Extension study of a statistical age prediction model for acrylic paints
Introduction
The contemporary art market is experiencing an unprecedented level of buying and selling activity that generates millions of dollars worldwide. As art prices rise, art-related crime is on the increase, making counterfeiting a multi-million dollar enterprise. The uncovering of fake artworks has a negative impact on the entire art market. That is why new technologies are being used to detect counterfeits that have already entered the art market as well as to prevent their sale and acquisition [1,2].
Non-destructive techniques such as Raman spectroscopy [[3], [4], [5], [6]], Fourier transform infrared spectroscopy (FTIR) [3], IR hyper spectral imaging (IR-HIS) [7,8], FTIR-attenuated total reflection (FTIR-ATR) [6,9], IR reflection [10] and scanning electron microscopy-energy dispersive X-ray spectroscopy (SEM-EDX) [6,9] are usually used as first option as they do not damage the work of art [3,6,9]. Among the techniques, FTIR-ATR spectroscopy is widely used for the characterisation of artwork materials [6,9,11] and for the study of the influence of pigments on the photodegradation of acrylic paints [[12], [13], [14], [15]].
In some cases, non-invasive techniques are not enough, so analytical techniques combined with statistical methods, such as multivariate chemometrics, have been widely used to extract the maximum and most representative information from the acquired data in an interpretable manner. Partial Least Squares (PLS) is an X→y regression methodology typically used for either predictive purposes or control monitoring. Despite the applications of partial least squares (PLS) for quantification of paint components in the restoration of artworks [16] as well as for the determination of their properties in quality control [17], there are not many studies that apply this method to the dating of works of art, although the applicability and effectiveness of this tool has been demonstrated in other forensic fields [18,19]. An important challenge that the PLS has to face is the eventual lack of robustness of the models when they want to be used for universal applications [20]. Since the setup of an optimal FTIR-ATR multivariate model is costly, once it has been developed it is expected to be valid for a long period of time.
In previous research by L. Ortiz-Herrero et al. [21], a non-invasive paint dating methodology was developed for the authentication of acrylic paints used in contemporary works of art. A multivariate chemometric method (orthogonal partial least squares (OPLS)) was applied to FTIR-ATR spectroscopic data acquired from Liquitex® brand aged paints containing the phthalocyanine green pigment (PG7) to develop age prediction models split into two temporal ranges (short- and long-term). The applicability of the predictive models was tested in artworks created with the Liquitex® brand by internationally recognised contemporary Basque artists. The correlation obtained between natural ageing and its equivalent to the accelerated one may have possible applications in the dating of artworks for up to at least 22 years preserved under study conditions and created with the same paint formulation. In addition, the loadings graph allows studying which variables influence the predictive models. From this graph, the ageing processes of the paint constituents by which the predictive models were characterised were studied. The short-term model was influenced by the degradation of the acrylic binder due to chain-scission reactions and oxidation phenomena as well as by the non-ionic polyethylene oxide surfactant due to the excision of the hydrocarbon chains. The long-term model was only characterised by the modifications undergone by the acrylic binder due to the disappearance of oxidized species and cross-linking reactions. Thus, it was concluded that the predictive models could potentially be applied to other brands of acrylic paints with the same n-butyl acrylate-methyl methacrylate (nBA-MMA) copolymer as well as to different Liquitex® brand paint colours, as pigment modifications during ageing seemed not to be a determining factor in the predictive models.
The aim of this article was to corroborate the above facts and to study the robustness and potential applicability of the predictive models in different acrylic paint colours and brands. For this purpose, three Liquitex® paint colours and a Vallejo® paint that shared the same organic pigment and acrylic resin as the paint used in the previous research were selected. The paints were exposed to slightly modified conditions of accelerated ageing, analysed by FTIR-ATR and subsequently applied to the predictive models to determine their robustness and feasibility based on the accuracy error values obtained for each paint sample. The acrylic paints were also characterised and the evolution of their ATR spectra throughout ageing was studied.
Section snippets
Paint samples and preparation
The choice of manufacturers and paint colours was made on the basis of publications and interviews with contemporary artists to find out the most studied and widely used materials. Different pigments were selected to check whether the type and colour of the pigment influenced the chemical behaviour and stability of the acrylic paints, and therefore whether it could affect the applicability of the predictive model. Commercial acrylic paints from Liquitex Acrylic Paint® (U.S.A.) containing Hansa
Characterisation by FTIR-ATR
The characterisation of FTIR-ATR spectra of Liquitex® paints containing PY3 and PB29 pigments was performed (Fig. 1). The characterisation of Liquitex® paint with PG7 pigment was previously carried out by L. Ortiz-Herrero et al. [21]. The two Liquitex® paints gave a characteristic nBA-MMA copolymer profile. A strong band at around 1727 cm−1 ascribed to the carbonyl stretching vibration was observed. The absorption bands at about 2955, 2934 and 2874 cm−1 were attributed to C–H bond stretching
Conclusions
In this study, four commercial acrylic paints from two different manufacturers were characterised by FTIR-ATR. Each acrylic paint showed characteristic modifications of its ATR spectrum throughout the accelerated ageing in spite of sharing the same paint constituents such as the acrylic binder, the surfactant and in two of the cases the organic pigment.
