895–913.
- 218. Di Turo F, Moro G, Artesani A, Albertin F, Bettuzzi M, Cristofori D, et al. Chemical analysis and computed tomography of metallic inclusions in Roman glass to unveil ancient coloring methods. Sci Rep. 2021;11(1):11187. pmid:34045535
- 219. Williams NR. The calculation of air density in various units *. Bull Am Meteorol Soc. 1949;30(9):319–20.
- 220. Lehmann EH, Hartmann S, Speidel MO. Investigation of the content of ancient tibetan metallic buddha statues by means of neutron imaging methods. Archaeometry. 2010;52(3):416–28.
- 221. Tansella F, Vigorelli L, Ricchiardi G, Re A, Bonizzoni L, Grassini S, et al. X-ray computed tomography analysis of historical woodwind instruments of the late eighteenth century. J Imaging. 2022;8(10):260. pmid:36286354
- 222. Piatak NM, Seal RR II. Mineralogy and the release of trace elements from slag from the Hegeler Zinc smelter, Illinois (USA). Appl Geochem. 2010;25(2):302–20.
- 223. Viadero R. The geochemistry of acid mine drainage. In: Encyclopedia of water. Wiley; 2019. p. 1–8.
- 224. Consani S, Carbone C, Dinelli E, Balić-Žunić T, Cutroneo L, Capello M, et al. Metal transport and remobilisation in a basin affected by acid mine drainage: the role of ochreous amorphous precipitates. Environ Sci Pollut Res Int. 2017;24(18):15735–47. pmid:28528499
- 225. Helmig D, Jackwerth E, Hauptmann A. Archaeometallurgical fieldwork and the use of a portable X‐ray spectrometer. Archaeometry. 1989;31(2):181–91.
- 226.
Heskel D. The development of pyrotechnology in iran during the fourth and third millenium B.C. Harvard University; 1982.
- 227. Roberts BW, Thornton CP. Archaeometallurgy in global perspective. In: Roberts BW, Thornton CP, editors. Archaeometallurgy in global perspective: methods and syntheses. New York (NY): Springer New York; 2014.
- 228. Gee C, Ramsey MH, Maskall J, Thornton I. Mineralogy and weathering processes in historical smelting slags and their effect on the mobilisation of lead. J Geochem Explor. 1997;58(2–3):249–57.
- 229. Tabelin CB, Corpuz RD, Igarashi T, Villacorte-Tabelin M, Ito M, Hiroyoshi N. Hematite-catalysed scorodite formation as a novel arsenic immobilisation strategy under ambient conditions. Chemosphere. 2019;233:946–53. pmid:31340422