Institute of Natural Sciences and Technology in the Arts (INTK), Academy of Fine Arts Vienna, Schillerplatz 3, 1010, Vienna, Austria.
Institute of Chemical Technologies and Analytics, Vienna University of Technology, Getreidemarkt 9/164, 1060, Vienna, Austria.
Anal Bioanal Chem. 2020 May;412(13):3187-3198. doi: 10.1007/s00216-020-02574-z. Epub 2020 Mar 14.
The ever-increasing speed of exchange of ideas, information, and culture allows contemporary art to be in constant growth, especially concerning the choice of artistic materials. Their characterization is not only crucial for the study of artistic techniques but also for research into the stability of the material and, consequently, the best preservation practices. For this aim, an analytical method should have the advantages of not requiring sample preparation, performing superficial micro-analysis, and obtaining detailed spectral information. For this study, laser-induced breakdown spectroscopy (LIBS) was employed. It was used for the identification of modern paints composed of inorganic pigments and organic binders, such as acrylics, alkyds, and styrene-acrylics. Principal component analysis (PCA) was used to classify the different pure materials, above all, the polymeric binders. To distinguish the paint mixtures, whose LIBS spectral results were more complex due to the pigment/binder interaction, a statistical method recently employed in the cultural heritage field was chosen, namely, random decision forest (RDF). This methodology allows a reduction of the variance of the data, testing of different training data sets by cross-validation, an increase of the predictive power. Furthermore, for the first time, the distribution of different inorganic pigments and organic binder materials in an unknown sample was mapped and correctly classified using the developed RDF. This study represents the first approach for the classification of modern and contemporary materials using LIBS combined with two different multivariate analyses. Subsequent optimization of measurement parameters and data processing will be considered in order to extend its employment to other artistic materials and conservation treatments.
思想、信息和文化的交流速度不断加快，使当代艺术不断发展，尤其是在艺术材料的选择方面。这些材料的特性不仅对艺术技术的研究至关重要，而且对材料稳定性的研究以及因此对最佳保存实践也至关重要。为此，分析方法应该具有无需样品制备、进行表面微分析和获得详细光谱信息的优点。为此，本研究采用了激光诱导击穿光谱（LIBS）。它用于鉴定由无机颜料和有机粘合剂组成的现代涂料，例如丙烯酸盐、醇酸树脂和苯乙烯-丙烯酸盐。主成分分析（PCA）用于分类不同的纯材料，尤其是聚合粘合剂。为了区分涂料混合物，由于颜料/粘合剂的相互作用，LIBS 光谱结果更加复杂，因此选择了最近在文化遗产领域中使用的统计方法，即随机决策森林（RDF）。该方法可以减少数据的方差，通过交叉验证测试不同的训练数据集，从而提高预测能力。此外，首次使用开发的 RDF 对未知样品中的不同无机颜料和有机粘合剂材料的分布进行了映射和正确分类。本研究代表了首次使用 LIBS 结合两种不同的多元分析方法对现代和当代材料进行分类的方法。为了将其应用于其他艺术材料和保护处理，将考虑进一步优化测量参数和数据处理。
