Portable NIR Applied to the Evaluation of Resin, Glue and Agglutinant used in Canvas Paintings
Melquiades, F. L.; Molari, R.; Appoloni, C. R.
DOI 10.5433/1679-0375.2025.v46.52197
Citation Semin., Ciênc. Exatas Tecnol. 2025, v. 46: e52197
Received: January 14, 2025 Received in revised for: April 30, 2025 Accepted: June 30, 2025 Available online: July 25, 2025
Abstract:
The aim of this work was to study various compounds commonly used in painting canvases by employing portable near-infrared spectroscopy (portable NIR) and to discriminate the compounds through exploratory statistical analysis of the spectra data. To this end, a painting specifically created for archaeometric analysis was examined. Two portable spectrometers with complementary wavelength ranges were combined for the measurements, one operating from 900 to 1700 nm (11100 - 5880 cm-1) and the other from 1350 to 2150 nm (7400 - 4650 cm-1). Portable NIR spectrometry identified characteristic spectral bands for each class of compound: waxes, binders, organic resins, and acrylic resins. Applying exploratory statistics via principal component analysis (PCA) allowed for material differentiation. This study demonstrated that combining NIR spectroscopy and PCA is a versatile tool for achieving consistent results in this application.
Keywords: archaeometry, canvas painting, NIR spectroscopy, exploratory analysis
Introduction
A canvas painting has different layers depending on the technique and particular effects the artist intends to impress. In a simplified description, a canvas structure is composed of a support (wood, fabric canvas), a preparation layer (glue, gypsum, etc), a pictorial layer (paint layers) and a coating layer (glaze and varnish). An artist employs different ingredients in each layer to produce a work of art. Considering that, the focus of this paper is to provide a technical contribution in the evaluation of some components from products used in the preparation layer and coating layer such as animal glues, wax, organic resins, acrylic resins, and agglutinants (Cabral, 1995; Colombini et al., 2022).
Conservation and restoration of a canvas require knowledge about the materials employed by the artist in its production. The raw materials, agglutinants, binders and pigments characterizations are of fundamental importance. Archaeometric methods based on spectroscopy techniques are non-destructive and recommended to be part of an initial canvas examination prior to micro-sample extraction. There are some options with X-ray, ultraviolet, visible and infrared radiation depending on the analyst’s interest or on the questions to be solved.
To evaluate binders based on organic resins (alkyd and vinyl), acrylic resins, lipids and proteins, the recommended fingerprint method is the near and mid-infrared spectroscopy. The attractive features of near-infrared spectroscopy (NIR) in reflectance mode such as rapidity, portability, non-invasive and contactless analysis, and cost-effective method have increased the interest towards its application in archaeometry studies . Molecular and structural information about the sample surface is obtained from electronic and vibrational transitions detected in the analysis. NIR spectrometry comprises the range from 900 to 2500 nm (11100 to 4000 cm-1). In this range, the spectrometer outputs are, in general, spectral bands from NH, CH, OH and CO combination and their overtone modes (Weyer, 1985; Dooley et al., 2017), which are related to proteins, lipids, organic resins and acrylic resins (Brunetti et al., 2016). The challenge is to interpret the combination of bands that characterize a family of binders or a specific compound.
Therefore, to contribute to the understanding and the characterization of binders applied in canvas painting, the aim of this work was to study, technically, different compounds commonly used in canvas painting using portable near-infrared spectroscopy (portable NIR) and, through the spectra, to discriminate them using exploratory statistical analysis. The measurements with portable devices were performed on several known products applied over a mimetic canvas prepared specifically for archaeometric research purposes.
