Discovering Hidden Painted Images: Subsurface Imaging Using

Dec 15, 2016 - We demonstrate for the first time the mapping capability of micro-spatially offset Raman spectroscopy (micro-SORS). The technique enabl...
2 downloads 11 Views 1MB Size
Article pubs.acs.org/ac

Discovering Hidden Painted Images: Subsurface Imaging Using Microscale Spatially Offset Raman Spectroscopy Alessandra Botteon,*,† Claudia Conti,† Marco Realini,† Chiara Colombo,† and Pavel Matousek*,‡ †

Consiglio Nazionale delle Ricerche, Istituto per la Conservazione e la Valorizzazione dei Beni Culturali (ICVBC), Via Cozzi 53, 20125, Milano, Italy ‡ Central Laser Facility, Research Complex at Harwell, STFC Rutherford Appleton Laboratory, Harwell Oxford, OX11 0QX, United Kingdom ABSTRACT: We demonstrate for the first time the mapping capability of micro-spatially offset Raman spectroscopy (micro-SORS). The technique enables to form noninvasive images of thin sublayers through highly turbid overlayers. The approach is conceptually demonstrated on recovering overpainted images in situations where conventional Raman microscopy was unable to visualize the sublayer. The specimens mimic real situations encountered in Cultural Heritage that deal, for example, with hidden paintings vandalized with graffiti or covered by superimposed painted layers or whitewash. Additionally, using a letter as a hidden image, we demonstrated the micro-SORS potential to reconstruct also a hidden writing covered, for example, with paper sheets that cannot be easily removed. Potential applications could also include other disciplines such as polymers, biological, catalytic, and forensic sciences where thin, highly turbid layers mask chemically distinct subsurface structures.

N

Raman spectroscopy (SORS)3 with microscopy translating SORS capability down from millimeter to micrometer scale and as such enabling, unlike for the parent SORS technique, the probing of thin highly turbid layers such as layers of paint in art. To date, the technique has been demonstrated to be capable of identifying the chemical makeup of turbid sublayers in art, biological samples, and polymers.4,5 Fluorescence rejection from overlayers has also been reported with this method.6 Here we demonstrated conceptually the ability to also form 2D images of sublayers obscured behind highly turbid overlayers. Although the recovery of spatial information has been demonstrated with macroscale SORS in mapping and tomographic imaging7−10 no such demonstration has been carried out with micro-SORS which involves spatial scale reduction by two to 3 orders of magnitude accompanied typically by very different optical regime.6 In general, several micro-SORS variants have been demonstrated to date. The most basic is defocusing micro-SORS,1 which can be practiced on existing Raman microscopes without any modifications, but provides only limited effectiveness (in terms of penetration depth, sensitivity, and discrimination against top layer) compared with a more advanced variant, full micro-SORS.11−13 The full micro-SORS requires some modification to the microscope in order to be practiced in its unrestricted form, but permits by up to an order of magnitude

oninvasive imaging through thin stratified turbid (diffusely scattering) layers is an analytical challenge encountered in a wide range of disciplines. An illustrative example is the recovery of hidden paintings by nondestructive means in Cultural Heritage. Other beneficial areas potentially include polymer, catalytic, forensic, and biological sciences, where imaging through highly turbid overlayers may also be beneficial. Although several tools currently exist to accomplish such tasks, including X-ray absorption spectroscopy, optical coherence tomography (OCT), and near-infrared spectroscopy (NIR), these have limitations restricting their applicability. For instance, X-ray imaging requires X-ray (atomic) contrast between imaged layers to be present, OCT requires physical contrast (eg refractive index change) between layers and NIR spectroscopy relies on chemically specificity in overtone/combination region of the spectrum to be present (molecular contrast) that may not always be present or adequately high. In terms of the ability to provide high chemical contrast both Raman and mid-infrared (MIR) spectroscopies are highly performing techniques due to their high molecular specificity. However, MIR, for its inherently high absorption, cannot effectively penetrate through a typical painted layer. In contrast, Raman spectroscopy potentially can, as it typically exhibits much lower absorption in typical media but is hampered, in its conventional form, by the turbidity of overlayers rendering subsurface probing of painted layers often unviable, too. Recently, a new tool in Raman microscopy overcoming the above limitation with turbid media has emerged, micro-SORS.1,2 The concept combines conventional macroscale Spatially Offset © XXXX American Chemical Society