Pigments were found to influence the short-term stability of acrylic paints when exposed to accelerated ageing, as they may act as promoters or
Funding
This research did not receive any specific grant from funding agencies in the public, commercial or not-for-profit sectors.
CRediT authorship contribution statement
L. Ortiz-Herrero: Investigation, Methodology, Validation, Formal analysis, Data curation, Writing - original draft, Visualization. I. Cardaba: Resources, Investigation. L. Bartolomé: Conceptualization, Supervision, Project administration, Writing - review & editing. M.L. Alonso: Writing - review & editing. M.I. Maguregui: Conceptualization, Supervision, Project administration, Writing - review & editing.
Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgments
The authors would like to thank the General Research Services (SGIker) of the UPV/EHU for the technological support. Authors also thank to Omaira de la Hera of the Faculty of Science and Technology of the UPV/EHU for her useful discussion. L. Ortiz- Herrero thanks UPV/EHU for the pre-doctoral fellowship.
References (32)
- et al.
Spectroscopic investigation of a ‘Virgin of Sorrows’ canvas painting: a multi-method approach
Anal. Chim. Acta
(2005) - et al.
Raman spectroscopic signatures of the yellow and ochre paints from artist palette of J. Matejko (1838–1893)
Spectrochim. Acta A Mol. Biomol. Spectrosc.
(2015) - et al.
Chemometrics approach to FT-IR hyperspectral imageing analysis of degradation products in artwork cross-section
Microchem. J.
(2017) - et al.
Hyperspectral imageing combined with data classification techniques as an aid for artwork authentication
J. Cult. Herit.
(2017) - et al.
Analytical characterisation of artist's paint systems based on emulsion polymers and synthetic organic pigments
J. Anal. Appl. Pyrolysis
(2018) - et al.
Interpretation of mid and near-infrared reflection properties of synthetic polymer paints for the non-invasive assessment of binding media in twentieth-century pictorial artworks
Microchem. J.
(2016) - et al.
Photostability and influence of phthalocyanine pigments on the photodegradation of acrylic paints under accelerated solar radiation
Polym. Degrad. Stabil.
(2017) - et al.
Spectroscopic methods for the identification and photostability study of red synthetic organic pigments in alkyd and acrylic paints
Microchem. J.
(2018) - et al.
Influence of phthalocyanine pigments on the photo-degradation of alkyd artists' paints under different conditions of artificial solar radiation
Polym. Degrad. Stabil.
(2016) - et al.
Accelerated UV ageing studies of acrylic, alkyd, and polyvinyl acetate paints: influence of inorganic pigments
Microchem. J.
(2016)
ATR-FTIR spectroscopy and quantitative multivariate analysis of paints and coating materials
Spectrochim. Acta Mol. Biomol. Spectrosc.
Calibration of paint and varnish properties: potentialities using X-ray spectroscopy and partial least squares
Chemometr. Intell. Lab. Syst.
DATUVINK pilot study: a potential non-invasive methodology for dating ballpoint pen inks using multivariate chemometrics based on their UV–vis-NIR reflectance spectra
Microchem. J.
Direct and indirect approaches based on paper analysis by Py-GC/MS for estimating the age of documents
J. Anal. Appl. Pyrolysis
OPLS multivariate regression of FTIR-ATR spectra of acrylic paints for age estimation in contemporary artworks
Talanta
Azo-pigments effect on UV degradation of contemporary art pictorial film: a FTIR-NMR combination study
Polym. Degrad. Stabil.
Cited by (4)
Unmixing and pigment identification using visible and short-wavelength infrared: Reflectance vs logarithm reflectance hyperspaces
2023, Journal of Cultural HeritageMultivariate (O)PLS regression methods in forensic dating
2021, TrAC - Trends in Analytical ChemistryCitation Excerpt :Since multivariate regression models are built from samples prepared and aged under controlled conditions, we would expect each forensic scenario to have its own tailored model. However, this is not always the case, as there are multivariate regression models that have been successfully applied to samples different from those used in the construction of the model and exposed to moderately different conditions, as Ortiz-Herrero et al. state in their study [31]. Ortiz-Herrero et al. [31] studied the robustness and feasibility of the short- and long-term ageing OPLS models of the Liquitex® paint brand when applied to date different brands and colours of acrylic emulsion paints.
MODELING THE WETTING OF TITANIUM DIOXIDE AND STEEL SUBSTRATE IN WATERBORNE PAINT AND VARNISH MATERIALS IN THE PRESENCE OF SURFACTANTS
2022, Eastern-European Journal of Enterprise Technologies