*3.2. Evaluation of the Handheld Raman's Applicability*

Raman spectra were not only interpreted for pigments' identification, but also evaluated for readability and quality. When dealing with the handheld Raman spectra that were acquired from our samples, we came across some of the issues that generally arise in the discussions about the applicability of handheld Raman spectroscopy [19].

A common problem which can lead to uninterpretable spectra is the occurrence of strong fluorescence that obscures the Raman scattering signal and affects the accuracy and sensitivity of the measurements [50]. As mentioned, the Bravo-Raman device uses sequentially shifted excitation, a patented fluorescence minimization method [19].

As an example, a paint-out containing a Hansa Yellow pigment was selected to demonstrate the result of the automatic fluorescence reduction. This important class of SOPs, which is commonly found in modern artworks [30], often exhibits strong fluorescence in Raman spectroscopy. Figure 5 shows the handheld Raman spectrum of the paint-out Talens geel citroen [Talens yellow lemon], where Hansa Yellow (PY3) was identified. The spectrum recorded with the microRaman was not further processed after acquisition and shows a distinct fluorescence, whereas the spectrum recorded with the handheld Raman appears to be corrected in this respect, thanks to the SSE approach and the two lasers. The spectrum presents intense peaks, although the spectral resolution is not very high, as demonstrated by the lack of a peak at ~1593 cm−1, which was present in the microRaman spectrum.

**Figure 5.** Handheld and microRaman spectra of paint-out K44 Talens Yellow Lemon.

Another issue we considered was the merging or shifting of peaks as a consequence of over excitation, when measuring with a high, non-adjustable energy. This was observed in the Handheld Raman analysis of the paint-out Talens Rood Donker (Talens Red Dark), containing PR3 (Figure 6). Peaks that are close to each other and similar in intensity seem to be smoothed. This could be due to the high laser power, which might be responsible for a shift in peaks to lower wavenumbers, peak broadening, or twin peaks merging (for example at 1621 cm−<sup>1</sup> and 1605 cm<sup>−</sup>1).

Another example of possible overexcitation is reported in Figure 7. Here the handheld Raman spectrum of paint-out K41 Rembrandtgroen [Rembrandtgreen] is shown where the green copper phthalocyanine pigment PG7 was identified. Analysing phthalocyanine pigments with Raman spectroscopy can lead to overexcitation and peak shifts [29,51]. Fremout et al. (2012) made this observation while analysing the related PB15 pigment and state that the exact reason for this phenomenon is unclear and that a reversible, heat induced deformation of the crystal structure is possible [29]. The handheld Raman and microRaman spectra seem indeed to be shifted when compared to the KIK-IRPA reference spectrum

(https://soprano.kikirpa.be, accessed on 30 June 2021 [30]) (Figure 7). However, the PG7 spectrum published by Schulte et al. fits our Raman spectrum [10]. These discrepancies might be explained by the spectral differences observed for dry pigment powder and paint systems. Defeyt et al. noticed peak shifts when comparing the Raman spectra of blue copper phthalocyanine PB15 as a dry pigment powder and mixed with a binder [43].

**Figure 6.** Handheld and microRaman spectra of paint-out K50 Talens red dark, mixed with barium sulphate (986 cm<sup>−</sup>1).

**Figure 7.** Handheld and microRaman spectra of paint-out K41 Rembrandtgroen and the KIK-IRPA reference spectrum for PG7.

Figure 8 reports the spectrum of Talens rose [Talens pink] in comparison with the reference spectra of the BON pigment lake Lithol Rubine, precipitated on two different substrates. This example was chosen to illustrate the possible effects of the minor spectral resolution of the handheld Raman device when examining closely related pigments. The KIK-IRPA references of Lithol Rubine PR57:1 (calcium carbonate salt) and PR57:2 (barium salt) show slight spectral differences, allowing a differentiation, given that the recorded spectrum is well defined. The spectral resolution of the Bravo device is 10–12 cm<sup>−</sup>1, which turns out to be almost three times lower than that of our microRaman device (Table 1). Pozzi et al. also noted that "the inferior spectral resolution compared with benchtop spectrometers may cause difficulties in differentiating among closely related molecules with similar Raman fingerprints, especially if the analyses have poor Raman cross section or their spectra are characterized by low signal-to-noise ratio" [19]. It was, however, shown that the handheld Raman spectrum of Talens rose [Talens pink] contains the small spectral features that allow the assignment to PR57:1. In this case, the minor spectral resolution of the handheld Raman proved to be sufficient to allow a clear identification.