Received: September 8, 2016 Accepted: November 29, 2016

A

DOI: 10.1021/acs.analchem.6b03548 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry Table 1. Investigated Samples sample

description

top layer

hidden layers

S1

regular-shape painting covered with graffiti

yellow spray: pigment, PY151; thickness, 250 μm

S2

graffiti painting covered with whitewash

whitewash: pigment, calcite; thickness, 120 μm

S3

irregular-shape painting covered with red paint

red paint: pigment, cadmium red; thickness, 180 μm

S4

painted writing covered with colored paper

colored paper: pigment, PV2; thickness, 180 μm

blue paint: pigment, phthalocyanine blue + calcite; thickness, 100 μm green paint: pigment, phthalocyanine green + titanium white; thickness, 70 μm red spray: pigment, PR170 and PY139; thickness, 30 μm yellow spray: pigment, PY151; thickness, 30 μm yellow paint: pigment, PY176; thickness, 30 μm white paint: pigment, titanium white; thickness, 50 μm blue paint: pigment, phthalocyanine blue; thickness, 30 μm whitewash: pigment, calcite; thickness, 250 μm

including an extremely fluorescent top layer. Many pigments employed in artistic production can exhibit a high level of fluorescence, a phenomenon associated with electronic excitation of the pigment molecule by incident photon and subsequent emission of fluorescence photon. This emission can mask Raman signal of pigment and preclude its detection. S4 sublayer was made to form a letter “F” painted using a blue paint (phthalocyanine blue) over a whitewash background (calcite). Two superimposed pieces of paper impregnated with a marker pen pink ink (PV2) were used as top layer. This specimen aims at showing micro-SORS ability to reconstruct a graphic sign, and even “read” through a turbid overlayer; the complexity of this

larger contrast improvement between the layers than defocusing micro-SORS.6,12 In this study, we demonstrate, for the first time, the ability to retrieve noninvasively 2D images of overpainted sublayers using full micro-SORS configuration which otherwise are inaccessible to conventional Raman microscopy. The specimens used mimic real situations encountered in Cultural Heritage dealing with hidden paintings. The study paves the way for discovering and reconstructing paintings by restorers which have been vandalized with graffiti or covered by superimposed painted layers or whitewash. Additionally, using a letter as a hidden image, we demonstrate the potential of microSORS to reconstruct a writing (or a graphic sign in general) covered, for example, by paper sheet(s) that cannot be easily removed.



EXPERIMENTAL SECTION Materials. The mock-up specimens used on these proof-ofconcept experiments comprised different paint layers and were prepared with an aim to gradually increase the complexity of the covered images (from S1 to S4), thus, demonstrating the microSORS capability to elucidate a range of different situations, from the easiest to more complex ones (see Table 1). S1 consists of two layers, the sublayer is made of a green painted area located next to a blue area. The blue area is a mixture of phthalocyanine blue and calcite; the green one is a mixture of phthalocyanine green and titanium white (rutile). Calcite and titanium white were added to dilute phthalo pigments. These were covered with a yellow spray layer (250 μm). This sample serves to reconstruct small and regular shape images and mimics a painting vandalized with graffiti. S2 sublayer is made of yellow and red spray (PY 151 and a mixture of PR 170 and PY 139, respectively) applied one next to each other without overlapping. The covering layer consists on a calcite-based paint layer. This specimen serves to illustrate the reconstruction of large and regular shape images and recreates the case of a graffiti painting covered with whitewash. In S3, an irregular shape spot (approximately 1 mm in size) was painted with a yellow paint (PY 176) over a white paint background (titanium white) and covered with a red paint (cadmium red). With this specimen we increased the degree of complexity by involving a hidden irregularly shaped drawing, Table 2. Integrated Bands phthalocyanine blue phthalocyanine green mixture of PR170 and PY139 PY151 PY176

1535−1525 cm−1 1542−1535 cm−1 1367−1357 cm−1 1586−1576 cm−1 1607−1590 cm−1

Figure 1. Raman spectra of pigments and colorants used in this study: (a) phthalocyanine green; (b) phthalocyanine blue; (c) PY151; (d) mixture of PR170 and PY139; (e) calcite; (f) titanium white; (g) PY176; (h) cadmium red; (i) PV2. Asterisks indicate the integrated bands. B

DOI: 10.1021/acs.analchem.6b03548 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

Figure 2. S1 optical microscope picture and chemical images: (a) optical microscope picture of the uncovered sample; (b) optical microscope picture of the covered sample; (c) M1; (d) M2; (e) M3. Scale bar = 500 μm.

