5 CONCLUSIONS AND FUTURE WORK

In simulated and actual paintings, the ultrasonic system successfully mapped flaws that could not be detected by other techniques such as radiography and infrared thermography. The system detected certain flaws such as splits, checks, delaminations, cleavage, and voids as well as imaging wood grain and knots in the wood and craquelure in the paint layer. Used in conjunction with other techniques, the system can give a more complete understanding of the condition of a painting, for example by locating flaws when they are still subsurface or by detecting blind cleavage.

Delaminations, cleavage, and voids were most effectively shown with the amplitude scans, with the unprocessed amplitude scans showing the degree of delamination or cleavage. The amplitude scans were better at defining splits in radially cut panels, while the phase scans were better at defining splits in tangentially cut panels. It is therefore useful to take both amplitude and phase measurements.

The ultrasonic system offers a new way of examining other objects, for example furniture and paintings on canvas. Its ability to detect splits, checks, delaminations, cleavage, or voids depends upon how well the ultrasonic signal can penetrate the materials and the thickness of a particular material. For some cases, further work is needed in the area of image recognition to improve the reliability and ease of use of the air-coupled ultrasound system. In particular, it is important that methods be developed to distinguish between images of natural anomalies (grain) and flaws.

The safety of using this technique should be addressed. The peak power density within the air is 4,000 W/m2, and therefore the particle displacement in the air is approximately 1 Angstr�m at 50 kHz (Qi and Brereton 1995). The power density and the particle displacement are reduced in the sample by having the transmitting transducer defocused on the back surface of the painting. The signal is further attenuated because of the impedance mismatch between the wood and the air. Moreover, the signal is also reduced before it reaches the painted surface because it enters from the back of the painting and travels through the wood, which is highly attenuating. Considering all these factors, the total power density at the paint layer is expected to be at least 10,000 times less than what is used in an ultrasonic cleaning bath. It is not recommended that the transmitting transducer be placed on the painted surface side, as the incident power densities may approach those of an ultrasonic cleaning bath. There has been no physical evidence to show that the experimental setup described in this paper causes deterioration; before the paintings were examined, repeated investigation of test samples took place. To apply this technique widely, further tests are needed to ensure that no material will be removed upon application of this method.

To take full advantage of air-coupled ultrasound as an examination technique, a useful next step would be the development of a more portable and economical system than the prototype described in this paper. Systems like the one used are currently available in a number of materials science laboratories in government, industrial, and university research institutes. As more research laboratories invest in this technology, it should become more accessible to conservators through collaborative work between research institutes and museums.

ACKNOWLEDGEMENTS

This work was funded by a predoctoral conservation science fellowship from the Conservation Analytical Laboratory at the Smithsonian Institution, in conjunction with the Materials Science and Engineering Department and the Center for Nondestructive Evaluation at the Johns Hopkins University. We would like to thank colleagues at both institutions for their invaluable advice and assistance. We are most grateful to Dr. and Mrs. Hans Goedicke and the National Museum of American Art for lending us the paintings. We would also like to thank Eric S. Boltz, Martin C. Renken, and Johanna Bernstein for their technical expertise and advice, as well as the reviewers for their very useful comments. The substance of this work is based on a paper in Nondestructive Characterization of Materials 6(Murray et al. 1994).