Tales from the lab beneath the Louvre

The Louvre’s particle accelerator has gone digital. The world’s most famous art museum has had a particle accelerator since the late 1980s, the crown jewel of a robust analytical laboratory dedicated to paintings, sculptures, and other cultural heritage artifacts. Major upgrades in recent years replaced all the optics, controls, and sensors—everything except the core acceleration module. The instrument can now fling protons, helium ions, and, soon, deuterons at priceless pieces of art to probe their elemental composition with all the precision and repeatability of modern vacuum science.

The lab is called the Center for Research and Restoration of the Museums of France, or C2RMF; it draws scientists looking to study art, just as the Mona Lisa, displayed 260 m to the east, draws more generalist art lovers. C2RMF’s star attraction is the Accélérateur Grand Louvre d’analyse élémentaire (AGLAE), which produces atomic emission spectra that enable researchers to tease out the elemental composition of materials. It’s joined by other X-ray equipment; spectroscopy and imaging tools that use visible, ultraviolet, near-infrared, and infrared light; magnetic resonance instruments; chromatographs; and mass spectrometers, all tucked underground and connected to the museum’s network of secret subterranean tunnels.

Detecting art fraud is the C2RMF’s major raison d’être. French law gives the government the right of first refusal on art that goes up for sale in the country. As part of that process, curators and historians at the Louvre must determine if the work is authentic. Analytical results from C2RMF scientists inform those decisions, says Gilles Wallez, a crystallographer at the center and a research professor at Sorbonne University.

The work is a collaboration, but the curators always get the final call, Wallez says. The case of a ceramic Chinese horse provides a good illustration of the dynamic.

The case of the ceramic Chinese horse

The seller claimed the sculpture was 2,000 years old, a product of China’s Han dynasty. The museum’s experts in ancient Chinese art had their doubts, however, and thought X-ray imaging could uncover hidden structural elements to help them decide. But previous work at the C2RMF and peer laboratories has shown that X-rays can also muddy chemical details that other techniques rely on to determine the age of ceramics.

One such method is thermoluminescence dating, which uses heat to release electrons that have been trapped in defects in a ceramic’s microscopic crystals, a process that emits light scientists can quantify. The electrons get jammed into these weird, quantum-mechanical pockets as they absorb ambient background radiation over time—more trapped electrons means more time has passed since the piece was made.

An X-ray scan of the whole horse would artificially boost the number of trapped electrons, making future thermoluminescence analysis useless or, worse, misleading. So the curators decided instead to have C2RMF scientists remove just a tiny chip from the bottom of one of the hooves—a wound no one would see when the horse was standing on display—and date the tiny fragment.

Anne Bouquillon, one of the lab’s experts on thermoluminescence, took the sample and obtained results that were consistent with the claimed age of the horse. Even with that supporting evidence, the museum curators still felt something was off. After much deliberation, they took a deep breath and asked the scientists for an X-ray scan of the whole piece, despite the concerns.

The risk paid off. “Inside, they saw iron rods, which were used to strengthen the ceramic, obviously recent,” Wallez says. “It was a fake.” Together, the curators and scientists eventually concluded that beyond not being ancient, the piece was an intentional fraud. Counterfeiters had found bits of legitimate Han-period pottery and stuck them on the horse’s heels, expecting laboratory sleuths to sample from that least-visible spot.

“You always have to keep in mind that the trading of counterfeit works of art is the same in income, let’s say, as it is for drugs,” Wallez says. “But people who are working in counterfeiting generally have higher education. And there is less risk of being shot.”

A significant shard

It might seem that the C2RMF would hire scientists steeped in art. But the Louvre attracts the best art historians and restorationists on the planet; art knowledge is in ready supply from those departments. What the lab needs is people with world-class sample-handling skills, people who can safely stick a specimen from the Mona Lisa into the beamline of a synchrotron.

Wallez and coworkers recently did just that, building on nearly 2 decades of study on the lead pigments underlying Leonardo da Vinci’s portrait masterpiece with a set of experiments at the European Synchrotron Radiation Facility, in Grenoble, France.

