Direct evidence for metallic mercury causing photo-induced darkening of red cinnabar tempera paints

The study addresses why and how red cinnabar (α-HgS) in artworks undergoes photo-induced darkening, a phenomenon that alters the visual perception of paintings and polychrome surfaces. Prevailing hypotheses attribute darkening to the formation of metallic mercury (Hg0), often invoking chloride-mediated pathways. Band-edge considerations for cinnabar as an n-type semiconductor (band gap ~2.0 eV; visible-light activation <620 nm) have been used to argue that direct photo-reduction to Hg0 is not feasible without chlorides. The research question is whether cinnabar can darken via a chloride-independent pathway and what reactions are responsible, especially in realistic paint systems with organic binders. The purpose is to test UV-aged cinnabar pigment and egg-yolk tempera to elucidate mechanisms considering semiconductor behavior and pigment–binder interactions. The work is significant for conservation science, informing strategies to diagnose and mitigate color change in heritage objects.

Multiple pathways have been proposed for vermilion/cinnabar blackening: (1) transformation to black metacinnabar (β-HgS); (2) formation of a thin metallic Hg layer, often considered the main cause based on the black appearance of colloidal Hg; and (3) chloride-involved mechanisms where chlorides act as catalysts or intermediates (e.g., HgCl2(ads), Hg2Cl2) that can be photoreduced to Hg0. Band-edge analyses place cinnabar’s valence and conduction band edges at ~2.02 and 0.02 V vs NHE (pH 2), while the HgS/Hg redox couple is −0.70 V vs NHE, suggesting that direct photo-reduction of Hg2+ to Hg0 is not energetically favored; however, mercury chloride species with potentials within the band gap (e.g., 0.48 and 0.27 V vs NHE) could be reduced via photo-induced electron transfer. Proposed chloride sources include impurities in wet-process vermilion; historical coatings (cera punica); sea spray; conservation varnishes/binders/consolidants; and museum housekeeping chemicals. Nonetheless, several reports observed darkening without detectable chlorides, including naturally aged mock-ups exposed outdoors and laboratory samples lacking halides (albeit with slower rates). These mixed findings motivate testing chloride-free scenarios and re-examining mechanisms including semiconductor photocorrosion and interactions with organic binders.

Materials: Natural Chinese cinnabar pigment (Hunan; 15–90 µm) and fresh egg yolk as binder. Egg yolk composition: ~50 wt% solids; 31.8–35.5 wt% lipids, 15.7–16.6 wt% proteins; minor K, P, Ca, Mg (≤1 wt%) and S, Fe, Na (≤0.1 wt%). Sample preparation: Pigment-only: 1 g cinnabar mixed with 3 mL Milli-Q water and applied on 33 mm Ø glass slides. Egg-tempera: cinnabar with egg yolk prepared by traditional methods and brushed in multiple layers on glass slides; pure egg yolk controls also prepared. Drying at ~20 °C, ~40% RH. Final paint layer thickness 0.15 ± 0.03 mm; binder content 14 wt% (dry); mock-up size 23 × 20 mm. UV aging: Samples irradiated for 2 months using a Pen-Ray mercury lamp (185–436 nm; primary 254 nm; ~2800 µW/m2 UV-C at ~2.5 cm). Aging in a 9 L glass chamber at 21 ± 2 °C and 80 ± 3% RH; RH maintained with Milli-Q water reservoir. Daylight excluded with aluminum foil. Under high RH, reactive species generated included ozone, OH•, O2•−, 1O2, and H2O2. Analytical techniques: - Imaging: Photography and optical microscopy (reflection) to document textural/chromatic changes. - Colorimetry: Portable spectrophotometer (D65, 10° observer, Ø 6 mm; SCI) with CIE L*a*b*; ΔE* = (ΔL*2 + Δa*2 + Δb*2)1/2; ≥10 measurements/sample. - µ-XRF mapping (50 kV, 600 µA) to assess elemental composition and impurity distribution. - FESEM-EDS (10−6 Pa; 3 kV SE, 10 kV BSE; 20 kV for EDS) to observe morphology, cracking, secondary phases, and detect Hg/S ratios; droplet behavior under beam/vacuum examined. - XRD (Cu Kα, Ni filter; 45 kV, 40 mA; 3°–60° 2θ; 0.05° 2θ/s) for phase identification, peak intensity changes, and crystallite size (Scherrer). - ATR-FTIR (400–4000 cm−1; 2 cm−1; 75 scans) on pure binder and paint chips to monitor lipid/protein oxidation and hydrolysis. - UV-Vis-NIR of egg yolk to assess optical absorption. - XPS (monochromatic Al Kα; survey 75 W/160 eV; HR 225 W/20 eV; 1.33×10−8 Pa) for Hg and S oxidation states and Cl detection; C 1s at 284.6 eV as reference; S 2p doublet constraints ΔBE=1.2 eV, 2:1 ratio; depth profiling by Ar+ etching (4 keV, 10 mA; 0–120 s; nominal etch rate 0.75 nm/min; raster 3.5 × 3.5 mm).

