Prior studies have shown the improved ability to identify artists’ pigments by combining results from X-ray fluores- cence (XRF), which provides elemental information, with reflectance spectroscopy in the visible to near infrared (400–1000 nm) that provides information on electronic transitions. Extending the spectral range of reflectance spectroscopy into the UV, 350–400 nm, allows identification of several white pigments since their electronic transi- tions occur in this region (e.g., zinc white and rutile and anatase forms of titanium white). Extending the range further into the infrared, out to 2500 nm, provides information on vibrational transitions of various functional groups, such as hydroxyl, carbonate, and methyl groups. This allows better identification of mineral-based pigments and some paint binders. The combination of elemental information with electronic and vibrational transitions provides a more robust method to identify artists’ materials in situ. The collection of both sets of spectral information across works of art, such as paintings and works on paper, allows generating a more complete map of artists’ materials. Here, we describe a 2-D scanner that simultaneously collects XRF spectra and reflectance spectra from 350 to 2500 nm across the surfaces of works of art. The scanner consists of a stationary, single pixel XRF spectrometer and fiber optic reflectance spectrome - ter along with a 2-D position-controlled easel that moves the artwork in front of the two detection systems. The dual- mode scanner has been tested on a variety of works of art from illuminated manuscripts (0.1 × 0.1 m ) to paintings as large as 1.7 × 1.9 m . The scanner is described and two sets of results are presented. The first is the XRF scanning of a large warped panel painting by Andrea del Sarto titled Charity. The second is a combined XRF and reflectance scan of Georges Seurat’s painting titled Haymakers at Montfermeil. The XRF was collected at 1 mm spatial sampling and the reflectance spectral data at 3 mm. Combining the results from the data sets was found to enhance the identification of pigments as well as yield distribution maps, in spite of the relatively low reflectance spatial sampling. The elemental and reflectance maps allowed the identification and mapping of lead white, cobalt blue, viridian, ochres, and likely chrome yellow. The maps also provide information on the mixing of pigments. While the reflectance image cube has 10–20× larger spatial samples than desired, the elimination of having to use two hyperspectral cameras to cover the range from 400 to 2500 nm makes for a low cost dual modality scanner. Keywords: Reflectance imaging spectroscopy, MA-XRF, X-ray fluorescence, FORS, Pigment mapping *Correspondence: firstname.lastname@example.org Department of Scientific Research, National Gallery of Art, Washington, DC 20565, USA Full list of author information is available at the end of the article © The Author(s) 2018. This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creat iveco mmons .org/licen ses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creat iveco mmons .org/ publi cdoma in/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Delaney et al. Herit Sci (2018) 6:31 Page 2 of 12 collection time of each modality is long. For example, Background 2 2 scanning an area of 1 m at 1 mm spatial sampling, and Fiber-optic reflectance spectroscopy (FORS) and X-ray 100 ms integration time would take approximately 28 h. fluorescence (XRF) spectroscopy are non-invasive point- Each additional point-based modality would require an based analytical methods that can be utilized in artists’ additional 28 h if collected at the same spatial sampling material identification. Both FORS and XRF spectros - and integration time. By collecting data from each point- copy are common and valuable tools in the analysis of based modality at the same time, the total collection time artists’ materials. FORS provides information regarding is significantly reduced. the electronic and vibrational transitions of molecules, The decision to co-collect reflectance spectra is based while XRF spectroscopy can be used to identify elemen- on a long demonstrated synergy between XRF, which tal composition . Both, however, have the same limita- provides elemental information, and reflectance spec - tions of other point-based analytical methods: (1) using troscopy, which provides molecular information (i.e. materials identified at one point to infer the presence electronic transitions and absorption features associ- of materials at another