The story of the significant shard starts in 2007, when the Mona Lisa was out of its frame for maintenance. The museum decided to take a small sample from a spot on the upper right of the painting, one that would be hidden by the frame, to study the chemistry of the paint used by the Renaissance polymath. The lab has been using the same 10 µg fragment ever since and expects to never be allowed to take another, Wallez says.

The stakes felt especially high when he and colleagues took a tiny section of the shard to Grenoble. The X-rays from a synchrotron are more brilliant and focused than those lab-scale equipment can produce. That intensity is helpful for getting detailed chemical information from very small samples using a common method known as powder X-ray diffraction.

Unfortunately, the glass capillary tubes that held the Mona Lisa shard and 19 other samples were not quite right for the instrument in Grenoble. The team had access to the beamline instruments for just 48 h and now had to break each capillary on-site and transfer the samples to new tubes. “I was frightened,” Wallez says. “I knew there was a sample of the Mona Lisa” in the batch. His colleagues refused to make the transfers, so the task fell to him as the most experienced crystallographer.

“I told them, ‘Do not tell me which sample is the Mona Lisa. I don’t want to know,’ ” Wallez recalls. One by one, he moved the shards to new capillaries. After he had completed a few, his colleagues all let out a relieved sigh and told him he had just finished the one from the Mona Lisa. In the end, the trip went well. The team found plumbonacrite in that shard, for example, which they weren’t expecting and which suggests that Leonardo used lead oxide in an alkaline formulation to thicken the oil in his primer (J. Am. Chem. Soc. 2023, DOI: 10.1021/jacs.3c07000).

Almost 20 years earlier, when the Mona Lisa lead work began, coauthor Elisabeth Ravaud was a new hire whom the C2RMF had recruited from the field of medicine, where she was a physician specializing in radiology and scientific imaging. Her first assignment: the curator in charge of Italian Renaissance paintings asked for an X-ray imaging study of the Mona Lisa.

As if working on the famous portrait right off the bat wasn’t enough, the pressure on Ravaud got more intense when she reached a controversial conclusion: Leonardo had mixed the pigment lead white with oil as the base coat for the portrait. Toxicity aside, lead makes excellent white pigments. It also shows up vividly in an X-ray image, making it easy for researchers to identify. But Italian painters at the time didn’t use lead white and oil to prepare wood for painting; they used gesso, a mixture of gypsum, rabbit skin glue, and water.

Lead wasn’t common in base coats of paintings until well after Leonardo’s time. One example, The Night Watch, was painted by the prototypical Dutch master Rembrandt in 1642, nearly 120 years after Leonardo’s death (Sci. Adv. 2023, DOI: 10.1126/sciadv.adj9394).

“The curator said, ‘You made a mistake. It’s not possible; you have to change your report,’ ” Wallez says. “For us, it’s a crime against science: [Ravaud] had to change the report and say the contrary of what she found.”

Despite the setback, Ravaud persevered at the C2RMF, becoming a sought-after expert on nondestructive imaging that reveals the chemical secrets of significant paintings. And she was right about the Mona Lisa’s lead base coat, which turned out to be one of several uncommon formulations that Leonardo most likely devised on his own (Sci. Rep. 2020, DOI: 10.1038/s41598-020-78623-5).

“It must happen that someone was the first one,” Wallez says. “It was Leonardo.”

Ravaud’s discovery eventually prevailed. In 2019, the Louvre published a book—L’Art de la Matière, l’Art et la Manière (The art of the material, the art and the manner), with Ravaud as the lead author—that includes a detailed description of Leonardo’s lead white tinkering. The 2023 synchrotron paper Wallez recently coauthored treats the use of lead in the Mona Lisa’s base coat as settled science. “It was, for her, something very satisfying to see,” he says.

Conservation collaboration

When they’re not doing forensic work on antiquities—legitimate or otherwise—the scientific staff members of the C2RMF spend a lot of time developing methods that will let them get results without damaging the art. Often, that means testing techniques on more mundane materials.

