A standard method in many conservation science labs is the use of Meth-Prep II (a methanolic solution of trifluoromethylphenyl trimethyl ammonium hydroxide) for the transesterification of triglycerides to methyl esters and esterification of free fatty acids (FAs). However, this method gives you the total fatty acid content of the sample, including free fatty acids, their oxidation products, fatty acids bound in metal soaps, and intact triglycerides. If you want to assess the extent of hydrolysis of an oil paint film, which might be related to water sensitivity, you need to use another method. What Joy has chosen to do is to modify the Meth-Prep II solution to prevent transesterification. She takes 400 μL of Meth-Prep II, evaporates it at 50 oC under nitrogen to remove the methanol, rinses it with toluene, and then adds 500 μL of toluene and 100 μL of t-butanol. T-butanol does not permit transesterification, but the methylation of free alcohol groups can still occur. So a sample can be analyzed with this modified solution- only the free fatty acids and fatty acids in metal soaps (non-glycerides or NG) will be methylated, and thus only they will quantified by GC-MS. Then the same sample can be re-treated with the normal Meth-Prep II solution to determine the fatty acids in triglycerides.
As with any quantitative GC-MS method there are potential issues. The sample needs to be well homogenized to ensure extraction and reaction. The efficacy in methylation of fatty acids soaps is not known, and so those species may not react fully and be underrepresented in the chromatograms. She also sees partial evaporation of her internal reference when the sample is retreated with unmodified Meth-Prep II and so that must be accounted for.
With all of these caveats in mind, they began to look at reference paints, including Bellini tube paints and handmade paints in cold-pressed linseed oil. For the new handmade paints the fatty acid profiles of the NG and the total FAs look similar. But pigments do make a difference- for yellow ochre and red iron oxide paints, the NG accounted for 50-57% of the total FAs, but for malachite and lead white paints the NG fraction was much lower, approximately 12%. This might be due to the formation of metal soaps and poor extraction of fatty acids from those complexes.
For water sensitive Winsor and Newton oils she found 17-37% NGs, although the high levels of palmitate present in cadmium paints may indicate the addition of palmitate soaps. However, there was nothing obvious about the fatty acid profiles that seemed to relate to the water sensitivity. Likewise, a water sensitive cadmium paint in a 1961 Jack Youngerman painting, while also having elevated palmitate levels suggesting palmitate soap addition, had nothing else unusual about the binder. To address this issue cadmium yellow and a chrome yellow paint were mixed with linseed oil, applied onto canvas and glass and artificially aged. The resulting cadmium yellow paint film was very water sensitive, while the chrome yellow film was not. However, there was no major difference between the NG ratios of the two films, and so something else must be rendering the paint sensitive. Furthermore, that sensitivity is not related to the palmitate soaps found in the commercial paints. So there’s more work to be done to determine what is going on, but in the meantime beware cadmium paints!
She had the opportunity to analyze a sample from MOMA’s Water Lilies by Monet, which has a very water sensitive surface. The total FA content of the samples were very low, indicating a very medium poor paint, consistent with Monet’s working method of removing media from his paints. The NG content depended strongly on the sample location and pigments present. The pink-orange paint contained no stearate or palmitate, while other samples contained very high palmitate in the NG fraction. Again, the origin of these differences cannot be explained at this time- but might be due to soap formation and incomplete methylation.
Additional objects tested with this methodology included a variety of Italian paintings ranging from 1595-1758 in date, a Byzantine painting, a Fayum portrait, and a Jackson Pollock painting. The Italian paintings had higher NG fractions than was observed for the tube paints and modern paintings, 22-66% of the total fatty acids, but the NG fraction did not correlate with the age of the material- so extent of hydrolysis cannot be used to determine the age of an object. One possible explanation for this is that paintings this old have likely been cleaned several times and this may have impacted the amount of free fatty acids. The Byzantine painting had a higher NG content of 40-90% and the high palmitate and stearate amounts suggested the formation of metal soaps. Joy found that the modified Meth-Prep method simplified the analysis of beeswax and resulted in less overlapped and confusing chromatograms. In 19th century beeswax she found palmitate and stearate, but in a sample of unpigmented wax from a Roman tomb, she detected neither of those species, implying that they had all evaporated with time. However, in the pigmented Fayum portrait both species were seen, which might be due to the formation of metal soaps, which would help preserve those species, or to the modification of the wax- a contentious subject in the art history field. The white sample from the Pollock painting contained no fatty acids at all, so they evaporated the Meth-Prep solution and reanalyzed it using a method developed for proteins and determined that it was actually a casein paint.
So this seems to be a promising method, and one that could fit easily into a laboratories work flow. Obviously there is more work to be done- particularly on determining efficacy of extraction of fatty acids from soaps and trying to determine how certain pigments create water sensitive films without altering the fatty acid ratios…
Skip to content
AIC Blog Archives: Conservators Converse
Former Blog of the American Institute for Conservation