Material and methods
In this study, a mockup painting, illustrated in Figure 1, prepared specifically for archaeometric studies was evaluated. The binder and pigments employed in the mockup canvas were commercial ingredients with known provenance, and some of them had known composition. It was previously evaluated by X-ray spectrometric methods for inorganic characterization of the pigments (Appoloni et al., 2023). The mockup had strips on the borders with individual materials, metal foils, antique pigments and modern pigments. The central area contained several combinations of the individual compounds used in the rectangles on the strips. The mockup was not subject to any type of aging or treatment. The strips on the left and upper sides of the mockup painting had no underlayer preparation and were divided into 44 rectangles. In the present study, the strip on the left side, comprising the chips from 1 to 19 and 24, were evaluated. The chips had 1 cm \(\times\) 2 cm dimensions approximately. Each one of them had solely one material applied directly on the canvas without any kind of preparation. They were classified as agglutinant, wax, organic resin and polymeric resin, as described in Table 1.
| No. | Binder | Binder Type | Chips | No. | Binder | Binder Type | Chips | |
| 1 | rabbit skin glue | agglutinant | 11 | methacrylate | acrylic resin | |||
| 2 | bone glue | agglutinant | 12 | acrylic resin | acrylic resin | |||
| 3 | gum arabic crystals | agglutinant | 13 | vinyl acetate | acrylic resin | |||
| 4 | microcrystalline wax | wax | 14 | polyamide | acrylic resin | |||
| 5 | beeswax | wax | 15 | damar resin | organic resin | |||
| 6 | carnauba wax | wax | 16 | casein | agglutinant | |||
| 7 | copal resin | organic resin | 17 | egg white | agglutinant | |||
| 8 | mastique resin | organic resin | 18 | egg yolk | agglutinant | |||
| 9 | lacca gum | agglutinant | 19 | linseed oil | agglutinant | |||
| 10 | polyvinyl acetate | acrylic resin | 24 | rosin resin | organic resin |
Two portable spectrometers with complementary wavelength ranges were combined for the measurements: the NIR-S-G1 (900 – 1700 nm or 11100 – 5880 cm\(^{-1}\)) and the NIR-M-R11 (1350 – 2150 nm or 7400 – 4650 cm\(^{-1}\)) (InnoSpectra Co., Hsinchu, Taiwan), both with a DLP micromirror array, a single InGaAs detector, with 12 nm resolution. The measurements were performed directly on the chips containing the different compounds. The spectrometers have a protective glass window and each one of them was positioned over each rectangle at 1 mm distance. The spectrometers were configured to register 30 scans in 20 s in each round and the average spectrum is registered. All the points were measured in duplicate and an average spectrum was computed for analysis.
The acquired raw spectra were used in the exploratory analysis, which was performed using principal component analysis (PCA). Prior to the modeling, the following preprocessing sequences were applied: Multiplicative scatter correction + First derivative Savitz-Golay with second order polynomial and 15 window channels + Mean Center.
Results and discussion
Spectra interpretation
As mentioned, the objective of this study is to contribute to a technical evaluation of the binders commonly applied in canvas painting evaluating known ingredients using a portable NIR spectrometer. Some characteristic bands appeared in all the chips.
The bands from O-H bonding related to water are well defined around 1430 nm (6993 cm\(^{-1}\)), 1480 nm (6757 cm\(^{-1}\)) and 1930 nm (5181 cm\(^{-1}\)) (Panero et al., 2018; Weyer, 1985), as observed in the spectra of Figure 2. In addition, the absorption band at 930 nm (10753 cm\(^{-1}\)) could be associated with the C-O band of lipids/oil (Tsenkova et al., 1999).
The wax samples were characterized through the characteristic absorptions of CH\(_2\) and CH combination bands and overtones at 1210 nm (8265 cm\(^{-1}\)), 1730 nm (5780 cm\(^{-1}\)) and 1760 nm (5682 cm\(^{-1}\)) (Ciofini et al., 2016; Furukawa et al., 2002; Vagnini et al., 2009; Longoni et al., 2022) as illustrated in Figure 2. The band at 2140 nm (4673 cm\(^{-1}\)) may be attributed to C-H deformation or \(\nu\)(C-H) and \(\nu\)(C=O) combination (Panero et al., 2018). Furthermore, the band at about 1180 nm (8475 cm\(^{-1}\)) may be related to C-O (Invernizzi et al., 2018), and around at 1400 nm (7143 cm\(^{-1}\)) there is a band from the O-H first overtone.