point can lead to errors, and (2) ated with vibrational modes) (1, 5–7). In these studies a point measurements do not show the spatial distribu- registration took place between the data collected from tion of materials. These limitations can be addressed by the XRF scanner and independently obtained reflectance constructing a scanner, in which each modality is moved image cubes collected using high efficiency (640–1000 in front of a painting, and spectra are collected at regu- spatial pixels along the slit) and high spatial sampling larly-spaced intervals. XRF imaging spectroscopy point hyperspectral cameras (0.16–0.3 mm /pixel). While the scanners, also known in the cultural heritage field as high spatial information provides useful information, Macro-XRF instruments, have been developed and used especially related to compositional changes, preparatory to produce two-dimensional elemental maps of paint- sketches and painterly techniques, the spectral informa- ings, stained glass windows and works on paper [2–4]. tion from limited spatial resolution systems are still of Separately hyperspectral cameras have been configured great value if the signal to noise ratio in the spectral data to provide reflectance image cubes of works of art to that is collected is large enough. Many remote imaging assist in pigment identification and mapping . spectrometers operate under these conditions, featuring At the National Gallery of Art, Washington, D.C., an sub-pixel extraction of spectral signatures of different XRF imaging spectroscopy scanner was constructed in materials within a single pixel. 2014 based on the design of the University of Antwerp’s For the dual-modal scanner a point-based FORS spec- scanner [2–4] but with one important modification. tral radiometer was chosen to cover the 350–2500 nm at Rather than moving the X-ray source and detector, the high spectral sampling (1 and 2 nm). The rationale for National Gallery of Art (Gallery) scanner utilizes a com- collecting spectra over this full range is because elec- puter-controlled, position control loop, two-axis easel to tronic transitions in the near UV/visible spectral region move the artwork, while the X-ray source and detector can be used to identify numerous inorganic and organic remain stationary. This allows large areas to be scanned pigments. The spectral features in the 1000–2500 nm (1.7 × 1.4 m ). This scanner has been used to collect XRF near-infrared region provide information on vibrational spectra and make elemental maps of several paintings transitions (combination and overtone features) of vari- [5–9], although this work was done with a lower power ous functional groups, such as hydroxyl, carbonate and X-ray source having a 65-micrometer spot size that has methyl groups, allowing for more complete identification been replaced as described here. Since the original devel- of pigments and even paint binders . opment of the Gallery’s XRF scanner, a variety of non- The information provided by FORS spectra (350– commercial, XRF scanners for cultural heritage science 2500 nm) nicely complements the elemental information have been developed based on laboratory X-ray sources. provided by XRF spectroscopy. At least two studies have Ravaud et al.  have developed a highly portable XRF shown the utility of collecting and making reflectance scanner without sacrificing performance. More advanced and elemental maps to identify artists’ materials either by scanners, such as one developed by Romano et al. , fusion of the data through automatic image registration which provide high scan rates with confocal X-ray optics, and direct comparison of reflectance and XRF spectra and on-the-fly fitting of the XRF spectra have been , or by fusion of pigment maps generated automati- constructed. cally and separately from the XRF and reflectance spec - One of the motivations in building a position-con- tral features . In the latter case, an improved final trolled easel that moves the artwork rather than the pigment assignment was achieved when the independ- detection system is to take advantage of multiplexing. ent pigment assignment maps from XRF and reflectance That is, co-collection of multiple point-based imaging spectroscopic imaging were mathematically combined. spectroscopy modalities is more efficient if the spectral Delaney et al. Herit Sci (2018) 6:31 Page 3 of 12 The construction of an integrated scanning system that The FORS collection system has an optical fiber bun - utilizes both point-based modalities will be discussed dle that feeds three spectrometers within the