(OPUS-IR software controlled) enabling controlled sideways movement to permit the setting of spatial offsets with respect to the Raman collection zone with a high accuracy and reproducibility in the positioning. Two types of measurements were carried out using the external probe: (1) first, when keeping the samples immobile with respect to the microscope objective during the measurements. This configuration allowed to acquire Raman spectra at different spatial offsets (Δs) and select, for each specimen, the most effective spatial offset rendering both high contrast between the layers and good signal-to-noise ratio for sublayer. (2) After that, the external probe was set to the above optimum spatial offset (700 μm for S1, S2, and S3; 600 μm for S4). The samples were then positioned on a micrometric sample stage to enable mapping. In this configuration laser illumination spot and Raman collection area maintained the same fixed distance and position in space, while sample was moved in x,y directions underneath. Three maps were acquired for each specimen: the first one was acquired at imaged position on the bottom layer, before the top layer was applied (M1). The second map was acquired at imaged position in the same area after the top layer was applied (M2). M1 and M2 were acquired with the conventional confocal Raman microscope configuration, ie without external probe, using the integral microscope optics (4× objective). The third map was then acquired on the complete sample with the external probe using micro-SORS, set to the optimum spatial offset (M3). The most effective spatial offset was chosen by moving the sample with offset steps ranging from 30 to 100 μm and identifying the one that provided the highest quality sublayer signal visibility. The maps were acquired with a laser power at the sample of 100 mW and 1200 grooves/mm grating in the detection system.

situation is elevated due to the fact that the Raman spectral features of the letter are partially overlapped with those of the covering material. All pigments used have acrylic media, except for the pink ink in S4 which contains a xylene based solvent and whitewash in S2 and S4, that was prepared using a mixture of calcium hydroxide and water (hydrated lime) resulting in hardening via carbonatation process forming calcite. All specimens were built up on top of a carbonatic stone as basic substrate. After the experiments, the specimens were cut to obtain cross sections; the paint layers thickness was then measured using an optical microscope (see Table 1). Methods. Micro-SORS measurements were carried out using a Senterra dispersive micro-Raman spectrometer (Bruker Optik GmbH) both in its conventional configuration (conventional Raman microscopy) and with small modifications (to enable full micro-SORS). With regard to the instrument modifications these included the addition of an external standard Raman probe (UniLabII, Bruker Optik GmbH), equipped with a 4× lens (a working distance of about 18 mm), to deliver the laser beam (785 nm excitation wavelength) to the sample (bypassing the microscope optics) at about 50° with respect to the incidence/ sample plane. The Raman signal was collected with a Peltier cooled CCD detector (1024 × 256 pixels) using standard microscope optics (Olympus BX51) and 4× objective (WD 18.5 mm, NA 0.1). This configuration resulted in an elongated illumination laser spot on sample surface. The spot size on sample surface was approximately 30 μm along x axis and 40−50 μm along y (elongated) axis. In this configuration its effect would only be noteworthy at the zero (“imaged”) position or at very small spatial offsets where it would be reducing the overlap between the collection and illumination zones and as such the signal contrast between the surface and subsurface layers. The external probe was mounted on the sample micropositioning stage C

DOI: 10.1021/acs.analchem.6b03548 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

Figure 3. S2 optical microscope picture and chemical images: (a) optical microscope picture of the uncovered sample; (b) optical microscope picture of the covered sample; (c) red spray M1; (d) yellow spray M1; (e) M2; (f) M3. Scale bar = 2 mm.



M1 and M2 were acquired using a 50 μm confocal pinhole and the micro-SORS measurements with a confocal slit of 50 × 1000 μm (M3). The maps covered an area ranging from approximately 3.5 mm2 to 1 cm2 with a step size along the y and x axes from 90 μm to 1 mm. The maps were acquired with total acquisition times ranging from 2 to 8 h (5−10 coadditions, 5−10 s integration time for each spectrum). OPUS (7.2) was used to generate the chemical images; the spectra were baseline corrected and the most informative band for each compound of the hidden layer was integrated considering the area above a straight line between two defined wavenumber limits (see Table 2).

RESULTS AND DISCUSSION

The reference Raman spectra of pigments and colorants used in this study are shown in Figure 1. In Table 2, the integrated bands for each compound are identified. For S1, the chemical images of M1 reproduce with good precision the actual distribution of the two paints (Figure 2c). The conventional Raman map acquired at imaged position after covering the paints with the yellow spray (M2) is, however, noninformative with respect to the sublayer: the different shades of colors in the chemical images are generated by random noise present in the integrated ranges (Figure 2d). The Raman signal D

DOI: 10.1021/acs.analchem.6b03548 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

Figure 4. S3 optical microscope picture and chemical images: (a) optical microscope picture of the uncovered sample; (b) optical microscope picture of the covered sample; (c) M1; (d) M2; (e) M3; (f) overlap of M3 and the optical microscope picture of the uncovered sample. Scale bar = 500 μm.

