Paint Relics on Middle Age Building Stones as Proxies of Commercial Routes and Artistic Exchanges: A Multi-Analytical Investigation

Fifty-four pieces out of 356 marble pieces deriving from the decorative and architectonic apparatus of the medieval monastic complex of S. Francesco of Castelletto (Genoa, Italy) preserve traces of varicolored paint layers. Microscopic samples of green, blue, red, pink, white, and yellow paint relics were collected by scalpel and analyzed by means of Scanning Electron Microscope coupled with Energy Dispersive Spectroscopy (SEM-EDS), μ-Raman, and Fourier Transform Infra-Red Spectroscopy with Attenuated Total Reflection (FTIR-ATR), to characterize pigments and binders. The combined results from the different techniques allowed verification that stone decoration in Genoa during the Middle Ages encompassed a calcite groundwork and the use of a mixture of oils and proteins (probably egg) to apply pigments. The assemblage of impurities within the pigment has been correlated with the provenance sites along the commercial continental (Hungary and France) and maritime (Sardinia, Cyprus, or Veneto) routes between the 13th and 15th centuries. Moreover, the investigation of the painted layer improved the characterization of the decorative techniques in use in Genoa during the Middle Ages.


Introduction
Studies over the last 50 years have demonstrated that the neoclassical perspective of absolute white marble sculptures from antiquity to modern times is not correct. Nowadays, it is well known that antique cultures (Egyptians, Greeks, and Romans) used to apply naturalistic polychromatic paint layers onto marble sculptures, although their symbolic value has still not been definitively assessed [1]. The use of paint on stone surfaces is also reported for the Middle Ages [2]. Here, bright colors not necessarily associated with naturalistic appearances [3] characterize architectural surfaces (portals, bas-reliefs, capitals, and columns, etc.), emphasizing the function of the building. In this perspective, the selection of pigments was also significant. In fact, depending on rarity and value, some pigments (e.g., natural ultramarine) were used only to represent deities and royalty.
Several theoretical essays of the Middle Ages (Pseudo Eraclio, Boileau, Cennini) report the habits of Middle Age painters and their know-how concerning materials used in art, stating that painting on stones was a minor art. Two treatises on the 13th century (Pseudo Eraclio, Boileau) report that polychromies on stone were the result of a succession of two (or three) ground layers of lead white, on top of which pigments dispersed in oil were applied [4]. Later on, Cennino Cennini (15th century) described the technique in use since Giotto, stating that pigments, usually obtained from minerals and rarely by alchemic processes, were reduced to powder and mixed with an oleic binder to be applied on stone surfaces, and were in use in the northern regions (Germany [5]). These technical essays report

Materials
The study of the pictorial layers was carried out on paint relics on nine marble fragments belonging to the Middle Age phase of the church, and stylistically dated c. the second half of the 13th century ( Figure 1; Table 1). Their function ranged from architectonic (e.g., pillar sections) to purely decorative (e.g., ornate arches), and originally provided the effect of colorful surfaces. Figure 1. Nine of the representative studied fragments. Scale bar 10 cm. The photographs are courtesy of the Superintendence of Archaeology, Fine Arts, and Landscape for the metropolitan city of Genoa and the provinces of Imperia, La Spezia, and Savona (Italian ministry of Cultural Heritage). Partial or total reproduction is forbidden without prior (written) permission.
The study was carried out with a multi-technique approach (Table 1) on microscopic samples (about 1-3 mm 3 ) collected by lancet. Samples were further subdivided so they could be analyzed with different instruments. For each sample, one part was used as collected (FTIR-ATR analysis), one part was embedded in epoxy resin to make cross sections (microscopy and Raman spectroscopy), and, finally, one part was mounted with carbon tape on a stub and was then coated to be conductive under SEM-EDS analysis. Paint traces were identified and documented in-situ at the repository with a digital portable microscope (Dinolite AM-7013MZT; Dino-Lite Europe IDCP B.V., Naarden, The Netherlands). The 50× to 200× magnification allowed the study of the painting layer characteristics and the stylistic details, including the brush strokes.

