Pigments—Arsenic-based yellows and reds

This review offers an update on arsenic-bearing minerals and pigments with the aim of serving as a guide for the study of Cultural Heritage materials in which these materials can be found. The different As-bearing mineral phases (realgar, pararealgar, orpiment, anorpiment, alacranite, dimorphite, bonazziite, uzonite, wakabayashilite, duranusite, arsenolite and claudetite) and some of their light-induced products are examined. The occurrence of As-sulfides and their trade, use, alteration and degradation are also reviewed. Finally, the analytical techniques commonly used for the identification of arsenic-containing pigments are discussed. The manuscript concludes with a good-practice guide and a summary of key concepts for use by those working in the field of cultural heritage.


Premise
This Topical Collection (TC) covers several topics in the field of study, in which ancient architecture, art history, archaeology and material analyses intersect. The chosen perspective is that of a multidisciplinary scenario, capable of combining, integrating and solving the research issues raised by the study of mortars, plasters and pigments ).
The first group of contributions explains how mortars have been made and used through the ages (Arizzi and Cultrone 2021;Ergenç et al. 2021;Lancaster 2021;Vitti 2021). An insight into their production, transport and on-site organization is further provided by DeLaine (2021). Furthermore, several issues concerning the degradation and conservation of mortars and plasters are addressed from practical and technical standpoints (La Russa and Ruffolo 2021;Caroselli et al. 2021).
The second group of contributions is focused on pigments, starting from a philological essay on terminology (Becker 2021). Three archaeological reviews on prehistoric (Domingo Sanz and Chieli 2021), Roman (Salvadori and Sbrolli 2021) and Medieval (Murat 2021) wall paintings clarify the archaeological and historical/cultural framework. A series of archaeometric reviews illustrate the state of the art of the studies carried out on Fe-based red, yellow and brown ochres (Mastrotheodoros et al. 2021), Cu-based greens and blues (Švarcová et al. 2021), As-based yellows and reds (this paper), Pb-based whites, reds, yellows and oranges (Gliozzo and Ionescu 2021), Hg-based red and white (Gliozzo 2021) and organic pigments (Aceto 2021). An overview of the use of inks, pigments and dyes in manuscripts, their scientific examination and analysis protocol (Burgio 2021) as well as an overview of glass-based pigments (Cavallo and Riccardi 2021) are also presented. Furthermore, two papers on cosmetic (Pérez-Arantegui 2021) and bioactive (antibacterial) pigments (Knapp et al. 2021) provide insights into the variety and different uses of these materials.

Introduction
The uses of arsenic and related mineral species are numerous and go far beyond the use that has been made of them as pigments. Arsenic has in fact been used as a poison but also as a medicinal remedy: proof of this are the numerous scientific works currently being carried out on arsenic in cancer treatments, using "old" compounds-well known to Chinese and Iranian medicine-and new pharmacological remedies (Bin et al. 2000;Miller et al. 2002;Nriagu 2002;Liu et al. 2008;Bujňáková et al. 2015;Darbandi and Taheri 2018).
Furthermore, arsenic has been and still is a fundamental component of several metal alloys (Mödlinger et al. 2018;2019;Sabatini et al. 2020 on ancient arsenical copper and bronze) and it is therefore a valuable geo-resource for the countries that extract it (chiefly China and Chile nowadays; Dill 2010). For example, the use of arsenic chalcogenides in optics and optoelectronics represents an ever-expanding field (Bonazzi and Bindi 2008;Jovanovski and Makreski 2020).
Much of the recent literature has focused on the geochemistry of arsenic in polluted contexts and the critical role of biochemical processes for the As cycling in nature (i.e., acid mine drainage and acid rock drainage; see, e.g., Lee et al. 2006, Dogan and Dogan 2007, Saunders et al. 2008, Henke 2009, Catelani et al. 2018, as well as on its use as a component in technologically advanced materials (especially glass and metal). In this review, instead, we focus on arsenic as a constituent of red and yellow pigments, used from ancient times to the modern era. In the first part, we will deal with the mineralogical description, occurrence, properties, alterations and uses of these minerals while in the second part, we will describe the analytical techniques used for the characterization of these pigments and propose a guide for best practice in archaeometry.

Arsenic and As-bearing types of minerals
The average abundance of arsenic in the Earth's crust is of 1.5-2 ppm (Plant et al. 2014). Arsenic is a constituent of several types of mineral deposits and is typically associated with those containing sulfides and sulfosalts. Arsenicbearing minerals include arsenates, sulfides, oxides and, to a lesser extent, arsenites, arsenides and metal alloys (see O'Day 2006 for a detailed overview on the chemistry and mineralogy of arsenic). Overall, they show all basic colors and bear variable amounts of As, frequently combined with Fe, Co, Ni, Cd, Cu, Ag, Au and Pb.
Minerals containing As can provide pigments with different colours when associated with other chromophores. For example, the green pigment (CaCu(AsO 4 )(OH); see Brecoulaki 2014, Buzgar et al. 2014), the blue pigment obtained from processing cobaltite (CoAsS; see Matin and Pollard 2017), in addition to the red and yellow pigments (realgar, As 4 S 4 , and orpiment, As 2 S 3 , respectively) which represent the main topic of this review.
Synthetic minerals are entirely man-made and are of primary importance in the investigation of pigments since their use is attested from the beginning of the history of painting (e.g., the production of cuprorivaite for Egyptian blue). As for As-bearing pigments, their preparation has been studied in detail and their use has been attested from the fifteenth century (see also Rötter et al. 2007).
Secondary phases are those formed by the alteration of either primary or artificial minerals (see Drahota and Filippi 2009 for a comprehensive review with an environmental focus). Consequently, they can occur both in the natural environment and in altered art works, depending on several factors including temperature, pressure, reactivity with other pigments and especially light exposure. Important secondary phases include the polymorphs arsenolite and claudetite (As 2 O 3 ), pararealgar (As 4 S 4 ) and the so-called χ-phase (see below).
Finally, it should be emphasized that in the pictorial context, all three types of minerals and related pigments may have been used in a more or less conscious way, regardless of their nature. For example, the use of yellow pigments based on orpiment (primary) rather than pararealgar (secondary) and vice versa may not always have been the result of a deliberate choice. Moreover, orpiment and arsenolite were long considered the only possible realgar alteration products (before discovering that As 2 S 3 cannot form in ambient air and that pararealgar is another possible form of As 4 S 4 ) and therefore the historical literature may not be completely reliable on this matter (see Corbeil and Helwig 1995;Trentelman et al. 1996).

Mineralogical arsenic-bearing phases found (or not yet found) in yellow and red pigments
This section provides information on the composition and structure of the numerous As-bearing sulfides and oxides that have been found in several types of pictorial works. Consequently, it is focused on As-sulfides while As-oxides discussions are limited to arsenolite and claudetite. As for sulfides, some phases such as duranusite, χ-phase and wakabayashilite have been added, even if they are either not widely attested or not yet recognized in works of art, because little known or recently discovered. Conversely, As 4 S 4 (II) (Kutoglu 1976) and jeromite (Lausen 1928) have been omitted since the former has never been found in nature and the latter has been discredited by the IMA Commission.
Among As-sulfides ( Fig. 1), the crystal structure of dimorphite, realgar, pararealgar, bonazziite, alacranite, uzonite and wakabayashilite consists of "a packing of cagelike, covalently bonded As 4 S n (n = 3, 4 and 5) molecules held together by van der Waals forces" (Bonazzi et al. 2006); conversely, orpiment and duranusite show layered structures. In the former group, the As:S ratio ranges from As 4 S 3 in αand β-dimorphites, to As 4 S 4 in the three polymorphs realgar, pararealgar and bonazziite, to As 4 S 4 and As 4 S 5 in alacranite and to As 4 S 5 in uzonite. Each phase is described below and the main features are reported in Table 1, while cage-like molecules and unit-cells are shown in Fig. 2.

