Uncovering the pigments and techniques used to paint the Berlin Wall

“This study highlights the powerful impact of the synergy between chemistry and deep learning in quantifying matter, in this case the pigments so fascinating for street art,” said study co-author Francesco Armetta.

In order to restore or conserve a work of art, it is important to gather information about materials and application techniques. But the painters of the Berlin Wall did not record this. In previous studies of other historical artifacts, scientists brought fragments and even whole objects into the lab and used Raman spectroscopy to identify the pigments on them without destroying the samples. Although handheld Raman devices can be used for field investigations, they lack the precision of full-scale laboratory equipment. So Armetta, Rosina Celeste Ponterio and colleagues wanted to develop an artificial intelligence algorithm that could analyze the output of a portable Raman device to more accurately identify pigments and dyes. In a preliminary test of the new method, they analyzed 15 paint chips from the Berlin Wall.

The researchers first zoomed in on the fragments and observed that they all had two or three layers of paint, with visible brush strokes. The third layer in contact with the masonry was white, which they believed was a primer used to prepare the walls for painting. Next, the researchers analyzed the wafers using a handheld Raman spectrometer and compared them to spectra collected from a commercial pigment spectral library. They determined that the main pigments in the sample were: azo pigments (yellow and red fragments), phthalocyanines (blue and green fragments), lead chromate (green fragments), and titanium white (white fragments). These results were confirmed by other non-destructive techniques, including X-ray fluorescence and fiber optic reflectance spectroscopy.

The researchers then mixed pigments from a commercial acrylic paint brand (used in Germany since the 1800s) with varying proportions of titanium white, trying to match the painter’s typical range of colors and tones. Researchers say knowing these proportions can help art conservators prepare suitable restoration materials. They used handheld Raman spectroscopy data of the mixture to train a machine learning algorithm to determine the percentage of pigments. The method showed that Berlin Wall paint chips contained titanium white and up to 75% pigment, depending on the chip analyzed and based on the hue. The researchers say these results show that their artificial intelligence model can provide high-quality information for art conservation, forensics and materials science in environments where it is difficult to bring laboratory equipment to the field.