ARTICLE pubs.acs.org/ac
Quantitative NIR Chemical Imaging in Heritage Science Linda Csefalvayova,† Matija Strlic,*,† and Harri Karjalainen‡ †
Centre for Sustainable Heritage, The Bartlett School of Graduate Studies, University College London, Gower Street (Torrington Place Site), London WCIE 6BT, United Kingdom ‡ Specim, Spectral Imaging Ltd., Oulu, Finland ABSTRACT: Until recently, applications of spectral imaging in heritage science mostly focused on qualitative examination of artworks. This is partly due to the complexity of artworks and partly due to the lack of appropriate standard materials. With the recent advance of NIR imaging spectrometers, the interval 10002500 nm became available for exploration, enabling us to extract quantitative chemical information from artworks. In this contribution, the development of 2D NIR quantitative chemical maps of heritage objects is discussed along with presentation of the first quantitative image. Further case studies include semiquantitative mapping of plasticiser distribution in a plastic object and identification of historic plastic materials. In the NIR imaging studies discussed, sets of 256 spatially registered images were collected at different wavelengths in the NIR region of 10002500 nm. The data was analyzed as a spectral cube, both as a stack of wavelength-resolved images and as a series of spectra, one per each sample pixel, using multivariate analysis. This approach is only possible using well-characterized reference sample collections, as quantitative imaging applications need to be developed, thus enabling spatial maps of damaged and degraded areas to be visualized to a level of chemical detail previously not possible. Such quantitative chemical mapping of vulnerable areas of heritage objects is invaluable, as it enables damage to historic objects to be quantitatively visualized.
W
orks of art are among the most complex objects to characterize because of their chemical and spatial inhomogeneity. Production methods, raw materials and additives, treatments, and various past environmental conditions induced complex routes of deterioration. For their successful conservation, it is essential that material properties of heritage objects are understood. However, their nature significantly restricts the possibilities for sampling. It is often difficult to ethically justify the use of destructive analytical techniques, and noninvasive analysis is essential where sampling is not permitted. Even if it is possible, many degraded objects are difficult to analyze without further damage or potential loss because of handling during the process of analysis. In this work, we explore the use of 2D NIR chemical imaging for visualization of damage on two types of most rapidly grading art objects: plastics and inks. Plastic materials represent a significant challenge to conservation as many synthetic materials are inherently unstable in the long term.1 Certain types are particularly susceptible to degradation and deteriorate rapidly requiring specialist care, regular monitoring, or even isolation. Among these are those manufactured from materials exhibiting autocatalytic degradation: cellulose esters, polyurethanes, and hydrocarbon polymers in particular.2 Iron gall ink drawings represent another type of rapidly degrading heritage objects, often also incredibly valuable. Popular with artists, writers and architects, iron-gall ink is abundant in paper-based collections, being the primary writing ink used from the 12th through to the early 20th Century in the West, with great variability in the ink composition. Unfortunately, the ink is infamous for its instability and corrosiveness as it induces r 2011 American Chemical Society
enhanced degradation of the writing or drawing support, which is a consequence of its acidity and content of transition metals.3 Acid-catalyzed hydrolysis of cellulose in paper is a rapid degradation process and depends on pH; the lower it is, the faster the degradation. In addition to acids, iron gall inks contain a substantial amount of transition metal ions promoting autoxidation of organic compounds present in the ink and in the paper. In both cases, being able to quantitatively image degraded areas of heritage objects would be of huge advantage to their successful conservation. During the past decade, spectroscopic imaging systems used in remote sensing, medicine or forensics have advanced dramatically and their use in heritage research46 significantly advanced identification and mapping of the distribution of authentic or added materials, examination of object condition, treatment evaluation, and enhancement of faded and illegible texts.7,8 However, the concept of hyperspectral imaging can be applied to any optical spectroscopic technique. Currently, infrared, Raman, fluorescence, UVvis, nuclear magnetic resonance, and X-ray photoemission spectroscopy imaging instruments are available,9,10 although not all have been explored in heritage science yet. Chemical imaging has been generally, though not exclusively, used to refer to vibrational imaging, which offers high chemical selectivity. The first spatially resolved IR spectra appeared in 1949, describing the use of a microscope coupled with an IR Received: December 26, 2010 Accepted: May 23, 2011 Published: May 23, 2011 5101
dx.doi.org/10.1021/ac200986p | Anal. Chem. 2011, 83, 5101–5106
Analytical Chemistry spectrometer.11 The first qualitative chemical maps collected with an FTIR microscope fitted with a moving stage were published in 1988, demonstrating the concept of spatial localization of chemical species using infrared spectroscopy.12 For some time held back by the lack of sensitive broadband array detectors, NIR imaging spectrometers have now also become commercially available.13 Working in NIR permits the use of refractive optics and the optical design allows users to adjust magnification simply by changing the image formation lens. Adaption to a wide variety of fields-of-view (FOV) and extreme tolerance to variations in sample geometry have only recently been fully exploited,1417 making NIR possibly the most robust imaging technique available. In addition, flatness of the sample is not a prerequisite as is mostly the case in Raman18 and mid-IR19 spectral imaging. With NIR hyperspectral imaging, tens of thousands of NIR spectra are collected in a measurement, each relating to a specific area or a pixel on the sample surface, resulting in spatially resolved information on the nature and quantity of chemical species. The immense quantity of data contained within a single hyperspectral image (3D hypercube) necessitates the use of chemometric data compression techniques to assist with image interpretation. Therefore, multivariate mathematical modeling techniques, such as Principal Component Analysis (PCA), are often applied to expose trends that would be otherwise undetectable.20 However, hyperspectral imaging data are also amenable to the application of conventional multivariate quantitative methods in which individual pixel spectra are analyzed using calibration data sets and techniques such as Partial Least Squares (PLS). To apply PLS directly to a hyperspectral image, individual Y-block (dependent variable) reference values are needed for all pixels of all calibration images.21 If such detailed data is unavailable, global or mean reference values can be used, if each reference value is properly matched with a mean spectrum from a single image or a large region-of-interest (ROI) representing the known sample. The resulting calibration model can then be used to make predictions at individual pixels of test images. The use of such ROI based models is explored in this paper. Hyperspectral imaging has made its way into heritage science in the spectral region