Realgar (As 4 S 4 )
Also referred to as α-As 4 S 4, it is a soft (1/2 on the Mohs scale) red, monoclinic, low-temperature (LT) phase. Although recognized since the eighteenth century (1747 in the IMA list), we must wait until the 1950s to witness the beginning of accurate crystallographic studies (Table 1; Ito et al. 1952;Hall 1966;Clark 1970;Mullen and Nowacki 1972;Roland 1972;Bryndzya and Kleppa 1988and 2003aKyono 2009). The occurrence of realgar will be examined in the section on the occurrence of As-sulfides further down, while the light-induced modifications are commented below, along with the description of the χ-phase.

Pararealgar (As 4 S 4 )
Formed by the alteration of realgar, this phase was first described by Roberts et al. 1980 as "powdery to granular fine-grained aggregates that replace realgar (…) yellow to orange-yellow with a resinous lustre, a bright yellow streak, Mohs hardness 1-1½, uneven fracture and no apparent cleavage". The two occurrences investigated by the authors (Gray Rock and Mount Washington copper deposit in British Columbia) were mainly associated with stibnite, although other Fe-, As-, Cu-and Sb-bearing phases were found (e.g., lepidocrocite, pyrite, arsenopyrite, arsenolite, stibnite, tetrahedrite). The crystal structure was determined by Roberts et al. 1980 andBonazzi et al. 1995. The first authors also fixed reaction rates and temperatures required to transform pararealgar into the HT β-phase (e.g., 5 min at 220 and 420 °C, 10 min at 300 °C) or the "original" LT α-phase (e.g., 1 day at 220 °C).

Alacranite (As 8 S 9 ) and bonazziite
Alacranite is a rare red mineral whose identification has had a troubled history for the variable use that has been made of α and β descriptors (as similarly happened to dimorphite). First, Hall (1966, pp. 16-24) defined three polymorphs of AsS: a) the high temperature (Bx +) orange form α-realgar (As 4 S 4 ), annealed at 276 °C; b) the intermediate temperature red (Bx-) form β-realgar and c) the low temperature orange-yellow (Bx-) form γ-realgar (As 4 S 4 ).
In the later literature, this initial nomenclature was frequently changed or differently used: (a) the α-phase has been sometimes used to indicate "normal" or "true" realgar; (b) the β-phase began to indicate the high temperature phase after Porter and Sheldrick (1972) but this denomination was not accepted by all authors (as pointed out by Douglass et al. 1992); and (c) the γ-phase was likely associated with pararealgar by Douglass et al. (1992). In the text that follows, the nomenclature used is the one reported by the individual authors. This choice can surely create some confusion but it was considered appropriate to shed light on terminological issues which, if not explained, could mislead the reader. For the same reason, it was deemed necessary to present the various information in the chronological order in which they were published.
In 1966, Roland (1966) fixed at 306(± 2 °C) the melting temperature of α-AsS. A few years later, Clark (1970) found the high temperature α-AsS ("high realgar") identified by Hall (1966) in a natural assemblage, namely the Alacrán Ag-As-Sb vein deposit in Chile. The same deposit also provided evidence of another type of "realgar" ("low realgar"), that the author defined "similar, but not identical to "normal" realgar, or β-AsS.
In 1972, Porter and Sheldrick (1972) purified tetra-arsenic tetrasulfide by sublimation and obtained both the mineral realgar, named α-As 4 S 4 , and a "previously unreported" hightemperature phase, named β-As 4 S 4 (Table 1). Later studies by Bonazzi et al. (1995) showed that the As 4 S 4 molecules of the latter phase are different from those of pararealgar.
Still in 1972, Roland (1972) explored the issue concerning the temperature of the reaction α-AsS ↔ realgar and provided new structural data on the high-temperature α-AsS phase ( Table 1). The inversion T°C was fixed between 239 and 263 °C, depending on composition and boundary conditions (e.g., preparation of the mix and presence of vapour).
In 1986, Popova et al. (1986) reported on the discovery of the high-temperature α-As 8 S 9 phase (Aльфa-cyльфид мышьякю; Table 1) identified by Clark (1970), in the hydrothermal mercury-antimony-arsenic mineralization at the Uzon caldera in Kamchatka (Russia). This natural hightemperature mineral was then named alacranite (Aлaнpaнит) and was approved by the IMA committee in that same year (Hawthorne et al. 1988).
The later study performed by Migdisov and Bychkov (1998) clarified the temperature intervals corresponding to In non-stoichiometric compounds the volume was found to linearly increase from β-As 4 S 4 to alacranite ss. as a function of the S content d In non-stoichiometric compounds the volume was found to linearly increase from β-As 4 S 4 to alacranite ss. as a function of the S content e Indexed by Bonazzi et al. (1996) f Bonazzi et al. 2003b suggested to identify the compound analysed by Burns and Percival (2001) as the β-As 4 S 4 unnamed phase and not as alacranite g Indexed by Bonazzi et al. (1996) h Obtained from natural realgar crystals, after 24-h exposure to polychromatic light (550-nm-long wavelength pass filter) i Obtained by heating a realgar sample from Shimen (China) at 295 °C for 24 h j Calculated from unit cell k Synthetized using As:S = 1:1, 500-600 °C l Obtained by oxydation of realgar realgar (70-95 °C), realgar + orpiment + uzonite + alacranite (50-75 °C) and orpiment + amorphous As sulfides (30-50 °C) assemblages and showed that the orpimentuzonite (Fig. 3, Eq. 1) and the uzonite-alacranite (Fig. 3, Eq. 2) replacements were both reversible transformations.
During the same period, Žáček and Ondruš (1997) identified two similar arsenic sulfides in a burning waste pile of the Kateřina colliery in Radvanice (Bohemia). The first one corresponded to "true" alacranite (sensu Popova et al. 1986) while the second one, named monoclinic As 4 S 4 , was found to be isostructural with alacranite but with a lower parameter b ( Table 1).
The studies proceeded in the following years with the identification of alacranite in the seafloor around Lihir Island (Papua New Guinea) by Burns and Percival (2001). Their study provided structural information on alacranite-defined as "the third polymorph of this composition, after realgar and pararealgar"-and led them to state that (a) the formula was As 4 S 4 rather than As 8 S 9 ; (b) it was "identical to that of synthetic α-As 4 S 4 "; and (c) it should have formed at relatively low temperatures, according to previous studies performed by Heinrich and Eadington (1986) and Migdisov and Bychkov (1998).
In 2002, Wilkin and Ford (2002) discovered that alacranite can also precipitate from solution through acid extraction and the following years marked a decisive step towards the definition of alacranite, β-phase and χ-phase (for the latter, see below the dedicated section).
The studies performed by Bonazzi et al. (2003a-b) took stock of the previous knowledge and definitively clarified the formulas and structure of these phases. Firstly, the As 8 S 9 formula was restored for alacranite s.s. and the distinction between the natural compounds As 8 S 9 (alacranite s.s.) and the unnamed β-As 4 S 4 phase was highlighted and clarified  (Table 1). Secondly, the existence of a continuous series between these two groups (realgar-type As 4 S 4 and uzonitetype As 4 S 5 ) was suggested (Bonazzi et al. 2003a;Bonazzi and Bindi 2008) and the gradual changes of the translation lattice simmetry from β-As 4 S 4 (C2/c; V = 803(2) Å 3 ) to alacranite s.s. (P2/c; V = 860(3) Å 3 ) were explained by the "simultaneous presence of As 4 S 4 (C2/c) and As 8 S 9 (P2/c) microdomains". The current terminology sometimes uses the plural term alacranites to describe this β-As 4 S 4 -As 8 S 9 series but the traces of the various passages (now outdated) which led to recognize alacranite as an ordered sequence of As 4 S 4 and As 4 S 5 molecules are still traceable, for example, in Strunz and Nickel (2001) and in the AMCSD (Downs and Hall-Wallace 2003).
In the past 10 years, the structural and vibrational properties and growth mechanisms of alacranite were further investigated (Gibbs et al. 2011;Pagliai et al. 2011) and new occurrences were reported by Mladenova (2000) and Bonazzi and Bindi (2008), in the Madzavoro deposit (eastern Rhodopes, Bulgaria) and the Sb--Hg deposit of Khaidarkan (Kyrgyzstan), respectively. Moreover, Mumford et al. (2013) demonstrated that alacranite (As 8 S 9 ) can be precipitated by a novel As(V)-respiring anaerobe strain MPA-C3.
More information on the high-temperature (HT) unnamed β-As 4 S 4 phase was instead acquired as part of the studies concerning the so-called χ-phase and the continuous involvement of Bonazzi and coworkers in the study of arsenic sulfides. These authors finally found a truly natural specimen of this phase within the collection of the Museo di Storia Naturale, Firenze, coming from the Khaidarkan deposit mentioned above. This orange mineral-which was then named bonazziite-was approved by the IMA (no. 2013-141) in the year 2013 and therefore constitutes the natural analog to the previously identified synthetic β-As 4 S 4 phase (Bindi et al. 2015a). Lastly, it is worth quoting Ledbetter et al. 2007 who discovered an arsenic-metabolizing organism able to produce β-realgar (sensu Muniz-Miranda et al. 1996).