Scanning Electron Microscopy Coupled with EDS
The SEM analyses were carried out using a Tescan Vega 3 LM scanning electron microscope equipped with an Apollo X detector and a Microanalysis TEAM Energy Dispersive Spectrometry (SEM/EDS: TESCAN, Brno, Czech Republic), installed at the Department of Earth Environment and Life Sciences (DISTAV) of the University of Genoa, Italy. Microphotographs were acquired with an optimized ratio between back scattered and secondary electrons in order to preserve chemical and topographic information. The EDS analysis were performed with an electron beam power of 20 kV and a working distance of 15 mm. The data were acquired under high-vacuum conditions on graphite sputter-coated three-dimensional samples.

µ-Raman Spectroscopy
The µ-Raman measurements were performed on three-dimensional micro-samples and on cross sections, at DISTAV, University of Genoa, with a Horiba Jobin-Yvon XploRA_Plus single monochromator spectrometer (HORIBA Ltd., Kyoto, Japan), equipped with 2400 groove/mm grating and coupled with an Olympus BX41 microscope (Olympus Italia S.r.l., Segrate, Italy; microscope objectives, 10× and 100×; spatial resolution, 2 µm). A solid-state laser source at 532 nm (maximum power 20 mW) was used for excitation. The spectra were collected between 80-1800 cm −1 and 3000-3400 cm −1 , but only the useful range is herein reported. The instrumental spectral resolution was 2 cm −1 , guaranteeing a peak position accuracy of 0.56 cm −1 . Neutral density filters were employed to keep the laser power at a low level (0.05-1 mW) on the samples to avoid laser-induced transformations. The spectra were collected with repeated acquisitions (10-15) of 3-10 s, according to signal to noise ratio. The spectrometer was calibrated on a pure silicon wafer (Raman peak at 520.5 cm −1 ). Peaks were identified and interpreted based on specialized databases reporting peak listings of minerals and art supplies in use from antiquity until the present day [21][22][23][24].

Fourier Transform Infra-Red (FTIR) Spectroscopy
FTIR spectroscopy was performed using a Perkin Elmer Spectrum 65 FTIR Spectrometer (Perkin Elmer, Waltham, MA, USA; spectral resolution of 0.5 cm −1 ) equipped with an attenuated total reflection (ATR) accessory (diamond crystal), a KBr beam-splitter and a deuterated triglycine sulfate (DTGS) detector. All spectra were recorded with 60 scans from 4000 to 600 cm −1 . Micro fragments of the samples were put directly in contact with the ATR diamond crystal, making sure that the outermost painted layer was correctly placed to be analyzed. FTIR absorption peaks were identified and interpreted according to databases specialized in artistic materials [21,24].  could be caused by silicate impurities, probably due to the burial phase, or to phosphate groups often present in the mixite structure [25].

Red
Traces of a light shade of orange-pink pictorial layer (0.2 mm thick; Figure 3d) from a double arch decorated with flowers and leaves (GPT'03 300-14) were addressed. The surface analysis evidenced a pictorial layer uniform in grain size and texture, although the color ranged from darker tones of red to light orange-pink (Figure 3a). The µ-Raman spectrometry (Figure 3h) allowed characterization of the main pigment as red ochre through its peaks at 224, 245, 293, 410, and 612 cm −1 (hematite) and 526 and 677 cm −1 (magnetite). In addition, quartz impurities (peaks at 125, 200, 262, 354, and 463 cm −1 ) were identified in the pigment, as in sample GPT'03 300-4a.
Carmine red brush strokes occurred on a concave face of the section of jamb of the arched portal or window GPT'03 300-51. The surface analysis by optical microscopy highlighted high homogeneity in the pictorial layer (Figure 3b). Based on the results of SEM-EDS (Figure 3e), FTIR (Figure 3g), and µ-Raman (Figure 3h) analyses, the pigment was interpreted as red ochre (hematite + clays; hematite peaks at 224, 245, 293, 410, and 612 cm −1 ; and silicate absorptions at 1032 and 3700 cm −1 ). EDS analysis also allowed the identification of rare grains of barite (Ba, S) as accessory mineral.  A light pink paint layer (0.15 mm thick; Figure 3f) also characterizes part of the phytomorphic decorations of the bas-relief arched niche GPT'03 109-1a. Isolated rare darker grains (Ø 0.05 mm), possibly acting as color centers, are dispersed in the paint layer (Figure 3c). The pigment was identified as burnt sienna (peaks at 211, 283, and 388 cm −1 ) by µ-Raman analyses (Figure 3h).

