Author: Chase Butler and Yonah Maor

Acoustic emission measurements use microphones, amplifiers, and computers to detect and record the release of elastic waves during stress relaxation processes within materials, such as crack propagation at both the macro and micro scales. This talk discussed how acoustic emission (AE) has been used to track these processes in wooden and stone objects under varying levels of relative humidity and outlined how such studies have been used to generate and validate RH guidelines.

During AE experiments, the microphones are positioned against objects mechanically, without the need for glues or clamps, and environmental noise can be determined when there is sufficient distance between two sensors. It is important to note that recording AE monitors internal stress-relieving processes in real-time but is not able to predict when damage may occur.

The AE from wooden cylinders was found to depend on both the change in RH and the rate at which this change occurred. Mild changes in RH applied over 48 hours, for example, did not lead to detectable acoustic emissions. Monitoring the AE of a wooden altarpiece in a church lead to the establishment of expanded RH guidelines of 35 – 60% RH.

AE studies have also been performed on clay-containing sandstones similar to those found in medieval cathedrals. The studies monitored the sandstone’s response to damaging wetting-drying cycles and detected a linear AE increase with the number of cycles. The results of this study, in conjunction with predicted climate data, were used to anticipate areas of Europe in which clay-containing sandstones may be at particular risk for damage.

Dr. Butler described in this talk the potential use of Fourier transform photoacoustic infrared spectroscopy (FTIR-PAS) to identify organic and inorganic pigments. In their commercially available device, the sample is placed in a gas filled cavity and exposed to periodic flashes of infrared light. The sample absorbs the radiation and heats up, causing the surrounding gas to expand. The expansions can be detected as sound waves by a microphone at the end of the cavity.

The technique has been used to obtain reproducible IR spectra of pure pigments in the range of 400 to 4000 cm^-1 with a resolution of 8 cm^-1. Samples from lab-prepared frescoes have also been analyzed. In some cases the background noise from the plaster overwhelmed the pigment spectra, but in others the spectra could be used for pigment identification. Dr. Butler mentioned the possibility for depth-profiling, which may allow for more complex samples to be analyzed. Such work has not yet been carried out.

This method can also add some useful peaks to a regular FTIR spectrum, making it another useful option in conservation science’s identification tool-kit. Only small samples are needed and minimal preparation is required, an advantage over the commonly used ATR (Attenuated Total Reflectance) technique that may call for the sample to be crushed.

This talk presented case studies to demonstrate the application of computed tomography (CT) scanning to archeological objects. According to Dr. Brown, CT is usually available in most hospitals and can often be used for free in the evening. JP recommended first taking a regular x-ray of the object to elucidate its general construction. He did not go into details about the technical aspects of CT and image processing; however, this topic was covered by Hai-Yen Nguyen’s talk earlier in the session (An Open-Source Workflow for the Visualization of CT Data in Art Conservation and Archaeology).

Resolutions of 0.3-1.0 mm can be obtained by CT. The results are presented in Hounsfield units (HU), a scaled measure of the attenuation of radiation due to the material. Water is defined to have 0 HU, while air is defined to have -1000 HU. Typical HU values for metal and bone are 3000 HU and 1000 HU, respectively. Many image-processing techniques, such as the application of false-color, rely on the different HU values of different materials.

Case 1: Animal mummy

The CT images clearly showed that a piece had been inserted into the mummy, probably to hold the head at a desired angle. The detail obtained of the skeleton allowed for the species to be identified as a type of gazelle common in Egypt.

Case 2: Polychrome Japanese sculpture

The CT showed the grain of the wooden object in great detail, enough that non-invasive dendrochronolgy could be possible. This piece demonstrated the problem of having highly attenuating materials in the object – the bright images caused by leaded glass eyes obscured some of the nearby details. However, false-color rendering was able to show areas of gesso and older conservation treatments on other areas of the sculpture.

Case 3: Moche pottery

One of the aims of this study was to check for the presence of organic residues inside the vessels. This was done by comparing the HU values of the pots to those obtained from samples of various modern food residues. Although the CT of one of the pots did not indicate that food residues were present, the image showed a pattern of holes suggesting that the vessel was designed to produce sound from blowing air. A second pot was simply a conch shell, and the CT confirmed the expected internal structure. Interestingly, material inside a third pot did indeed exhibit attenuation values matching those of the test food samples. The material was subsequently collected on a swab, and SEM images indicated that the organic material was likely charred plant stems.

Case 4: Restored archeological stucco (possibly Sasanian)

The sculpture depicted the head of a king. The crown portion of the object, the shape of which could be used to identify the specific king, had been largely restored. Although it was possible to visualize only the original material by manual segmentation of the restored portions, not enough of the original remained for the identity of the king to be determined.