Uzonite (As 4 S 5 )
H.J. Whitfield synthesized tetra-arsenic pentasulfide and solved its crystal structure in 1973 (Whitfield 1973b). In the year 1984, the IMA Commission approved uzonite as a new mineral (no. 1984-027) based on Popova and Polyakov (1985) who found this phase at the Uzon caldera (Russia). Twenty years later, Bindi et al. (2003) took up the studies previously carried out on Russian materials and provided the crystal structure of uzonite from the type locality. Uzonite is a soft (1.5 on the Mohs scale), yellow mineral which is not affected by light exposure (Bindi et al. 2003) and it is also found in Chinese and Indian medicinal remedies (Furuta and Sato 2016;Sharma et al. 2017).

Wakabayashilite [(As,Sb) 6 S 9 ][As 4 S 5 ]
This is a rare mineral characterized by As 4 S 5 molecules similar to those of alacranite and uzonite. It was first identified by Kato et al. (1970) at the Nishinomaki mine (30 km west of Takasaki, Gunma). After Kato's work, the crystal structure was investigated by Scott and Nowacki (1975), using a realgar specimen from the White Caps mine (Manhattan, Nevada-W.W. Pinch of Rochester collection, New York) and then solved by Bonazzi et al. (2005), similarly using acicular crystals from the White Caps mine (Museo di Storia Naturale collection, Firenze). The chemical formula was modified to [(As,Sb) 6 S 9 ][As 4 S 6 ] by the former authors and then to [(As,Sb) 6 S 9 ][As 4 S 5 ] by the latter authors. Bonazzi et al. (2005) went into detail and described this phase as constituted by two structural units: (1) [M 6 S 9 ] bundlelike unit (M = As, Sb) and (2) As 4 S 5 cage-like molecules. The investigation later performed by Bindi et al. (2014a) confirmed the chemical formula provided by Bonazzi et al. (2005) and highlighted some analytical problems of the previous literature. Minor amounts of Cu, Zn and Tl were found to be common in several specimens and, for these cases, the new formula M x [(As,Sb) 6 S 9 ][As 4 S 5 ] (with M = Cu, Zn, Tl and 0.00 < x < 0.13 a.p.f.u.) was provided ). Lastly, light-exposure experiments were found to be irrelevant since they did not affect the crystal structure of wakabayashilite (Bonazzi et al. 2005).

Orpiment and anorpiment (As 2 S 3 )
Orpiment, from the Latin word auripigmentum, is a yellow phase which is currently exploited for semiconducting and optical applications. The first description of this mineral is lost in the history of mineralogy while the structural determinations were reported by Morimoto (1949Morimoto ( , 1954 and Mullen and Nowacki (1972).
The greenish yellow anorpiment was approved by IMA in 2011. This phase is the triclinic dimorph of orpiment and was discovered at the Palomo mine in Peru (Castrovirreyna Province, Huancavelica Department) by Kampf et al. (2011). The mine is located in a volcanic area (andesitic lavas, pyroclastic rocks, tuffs and breccias) and the main ore minerals include pyrite, galena, sphalerite and chalcopyrite. Anorpiment was characterized as "a low-temperature hydrothermal mineral associated with dufrénoysite, muscovite, orpiment, pyrite, and realgar" and its structure was solved by its discoverers. Johan et al. (1973) discovered this orthorhombic phase in the Duranus deposit (close to Mt. Séréna, Alpes-Maritimes, France) in association with native arsenic and realgar. The structure has been studied in detail by Kyono (2013) and Bonazzi et al. (2016). The former hypothesized As 4 S molecular geometry while the latter demonstrated a layered structure that "consists of an alternated stacking along the [010] axis of two kinds of corrugated layers which match the As 2 and As 2 S composition, respectively" (Bonazzi et al. 2016). Duranusite has also been found as a biogenic product by dissimilatory metal-reducing bacterium Shewanella sp. HN-41 (Lee et al. 2007).