FS3 spec- here. tral radiometer and is mounted normal to the painting surface. The optical fiber is 1 mm in diameter with a Experimental divergent collection angle of 0.43 rad (25°) thus giving a Physical setup spot size of ~ 3 mm in diameter at a standoff distance of The National Gallery of Art integrated scanner consists 4 mm. The illumination system uses a tungsten halogen of a high-precision, two-axis, computer-controlled easel, light source that diffusely illuminates a 2 × 3 cm ellipti- and point collection XRF and FORS instruments. The cal spot at 45° from the painting normal, which can be easel was constructed by SmartDrive, UK (now SatScan baffled down to ~ 1 cm . The light level is approximately Ltd) and has two linear encoders that provide x, y posi- 5000 lux and the total exposure time is approximately tional information and allow continuous scanning of 7 s at any scanned spot on the painting. When inter- a 1.7 × 1.4 m area with step size of 0.01 mm. The XRF nally triggered, or free running, the spectral radiometer instrument consists of a rhodium X-ray source (XOS, acquires spectra at 10 Hz. However, when externally trig- East Greenbush, NY, USA) with converging polycapillary gered, the collection and saving of spectra is limited to optic, and a silicon drift detector with a digital pulse pro- 1 Hz. For the conditions here the spectrometer was trig- cessor (Vortex-90EX, SII DPP, Hitachi High Technologies gered at 1 Hz which allowed the averaging of four spectra Science America, Inc.). The FORS instrument consists each having a 134 ms integration time along with the sav- of an optical fiber spectroradiometer (ASD FS3) and an ing of the average spectrum. The light source has a peak external tungsten halogen lamp source (Plant Probe), spectral radiance of 0.18 W/m -sr-nm at 1000 nm. The both from Malvern PANalytical, USA. standard deviation from a 99% diffuse reflectance stand - The configuration of the XRF and FORS sources and ard with 134 ms integration time and 4 averages over a detectors are shown in Fig. 1. An attenuated diode laser small area (~ 30 pixels) in units of reflectance were 0.57% (Thor labs, NJ) and camera with a 25 EFL mm lens is standard deviation from 350 to 400 nm, 0.05% from 400 used to monitor the area being scanned as well as the to 1700 nm and 0.14% from 1700 to 2300 nm. distance between the artwork and the XRF detector. The The computer controlled easel system consists of a 2-D top-level system performance detection parameters for aluminum easel, controller, and GUI running on a desk- both modalities are summarized in Table 1. top computer. The easel is constructed from 8 × 4 cm In the initial configuration the X-ray source used was extruded aluminum and is mounted vertically and off - the rhodium tube of ARTAX or Bruker AXS Inc instru- set 1 m from the wall. The painting is mounted on a car - ment operating at 50 kV and 0.5 mA. This source has riage having dimensions of 1.6 by 1.7 m that has three been replaced with a higher current rhodium X-ray exci- independent clamps that hold the painting from the top. tation source also operated at 50 kV but with a current of The painting is in contact with small clips on the front 0.75 mA as noted above. The X-ray beam is aligned nor - of the lower ledge of the carriage to ensure the painting mal to the painting surface, and the X-ray detector is ori- is aligned in a plane that is normal to the XRF source. ented at 45° from the normal. The illumination spot sizes The carriage can accommodate paintings weighting up can be adjusted from 0.1 to 1 mm in diameter separately to 70 kg. The easel allows for continuous scanning of an from the position of the detector. The front edge of the area of 1.7 × 1.4 m (horizontal by vertical). Two opti- XRF detector collection tube is adjusted to be 5–10 mm cal encoders, one vertical and one horizontal provide from the painting surface, depending on the height of positional information at the micron level. Repeat posi- the impasto. Since the distance from the X-ray source tional precision is ± 50 micrometers or better. The easel to the painting surface controls the spot size, the entire is controlled by a GUI which allows the user to specify XRF sensor (source, detector and laser and monitoring the dimensions of the area of the art work to be scanned camera) is mounted on a translational table with manual and what scan mode to use; “step and hold” for framing micro-motion adjustment to set the separation between array cameras, “line scan” for push-broom hyperspectral the XRF sensor and the painting. Since