2 EXPERIMENTAL

2.1 Instrumentation

SPECTRAL ANALYSES WERE MADE on a Digilab 15-E Fourier Transform Infrared (FT-IR) spectrometer equipped with a Motorola 3200 computer. The spectrometer was purged with dry clean, CO2-free air from a Balston 75-60 filtration unit. A Mercury-Cadmium-Telluride (MCT) detector, cryogenically cooled, was used to examine the region from 4000-500 cm−1 (2.5–20 μm). Each spectrum is the sum of two hundred scans collected at a resolution of 4 cm−1.

2.2 Samples

Reference resins were obtained from A.F. Suter & Co and the conservation laboratories at the J. Paul Getty Museum. Homogeneous samples were obtained by crushing and thoroughly mixing a 2–3 gram portion, which was then stored in a glass vial prior to analysis.

Naturally aged resin samples were obtained from the Harvard Art Museums' collection of Gettens and Stout materials. The resins were films (<20 μm thick) on glass plates. The dates on these plates were from 1933–1937. It is assumed that the plates, which were labeled as controls, were stored in a drawer during the last 5 decades. A 50 mm2 section of each resin was removed from the plate with a scalpel and placed in a clean glass vial with a Teflon™-lined cap.

The unknown sample set was composed of finishes on furniture pieces from the J. Paul Getty Museum. Each piece was of 18th century European manufacture and was typically of veneer on oak. Samples were arbitrarily chosen from obscure areas on the furniture, near or under mounts, in the rear or on the legs. A scalpel was used to scrape barely visible amounts of sample (<100 μg) from clean, surface layers of the finish. This minimized the chance of contamination from dirt and wood particles which can interfere with the interpretation of the resin spectra. In a multilayered finish, this sampling method selectively removes the more recent coatings. In these cases, more than one sample may be scraped from different depths.

2.3 Sample Preparation

Samples were prepared as KBr pellets in either the standard 13 mm size or 1.5 mm micropellet size. Approximately 1 mg of a crushed bulk resin sample was used for the 13 mm KBr pellets. These standard sized pellets were used in the tests on variability due to sample preparation, resin supplier and resin refinement. Additionally, the 13 mm pellets were used for the resin mixtures; each mixture had a total weight of 1 mg with the individual components weighing from 0.01 mg to 1.20 mg. The 1.5 mm micropellets were prepared from approximately 30 μg of sample in a glove bag under a dry nitrogen purge. The micropellets were mounted in a Harrick Scientific 4� beam condenser for infrared analysis. The natural and artificially aged samples, the solvent dependency study and the furniture finish samples were all analyzed by this micropellet technique. Because the sample sizes were sufficient and the samples were prepared in the purged glove bag, none of the possible problems with KBr pellets (external contamination, water absorption, sample reactions) applied to this study.

For the solvent dependency study, 5 wt/v% solutions were prepared with sandarac, mastic and rosin in ethanol, turpentine and mineral spirit solvents and with shellac and copal in ethanol. After 48 hours, the solutions were filtered through one layer of cheesecloth, then 20–30 ml of each solution was cast onto clean thin sheets of aluminum using 6“�6” Teflon™ molds (Hansen and Taketomo, 1988). The films were air-dried for approximately one week, then analyzed.

For the accelerated deterioration studies, a thin layer of four resins in ethanol (mastic-20 wt/v%, rosin-20 wt/v%, sandarac-33 wt/v% and shellac-33 wt/v%) were painted with an ethanol-cleaned paint brush onto clean, polished-surface aluminum plates. These samples were allowed to dry overnight in a fume hood, then placed in a Weather-Ometer with controlled light, heat and humidity conditions. The samples were irradiated with a xenon lamp set at 0.35 W/m2 (340 nm) with inner and outer borosilicate filters simulating outdoor sunlight (Brennan, 1987). The black panel temperature was 53�C and the relative humidity was 50%. The total exposure for the samples ranged from 0 to 151 kJ (151 kJ is equivalent to 120 hours exposure at an irradiance of 0.35 W/m2 at 340 nm). The resin controls were kept in the dark for an equivalent time period.

2.4 Computer Methods

For each sample, the computer printed out band position values for peak maxima. In regions where peaks are overlapped, a deconvolution program was used to elucidate the individual features. The program operates on two or more overlapping bands to reduce the line width of the individual components and therefore improve the resolution of each line.

Spectral subtraction methods were used to separate components of a mixture. Intense bands were used to identify major components whose pure spectra were then multiplied by a scaling factor and subtracted out. After spectrally removing the bands corresponding to the major component, a scale expansion was used to enhance the remaining bands and identify the minor components. Following the same general scheme, computer addition of individual spectra was used to simulate spectra of mixtures, such as varnish recipes, which were then stored in a search library for access by the computer.