Several interesting points were raised during the question portion of the talk. It is known that x-rays affect the results of thermoluminescence (TL) dating, though it is not known to what extent. As a precaution, JP recommends removing a sample before CT scanning if TL may be performed later. Another interesting question regarded recommendations for approaching a hospital. JP suggested contacting the chief radiologist first, or ideally, finding a teaching hospital with a research radiologist. He mentioned that eventually an administrative/financial person will also need to be contacted and that having an exciting story about your proposed work will increase the chances that the hospital allows use of their CT machines.

This talk discussed how open-source image-processing software can be used to manipulate data obtained from computed tomography (CT) scans of objects. In this technique, radiographs are taken around an axis of rotation, and the three-dimensional volume of the object is virtually reconstructed. From JP Brown’s talk (Medical Computed X-Ray Tomography and Volumetric Reconstruction for the Technical Examination of Organic/Composite and Ceramic Objects), we learned that it is relatively easy to gain free access to CT equipment at local hospitals. However, as  Hai-Yen pointed out, the proprietary software for processing the data can cost around $15,000. By using several open-source programs, she and her co-workers were able to obtain quality images highlighting various aspects of the object under study.

CT images of a corroded metal artifact in a tub of water were presented at various stages of processing with different software programs. A median filter was applied in ImageJ to reduce noise. Most other processing applications were performed in ImageVis3D, including thresholding, slice analysis, and manual segmentation. The 16-bit raw tiff file formal was considered the most user-friendly for transfer between the software systems.

After the initial data processing, Hai-Yen demonstrated the use of false color to clearly show different types of material and used clipping (masking) to isolate certain features. Once a complete 3D rendering is obtained, it could potentially be used to create a physical model of the artifact without the need to dry out and clean the original.

Disclaimer: Neither of us bloggers has ever done CT or this type of image processing, so we may have missed salient details. Feel free to add information in the comment section below.

Dr. Rui Chen has been developing optical sensors using silver nanoparticles for the detection and quantization of sulfurous gases. These sensors are meant to serve as an improvement over the metal coupons used in the Oddy test, as silver nanoparticles react faster and with higher sensitivity than bulk silver. Additionally, their reaction with hydrogen sulfide causes the nanoparticles to lose their color, so the kinetics of the reaction can be monitored spectrophotometrically with the decrease in the absorbance spectrum.

To make the sensors, spherical yellow nanoparticles are assembled as a monolayer on a glass surface with a polyethylenimine linker. When the monolayer films are exposed to hydrogen sulfide gas, the observed decrease in absorbance is faster when the concentration of hydrogen sulfide is higher. At concentrations of 10 ppm hydrogen sulfide, for example, the reaction is complete within about an hour, while completion occurs within 4 to 6 minutes at concentrations of 100 ppm.

Furthermore, Dr. Chen has found a linear relationship between the first-order rate constant for the reaction and the concentration of the gas. Thus, an unknown concentration of hydrogen sulfide gas can be calculated based on the reaction rate observed between the gas and the nanoparticle-based sensor.

The sensors have been applied to wool samples aged under UVB light. These samples exhibited the highest emissions within the first 400 hours of aging, with concentrations of hydrogen sulfide reaching nearly 600 ppb per gram of wool. The sensors have also been applied to photograph cases and a box containing naturally aged silk fibers; four of the five objects tested exhibited some emission of hydrogen sulfide to give concentrations that ranged from 33 ppb to about 150 ppb. Future work on this project will involve the development of a user-friendly protocol for the application of the sensors.

The ability of conservators to accurately describe surfaces is integral to effective documentation and communication in the conservation field. While changes in color can be accurately described by the Munsell and CIELAB systems, changes in the appearance of a surface texture are often described by vague and indefinite terms. The work described by Dr. Whitmore arose from the desire for a new vocabulary for describing the surface texture and appearance of regularly-patterned surfaces such as canvas. His talk presented a background of how we perceive surface texture and some mathematical analyses that can be performed to yield qualitative metrics for describing certain textures.

Texture is not only an artifact of the topography of a surface, but also a result of the illumination of this topography; the appearance of a surface depends on the position and directionality of the light source as well as the distance of the light from the surface. As viewers, we perceive texture from the resulting distribution and contrast of light and dark areas.

The limitations of our perception can be mapped by a contrast sensitivity function, which relates our ability to perceive patterns of varying degrees of contrast with the size of the pattern and the distance at which it is viewed. Changes to texture are often accompanied by changes in contrast; a weathered surface, for example, exhibits decreased contrast between its light and dark areas. In periodic patterns, the contrast can be quantitatively measured by a gray-level correlation matrix analysis. Once this contrast value has been determined, the contrast sensitivity function can be used to calculate the maximum viewing distance at which a pattern of the given contrast can be perceived. This “maximum visibility distance” may serve as a useful metric with which to describe periodic surface textures.

The analysis has been successfully applied to canvases that had been subjected to extreme treatment, and further work will examine the degree of textural changes caused by common conservation treatments. Future work may also investigate other mathematical and image processing analyses for application to different surface textures.