Artificial As-based pigments
The use of arsenic-based artificial pigments is attested in numerous documents ranging from ancient treatises to encyclopedias and pharmaceutical prontuary (Grundmann and Rötter 2007). However, a problem that in many cases remains open concerns the natural or artificial origin of the pigments cited by the authors. For example, Pliny (Naturalis Historia, 35, 12, 30 1 ) listed the auripigmentum among natural pigments, clearly distinguishing it from those that "finguntur" (i.e. the artificial ones), among which realgar (sandaraca) is included. However, the problem is not limited to ancient times and, in fact, it is possible to find orpiment mentioned as an artificial pigment by Cennino Cennini (Il libro dell'arte, chapter 47 2 ) as well as quasi-modern texts in which orpiment is erroneously identified (on the basis of the yellow color) instead of pararealgar or other yellow As-sulfides.
The production of artificial orpiment is documented in various literature sources (e.g., Dossie 1764, Rose 1916, Merrifield 1967, reporting on the 1431 manuscript of Jehan Le Begue) and it has been reconstructed as a 4-steps process: (1) roasting of arsenic ores such as löllingite (FeAs 2 ) in iron boilers to produce arsenolite (Fig. 3, Eq. 3); (2) collection and selection of the formed arsenic crusts and elimination of impurities through sublimation; (3) heating of the resulting compound to obtain a light yellow arsenolite glass; and (4) addition of sulfur (3 wt%) to convert the former glass into a yellow glass (with ~ 97 wt% arsenolite) called 'gelber arsenik' (Grundmann and Rötter 2007).
The production of these pigments for artistic purposes has long been investigated, starting from Wallert (1984) and West FitzHugh (1997); however, the first systematic work on this topic is due to Grundmann and Rötter (2007) who experimentally tested the production cycle and provided the necessary information to distinguish natural from artificial pigments. They prepared five series of tests, using two different wet-process precipitations (using thioacetamide for series I and hydrogen sulfide for series II) and three different dry-process cakes and/or sublimates (using natural orpiment for series III, natural orpiment + synthetic arsenolite for series IV and synthetic arsenolite + sulfur for series V). The first series of experiments (wet process I and II) produced two golden yellow amorphous powders, partially different from each other in color distribution, shape and state of aggregation but indistinguishable under the microscope if mixed. The second series of experiments (dry process III-V) showed that orpiment is formed only if orpiment itself is used as a raw material either by roasting (series III) or by roasting and sublimating (series IV). In these cases, the product also includes colourless arsenolite and, occasionally, alacranite members. In the absence of orpiment (series V), a mixture of colorless aggregates of arsenolite and yellow glassy drops and spherules (amorphous) are produced. The same authors claimed that the production of artificial Asbased pigments may have been born in conjunction with the production of vermilion through the roasting of sulfur and arsenic ores and that the arsenolite smelters made it popular during the sixteenth century.
It has to be added, however, that this mining activity was initially aimed at gold extraction, along with silver and other precious metals, and it may have been this very activity that triggered the production of a by-product such as a pigment. The frequent natural association of arsenic with gold-bearing deposits (see the section on the As-sulfides occurrence below) suggests a close link between the beginning of the two processes. Moreover, arsenic is still nowadays a by-product of pyrometallurgical processes for copper and gold.
Proceeding further, Grundmann and Richter (2008), added a terminological note, specifying that "orpiment glass (g-As 2 S 3 )" should be used for arsenic sulfide particles molten and solidified from natural orpiment" while "arsenic sulfide glass (g-As x S x )" should be used for all other types of compounds of these series.
Additional experiments have been performed by Vermeulen et al. (2019) who synthetized (a) four compounds with different As:S ratio (40:60, 34:66, 30:70 and 25:75) and final colors (from orange to light yellow), using the meltquenching method (800 °C for 48-h cold water; dry process) and (b) two yellow amorphous arsenic sulfide compounds (suitable for painting), using the wet-process (hydrochloric solution of arsenic trioxide treated with a stream of hydrogen sulfide-pyrite and thioacetamide). For all these compounds, the reaction series (Eqs. 4-6) and are provided in Fig. 3.
On a similar topic, a year earlier, Vermeulen et al. (2018a) had identified both orpiment and amorphous arsenic sulfide on two works of art (seventeenth to eighteenth centuries) and hypothesized that pararealgar was the starting material from which the amorphous arsenic sulfide was produced.

[On furnaces] "
The furnace for subliming king's yellow must have a sand-pot; as the heat of the naked fire would be too great" (Dossie 1764, p. 20). [On 'Common orpiment'] "When purified by subliming, it becomes the king's yellow" (Dossie 1764, p. 98). [On sublimation] "Sublimation is the saising solid bodies in fumes, by means of heat: which fumes are afterwards collected by condensation, either in the upper part of the same vessel where they are raised, or in others properly adapted to it for that purpose" (Dossie 1764, p. 30. [On King's yellow] "This pigment must be prepared by mixing sulphur and arsenic by sublimation, which may be done in the following man-King's yellow among the products imported from London and proposed a connection between the name of this pigment and the Arabic nomenclature where orpiment and realgar are referred to as "the two kings." Less is known about realgar production. Pliny probably confuses realgar with minium since he includes the production of sandaraca with that of cerussa (Naturalis Historia 34, 54, 175-176 4 ) but immediately after he adds a detail that could instead be true when he states that "it is found both in gold-mines and in silver-mines. The redder it is, the more pure and friable, and the more powerful its odour, the better it is in quality" (Naturalis Historia 34, 177, 55 5 ). Undoubtedly, this description takes us back to the gold deposits and seems to contradict the artificial nature of the pigment (which probably referred to minium). To support this hypothesis, we can quote the words in Pliny's following book (Naturalis Historia 35, 22, 39 6 ) in which he returns to the subject and specifies that there is also another type of sandaraca "prepared by calcining ceruse in the furnace." The contemporary Dioscorides also seems to refer to both a natural and an artificial origin in his De Materia Medica (5, 121-122): "Arsenicum is found in the same mines as sandaraca. The best is crusty, gold in colour and with crusts like fish scales (as it were) lying one over another, and it is not mixed with any other material. That which is found in Mysia (which is in Hellespont) is like this. There are two kinds: one as mentioned, and the other in clumps and like sandaraca in colour. It is brought from Pontus and Cappadocia. It is roasted as follows: place it in a new ceramic jar, set it over burning coals and stir it continuously; and when it is red-hot and has changed colour, cool it, beat it finely, and put it in jars. It is antiseptic, astringent, and scab forming with a burning, strong, biting strength, and it is one of those medicines that repress abnormal growths and make the hair fall out.
Sandaraca is most highly valued which is fully red, brittle, easily pounded into small pieces, and pure -looking like cinnabar in colour, and also having a sulphurous smell. It has the same properties and method of roasting as arsenicum. Used with rosin it heals loss of hair [alopecia], and with pitch it removes leprous nails. It is good with oil for lice infestations, and with grease it dissolves small swellings. It is also good with rosaceum for ulcers in the nostrils and mouth, and for other pustules, as well as for venereal warts. It is given with mead [honey wine] to those who spit up rotten matter, and it is made into an inhalant with rosin for old coughs, the smoke drawn through a funnel into the mouth. Licked in with honey it clears the voice, and it is given with rosin in a catapotion [pill] to the asthmatic." (Translation from Osbaldeston and Wood 2000).
Undoubtedly, the interest of Dioscorides is exclusively medical and pharmaceutical but the information about the production of arsenic and realgar can still be considered reliable.
Still browsing among ancient authors, Cennino Cennini (Il libro dell'arte, chapter 48 7 ) informs about the highly toxic nature of the "risalgallo" but does not dwell on its nature or preparation. Conversely, Merrifield (1967), reporting on a seventeenth century manuscript, includes "red orpiment" as a product of the melting, cooling and grinding of orpiment.
West FitzHugh (1997) listed a few examples of modern production techniques such as heating a mixture of pyrite and arsenopyrite to obtain a mixture of realgar and orpiment, the fusion of sulfur and arsenic in the presence of carbon or the heating of orpiment with arsenic.
Lastly, Grundmann and Rötter (2007) inform us of the production run by the Güttler arsenolite smeltery, where a mixture of löllingite and sulfur was heated to obtain a red sublimation product.

