Microchemical analysis of Leonardo da Vinci's lead white paints reveals knowledge and control over pigment scattering properties

Leonardo da Vinci, a pivotal Renaissance figure, is celebrated for his innovative painting techniques, particularly sfumato. However, his methods remain largely undocumented. Chemical analysis of his artworks offers a pathway to understanding his materials and techniques. This study focuses on a sample from *The Virgin and Child with St. Anne*, one of Leonardo's final masterpieces, painted between 1503 and 1519. The painting's restoration at the Centre de Recherche et de Restauration des Musées de France (C2RMF) in 2007 provided an opportunity to collect microsamples. One sample, taken from the background near the transition between a brown landscape and blue mountains, is the subject of this analysis. The sample's stratigraphy reveals a blue hue achieved using a small quantity of ultramarine dispersed in an almost entirely white layer (layer β), overlying a white imprimitura (layer α) and a gesso ground layer. A more complex upper layer (layer γ) also exists. The white material in layers α and β is identified as lead white, a common pigment. Lead white is a mixture of cerussite (C) and hydrocerussite (HC), formed via pH-dependent processes during synthesis. While the synthesis method (stack process) was consistent across time, the resulting pigment quality varied due to post-synthesis treatments. Leonardo's choices regarding pigment type and application remain unknown and can only be inferred through such detailed chemical analysis. This study seeks to quantify the cerussite/hydrocerussite ratio in Leonardo's lead white to elucidate his painting techniques and aesthetic aims. Traditional X-ray diffraction (XRD) methods pose challenges for precious samples, while previous photoluminescence (PL) studies have limitations in micro-imaging applications. This study explores deep-UV excited hyperspectral photoluminescence microspectroscopy as a less invasive alternative, validated by synchrotron XRD.

Previous research on Leonardo da Vinci's techniques has focused on his sfumato technique and the materials he employed. Studies using X-ray fluorescence spectroscopy have helped reveal aspects of his layering techniques. However, detailed chemical analysis of his pigment choices, especially regarding the variations in lead white, has been limited. Prior studies have utilized synchrotron-based high-angle resolution X-ray diffraction (SR-HR-XRD) to quantify cerussite and hydrocerussite in lead white pigments from other Old Masters' paintings, providing information about the phase composition and average crystallite size. However, this technique is radiation-intensive and is thus problematic for delicate art samples. Photoluminescence (PL) spectroscopy offers a less damaging method but has yet to be fully implemented in high-resolution micro-imaging for lead white analysis. The current study integrates these techniques to overcome the challenges encountered in previous research. This combined method allows for the precise characterization of Leonardo's lead white pigment in a more sensitive, less damaging way while still providing detailed spatially-resolved data on the sample's composition.

The research employed two primary analytical techniques: synchrotron high-angle resolution X-ray diffraction (SR-HR-XRD) and synchrotron deep UV-excited photoluminescence (SR-µ-PL). For SR-HR-XRD, two fragments (approximately 50–80 μm) from layers α and β were analyzed at the ID22 beamline (ESRF, Grenoble, France). The samples were rotated in a capillary in front of a 1×1 mm² beam at 35 keV. Data acquisition times ranged from 1 to 4 hours. Rietveld refinement of the XRD patterns, using the FullProf Suite, provided quantitative phase analysis (cerussite and hydrocerussite content) and information on crystallite morphology and dimensions. The Williamson-Hall (WH) method was used to assess crystallite size and strain. For SR-µ-PL, the St. Anne sample was analyzed at the POLYPHEME end-station of the DISCO beamline (Synchrotron SOLEIL). Hyperspectral photoluminescence micro-imaging produced full emission spectra (400–680 nm) with 250 nm excitation. A 30 × 20 pixel map (3 µm step size) covering a 90 × 60 μm area was recorded. A spectral unfolding procedure was employed to separate the emission bands from cerussite, hydrocerussite, and the binding medium. This method involves fitting Gaussian functions to the average emission spectrum to determine the spectral position of the main emission bands, followed by a model refinement with soft limits to fit individual spectra and determine emission amplitude matrices. The ratio of emission amplitudes for cerussite and hydrocerussite, corrected for potential effects of local absorption and reabsorption, served as a semi-quantitative measure of the HC/C ratio. This was validated with the HR-XRD data.

SR-HR-XRD analysis revealed a significantly higher hydrocerussite content in layer α (65 ± 2 w%) compared to layer β (46 ± 2 w%). Considering potential contamination of the layer α sample with some layer β material, the actual hydrocerussite content in layer α is likely higher. The Williamson-Hall analysis indicated that hydrocerussite in layer α formed platelets approximately 110 ± 20 nm thick, with a 10:1 aspect ratio. Cerussite particles in layer α were larger than 500 nm. SR-µ-PL analysis confirmed the different lead white compositions in layers α and β. The 2.03 eV emission band (attributed to HC) was dominant in layer α, while the 2.94 eV band (attributed to C) was more prominent in layer β. A metric (Γ) relating the ratio of emission amplitudes to the HC/C ratio, derived from PL, was consistent with the HR-XRD results. Comparing these results to data from other Renaissance paintings, the researchers found that the lead white in layer β showed signs of post-synthesis treatment, likely involving an acidic treatment that reduced crystallite size and shifted the composition towards cerussite. This interpretation is supported by notes in Leonardo's Arundel Manuscript 263, which lists orders for two different lead white pigments with varying prices. Further analysis of the reflectance spectra showed that the sub-micrometric lead white in layer β exhibited a pronounced decrease in reflectance towards the red part of the visible spectrum, implying increased Rayleigh scattering and a potential blue hue. This observation connects to Leonardo's writings on the blue color of the atmosphere and the potential use of lead white to achieve similar effects.

The findings demonstrate that Leonardo da Vinci possessed a sophisticated understanding of lead white pigment properties and intentionally used different lead white types to achieve specific optical and aesthetic effects. The use of a lead white with smaller particles in layer β might have been aimed at either manipulating the rheological properties of the paint for thin glaze layers or to induce Rayleigh scattering, producing a subtle blue hue, consistent with the atmospheric perspective in that part of the painting. The manuscript evidence further supports the idea that Leonardo purchased different lead white varieties, suggesting a level of material control exceeding typical artistic practices. The study validates the use of hyperspectral photoluminescence as a less invasive tool for analyzing valuable historical artworks.

This study reveals Leonardo da Vinci's deliberate use of different lead white grades in *The Virgin and Child with St. Anne*, highlighting his advanced knowledge of pigment properties and their manipulation to achieve specific aesthetic outcomes. The combined HR-XRD and SR-µ-PL analysis, supported by historical evidence, establishes the efficacy of this approach for investigating artistic techniques and validating the hypothesis that post-synthesis treatment was used to produce a lead white with desirable scattering properties. Future research should analyze more samples from Leonardo's works to determine the prevalence of this specific lead white type and further explore the potential of combined techniques like PL and Raman microscopy for analyzing pigments in historical artworks. The development of non-invasive macro-scale analysis techniques is also key for obtaining broader insights into pigments and their use in historical art.

The analysis is based on a single sample, which limits the generalizability of the findings to the entire painting or to Leonardo's entire body of work. The semi-quantitative nature of the PL analysis requires further investigation for improved quantification. The micro-sampling nature inherently restricts the volume of material that is investigated and does not necessarily reflect the entire painting.