the system is used scanners and “raster scan” for single pixel sensors such only for ‘qualitative mapping’ the integrated count rate as the XRF and FORS sensors described here. In raster is the key performance metric to compare system effi - scan mode the sample size at the painting is needed along ciency. Using a Röntec bronze standard (Cu 76.8%, Sn with the scan rate; for reasons of efficiency, scanning is 8%, Pb 12%, Zn 1.1%, Sb 0.5%, As 0.02%) an integrated done in a boustrophedonical (serpentine) manner. In all rate of 900 k counts in 1 s was measured for the Rh tube scan modes, a text file is produced by the easel software set to a 1 mm diameter spot size, 50 kV and 0.75 mA. containing all the x, y positions (in the plane of the art work) at which the triggers to the camera or instruments Delaney et al. Herit Sci (2018) 6:31 Page 4 of 12 Fig. 1 ( Top) a photograph of the view from the painting of the XRF and FORS point measurement collection head showing the X-ray source, X-ray detector and fiber optic reflectance sensor and light source. (Bottom) a photograph showing the painting on the computer controlled easel along with the diffuse white standard and the XRF and FORS instruments were sent. This positional information allows the spatial the trigger TTL pulse based on 2-D positions and not on re-assembly of the image data. Since the positional infor- time stamp data, for the collection of the FORS and XRF mation is from two linear encoders physically attached to spectra. The XRF detector used here requires a software the easel frame, it is absolute, not relative. trigger; a microcontroller (Arduino ATmega2560) was A systems-level block diagram of the XRF and FORS programmed to send a signal to the desktop computer scanner is shown in Fig. 2. The easel controller provides via a serial (USB) connection. A second microcontroller Delaney et al. Herit Sci (2018) 6:31 Page 5 of 12 Table 1 Detection parameters Mode Spectral range Spectral sampling Spectral response Spatial sampling XRF and FORS XRF only integration FWHM diameter (mm) integration times (s) times (s) Reflectance 350–2500 nm 1.4, 350–1050 nm 3 @ 700 nm 3–5 1 NA 2, 1000–2500 nm 10 @ 1400 nm XRF 1.6–25 keV 13.7 eV/channel 165 eV @ Mn K alpha 0.1–1 0.33 0.01–2 Fig. 2 Block diagram showing the key instruments and interfaces of the XRF and FORS scanner (Arduino ATmega2560) for FORS triggering is used to interpolation function is used to resample the raw data divide down the triggers output from the easel given that to a regularized (x, y) grid. Since the ideal interpolator for the FORS spectrometer collection rate is limited to 1 Hz. such a case would be a sinc function, the use of a bi-cubic Typically the XRF scanner is operated at 3 Hz and the function introduces little error. FORS spectrometer at 1 Hz. The final XRF and reflectance image cubes can be Two graphical users interfaces (GUI) are used to set up exploited directly. However, the determination of ele- and run the collection of the XRF and reflectance spec - mental maps from the XRF data cube, more specifically tral data sets. The easel GUI sets the spatial positions emission maps, obtained from integrating the counts to scan the artwork, along with the desired pixel size at from each detected emission peak has been found to be the painting and scan rate (i.e. pixel size at the artwork most useful [12–14]. To ensure the broad background divided by the XRF integration time). This can be varied and the response function of XRF detectors, which allow from ~ 1 mm/s to 50 mm/s (safety limit). A separate cus- for contributions from more than one emission peak to tom GUI for monitoring the XRF spectra collection dis- be collected in a specific energy bin, do not lead to errors plays the sum spectrum at the end of each scan line and in the maps due to elemental overlaps, various fitting a map of the emerging image from the intensity at one strategies and software have been developed [12–14] and emission energy value. The vendor supplied FORS GUI is are commonly used. used to see each average spectrum saved. As the National Gallery of Art’s XRF scanner was ini- tially developed in 2014, more empirical fitting software Formation of XRF and reflectance image cubes was programmed at the time in which each XRF spec- The raw XRF and FORS image cubes are created directly trum is fit to a sum-of-Gaussian-peak group functions from the acquired spectra and from the positional infor- . While such an approach is not new, it has certain mation obtained from the scanner. As in all scanning sys- tems where the scanner