Figure 5. Schematic illustration of three different situations encountered during full micro-SORS mapping (M3) of S3; (a), (b), and (c) represent a cross-section view of top (red) and hidden (yellow) layers: the pathway of the laser and the generated signal are shown; (d) represents the overlapping between M3 and optical microscope image (scale bar = 500 μm).

the other hand, the yellow spray distribution in M3 is less precise, as shown in Figure 3f, with some upper segments missing. This may be due to the layer thickness nonuniformity: in the “missing” area the yellow layer is thinner and this is reflected in a less intense signal of the pigment in both M1 and M3 images. In an earlier study the capability of micro-SORS to recover the chemical makeup of sublayers in the presence of a fluorescent top layer was also demonstrated.6 In this work, the investigations on S3 analogously demonstrate the ability of obtaining a full 2D chemical image of a hidden figure obscured by an intensely fluorescing overlayer. Moreover, the method was proved to be capable of detecting and reproducing, with some approximation, a small dimension (in the range of one millimeter) and irregularly shaped painted object. M2 is dominated by the strong fluorescence signal of cadmium red and the image is entirely obscured, as is shown in Figure 4d; on contrary, in M3 the chemical image obtained by integrating the most intense signal of the yellow paint shows the original shape of the spot, with some imprecisions present: its upper part is not detected in M3 (Figure 4e). Again, this is ascribed to the fact that the layer is thinner in this area as was confirmed after samples were cut and layers’ thickness was measured; in any case, the fact is also evident from M1 map, where a less intense signal is detected from this part of the image.

observed in the spectra of M2 is entirely due to that of the upper layer, made of acrylic yellow spray. On the contrary, when the map is acquired with micro-SORS (700 μm offset, M3) the signals of phthalocyanine blue and green clearly appear, while the yellow layer signal is less intense. In the chemical images, the pattern of green and blue is recognizable although it is less spatially resolved compared to M1 as some blurring occurs (Figure 2e). This is expected and is due to laser and Raman photon diffusion processes within sample and noise present in spectra as well as overlap present between the two principal bands. Additional contribution to the blurring may also originate from potential nonuniform thickness (and density) of layers. Calcite and titanium white were used only to dilute phthalo pigments and were not targeted in these measurements. To demonstrate the micro-SORS capability to form chemical images over larger areas, S2 specimen was prepared and mapped. In M2 only, the signal of calcite (upper layer) is visible in the most of the measured points. Occasionally, a weak signal of the sublayer appears (see Figure 3e). This is explained by layers thickness irregularity or potential presence of porosity in the top layer enabling the detection of the sublayer at certain locations using conventional Raman microscopy. With respect to the red spray signal, a very good match between M3 (Figure 3c,d) and M1 (Figure 3f) is observed. On E

DOI: 10.1021/acs.analchem.6b03548 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

Figure 6. S4 optical microscope image and chemical images: (a) optical microscope image of the uncovered sample; (b) optical microscope image of the covered sample; (c) M1; (d) M2; (e) M3. Scale bar = 3 mm.

blue 1525−1535 cm−1 band (Figure 6). M1 reproduces a welldefined F letter. The pink covering material Raman spectrum yields an overlapped band located at 1530 cm−1, as such M2 chemical image displays a spurious information with regards to the hidden blue paint distribution (Figure 6d). M3 chemical image reveals the hidden painting and F letter is again observable (Figure 6e). As shown, the overlapping of the Raman bands between the two Raman spectra does not limit the applicability of micro-SORS. Calcite, that is used as background in the sublayer, is not detected in M3. This is probably due to the fact that calcite is at a lower position compared to the blue pigment and also exhibiting a lower Raman scattering cross section compared with phthalocyanine blue (approximately by a factor of 8).