- No chlorides were detected in the cinnabar pigment or paint mock-ups by µ-XRF, FESEM-EDS, or XPS. - UV-aged pigment (no binder): • Visual/microscopic darkening and formation of a dense yellow surface layer after 2 months. • XRD detected mercury sulfate hydrate (HgSO4·H2O) and basic mercury(II) sulfate (schuetteite, Hg3(SO4)O2) after 4 weeks. Cinnabar Bragg peak intensity decreased by ~50% after 2 months; crystallite size decreased by 9 ± 2% (strain-related broadening). No metacinnabar or chloride phases detected. Minor gypsum formed (from calcite sulfation). • FESEM revealed crack development and extensive coverage by star-shaped aggregates of ~500 nm plates (HgSO4·H2O) and larger 5–10 µm plate-like crystals (schuetteite), with Hg/S atomic ratios ~1 and ~1.5, respectively. Occasional nano-droplets presumed to be Hg0 evaporated under electron beam/vacuum. • XPS S 2p showed ~60 at% sulfate (168.4 eV), stable upon Ar etching; remaining S at ~162.6 eV attributed to a metal-deficient sulfide layer. Hg 4f7/2 comprised ~75–80 at% at 100.8 eV (Hg in HgS/mercury sulfate) and an oxidized Hg species at ~102.4→102.0 eV upon etching; Hg0 was not confirmed by XPS, possibly due to evaporation under UHV. - UV-aged tempera paint (with egg yolk): • Pronounced darkening without a visible yellow sulfate coating. Colorimetry: ΔE* = 16.0 ± 0.7 with decreases in L*, a*, b* (darker, bluish-greenish). • FESEM showed severe binder degradation and nano-to-micrometer Hg droplets (<100 nm to 1.5 µm) on the surface that persisted under high vacuum and electron beam; apparent contact angle reduced to ~105° vs ~140° on polished cinnabar, suggesting enhanced interaction with oxidized organics. • XRD did not detect Hg sulfate phases or peak broadening; gypsum and possibly goethite and zinc sulfate were observed (oxidation/sulfation of accessory minerals). • XPS detected surface sulfates (S 2p3/2 ~168.5 eV) with an initial S(sulfide):S(sulfate) ratio ~1; after 120 s Ar etch (~1.5 nm), sulfate decreased to ~20%, indicating a thin sulfate layer below XRD detection. Hg 4f7/2 remained at ~100.2 eV, consistent with weakly bonded elemental Hg. - ATR-FTIR showed accelerated photo-oxidation of the binder in the presence of cinnabar: rapid loss of unsaturated lipid band at 3006 cm−1 (disappearing by day 2 with cinnabar vs day 9 in pure egg yolk), decreases in CH bands (2955, ~2920, ~2850 cm−1), decrease/shift of ester C=O band (1740→1733 cm−1) and emergence of free fatty acid bands (around 1705–1711 cm−1). Protein amide changes indicated oxidation/aggregation; shifts in amide II suggested protein–Hg2+ complexation. - Mechanistic evidence: Cinnabar undergoes photocorrosion under light and high RH, oxidizing S2− to SO4 2− (forming HgSO4·H2O and Hg3(SO4)O2). Redox potentials for mercury sulfate and Hg(I/II) species (0.83 and 0.80 V vs SHE) lie within cinnabar’s band gap, enabling subsequent photo-induced electron transfer to Hg0. In pigment, extensive sulfate formation (especially schuetteite) sequesters Hg2+, limiting Hg0. In paint, the organic binder both limits sulfate formation (hole scavenging, physical/photochemical filtering) and provides oxygenated functional groups that chemisorb Hg0, resulting in stable Hg droplets and visible darkening. The entire process proceeds without chlorides.