Slight differences in the agglutinant samples are noted from \(1650\) to \(1800\) nm (\(6060\) to \(5555\) cm-1) as presented in Figure 3. Casein presents a characteristic band at \(1680\) nm (\(5917\) cm-1). Arabic gum and linseed oil have a band at \(1760\) nm (\(5682\) cm-1) from CH\(_2\) first overtone. Lacca gum and egg yolk have bands at \(1700\) nm (\(5930\) cm-1) and \(1730\) nm (\(5780\) cm-1) related to the 1st overtone CH\(_2\) (Furukawa et al., 2002; Vagnini et al., 2009). The band at about \(1180\) nm can also be related to C–O. At \(1210\) nm (\(8265\) cm-1) a band is observed, referring to the 2nd harmonic of \(\nu(\text{C--H})\). The band at \(2050\) nm (\(4878\) cm-1) may be ascribed to \(\nu(\text{NH}) + \delta(\text{NH})\) (Vagnini et al., 2009), and the band around \(2140\) nm (\(4673\) cm-1) may characterize the C–H deformation or \(\nu(\text{C--H})\) and \(\nu(\text{C{=}O})\) combination (Panero et al., 2018).
Organic resin presented bands at \(1180\) nm (\(8475\) cm-1) due to the second overtone of CH\(_2\), see Figure 4. There is an indication of a band at \(1700\) nm (\(5882\) cm-1) related to the first CH\(_2\) overtone. The bands at \(1730\) nm (\(5780\) cm-1) and \(2140\) nm (\(4673\) cm-1) are due to C–O and CH\(_2\) bonds (Longoni et al., 2022). The mastic resin was the only one with a pronounced band at \(1630\) nm (\(6135\) cm-1) due to the CH\(_2\) first overtone (Vagnini et al., 2009). In addition, the band at \(1760\) nm (\(5682\) cm-1) is also related to the CH\(_2\) first overtone.
Acrylic resin set has \(5\) different samples as illustrated in Figure 5. Polyamide and vinyl acetate are characterized by \(1210\), \(1730\) and \(1760\) nm (\(8265\), \(5780\), \(5682\) cm-1) bands corresponding to CH\(_2\) absorption and CH combination bands and overtones (Panero et al., 2018). However, polyamide has an extra band at \(2050\) nm which corresponds to one of the NH bands from amide (Unger et al., 2011) and clearly differentiates this sample from the others. Vinyl acetate has the highest bands at \(1730\) and \(1760\) nm (\(5780\) and \(5682\) cm-1). The bands around \(1550\) and \(1570\) nm (\(6452\) and \(6370\) cm-1) may be related to the vibrational N–H bands, and around \(2140\) nm (\(4673\) cm-1), the observed band may be characterized by the C–H deformation or \(\nu(\text{C--H})\) and \(\nu(\text{C{=}O})\) combination (Panero et al., 2018). Around \(1400\) nm (\(7143\) cm-1) there is a band from the O–H first overtone (Tsenkova et al., 1999). Finally, according to (Rufino & Monteiro, 2003), the band close to \(1680\) nm (\(5952\) cm-1) may be related to the C–H/CH\(_2\), CH\(_3\) stretch.
In general terms, the spectrometers employed were able to identify the different groups of elements. Table [tab:2] presents a synthesis of the bands identified by the portable instrument that characterize each chip ingredient. Some important differences can be noticed. For instance, the \(2050\) nm band is present only in the spectra of rabbit skin and bone glues, casein, egg yolk, and polyamide; the \(1585\) nm band is present only in the spectra of Copal, Damar, mastic, and Rosin resins; the \(1630\) nm band is present only in the spectrum of mastic resin; the \(1550\) and \(1570\) nm bands are present only in the spectrum of polyamide; the \(1400\) nm band is present only in the spectra of the three waxes and polyamide.