The χ-phase and the light-induced variations in arsenic sulfides
Despite the fact that other studies on light-induced transformations of arsenic sulfides had already been published (e.g., Bhatnagar and Rao 1923;Cahoon 1965;Keneman et al. 1978;Stodulski et al. 1984); Douglass et al. (1992) first proved that the radiation wavelength is crucial for the transformation of realgar into pararealgar. In detail, the authors demonstrated that if realgar is exposed to UV and IR radiation or wavelengths shorter than 500 nm, no transformations occur. On the other hand, wavelengths greater than 560 nm (esp. above 610 nm) slow down the reaction rate, as they pass through the crystals without being absorbed "and little energy is being transferred to break bonds".
The research performed by Douglass (Table 2) and co-workers also led to the identification of an intermediate phase precursor to the formation of pararealgar, named χ-phase (Table 1), which formed after the LT α-phase (realgar, Fig. 3, Eq. 7) with variable times depending on which radiation wavelength was used. Moreover, the χ-phase was also detected in the reverse reaction as a precursor of the HT β-phase (Fig. 3, Eq. 8), performed at temperatures of 175 and 190 °C. Considering that "normal" β-phase (i.e., not involved in the reverse reaction) took several weeks (at 240 °C) to transform into α-phase (realgar), the ~ 2 days it took for the reaction (Fig. 3, Eq. 8) to complete suggested that χ-phase played a role in accelerating the reaction.
Douglass' studies were immediately received in an archaeometric environment by Trentelman et al. 1996 who warned about the veracity of the current association red = realgar, yellow = orpiment and presented for the first time the identification of the χ-phase in the sixteenth-century (ca. 1547) painting 'The Dreams of Men' by Tintoretto.
In the same year, Bonazzi and coworkers added an important piece of information as they clarified some aspects of the crystal structure of the χ-phase obtained from light-induced alteration of both realgar and the HT phase β-As 4 S 4 (Tables 1 and 2) . The authors observed that, increasing light-exposure, the formation of the χ-phase was "preceded by a strong anisotropic increase of the unit-cell volume" and determined that while in realgar it is a and c sin β which linearly increase (with b remaining unchanged), in the β-As 4 S 4 phase it is instead b and c sin β that increase (with a decreasing) and the χ-phase so occurs after 1560 min. Last but not least, the authors first suggested to consider the χ-phase as an "expanded, less-ordered β phase" (Bonazzi et al. , 2006, probably characterized by a lower molecular symmetry, as demonstrated also by the results provided by Muniz-Miranda et al. (1996) (Table 2, Fig. 4A).
A few years later, Bindi et al. (2003) tested uzonite and the negative evidence they obtained (i.e., no transformation) suggested to the authors that only As 4 S 4 molecules are affected by light exposure while the As 4 S 5 molecules of uzonite are not. This new data led them to speculate that (a) light induces an increase of sulfur in the structure; (b) the increase of sulfur induces an increase of As 4 S 5 molecules that, in turn, explains the increase of the unit-cell volume; and (c) likely, As 4 S 5 molecules were not able to incorporate additional sulfur as the As 4 S 4 were instead. The Eq. 9 (Fig. 3) was then proposed. Kyono et al. (2005) followed a similar path and found comparable results as to the variation of the a and c sin β parameters and the unit-cell volume expansion, thus substantiating previous results. However, the authors further claimed that Eq. 9 does not necessary need light exposure to happen since small amount of oxygen are sufficient to trigger the reaction and, contrary to Bonazzi et al. (1996), suggested that the volume expansion of the unit-cell was only due to variable intermolecular distances. Moreover, although lacking experimental evidence (especially on S migration), they described the realgar transformation into pararealgar as the cycle illustrated in Fig. 4B: (a) free S atom combines with realgar (As 4 S 4 ) and produces the As 4 S 5 molecule; (b) the As 4 S 5 molecule releases an S atom and becomes pararealgar (As 4 S 4 ). The cycle can be further by the series of equations Eqs. 10-13 in Fig. 3 (from Naumov et al. 2007).
Following these results, Ballirano and Maras (2006) deepened the kinetics of the process and suggested that (a) the formation of the χ-phase and the expansion of realgar occur contemporaneously (not one before the other); and (b) the production of an intermediate As 4 S 4 + x phase occurs "because it represents the only sulfide phase that is able to sustain the presence of As 4 S 5 -type cages" (Table 2).
In the same year, Bonazzi et al. (2006) published a new paper on light-induced changes in As-sulfides where they investigated the reaction that led to the transformation of the β-As 4 S 4 -As 8 S 9 series and the synthetic β-As 4 S 3 phase into pararealgar (Table 2). Experimental data demonstrated that light-induced changes in these phases consist in a marked increase of the unit-cell volume in all crystals of the β-As 4 S 4 -As 8 S 9 series but not in alacranite s.s. (stoichiometric As 8 S 9 ) and synthetic β-As 4 S 3. To explain the reasons for these diversified or imperceptible increases, the authors considered both the different types of molecules and their different packings, i.e. the two factors that determine the ability of the molecules to incorporate the sulfur necessary for the conversion from As 4 S 4 to As 4 S 5 . Their conclusions, sometimes confirmatory or hypothetical, can be summarized as follows: (1) the increase of the exposure time provokes the increase of As 4 S 5 molecules in the structure, due to the incorporation of sulfur by As 4 S 4 that transforms into As 4 S 5 (according to Eq. 12); (2) the negligible expansion of the stoichiometric alacranite unit-cell is tentatively explained by excluding alacranite microdomains (P2/c) from the reaction and thus implying that only β-As 4 S 4 microdomains (C2/c) are involved in this light-induced alteration process; (3) only the As 4 S 4 molecule is able of incorporating sulfur while the As 4 S 3 and As 4 S 5 molecules are not; (4) the complete transformation from β-As 4 S 4 (C2/c) to pararealgar (As 4 S 4 ; P2 1 /c) does not imply a complete rearrangement since the packing Obtained from a realgar sample from Shimen (China), through grounding and heating "at 295 °C in an evacuated silica tube in a horizontal furnace for 24 h"  of As 4 S 4 molecules is nearly identical in their structures; (5) realgar has the same As 4 S 4 molecule of β-As 4 S 4 ; however, the increase in unit-cell volume of β-As 4 S 4 is much higher than that in realgar. The limited increase in the volume of realgar can be due to an equally limited conversion of As 4 S 4 to As 4 S 5 or to pararealgar-type As 4 S 4 molecules; (6) β-As 4 S 4 (C2/c) microdomains in the alacranite series seem unable to incorporate sulfur since the conversion to pararealgar was partial (and none for As 8 S 9 (P2/c) microdomains); (7) the χ-phase consists of a "β-As 4 S 4 structure (C2/c) with As 4 S 5 molecules randomly substituting for As 4 S 4 " (Bonazzi et al. 2006). These works marked a milestone in the history of studies on this subject and, after them, the overall picture has remained essentially unchanged despite further details being gradually added. The newest achievements can be summarized as follows.
-A new orthorhombic phase (~ As 4 S 5 ), produced by the light-induced alteration of a natural As 4 S 4.35 crystal, was  Fig. 1). B The cyclic transformation described by Kyono et al. 2005 (modified from reference Fig. 6). As specified by the authors, the fact that the As-As bonds are weaker than the As-S bonds, causes the free S atom to be inserted between the atoms of As. C The alteration of β-As 4 S 4 when non/exposed to air and the following heat treatment investigated by Zoppi and Pratesi 2012 (modified from reference Fig. 5). D The alteration of realgar when non/exposed to air and the following heat treatment investigated by Pratesi and Zoppi 2015 (modified from reference Fig. 4) discovered by Bindi and Bonazzi (2007) (Tables 1 and  2). -Kyono 2007 verified that "the exposure time required for the phase transformation increases gradually as the light intensity decreases" (Table 2). -The evidence of additional sulphur atoms was finally observed, thus substantiating and detailing the mechanism described by Kyono et al. (2005). The series of experiments performed by single-crystal X-ray photodiffraction (Naumov et al. 2007), Fourier transform infrared (FTIR) and Raman spectroscopies (Bujňáková et al. 2015) and steady-state single-crystal XRD (Naumov et al. 2010) contributed to directly observe the light-induced realgar → pararealgar transformation at the atomic scale and to visualize "the migration" of the sulphur atom (see also Jovanovski and Makreski 2020) ( Table 2). -The importance of the presence of oxygen in the transformation of both the β-As 4 S 4 phase and realgar was highlighted by Zoppi and Pratesi (2012) and Pratesi and Zoppi (2015), respectively ( Table 2, Fig. 4C, D). In Fig. 3, the reaction formulas provided by Pratesi and Zoppi (2015) for the alteration of β-As 4 S 4 in air (Eq. 14), the light-induced alteration of realgar in air (Eq. 15 where 0 < x < 1) and the heat-induced transformations of pararealgar obtained in the absence of air (Eq. 16, reporting the phases in the order b-e of appearance) are provided. -The natural analog of the HT synthetic β-phase (i.e. bonazziite) was discovered by Bindi et al. (2015a, b, c).