Black
The sculpted grapes cropping up from the racemes in GPT'03 109-1b preserve traces of violet-black color. In-situ microscopy evidenced a bluish grey ground-layer with small irregular black particles as color cores (Figure 4a,c). The µ-Raman spectrum (Figure 4f) allows identification of the pigment grains as carbon black (broad peaks at 1333 and 1604 cm −1 ).

Yellow
A fragment of a decorated jamb of arched portal or window (GPT'03 300-42a) showed remnants of a yellow paint layer on at least two sides. The surface exhibited a homogeneous thin layer (Figure 5a). Microtextural analysis showed microscopic pigment powders (1-10 µm) dispersed in an organic binder (Figure 5b,c). The pigment is identified as a lead-based pigment (probably massicot; peaks at 144 and 283 cm −1 ) mixed with yellow ochre (goethite peaks at 303, 430, and 675 cm −1 and EDS peaks of Fe, Al, and Si) (Figure 5f). FTIR absorption bands at 1032 and 3700 cm −1 confirmed the presence of silicates (Figure 5e). The µ-Raman (Figure 5f) also allowed the identification of gypsum (peaks at 1011 and 1037 cm −1 ). Furthermore, EDS evidenced rare barite (Ba, S) grains, probably related with the ochre.

Ground Layers
A portion of white layer, probably the ground layer for subsequent decoration, was sampled from a decorated double arch (GPT'03 300-4b) (Figure 4b). The micro-texture evidenced three overlapping layers (Figure 4d), determined as Ca/Mg carbonate layers (see also, the Raman spectrum, Figure 4e), with minor occurrences of silicates and Fe-oxides. The same composition characterized the ground layers in GPT'03 300-141.  Figure 6). In particular, by computer based modifications, it was possible to subtract the signal of the inorganic fraction (i.e., mainly calcite and azurite) from the spectra of sample GPT'03 300-9 ( Figure 6), improving the signal of the absorptions due to lipoid chains (esters-triglycerides absorption at 1730 cm −1 ) and proteinaceous materials (amides transitions at 1600 and 1300 cm −1 ). Figure 6. FTIR spectra of GPT'03 300-9. Comparison between the original spectrum and that originated by software after carbonate subtraction.

Weathering Products
Secondary processes as precipitation of halite, gypsum, and secondary calcite on the surfaces of the majority of the pieces were inferred by SEM-EDS (Figure 7a,b). Moreover, in GPT'03 300-42a, a crust of Ca oxalates (weddellite or whewellite) occurred between the pictorial coat and the marble substrate (Figure 7c).

Blue
Blue paint layers proved to be made up mainly of azurite. This mineral has been used as a pigment since antiquity, and several extraction sites are documented. Antique treatises report that copper carbonate was extracted mainly in Hungary and Germany, and in Chessy (near Lyon, France), Sardinia, and Tuscany (near Siena, Italy) [5,[26][27][28]. The occurrence of accessory monazite and clinoatacamite, malachite, barite, and mixite, associated with azurite in the addressed paint layers were used to infer a plausible pigment provenance [29,30]. The associated atacamite and clinoatacamite can be interpreted as intentional additions to the mixture, accessory minerals proxies of the ore deposit, or weathering product [31]. In fact, the chlorides, atacamite and clinoatacamite, already used as pigments by Egyptians and easily synthetized by alchemic processes, originate naturally in the azurite deposits and, after pigment application, as weathering products in the presence of Cl compounds. Also, the rare grains of malachite associated with azurite in natural outcrops and originating as its natural decay product in wet environments, could be naturally associated to calcite or intentionally added to obtain a desired color. Mixite has already been found in Italian artworks as impurity associated with azurite [28,30] and hypothesized as a provenance indicator [30]. By joining the historical data of the coeval technical essays, and its occurrence with azurite, the possible provenance for the pigment was restricted to Germany, Hungary, and, probably, the Czech Republic [30,32]. Moreover, the presence of monazite grains, a ubiquitous mineral phase, and pyromorphite could provide provenance indications, although no literature is reported. However, coupling historical data and mining information [5,27], and considering the mineralogical association of azurite, lead minerals, and monazite, Sardinia is a possible source [33].
The match between the analysis of the paint layers of GPT'03 300-44 and GPT'03 300-98 suggests that the fragments are part of the same decorative piece, as already hypothesized after the stylistic analysis.