does not stop for each spectral acquisition operation, the actual position at which the The XRF spectral fitting and map-making software tool as well as the con- data is acquired varies with scan speed. Thus, the samples trol GUI for collecting the XRF cubes is available as open source software may not be collected on a regular fixed grid. A bi-cubic and can be obtained by emailing the corresponding author. Delaney et al. Herit Sci (2018) 6:31 Page 6 of 12 advantages of being fast and requiring little user interac- having the characteristic feature/s of the pigment of tion. The user needs to identify the elements expected, interest. Typical threshold angle values (i.e. the values obtained by examining the raw cube and using a priori below which a “close match” is assigned) ranged from knowledge of the pigments available in the time period 0.02 to 0.12 radians. Such maps can be compared with the artwork originates from. Since the energy of each ele- the XRF elemental maps in order to assign pigments to ment’s emission lines is known (Ε ), and the XRF detec- them. Note, the limiting angle is found to vary for each tor response function width as a function of peak energy reflectance spectrum mapped. is also known (ω , the peak full width at half maximum The reflectance maps presented here were calculated (FWHM)), the position and width of the Gaussians are using SAM with a portion of the spectral range that considered to be known a priori. Only the amplitudes contained the key spectral feature/s characteristic of the (A ) and offsets (C ), which represent the local baseline material to be mapped. The list of pigments and spectral g g below the peak/s, are unknown and best-fit values for ranges are as follows: lead white 1435–1455 nm, ver- these parameters need to be determined. To ensure a milion 550–630 nm, red ocher 410–904 nm, cobalt blue robust fit, however, the width of the response function 640–1865 nm, and viridian 450–1000 nm. (ω) is also fitted, with the simplification that ω is locally constant over narrow energy ranges (we fit a single value, Results and discussion rather than one value per element, in regions of overlap). Two examples of hyperspectral image cubes collected u Th s these values are determined by nonlinear regression with the dual-mode scanner are presented. The first of the function, involves the collection of only XRF spectral image cubes from a large warped panel painting and the second a −4 ln (2) E−Eg ( ) small panel painting where both XRF and reflectance f E, A , ω, C; E = A e + C 1..G g g g spectral image cubes were recorded. g=1 Elemental maps were obtained from Andrea del Sarto’s (1) Charity, (before 1530), in the collection of the National to the spectral data, where G is the number of emission Gallery of Art, Washington, DC (Fig. 3). The painting peaks in the spectral region being fit. is 1.2 × 0.93 m in size and is on an un-cradled thick XRF elemental image maps for each emission line rep- wooden panel, which is warped and is roughly barreled resent the area under the fitted Gaussians to the emission shape, curving away towards the edges and corners. The peaks. The presence of pigments are typically inferred panel was mounted vertically on the easel, and held in based upon the elemental composition, and also using place by three vertical clamps at the top of the painting. the visual color to guide the assignment. The painting is in contact with three small clips on the Since the FORS instrument was calibrated to apparent lowest ledge of the easel carriage. This allowed the paint - reflectance using a diffuse 99% white reflectance stand - ing to remain in the same place for the collection period ard prior to the data collection, no further calibration required. Over the area of interest, 0.7 × 0.72 m , the was required before making material maps. Maps of pix- high point of the painting occurs in the center and the els with similar reflectance spectra are made using algo - panel curves away by − 6 mm at the bottom left corner rithms such as the spectral angle mapper (ENVI) as done and − 5 mm on the bottom right corner (z-axis direction, here. In brief, the reflectance spectrum at each pixel can normal to the painting). be represented as a point in n-dimensional space where A sensitivity test using the copper peak (Cu-Kα) inten- n is the number of spectral bands in the spectral region sity from a spot on the painting was used to determine being analyzed. A vector can be drawn from the spec- the change in signal with displacement of the XRF sensor trum point through the origin. The