The occurrence of AS-sulfides and some notes on their trade
Among the hundreds of sulfides known nowadays (see Vaughan and Corkhill 2017 for a detailed overview on sulfides), only pyrite (FeS 2 ), pyrrhotite (Fe 1-x S), galena (PbS), sphalerite [(Zn, Fe)S] and chalcopyrite (CuFeS 2 ) were classified as rock-forming minerals, based on their abundance (Bowles et al. 2011). As-sulfides are less abundant but the concentration of minable ore metals makes them important georesources but harbingers of potential air, water and soil pollution as the other metal sulfides. Primary As-sulfides prevail over the other groups of As-bearing minerals and mainly occur in magmatic and hydrothermal environments (Eary 1992;Drewniak and Sklodowska 2013). However, the conditions for their formation are different. For example, high temperatures are commonly required for the formation of arsenopyrite, which is thus typically found in pegmatites, contact metamorphic rocks or, more rarely, in basalts (Drewniak and Sklodowska 2013). Microbial activity can even influence the formation of realgar, pararealgar and alacranite at ambient temperatures in sulfate-reducing environments (O'Day et al. 2004). In this regard, the studies performed by Newman et al. (1997) demonstrated that the bacterium Desulfotomaculum auripigmentum precipitates As 2 S 3 both intra-and extracellularly, due to its ability to reduce As 5+ to As 3+ and S 6+ to S −2 . A few years later, Huber et al. (2000) provided evidence of a new specie-the pyrobaculum arsenaticumthat "can precipitate realgar over its temperature range of growth" (i.e. 68-100 °C) and Demergasso et al. (2007) discussed the microbial involvement in the (abiotic) precipitation of arsenic minerals in borate deposits.
This topic is beyond the scope of this review but it should be pointed out that the literature is vast because it concerns important environmental remediation studies. The articles cited above are only examples and can be used for a first approach to the subject. To return instead to the geographical occurrence of these phases, it immediately becomes evident that, despite literature reports that the deposits exploited in historical times are located in the Carpathians, Caucasus, Gerogia, Armenia and Azerbaijan (Grundmann and Rötter 2007), the geological literature pays little attention to these occurrences while privileging deposits in which arsenic ores are associated with gold deposits. The frequent occurrence of arsenic and realgar in geothermal fluids and Carlin-type gold deposits (see, e.g., Rytuba 1986, Ballantyne and Moore 1988, Boyle 1987, Arehart 1996 triggered, in fact, a large production of literature that we have tried to summarize in Table 3. Data included in Table 3 are certainly partial because the selected literature includes only particularly large deposits that have been the subject of systematic studies while it excludes minor deposits that may have been exploited in antiquity. Mindat.org, for example, lists 796 occurrences (in 57 countries) only for realgar and these too are to be considered incomplete since not all countries have been studied with the same level of detail: Afghanistan (1), Albania (6) (19), Slovenia (2), South Africa (10), Sudan (1), Sweden (2), Switzerland (7), Taiwan (2), Tajikistan (1), Turkey (17), UK (2), Ukraine (2), USA (221), Uzbekistan (1), Vanuatu (1). However, while a review of all occurrences was beyond the scope of this review, we instead found it useful to show the most common mineralogical associations.
Other sources such as those mentioned by Dioscorides in Mysia, Pontus and Cappadocia (see above) or by Schafer (1955) in Hungary, western Georgia, Kurdistan, Shiraz, Takht-i-Sulaiman and Mt. Demawand in Persia are more difficult to localize and/or describe.
Despite these limitations, however, the data collected are sufficient to demonstrate three different aspects: (a) the wide availability of these As-sulfides, (b) the minerals with which they are generally associated; and c) the prevalence of realgar, orpiment and arsenopyrite compared to the other As-sulfides and As-oxides. Another consideration that can be made is that, apart from native elements, realgar is primarily associated with orpiment, arsenopyrite, chalcopyrite, cinnabar, galena, marcasite, pyrite, sphalerite, stibnite and gangue minerals such as barite, calcite, chalcedony, quartz and dolomite. The absence of pararealgar and sometimes also of orpiment in the literature sources should not be considered representative as it is legitimate to imagine that where realgar is, these two phases can also be present.
As for ancient trades, information is scarce. Pliny not only mentioned the island of Topazos in the Red Sea as a source of both sandaraca and ochra Iubra but also specified that neither of them were imported to Rome from that place (Naturalis Historia 35, 39, 22). Therefore, it is not well understood whether this news can be considered reliable since Topazos-i.e. the Zabargad Island (St. John's in English)-is well known for being the likely source of Egyptian peridot 8 and may therefore have exercised a certain suggestion. However, it is a quaternary volcanic island where realgar can therefore be present.
Also with reference to the trades mentioned by Dioscorides, we have no archaeological evidence to prove it or any useful arguments to discredit his information. On the other hand, most of the territories he mentioned have actually proved to be distribution areas of realgar (Table 3).
Lastly, Schafer (1955) claimed that Egyptian orpiment was thought to be of likely Persian origin while a Syrian, Cappadocian, Mysian, Pontic and Carmanian provenance was reconstructed for orpiment traded in Mediterranean Europe. The same Author stated that medieval painters imported their pigments chiefly from Asia Minor, while several sources at Burma, Khotan, Kucha, Aqsu, Malayse, Champa and Japan could have supplied the Chinese and, overall, Far-Eastern painters.