Red
Various ochres were identified as major components of red and pink paint traces. This type of pigment has been widely employed for centuries across Europe, due to its widespread availability in nature. Recent studies [34] linked mineral impurities and Raman peak shifting to ochre provenance: red ochre rich in hematite and magnetite with quartz impurities were related with a provenance from France (Puisaye or Vaucluse [34]). Conversely, barite impurities were recently associated with ochre deposits from the Ligurian area [35].

Black
Since antiquity, carbon blacks have been widely used. In fact, their simple production processes mean they can be produced locally, reflecting the culture and environment of each region [21]. The relative intensity of the peaks and the position of the secondary peaks suggest that the pigment applied was vine black [36,37], widespread in the Mediterranean region. It is likely that combusted vegetal remains, probably vine, were added to red ochres (in GPT'03 300-51) and azurite (in GPT'03 300-9) to produce darker shades.

Yellow
Analytical results on GPT'03 300-42a agree with Middle Age essays on pictorial art, where the reported yellow pigments were yellow ochre, orpiment, and lead based pigments including litharge, massicot, and lead-tin yellow. The alchemic synthesis for these pigments does not allow useful hints in provenance assessment.

Green
Green earths originate in glauconite and/or celadonite [38,39], associated with different geological environments. In fact, celadonite occurs as an alteration of basaltic rocks, whereas glauconite is derived from alteration of marine clays and sandstones [31,40,41]. Based on historical art supply data from the coeval technical essays, mineral identification helped in inferring provenance for the used green earth pigments. Spectroscopic techniques allowed the identification of celadonite as the major/sole component of the pigment [38]. Therefore, two provenance sites, assessed extraction localities for celadonite from antiquity to the present day, are possibly Mount Baldo (Verona, Italy) and Cyprus (Greece).

Ground layers
The exposed ground layer was analyzed in GPT'03 300-4b. The results of EDS microanalyses are comparable with those of the local production of lime mortar. In fact, since the Middle Ages in Genoa, quarries and furnaces for lime production were active in the area of Mount Gazzo (Sestri Ponente, Genoa) [42]. The production exploited the Triassic dolostone formation that also contains traces of Si, Al, and Fe, as found in the addressed samples [42].

Binders
The FTIR spectrometry evidenced the joint occurrence of lipoid chains and amides, as in previous studies [8]. The signal due to the organic fraction is weak compared to that of the pigment, probably due to ageing of the lipoid and protein structures of the binder. In fact, the chemistry of the substrate (i.e., calcium carbonate) and the nature of the pigments (e.g., copper carbonate, lead oxide, etc.) can induce chemical reaction on the organic medium [43].

Weathering Products
Most pieces show superficial accretional weathering (gypsum and secondary calcite) whose origin cannot be assessed, as the outdoor exposure for 500 years and the burial time interval could both have contributed [20].
In GPT'03 300-42a, a crust of oxalates probably originated from the reaction between the marble substrate and the decomposed organic fraction, in turn derived from the interaction between the lipoid-proteinaceous binder and the metal oxide pigment (massicot) [11,16,44].

Conclusions
The multi-technique approach allowed the characterization of the ground layers and the pigment relics of the pictorial layers. The first are usually composed of locally-produced lime, but, herein, no organic binder was preserved enough to be characterized. Conversely, although extremely aged, it was possible to assess that pigments were mixed with lipo-proteinaceous binders.
Even if the monastic complex was relevant in Genoa during medieval times, no valuable pigments, such as cinnabar or natural ultramarine, were identified. A possible reason for this is the fact that it was a Franciscan monastery vowed to poverty, or that the analyzed paint relics were found on ornamental fragments and not on statues or on fine decorations.
The lack of evidences for local sources of azurite, green earths, or massicot supports the fact that these art supplies were imported along the many maritime trade routes connecting the Genoa harbor to the rest of Europe, the coast of the Mediterranean up to the Black sea, and the Far East [45].
The two different mineralogical associations with azurite suggest that the pigment had different origins, pointing to a thriving market of art supplies during the Middle Ages. The lack of archive information on pigment trades during the Middle Ages, and the scarcity of mineralogical studies describing the extraction sites for art materials, make it difficult to definitively recognize the provenance of the analyzed pigments. However, based on the analytical and bibliographical data, it is possible to claim that the city of Genoa was a trade destination for both continental routes (Hungary and France) and maritime commerce (Sardinia, Cyprus, and Veneto).