The findings demonstrate that cinnabar darkening can proceed through a chloride-independent, semiconductor-driven pathway. Photocorrosion of HgS under UV/high RH generates holes that oxidize surface sulfide to sulfate, forming mercury sulfate phases. Because the redox potentials of mercury sulfate/Hg species fall within the band gap of cinnabar, photo-excited electrons can reduce these to Hg0 in sequential steps. This explains the observation of Hg droplets on UV-aged paint. In pigment-only samples, abundant sulfate formation (notably schuetteite) acts as a sink for Hg2+, restricting Hg0 generation and leading to yellow surface layers and cracking due to volume increases. In tempera paint, the organic binder competes for photo-generated holes (promoting binder oxidation) and acts as a partial protective/photochemical filter, thereby limiting sulfate formation to a thin surface film, yet permitting sufficient Hg0 generation and stabilization via chemisorption to oxidized functional groups on organics. This directly addresses the research question by providing mechanistic and analytical evidence that metallic Hg is produced and responsible for darkening without requiring chlorides, reconciling cases where chlorides are absent in altered artworks.

This study provides direct, unambiguous evidence of metallic mercury droplets on UV-aged cinnabar tempera paint surfaces and establishes an alternative, chloride-free pathway for cinnabar darkening. The pathway involves (1) photooxidation of HgS to mercury sulfates via photocorrosion and (2) subsequent photo-induced electron transfer reducing mercury sulfate/ions to Hg0. In paint systems, the organic binder accelerates its own photo-oxidation, limits extensive sulfate growth by scavenging holes, and supplies adsorption sites that stabilize Hg0, producing visible darkening. These insights inform diagnosis and conservation of cinnabar-based artworks where chlorides are absent. Potential future research could systematically investigate wavelength- and humidity-dependent kinetics under museum-relevant conditions, explore different binders and additives influencing hole scavenging and Hg0 sorption, and assess mitigation strategies to inhibit photocorrosion or Hg0 stabilization.

- Aging conditions used a UV-C rich source and high relative humidity (80 ± 3%), which are more aggressive than typical display conditions; although visible light can activate cinnabar, rates under milder conditions may differ. - XPS could not unambiguously detect Hg0 on pigment-only samples, likely due to evaporation of loosely bound Hg0 under ultra-high vacuum; identification of an oxidized Hg species at ~102 eV lacked definitive reference assignment. - XRD did not detect mercury sulfate phases in paints, implying very thin layers near or below detection limits; quantitative thickness/composition of these layers were inferred from XPS depth profiling. - The role of trace impurities and accessory minerals (e.g., calcite, pyrite, sphalerite) in sulfation/oxidation was observed but not fully quantified regarding their impact on HgS photocorrosion and darkening. - The study focused on one pigment source and one binder (egg yolk); generalization to other cinnabar sources, formulations, and binders requires further testing.