**Figure 8.** Handheld Raman spectrum of paint-out Talens rose and Raman spectra of the KIK-IRPA PR57:1 and PR57:2 references.

Two paint-outs (K37 Paul Veronesegroen [Paul Veronese Green] and K37 Dekgroen [Dekgreen]) were revealed to be rather sensitive to the handheld Raman lasers. The aftermeasurement check of the surfaces showed local burns, resulting in a small (~0.5 mm) black discoloration. According to Talens' recipes, these samples should contain a green SOP, but the X-ray fluorescence (XRF) measurements identified a copper arsenic pigment, which is rather sensitive to Raman laser irradiation. The analysis of these paint-outs highlights a strong disadvantage of the handheld Bravo, namely the non-adjustable laser energy, which is about 50 mW, more than two times higher compared to the microRaman instrument we used. The laser power can, however, be reduced by increasing the working distance to 1–2 mm. In this way, potential changes or thermal damage can be minimized or even prevented. Pozzi et al. [19] address this problem by proposing preventive measurement settings for sensitive samples. Other recommended options to avoid damage are the placement of a neutral density filter in front of the laser spot or the use of a defocusing tip.

It is worth mentioning that no SOP-containing paint-out showed any comparable damage. Since the sensitivity of emerald green is known, preliminary elemental analysis with XRF spectroscopy would have raised attention to this risk and demanded preventive action by adjusting the measurements settings.

### **4. Conclusions**

This study demonstrated that identification of SOPs with a non-invasive approach in modern oil paint systems, also in the presence of a varnish layer, is possible. The handheld Raman was able to identify a large variety of SOPs—belonging to eight different pigment groups—within the historic varnished paint-outs of a leading artists' paint manufacturer and provided in the meantime a very useful insight into the use of early SOPs in modern artists' oil paints dated 1932–1950.

This research intends to provide a basis for future research into the non-invasive identification of SOPs in pre-1950 works of art. Detailed knowledge about the type and the characteristics of modern SOPs contributes to a better understanding of artists' materials from this period and helps to assess the restoration and conservation needs. Setting these pigments into their art technological context will contribute to studying the history of certain works of art and add to the knowledge available to detect forgeries.

Some of the SOPs identified within this study have not yet been analytically detected within works of art from before 1950. This outcome emphasizes the importance of further and intensive study of the use of SOPs in modern art. On the one hand, the knowledge of how to analyse SOPs without taking samples in a scan-like examination setup opens up new possibilities for their further investigation. On the other hand, this non-invasive approach can be used to clearly determine adequate spots for sampling, when necessary.

The advantages and drawbacks of the use of handheld Raman spectroscopy in the investigation of SOPs in artists' materials were considered. Thanks to the efficient suppression of fluorescence, varnished paint systems could be analysed and it was even possible to identify the mixtures of SOPs. The rather low resolution was a limiting factor, but turned out to be not too problematic for this set of samples. In some cases subtle spectral differences could still be detected, enabling a distinction between pigment substrates. The issue of non-adjustable energy was revealed to only be risky for one inorganic pigment, but this can be addressed with other non-invasive complementary techniques, such as handheld X-ray fluorescence spectroscopy. Another challenge is the lack of a camera for exact positioning. In this study a Melinex®template was successfully used for the analysis of the paint-outs, but the examination of specific and/or detailed paint areas in artworks could be hindered or difficult, depending on the surface and type of object.

**Author Contributions:** Conceptualization, R.P., I.D.v.d.W., and K.J.v.d.B.; methodology, R.P., I.D.v.d.W., and K.J.v.d.B.; investigation, R.P.; writing—original draft preparation, R.P., I.D.v.d.W., and K.J.v.d.B.; writing—review and editing, R.P., I.D.v.d.W., and K.J.v.d.B.; supervision, I.D.v.d.W. and K.J.v.d.B. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Data available on request. The data presented in this study are available on request from the corresponding author.

**Acknowledgments:** The authors would like to thank Royal Talens for allowing access to their production archive and the Historical collection of dyes of the TU Dresden for providing reference samples, as well as Art Ness Proaño Gaibor, Sanne Berbers (Cultural Heritage Laboratory, Cultural Heritage Agency of the Netherlands, Amsterdam, the Netherlands) and Brynn Sundberg (University of Amsterdam, Amsterdam, the Netherlands) for conducting conclusive UHPLC-MS analyses. Furthermore we would like to thank the SOPRANO network, especially Wim Fremout (KIK-IRPA), for providing reference spectra.

**Conflicts of Interest:** The authors declare no conflict of interest.