The use of arsenic-based pigments
A good review of the recorded occurrence of orpiment and realgar as pigments was published by West FitzHugh in (1997) as part of the third volume of the Artists' Pigments series. The review spans publications, analyses and personal communications from the 1960s until the mid-1990s and lists the occurrence of orpiment and realgar in objects and collections around the world, from the 1 st Dynasty in Ancient Egypt (31 st century B.C.) to 19 th century Europe. Other, more partial reviews were published by Orna in 2015 and Scott in 2016, on mineral and Egyptian pigments (including arsenic pigments), respectively.
As expected, most of the historical records of arseniccontaining pigments refer to orpiment and realgar. West FitzHugh does mention other arsenic sulfide minerals such as pararealgar, alacranite, duranusite and dimorphite in her Artists' Pigments chapter, but does not document any occurrence of them as pigments, with the exception of samples taken from Tintoretto paintings. However, she rightly suggests that there may well be art objects where pararealgar may have formed over time as a result of light-induced degradation of an original realgar-containing pigment. Since then, pararealgar has been detected on many objects, both as a degradation product but also as a yellow pigment in its own right-assuming artists in the past would not have been able to tell the difference between mineral orpiment and mineral pararealgar (Corbeil and Helwig 1995;Trentelman et al. 1996;Clark 2000, Mazzeo et al. 2004;Casadio et al. 2005 and references therein, Burgio et al. 2007;Burgio et al. 2008).
An intermediate phase between realgar and pararealgar (the so-called χ-phase) has also been detected on various types of artefacts, from Venetian paintings (Trentelman et al 1996), to 15 th century Greek icons (Burgio et al. 2003), to ancient Egyptian papyri (Daniels and Leach 2004), to 9 th century stuccoes from Samarra, Iraq (Burgio et al. 2007).
Despite the rarity of alacranite, several authors identified it in different types of works of art: Gard et al. (2020) used Raman microscopy to document it on an Egyptian cartonnage from the Ptolemaic period (305-30 BC), together with orpiment, bonazziite, pararealgar and uzonite. Luo et al. (2016) detected it in Japanese woodblock prints (1864-1895) and suggest its presence is a consequence of the use of synthetic arsenic glass.
The artificial nature of alacranite has been frequently discussed and Grundmann et al. 2011 suggest that many of its occurrences in works of art may be of artificial origin.
It has long been suspected that many arsenic-containing pigments on works of art are of synthetic origin, since mentions of and recipes for them have been available since antiquity. However, it was not until very recently that their occurrence was proved and recognized. Japanese woodblock prints have been studied extensively and were shown to contain synthetic arsenic sulfides (Vermeulen and Leona 2019). Synthetic arsenic pigments were also detected in a French rococo oil painting and Japanese leather paper , and arsenic sulfide glass was found in many early 18 th -century designs for woven silk made by designer and weaver James Leman in London (Burgio et al. 2019).

Alteration and degradation
As mentioned in "Arsenic-based minerals and pigments" and "The use of arsenic-based pigments" sections of this chapter, arsenic-containing minerals are intrinsically unstable especially when exposed to light. It is not uncommon to see realgar mineral specimens develop a yellow 'dusting'-historically believed to be orpiment, but now known to be pararealgar or one of the intermediate phases-and orpiment specimens lose their vivid yellow character and become duller and greyer, due to the formation of arsenolite.  Stolburg and Dunning 1985, Groff et al. 1997, Groff 2018Groff 2019;Bowell et al. 1999;Cline and Hofstra 2000;Cline 2001;Tosdal et al. 2003*Gold Bar Mining Districts, Nevada Calcite, barite, dolomite, orpiment, quartz, stibnite Yigit et al. 2006 *Cortez Hills deposit, Nevada Aktashite, chalcopyrite, christite, pyrite, sphalerite Maroun et al. 2017 Realgar is always present although never listed while native elements have been omitted. In the second column, localities marked with * correspond to deposits where also Au was found but not necessarily mined a "One of the largest and highest grade stratabound Carlin-type gold deposits in China" (Tan et al. 2015) b "The largest realgar deposit in Asia (…) mined for more than 1500 years" (Zhu et al. 2015); "worked since at least the sixth century AD primarily for medicinal arsenic compounds (important in Chinese medicine)." (Wilson 2007) c "A realgar specimen has a δ 34 S value of − 32.9, consistent with formation of sulfide by bacterial reduction of sulfate" (Miranda-Gasca et al. 1998) d Thallium mineral are marked in bold e "One of the world's leading producers of realgar crystals" (Stolburg and Dunning 1985) This awareness is reflected in a recent increase in studies concerning the alteration of As-bearing phases in works of art. The fact that some arsenic sulfides are altered by simple exposure to light makes artworks particularly sensitive when displayed, especially considering that other factors such as temperature, humidity and pollutants can contribute to the damage/alteration process.
The development of suitable lighting systems has been the focus of the experimental tests performed by Macchia et al. (2013); Blades et al. (2017) and Jo et al. (2019). Starting from the base concept that several pigments are semiconductors, Anaf et al. (2015) applied a thermodynamic approach to investigate their degradation phenomena and flagged up the pH of the environment as a crucial factor for the preservation of paintings. Keune et al. (2015Keune et al. ( , 2016 and Vermeulen et al. (2016) investigated the migration of several degraded arsenic sulfides in painting. In particular, Keune and coworkers provided the first evidence of water transport in the paint system and clearly visualized how the arsenicbased products of light-induced alteration (arsenites and arsenates) are mobilized and dispersed through the depth of the paint, along with being absorbed by chalk, gypsum, and other pigments such as lead white and iron-based earths. Their studies-chiefly performed by means of synchrotron radiation X-ray micro-analyses (As-K edge micro-X-ray absorption near edge structure)-also helped hypothesizing a 5-step process for arsenic trioxide: (1) formation; (2) dissolution in water, (3) oxidation to arsenic pentaoxide, (4) the reaction with other ions (e.g., Pb and Ca) and, finally, (5) deposition as insoluble arsenate. Vanmeert et al. (2019) found mimetite and schulténite as decay products of realgar and orpiment (for lead-based pigments see Gliozzo and Ionescu 2021). Similarly, Hradil et al. (2014) found mimetite as the alteration product of orpiment and minium.
The review compiled by Coccato et al. 2017 summarize the state of the art on pigments' stability and conservation treatments.
On a similar subject, the importance of binders as stabilizers or otherwise was evaluated in the 1-year-long irradiation experiments performed by Feneis and Steinbach (2018) using hydroxypropylcellulose; and in the experiments performed by Vermeulen et al. (2018b) using gum arabic (suitable stabilizer), animal glue, egg white, linseed oil (intermediate impact) and egg yolk (worst binder in terms of pigment degradation).
Finally, the alteration of realgar must also be appropriately evaluated in an experimental setting due to its potential sensitivity to irradiation (see the "Analytical methods" section).