| Material | Bands (nm) | ||||||||||||
| 1180 | 1210 | 1400 | 1550 | 1570 | 1585 | 1630 | 1680 | 1700 | 1730 | 1760 | 2050 | 2090 | |
| Microcrystalline wax | x | x | x | -- | -- | -- | -- | -- | -- | x | x | -- | x |
| Bees wax | x | x | x | -- | -- | -- | -- | -- | -- | x | x | -- | x |
| Carnauba wax | x | x | x | -- | -- | -- | -- | -- | -- | x | x | -- | x |
| Rabbit skin glue | x | x | -- | -- | -- | -- | -- | -- | -- | x | -- | x | x |
| Bone glue | -- | x | -- | -- | -- | -- | -- | -- | -- | -- | -- | x | x |
| Gum arabic crystals | -- | x | -- | -- | -- | -- | -- | -- | x | -- | x | -- | x |
| Lacca gum | -- | x | -- | -- | -- | -- | -- | -- | x | x | -- | -- | x |
| Casein | x | x | -- | -- | -- | -- | -- | x | -- | -- | -- | x | -- |
| Egg white | -- | x | -- | -- | -- | -- | -- | -- | x | -- | -- | -- | x |
| Egg yolk | x | x | -- | -- | -- | -- | -- | -- | x | x | -- | x | -- |
| Linseed oil | -- | x | -- | -- | -- | -- | -- | -- | x | -- | x | -- | x |
| Copal resin | x | -- | -- | -- | -- | x | -- | -- | x | x | x | -- | x |
| Mastic resin | x | -- | -- | -- | -- | x | x | -- | x | x | x | -- | x |
| Damar resin | x | -- | -- | -- | -- | x | -- | -- | x | x | x | -- | x |
| Rosin resin | x | -- | -- | -- | -- | x | -- | -- | x | x | x | -- | x |
| Polyvinyl acetate | x | -- | -- | -- | -- | -- | -- | x | -- | x | -- | -- | x |
| Methacrylate | x | -- | -- | -- | -- | -- | -- | x | -- | x | -- | -- | -- |
| Acrylic resin | x | -- | -- | -- | -- | -- | -- | x | -- | x | -- | -- | x |
| Vinyl acetate | x | x | -- | -- | -- | -- | -- | -- | -- | x | x | -- | x |
| Polyamide | -- | x | x | x | x | -- | -- | -- | -- | x | x | x | -- |
Note: “x” indicates the presence of a band, and “–” indicates the absence of a band.
Exploratory data analysis
The PCA with all the spectra considering 4 PCs accomplished 86.5% of explained variance. However, the first two PCs plot, which explains 70.5% of the data variance, represents the global discrimination reached by this analysis.
In a general view, the score plot, Figure 6, shows that the wax samples are well separated in PC1 positive direction, acrylic resin samples are spread along PC2 direction due to their multiple bands at \(1600\), \(1650\) and \(2060\) nm. Polyamide is the one with the lower intensity of these bands and because of that it appears in PC2 negative direction. The organic resin samples form a concise grouping in the PC1 negative direction.
The proteinaceous tempera, i.e., the agglutinants, are at the third quadrant of the score plot and have an influence of PC1 and PC2 negative direction, specifically \(1740\), \(1905\) and \(1200\) bands. The wax samples are the most prominent group in the score plot. It is positioned in the extreme positive direction of PC1, separated due to the higher intensity in bands of \(1720\) nm and \(1200\) nm according to the loadings plot presented in Figure 7.
Conclusions
Portable NIR spectrometry enabled the differentiation between the binder families evaluated, i.e., Wax, Agglutinants, Organic Resin and Acrylic Resin. The characteristic bands of each class of organic compounds were identified, and the PCA allowed their discrimination. The portable NIR used in this study is an effective low-cost tool for the characterization of molecular compounds in canvas paintings and may be a complementary tool to X-ray fluorescence, which is widely used in archaeometry for the determination of inorganic components, especially pigments.
Author contributions
F. L. Melquiades participated in conceptualization, data curation, formal analysis, funding acquisition, investigation, methodology, supervision, and writing – original draft. R. Molari participated in formal analysis, methodology, validation, and visualization. C. R. Appoloni participated in supervision, validation, and writing – review and editing.
Conflicts of interest
The authors declare no conflict of interest.
Acknowledgements
The authors acknowledge Dr. Márcia Rizzo for the mimetic preparation. Also the support from CNPq (grant number 306309/2023-8 and project number: 404214/2021-5), INCT-FNA (grant number 464898/2014-5).
