Testing the sample and Raman fingerprints
The first stage of the June 3, 2010 test of David Robertson’s Autumn Harbour, required taking a tiny sample from the painting,. These samples are usually a fleck of a few microns (millionths of an inch), which can then be tested to ensure the lasers are set at the correct level assuring no danger of damage to the painting. (Robertson extracted the sample himself prior to arriving at the conference. He did not allow anyone else to touch his purported Harris before, during, or after the test.)
Here’s how ProSpect* Scientific describes the ‘rehearsal’ test on the paint chip,
Tests on this chip were done simply to ensure we knew what power levels were safe for use on the painting. While David R stated he believed the painting was oil on canvas without lacquer, we were not entirely certain of that. Lacquer tends to be easier to burn than oil pigments and so we wanted to work with this chip just to be entirely certain there was no risk to the painting itself.
The preliminary (rehearsal) test resulted in a line graph that showed the frequencies of the various pigments in the test sample. Titanium dioxide, for example, was detected and its frequency (spectra) reflected on the graph.
I found this example of a line graph representing the spectra (fingerprint) for a molecule of an ultramarine (blue) pigment along with a general explanation of a Raman ‘fingerprint’. There is no indication as to where the ultramarine pigment was obtained. From the WebExhibits.org website featuring a section on Pigments through the Ages and a webpage on Spectroscopy,
Raman spectra consist of sharp bands whose position and height are characteristic of the specific molecule in the sample. Each line of the spectrum corresponds to a specific vibrational mode of the chemical bonds in the molecule. Since each type of molecule has its own Raman spectrum, this can be used to characterize molecular structure and identify chemical compounds.
Most people don’t realize that the chemical signature (spectra) for pigment can change over time with new pigments being introduced. Finding a pigment that was on the market from 1970 onwards in a painting by Jackson Pollock who died in 1956 suggests strongly that the painting couldn’t have come from Pollock’s hand. (See Michael Shnayerson’s May 2012 article, A Question of Provenance, in Vanity Fair for more about the Pollock painting. The article details the fall of a fabled New York art gallery that had been in business prior to the US Civil War.)
The ability to identify a pigment’s molecular fingerprint means that an examination by Raman spectroscopy can be part of an authentication, a restoration, or a conservation process. Here is how a representative from ProSpect Scientific describes the process,
Raman spectroscopy is non-destructive (when conducted at the proper power levels) and identifies the molecular components in the pigments, allowing characterization of the pigments for proper restoration or validation by comparison with other pigments of the same place/time. It is valuable to art institutions and conservators because it can do this. In most cases of authentication Raman spectroscopy is one of many tools used and not the first in line. A painting would be first viewed by art experts for technique, format etc, then most often analysed with IR or X-Ray, then perhaps Raman spectroscopy. It is impossible to use Raman spectroscopy to prove authenticity as paint pigments are usually not unique to any particular painter. Most often Raman spectroscopy is used by conservators to determine proper pigments for appropriate restoration. Sometimes Raman will tell us that the pigment isn’t from the time/era the painting is purported to be from (anachronisms).
Autumn Harbour test
Getting back to the June 3, 2014 tests, once the levels were set then it was time to examine Autumn Harbour itself to determine the spectra for the various pigments. ProSpect Scientific has provided an explanation of the process,
This spectrometer was equipped with an extension that allowed delivery of the laser and collection of the scattered light at a point other than directly under the microscope. We could also have used a flexible fibre optic probe for this, but this device is slightly more efficient. This allowed us to position the delivery/collection point for the light just above the painting at the spot we wished to test. For this test, we don’t sweep across the surface, we test a small pinpoint that we feel is a pigment of the target colour.
We only use one laser at a time. The system is built so we can easily select one laser or another, depending on what we wish to look at. Some researchers have 3 or 4 lasers in their system because different lasers provide a better/worse raman spectrum depending on the nature of the sample. In this case we principally used the 785nm laser as it is better for samples that exhibit fluorescence at visible wavelengths. 532nm is a visible wavelength. For samples that didn’t produce good signal we tried the 532nm laser as it produces better signal to noise than 785nm, generally speaking. I believe the usable results in our case were obtained with the 785nm laser.
The graphed Raman spectra shows peaks for the frequency of scattered light that we collect from the laser-illuminated sample (when shining a laser on a sample the vast majority of light is scattered in the same frequency of the laser, but a very small amount is scattered at different frequencies unique to the molecules in the sample). Those frequencies correspond to and identify molecules in the sample. We use a database (on the computer attached to the spectrometer) to pattern match the spectra to identify the constituents.
One would have thought ‘game over’ at this point. According to some informal sources, Canada has a very small (almost nonexistent) data bank of information about pigments used in its important paintings. For example, the federal government’s Canadian Conservation Institute (CCI) has a very small database of pigments and nothing from Lawren Harris paintings [See the CCI’s response in this addendum], so the chances that David Robertson would have been able to find a record of pigments used by Lawren Harris roughly in the same time period that Autumn Harbour seems to have been painted are not good.
In a stunning turn of events and despite the lack of enthusiasm from Vancouver Art Gallery (VAG) curator, Ian Thom, on Wednesday, June 4, 2014 the owner of the authenticated Harris, Hurdy Gurdy, relented and brought the painting in for tests.
Here’s what the folks from ProSpect Scientific had to say about the comparison,
Many pigments were evaluated. Good spectra were obtained for blue and white. The blue pigment matched on both paintings, the white didn’t match. We didn’t get useful Raman spectra from other pigments. We had limited time, with more time we might fine tune and get more data.
One might be tempted to say that the results were 50/50 with one matching and the other not, The response from the representative of ProSpect Scientific is more measured,
We noted that the mineral used in the pigment was the same. Beyond that is interpretation: Richard offered the view that lapis-lazuli was a typical and characteristic component for blue in that time period (early 1900’s). We saw different molecules in the whites used in the two paintings, and Richard offered that both were characteristic of the early 1900’s.
* ‘ProsPect’ changed to ‘ProSpect’ on June 30, 2014.
ETA July 14, 2014 at 1300 hours PDT: There is now an addendum to this series, which features a reply from the Canadian Conservation Institute to a query about art pigments used by Canadian artists and access to a database of information about them.