Analytical methods
The presence of an arsenic-containing yellow, orange, or red pigment can be easily confirmed in principle with any simple elemental technique-any scientific method that can reveal which elements are present on an object or in a sample. The most commonly used methods of this sort include X-ray fluorescence (XRF, see Bezur et al. 2020), scanning electron microscopy with energy-dispersive X-ray analysis (SEM-EDX, see Vermeulen et al. 2016) and laser-induced breakdown spectroscopy (LIBS, see Analytical Methods Committee, AMCTB No 91 2019; Westlake et al. 2012;Burgio et al. 2001). XRF can be used non-destructively and in situ, and usually provides qualitative or semi-quantitative information, making it clear if the arsenic-containing pigment is one of the main constituents or a minor or trace component in the object or sample. Data accuracy, however, decreases from laboratory to portable instrumentation. SEM-EDX usually involves sampling (although small objects can also be analyzed whole, and there are cases where samples can be reused for other investigations) and can provide semi-quantitative information. It is worth mentioning that there are two types of SEM equipment, field emission SEM (FE-SEM) and variable pressure or environmental SEM (ESEM), which allow the analysis of fragments small enough to enter the chamber and, net of calibration, provide a much more accurate analysis the flatter the sample surface is. LIBS entails the vaporization of a microscopic portion of an object or sample but can be done in situ. More detailed information about XRF and SEM-EDX can be found in Burgio (2021) in this TC.
However, if the exact arsenic species has to be identified, a molecular or mineralogical technique needs to be used. These can be invasive, minimally invasive or noninvasive, destructive or non-destructive, and can be used in situ or may require sampling. The ones most commonly used in cultural heritage since the late 1990s are Raman microscopy (Smith and Clark 2001;Analytical Methods Committee AMCTB No. 67 2015, Casadio et al 2016; Fourier-transform infrared spectroscopy (Vermeulen et al 2016), diffuse reflectance spectrophotometry (Analytical Methods Committee, AMCTB No 75 2016), optical microscopy (OM) and X-ray diffraction (XRD, see Cloutis et al. 2016;Lutterotti et al. 2016). Synchrotron radiation sources have allowed access to high-energy X-rays (with tunable depth capabilities, higher than those provided by conventional instrumentations) and have boosted the potential of many scientific techniques (Keune et al. 2015;Vermeulen et al. 2016;Janssens et al. 2016;Cotte et al. 2018).
Among the molecular techniques, the past two decades have asserted the preeminence of Raman microscopy: it is non-destructive, can be used in situ and on individual particles as small as 1 μm across, and provides the unambiguous molecular identification of the species under investigation (Clark 1995(Clark , 2007Smith and Clark 2004;Bersani and Lottici 2016;Casadio et al 2016). It was Raman microscopy that proved that often the yellow arsenic-containing pigment on works of art, manuscripts and museum objects was not orpiment but pararealgar or one of its polymorphs (Trentelman et al. 1996;Clark and Gibbs 1997). Pararealgar and other polymorphs were also found as lightinduced degradation products on the surface of realgar mineral specimens (Burgio 2000, unpublished doctoral thesis data, see Fig. 5). And it was still Raman microscopy which helped identifying and characterizing arsenic glass used as a pigment (Vermeulen et al. 2015;Vermeulen and Leona 2019;Burgio et al. 2019).
Among mineralogical techniques, polarized optical microscopy on thin sections or dispersions has been long used for the identification of the phases. However, the destructiveness of the technique and the parallel development of other, more effective techniques, has limited its use in recent years. On the other hand, X-ray diffraction remains the main techniques for the identification of minerals and pigments. Moreover, while the structural study of new minerals necessarily requires sample preparation and laboratory instrumentation, the identification of known minerals and pigments can be carried out both in a destructive and non-destructive way. In the latter case, some recently developed portable XRD (sometimes combined with fluorescence; Gómez-Morón et al. 2016;de Viguerie et al. 2018), the scanners (MA-XRD; see Vanmeert et al. 2018;2019;De Keyser et al. 2017;Uhlir et al. 2019;De Meyer et al. 2019a, b), and the XRD beamlines present at the synchrotrons (Vanmeert et al. 2015;Van Der Snickt et al. 2012;Van Der Snickt et al. 2012;Salvant et al. 2018;De Meyer et al. 2019a, b) have been used for the investigation of artworks. Moreover, X-ray diffraction allows both the incontrovertible identification of the pigment (qualitative analysis) and the quantitative determination of the various phases (QPA) by means of Rietveld analysis (Bish and Post 1993).
The XRD patterns of the main mineralogical phases/pigments are provided in Fig. 6.

Good practice guide-analysis protocol
In choosing the operating procedure, various factors must be taken into account: (1) operator safety; (2) preservation of the artwork; (3) type and accuracy of the data to be obtained.

Operator safety
It is more likely than not that arsenic pigments may be present in historical art objects; it is therefore advisable to adopt all the relevant health and safety precautions to avoid any risks for anyone who can come in contact with a potentially hazardous material.

Artwork preservation
Many of the studies discussed earlier in this chapter demonstrate that most naturally occurring arsenic-containing species are sensitive to light-even exposure to natural sunlight can trigger their molecular transformation and often cause a change in colour. With this in mind, analytical techniques should be chosen and employed with care and a specific protocol in mind, so as to avoid any experimental conditions which may affect any arsenic-containing pigment present. FT-Raman spectra of a mineral sample of naturally degraded realgar. a Raman spectrum of a yellow crystal, identical to that of pure pararealgar. b Raman spectrum of an orange-red crystal, corresponding to the χ-phase, possibly mixed with small amounts of other polymorphs. c Raman spectrum of realgar. All spectra with 1064 nm excitation, 8 mW, resolution 1 cm −1 for a and b 0.5 cm −1 for c  Whitfield (1970), B Whitfield (1973a, b), C and I Mullen and Nowacki (1972), D Bonazzi et al. (1995), E Bonazzi et al. (2003a, b), F Bindi et al. (2015a, b, c), G Bindi et al. (2003), H Bonazzi et al. (2005), J Kampf et al. (2011), K Ballirano and Maras (2002), L Pertlik (1978a). Graphics by Mercury software For example, Raman microscopy is often the technique of choice when looking at the fast, non-destructive, nonintrusive characterization of arsenic-containing pigments in cultural heritage, but as a rule of thumb it is best to use the lowest available power density at the sample, and the longest available excitation wavelength. For example, studies performed in the late 1990s (Burgio 2000, unpublished PhD thesis data) showed that the irradiation of realgar with a low-power 514.5 nm laser (green) almost instantly produces a material the spectrum of which can be consistently interpreted as a 1:4 mixture of realgar and pararealgar. Another degradation product in these conditions is orpiment, As 2 S 3 , indicating that partial oxidation can occur at this wavelength.
A red laser (such as a He:Ne laser at 632.8 nm or a 647.1 Ar + laser) is safer, but only at low power. If the power density is too high, realgar easily converts into pararealgar and its polymorphs. Even safer are far red (785 nm) or near infrared (1064 nm) lasers.

Type of data and their accuracy
As mentioned earlier in this chapter, the analytical approach is different depending on whether the purpose of the analysis is to verify the presence of arsenic in general or the specific composition of any arsenic-containing pigments.
For the former, any of the elemental techniques described in "Analytical methods" section can be used. For the latter, a molecular technique will be needed. At this point, however, it seems useless to add that a mere identification of the chemical element can affect the quality of the overall research which, consequently, is incomplete and in no way definitive.
For this reason, Raman spectroscopy and/or X-ray diffraction should be preferred (if not considered mandatory) for research that aims to be considered accurate and complete. The complementary use of imaging techniques should be also desirable to obtain an overall picture of pigment distribution. For research, where there is both the need to operate non-destructively and the necessity to operate with a highenergy and high-penetrating beam, synchrotron techniques (SR-XRD, SR-XRF, XAS) can provide the necessary help.

Concluding summary of key concepts
• Contrary to what has often happened historically, it is not possible to deduce the identity of the arsenic-containing species simply from the color of the pigment or pigment mixture • Arsenic-based minerals and compounds are numerous and can be found to be associated with realgar and orpiment in the ore sources. Therefore, it is necessary not only to pay due attention to "new" phases that are discovered but also to test the possible presence of other phases already known. It is not unheard of, in fact, that rare phases in nature are found in cultural materials. • The choice of the analytical methodology must take into account the three criteria developed previously, namely the safety of the operator, that of the artwork, and the overall quality of the resulting data. • As a first corollary to the previous point, care should be taken not to employ experimental conditions that can alter the composition of the pigments analyzed and convert one arsenic-containing pigment into another. • As a second corollary, in the wide range of instruments available, precedence must be given to those that are able to incontrovertibly identify the pigment.
Authors' contributions Chapters "Arsenic-based minerals and pigments", "The occurrence of As-sulfides and some notes on their trade" by EG. Chapters "The use of arsenic-based pigments", "Alteration and degradation", "Analytical methods", "Good practice guide -analysis protocol", "Concluding summary of key concepts" by LB. Chapter "Introduction" by EG-LB. Language revision by LB.
Funding Open access funding provided by Università degli Studi di Bari Aldo Moro within the CRUI-CARE Agreement.
Availability of data and materials Data sharing is not applicable to this review article as no new data were created or analyzed in this study.

Conflicts of interest
The authors declare that they have no conflicts of interest.
Code availability Not applicable.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.