References
Appoloni, C. R., Lopes, F., Rizzutto, M. de A., Neiva, A. C., Ikeoka, R. A. & Rizzo, M. de M. (2023). PIXE and pXRF Comparison Analysis of a Mockup Canvas Painting. e47604. https://doi.org/10.5433/1679-0375.2023.v44.47604
Brunetti, B., Miliani, C., Rosi, F., Doherty, B., Monico, L., Romani, A. & Sgamellotti, A. (2016). Non-invasive investigations of paintings by portable instrumentation: The MOLAB experience. 374(10), 1-35. https://doi.org/10.1007/s41061-015-0008-9
Cabral, J. M. P. (1995). Exame científico de pinturas de cavalete. 60-83.
Ciofini, D., Striova, J., Camaiti, M. & Siano, S. (2016). Photo-oxidative kinetics of solvent and oil-based terpenoid varnishes. 47-61. https://doi.org/10.1016/j.polymdegradstab.2015.11.002
Colombini, M. P., Degano, I. & Nevin, A. (2022). Analytical Approaches to the Analysis of Paintings: An Overview of Methods and Materials. Springer. 95-111. https://doi.org/10.1007/978-3-030-86865-9_3
Dooley, K. A., Coddington, J., Krueger, J., Conover, D. M., Loew, M. & Delaney, J. K. (2017). Standoff chemical imaging finds evidence for Jackson Pollock’s selective use of alkyd and oil binding media in a famous ‘drip’ painting. 9(1), 28-37. https://doi.org/10.1039/C6AY01795A
Furukawa, T., Kita, Y., Sasao, S., Matsukawa, K., Watari, M., Šašic, S., Siesler, H. W. & Ozaki, Y. (2002). On-Line Monitoring of Melt-Extrusion Transesterification of Ethylene Vinylacetate Copolymers by near Infrared Spectroscopy and Chemometrics. 10(3), 195-202. https://doi.org/10.1255/jnirs.335
Invernizzi, C., Rovetta, T., Licchelli, M. & Malagodi, M. (2018). Mid and Near-Infrared Reflection Spectral Database of Natural Organic Materials in the Cultural Heritage Field. 2018(1), 1-16. https://doi.org/10.1155/2018/7823248
Longoni, M., Genova, B., Marzanni, A., Melfi, D., Beccaria, C. & Bruni, S. (2022). FT-NIR Spectroscopy for the Non-Invasive Study of Binders and Multi-Layered Structures in Ancient Paintings: Artworks of the Lombard Renaissance as Case Studies. 22(5), 2052. https://doi.org/10.3390/s22052052
Panero, J. S., Silva, H. E. B. da, Panero, P. S., Smiderle, O. J., Panero, F. S., Faria, F. S. E. D. V. & Rodriguez, A. F. R. (2018). Separation of Cultivars of Soybeans by Chemometric Methods Using Near Infrared Spectroscopy. 10(4), 351. https://doi.org/10.5539/jas.v10n4p351
Rufino, E. S. & Monteiro, E. E. C. (2003). Infrared study on methyl methacrylate–methacrylic acid copolymers and their sodium salts. 44(23), 7189-7198. https://doi.org/10.1016/j.polymer.2003.08.041
Tsenkova, R., Atanassova, S., Toyoda, K., Ozaki, Y., Itoh, K. & Fearn, T. (1999). Near-Infrared Spectroscopy for Dairy Management: Measurement of Unhomogenized Milk Composition. 82(11), 2344-2351. https://doi.org/10.3168/jds.S0022-0302(99)75484-6
Unger, M., Ozaki, Y. & Siesler, H. W. (2011). Variable-Temperature Fourier Transform Near-Infrared (FT-NIR) Imaging Spectroscopy of the Diffusion Process of Butanol(OD) into Polyamide 11. 65(9), 1051-1055. https://doi.org/10.1366/11-06309
Vagnini, M., Miliani, C., Cartechini, L., Rocchi, P., Brunetti, B. G. & Sgamellotti, A. (2009). FT-NIR spectroscopy for non-invasive identification of natural polymers and resins in easel paintings. 395(7), 2107-2118. https://doi.org/10.1007/s00216-009-3145-6
Weyer, L. G. (1985). Near-Infrared Spectroscopy of Organic Substances. 21(1-2), 1-43. https://doi.org/10.1080/05704928508060427