spectral angle mapper in the z-axis. At the optimal z-position a count peak rate (SAM, ENVI Software) routine returns the angle between of 28 k counts per second (cps) was measured. A set of the spectra of each pixel in the image cube and the refer- measurements when z was varied yielded a slope of 1589 ence spectrum. A small angle means a close match and a cps per mm displacement in z. The count rate increased high intensity value in the SAM false color images. The when moving towards the painting and decreased when limiting angle for the SAM images was determined by moving away from it. Setting an acceptable variation examining the histogram of angles returned by the SAM of ± 10% meant that the areas scanned should remain routine. A typical threshold angle was set to be on the ± 2 mm from the nominal position in z. Six rectangles rising edge of the histogram or the first peak of a multi- of different dimensions on the painting were determined peaked histogram. The spectra of the pixels obtained and scanned over several days with the nominal z posi- from these settings were examined to ensure the selected tion adjusted for each. The data collection was done with angle, used by the SAM routine, identified only spectra a 1 mm diameter spot size and 100 ms integration time, Delaney et al. Herit Sci (2018) 6:31 Page 7 of 12 Fig. 3 (Left) color image of Andrea del Sarto’s Charity, Samuel H. Kress collection, National Gallery of Art, Washington D.C. Photo Mr. G. Williams, (middle) Copper (Cu Kα) map, (right) inverted natural log of Iron Kα map. Dark areas denote where iron was found. Note in the Cu Kα map the head of a man can be seen in reserve (red arrow), in the area of the chest and chin of the boy in the middle of the painting or a scan speed of 10 mm/s. Because the easel provides cube and XRF image cube were constructed. A false- a text file of absolute positions in x, y of all the spectra color image obtained from the reflectance image cube is collected, the six data collections could be directly com- given in Fig. 4, along with reflectance and XRF spectra bined to make the full image cube . from the same site, i.e. the shoulder of the woman in the Elemental maps (Fig. 3) calculated from the image cube blue shirt. These two point-based spectra demonstrate using Eq. 1 show a good signal to noise ratio was obtained the complementary nature of reflectance and XRF spec - by the scanning system, giving rise to uniform integrated troscopies. The XRF spectrum is dominated by elemen - maps. Of particular interest in the XRF elemental maps is tal emission from lead and cobalt, with less intense XRF the reserve left for the head of a man in the copper map emission from iron and chromium. The associated reflec - (Fig. 3) not apparent in the finished painting. The reserve tance spectrum confirms the blue color is dominated by appears to be behind the chest and chin of the central cobalt blue . Specifically, the broad absorption in the child. The painting was XRF scanned along with the col - near infrared from approximately 1160–1600 nm as well lection of an infrared reflectogram (150 dpi) for com - as small absorption features in the visible (approximately parison with a del Sarto painting titled Borgherini Holy 545, 585, and 625 nm) is characteristic of cobalt blue. Family, in the collection of Metropolitan Museum of Art. These features originate from the ligand-field transitions In that painting, St John the Baptist is present as a man between the d–d orbitals of Co(II) in a pseudo-tetrahe- whereas in Charity he is a young boy. Comparison of the dral configuration. Cobalt blue does not absorb much XRF maps and infrared reflectograms of both paintings light between 700 and 1000 nm and thus displays a sharp show that these works are closely related as is discussed increase in reflectance at around 690 nm with a local by Bayer et al. . reflectance maximum near 850 nm. In the reflectance As a second example, a portion of the painting Hay- spectrum shown, however, a broad absorption from 660 makers at Montfermeil (total painted area 24.8 × 15.6 cm, to 1120 nm indicates that another pigment is present. c. 1882) by Georges Seurat was scanned (17.7 cm Iron ochre and/or umber absorb in this spectral region, width × 16.8 cm height) and both XRF and reflectance and are likely affiliated with the small iron peak seen in spectra were collected. The dimensions of the area to be the XRF spectrum. The strong near-infrared absorption scanned, the spot size (1 mm dia. for XRF) and the scan of cobalt blue in the reflectance spectrum nearly masks