The white paint (titanium white) used as background in the sublayer (third layer) is not detected in M3. This is probably due to the fact that the white pigment is at a lower position compared to the yellow spot. In addition, rutile most intense bands are located at 443 and 610 cm−1 where the fluorescence is the most severe making its detection more challenging. The above observations point at limitations of micro-SORS approach as too thin sublayers or too thick, intensely fluorescent or intensely Raman scattering overlayers can render this technique ineffective. In Figure 4f, M1 and M3 are shown overlapped on top of each other. We note a small shift between the images. This is due to the fact that in full micro-SORS the maximum signal originates from around a midpoint zone (in a typical scenario) between the laser illumination point and the Raman collection area, which themselves are separated by 700 μm (see Figure 5). This effect leads to a shift to the image in the direction of the laser illumination zone by approximately a half of the spatial offset used with respect to the actual position of the object under the Raman microscope objective (for typical sublayers). It is worth pointing out that the image shift occurs in all specimens, although in S3 is particularly evident due to the shape and size of the spot; similarly, in S1 the shift is more clearly recognizable than for S2 or S4 due to the small dimension of S1 map. S4 sample was assembled and mapped to demonstrate that the method is capable of also visualizing a hidden writing. The chemical images were obtained by integrating phthalocyanine



CONCLUSIONS The study has demonstrated that micro-SORS is not only capable of recovering chemical information on the constitution of buried sublayers with diffusely scattering samples but also forming 2D chemical images, also in situations where conventional Raman and optical microscopy is ineffective in rendering these due to obstruction by diffusely scattering overlayers. The technique was conceptually demonstrated on specimens mimicking real situations encountered in Cultural Heritage. Additionally, using a letter as a hidden image, it was demonstrated that hidden writing can also be retrieved by this F

DOI: 10.1021/acs.analchem.6b03548 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry method. Other potential applications may include mapping of polymers, biological, catalytic, and forensic samples in situations where highly turbid layers mask chemically distinct subsurface structures with thin stratified samples. Some limitations of micro-SORS approach were also identified; for example too thin sublayers, compounds with too small Raman scattering cross sections or too thick, intensely fluorescent or intensely Raman scattering overlayers can preclude the detection of sublayer. This study represents the first conceptual demonstration of micro-SORS imaging approach. To establish the real potential and limitations of the proposed approach, the method needs to be tested more thoroughly on real objects of art.



AUTHOR INFORMATION

Corresponding Authors

*E-mail: [email protected]. *E-mail: [email protected]. ORCID

Alessandra Botteon: 0000-0002-0948-5767 Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS We would like to thank “Arte e mestieri” company (Casarsa della Delizia - PN) for providing the hydrated lime (Natural Calk) used for samples preparation.



REFERENCES

(1) Conti, C.; Colombo, C.; Realini, M.; Zerbi, G.; Matousek, P. Appl. Spectrosc. 2014, 68, 686. (2) Matousek, P.; Conti, C.; Realini, M.; Colombo, C. Analyst 2016, 141, 731. (3) Matousek, P.; Clark, I. P.; Draper, E. R. C.; Morris, M. D.; Goodship, A. E.; Everall, N.; Towrie, M.; Finney, W. F.; Parker, A. W. Appl. Spectrosc. 2005, 59, 393. (4) Conti, C.; Realini, M.; Colombo, C.; Matousek, P. J. Raman Spectrosc. 2015, 46, 476. (5) Conti, C.; Realini, M.; Colombo, C.; Sowoidnich, K.; Afseth, N. K.; Bertasa, M.; Botteon, A.; Matousek, P. Anal. Chem. 2015, 87, 5810. (6) Conti, C.; Botteon, A.; Colombo, C.; Realini, M.; Matousek, P. Analyst 2016, 141, 5374. (7) Schulmerich, M. V.; Finney, W. F.; Fredricks, R. A.; Morris, M. D. Appl. Spectrosc. 2006, 60, 109. (8) Schulmerich, M. V.; Dooley, K. A.; Vanasse, T. M.; Goldstein, S. A.; Morris, M. D. Appl. Spectrosc. 2007, 61, 671. (9) Demers, J. L. H.; Davis, S. C.; Pogue, B. W.; Morris, M. D. Biomed. Opt. Express 2012, 3, 2299. (10) Stone, N.; Kerssens, M.; Lloyd, G. R.; Faulds, K.; Graham, D.; Matousek, P. Chem. Sci. 2011, 2, 776. (11) Di, Z.; Hokr, B. H.; Cai, H.; Wang, K.; Yakovlev, V. V.; Sokolov, A. V.; Scully, M. O. J. Mod. Opt. 2015, 62, 97. (12) Conti, C.; Realini, M.; Colombo, C.; Matousek, P. Analyst 2015, 140, 8127. (13) Buckley, K.; Atkins, C. G.; Chen, D.; Schulze, H. G.; Devine, D. V.; Blades, M. W.; Turner, R. F. B. Analyst 2016, 141, 1678.

G

DOI: 10.1021/acs.analchem.6b03548 Anal. Chem. XXXX, XXX, XXX−XXX