A few years ago, Wallez says, some colleagues went to an auction with an unusual angle. They had authenticated a certain piece as a legitimate 18th-century work typical of the time in its materials and methods but without any artistic or historical importance. “They bought it, brought it back here, and cut it into pieces to do experiments,” he says.

“You can work on mock-ups. We know what the oil is made of; we know what the pigments are made of, but you have to check it on the real thing,” Wallez says. “So, sorry for this painting; it was a sacrifice for science.”

One way in which the team is using those slightly sad slices is to test a laser tomography technique that could gently remove varnish. Day-to-day restoration work happens in a different underground Louvre laboratory. But as the C2RMF’s full name suggests, part of its work is to inform and improve the techniques used to maintain artwork. Paintings often have a topcoat of transparent varnish, which artists use to protect their pigments from oxygenation. But over time, many varnishes grow yellowed and dark, so conservators have to remove and replace the layer.

It’s a painstaking process that involves rubbing the artwork with a solvent-soaked cotton swab a centimeter at a time, stopping as soon as the cotton shows the slightest discoloration. The new laser tomography technique the C2RMF is testing has two conceptually straightforward steps. First, a laser scan precisely maps the thickness of the varnish across the surface of the painting; in a second pass, a cleaning laser vaporizes the varnish according to that depth map.

It’s a cool idea that could be safer and faster than the Q-tip method, Wallez says, but the restoration lab wants to see reams of test data before it tries the approach on anything of value.

Method development goes beyond making restoration easier and into creating replicas of centuries-old materials. This can tease out where and how real masterpieces were created, and analyzing these aboveboard counterfeits informs the lab’s work in fraud detection and in open-ended inquiry, says C2RMF senior researcher Thomas Calligaro. “You get something about the process that is used to produce it,” information that is impossible to get otherwise, he says. “A kind of hint of how it was done.”

To this end, Wallez has recently been replicating historical methods of synthesizing lead white. He spent almost a year deciphering recipes found in manuscripts and analyzing old tubes of dried paint found in the basements of historic homes, trying to match the color, texture, and crystalline composition of the base coats Leonardo developed in his workshops (Heritage Sci. 2024, DOI: 10.1186/s40494-023-01082-4). It’s a work in progress. “My white is not very white. It’s very thick; it’s very viscous. So no problem there. I have plumbonacrite. But it’s yellowish,” Wallez says. “I still try. It’s not easy to walk in such footprints. Leonardo and Rembrant, they are great chemists.”

Accelerating art investigations

Still, on the day C&EN visited, Calligaro was firing the particle accelerator at a reproduction paint that Wallez had prepared. Calligaro was using a method that shoots beams of high-energy protons at a pinpoint spot, causing atoms in the crosshairs to emit X-ray fingerprints that reveal their elemental identities. The method, known as particle-induced X-ray emission, generally does not damage inorganic materials such as metals and ceramics.

But Calligaro isn’t sure about paintings. “You have the presence of organic constituents like linseed oil. Or varnish. It can be a lot of things. Egg yolk. It depends,” he says. He has seen the proton beam damage lead white, and he recently observed a strong reaction from barium sulfate. “Why is it so sensitive? I have no idea,” he says.

The accelerator’s upgrades have expanded its capabilities. Better detectors enable researchers to get more data while exposing the artwork to less radiation. Automations and computerized controls make it possible to run the accelerator 24 h a day, more than doubling the time slots available for the in-demand instrument. Wallez says that a new particle source that uses deuterium is in final testing and will be ready to use in the next few months—a development that will add gamma ray experiments to the C2RMF’s repertoire.

The renewed particle accelerator and the suite of associated instruments is turning qualitative observations into quantitative measurements. C2RMF researchers can now collect data with enough detail and precision to unleash quantum mechanical calculations on the study of cultural heritage materials. “You must go one step beyond. You must really measure, quantify things,” Calligaro says. “And this is what we want to do.”