rate (3 mm/s) were entered into the easel GUI control- the weak absorption at 1445 nm, typical of the hydroxyl ler along with the positional starting point. The total scan group of basic lead white , unlike the XRF spectrum, time was 3 h. Using the positional information from the in which the lead peaks were the most intense. Analyzing computer-controlled easel, both the reflectance image the information from both modalities provides a more Delaney et al. Herit Sci (2018) 6:31 Page 8 of 12 Fig. 4 ( Top left) detail color image of Georges Seurat’s Haymakers at Montfermeil (1882). Collection of Mr. and Mrs. Paul Mellon, 2014.18.48, National Gallery of Art, Washington, D.C. Photo Mr. G. Williams, (Bottom left) false-color image (450, 550, 620 nm) created from the FORS cube, ( Top right) single spectrum from the scanned XRF data (at the point marked with a red dot), (Bottom right) single spectrum from the scanned FORS data (at the point marked with a red dot) complete picture of the pigments used (in this example, As shown below, many of the ‘dark’ areas in the reflec - cobalt blue, lead white and an iron oxide pigment). tance map correspond to the blue-colored areas of the Analysis of both the XRF and reflectance image cubes artwork. These blue areas are painted with cobalt blue, revealed evidence supporting that lead white was used which has a strong absorption in the same near-infrared throughout the entire painting. The XRF Pb-L map spectral region as basic lead white. Thus, the presence (Fig. 5b) shows the presence of lead across the painting. of cobalt blue may hinder the detection of lead white if The lead observed here is likely to be associated with using only reflectance spectroscopy. The reflectance commercial lead white (mostly basic lead(II)-carbonate) spectra and maps are still useful though, because when or possibly red lead. The reflectance spectrum of basic the 1446 nm absorption feature is detected, it confirms lead(II)-carbonate has a characteristic narrow absorp- that the XRF-detected lead is present as lead white. In tion feature at 1446 nm from a vibrational overtone asso- addition, the reflectance spectra and maps provide a bet - ciated with the hydroxyl group  (Fig. 5c). A map of ter indication of the pigments mainly responsible for the this hydroxyl feature (Fig. 5d) shows the presence of lead observed color of the painting, which is more difficult to white in many of the same areas as the XRF map. For discern if only looking at the XRF elemental maps. Note example, the white blouse of the woman to the right and that zinc white was ruled out as a possible white pigment, the lighter blue passage of the skirt of the woman to the as neither x-ray fluorescence of zinc nor a reflectance left are painted using lead white. The areas in the painting transition edge near 380 nm (indicative of zinc white’s having little or no lead white correspond to areas in the band gap absorption) was detected. This is an advantage reflectance map of which the reflectance spectra show of extending the reflectance scanner to 350 nm as zinc little or no absorption at 1446 nm (Fig. 5d). The lack of white and both forms of titanium dioxide can be detected absorption at 1446 nm in the reflectance spectra is either on the surface of the painting . due to a lack of basic lead white or the presence of a more The iron (Fe) elemental map (Fig. 6b) shows the pres- strongly absorbing pigment in this spectral region. This is ence of iron in the leaves on the ground and on the an advantage of having both modalities, as the XRF map shadowed skirt of the woman in the blue shirt. Since shows lead, presumably from lead white, across many of iron is used in ochres and umbers, further specificity the spatial areas that appear dark in the reflectance map. can be obtained by looking for characteristic absorption Delaney et al. Herit Sci (2018) 6:31 Page 9 of 12 Fig. 5 a Detail color image, (b) The XRF elemental map of lead (Pb) L-alpha emission, (d) reflectance map of hydroxyl feature of lead white at 1446 nm, where “white” and “gray” areas indicate the presence of the 1446-nm lead white absorption and “black” areas indicate the lack of this spectral feature. Map obtained using SAM algorithm and equalization of histogram applied. c Reflectance spectra from “white” (bottom of plot) to “black” (top of plot) areas in reflectance map features in the reflectance spectra. The reflectance spec - shirt does contain mercury and the associated reflec - trum (Fig. 6d, green spectrum) has a reflectance transi - tance spectrum shows a sharp transition edge at 585 nm tion edge at around 560 nm and broad absorptions near indicative of vermilion (Fig. 6d, red spectrum). It should 640 and 850 nm that result in a local reflectance peak at be noted that the transition edge of vermilion has been 760 nm, which is characteristic of the iron oxide hematite found to vary from 580 to 600 nm, but it is sharp (with [1, 6, 19]. Using this reflectance spectrum to query the a first derivative FWHM of less than 40 nm) . The red cube for spectral matches from 400 to 900 nm with the areas in the reflectance map (Fig. 6e) are a match to the spectral angle mapper algorithm shows that the reddish reflectance spectrum of vermilion and occur for the same brown leaves in the foreground were painted with hema- hand that was identified in the XRF data as containing tite (Fig. 6e, green). mercury. Some of the slight discrepancies between the Interestingly, the elemental map (Fig. 6c) for mercury XRF elemental maps and the reflectance maps may occur (Hg), likely from vermilion (HgS), shows little mercury is because the XRF maps can show the distribution of pig- associated with the reddish leaves, and in the few areas ments below the surface (depending on matrix effects), of the leaves where some is detected, the associated whereas the reflectance maps show pigments that are reflectance spectrum is dominated by the spectral fea - closer to the surface because reflectance spectral features tures of hematite. The hand of the woman in the white in the visible were used for mapping. Delaney et al. Herit Sci (2018) 6:31 Page 10 of 12 Fig. 6 a Detail color image, XRF elemental maps of (b) iron, Fe, and (c) mercury, Hg, (d) reflectance spectra containing hematite (green), vermilion (red) and associated (e) false-color map obtained using the SAM algorithm with the intensity values linearly stretched Fig. 7 a Detail color image, XRF elemental maps of (b) cobalt, Co, and (c) chromium, Cr, (d) reflectance spectra of cobalt blue (blue) and viridian (green) and associated (e) false-color image map obtained using the SAM algorithm with the histogram equalized Delaney et al. Herit Sci (2018) 6:31 Page 11 of 12 Fig. 8 a Detail color image of the trees at the top of the painting. b False-color image (RGB, 1250, 850, 650 nm), c reflectance spectra of viridian (yellow), cobalt blue (blue), and the combination of viridian and cobalt blue: less cobalt blue to viridian (red), more cobalt blue than viridian (black). Relatively pure viridian was used for the grass, but a mixture of viridian and cobalt blue was used for the trees. The arrow points to the absorption shoulder characteristic of viridian. XRF elemental maps of d cobalt, Co, and e chromium, Cr. The cobalt (Co) elemental map (Fig. 7b) reveals a wide Co-containing pigment used in these regions is cobalt use of a cobalt-based pigment or pigments in the com- blue. position; for example, in the blouse of one woman and The chromium (Cr) elemental map (Fig. 7c) matches up the skirt of another, as well as areas in the leaves, grass, with the green field, a few green leaves among the red- and trees at the top edge of the painting. Possible cobalt- brown leaves, and into the tree line where cobalt was also based pigments that could have been used include cobalt detected. The possible green pigments include chromium blue, cerulean blue, and even smalt, although this would oxide and viridian (hydrated chromium (III) oxide). The have been unlikely in the nineteenth century. Since no tin reflectance spectra from the green field have an absorp - (Sn) was found in the elemental maps, cerulean blue is tion feature at 705 nm and a sharp inflection point at not present, but smalt or cobalt blue are possible. Each 750 nm, which matches well with viridian green and of these pigments has a distinct reflectance spectrum not with chromium oxide . The reflectance map for associated with electronic transitions in the near infrared viridian green shows its presence in the grass but not in . The characteristic reflectance spectra from the deep the greenish portion of the trees at the top of the paint- blue-colored areas of the painting (Fig. 7d) have a broad, ing where Cr was detected in the XRF map. This material near-infrared absorption from ~ 1227 to 1506 nm that discrepancy in the trees was further examined in Fig. 8. indicate a match to cobalt blue. The spectra also show Examination of a false-color image (Fig. 8b) constructed weak visible absorption bands associated with ligand- from the reflectance image cube reveals that the blue- field transitions of cobalt blue as well (546, 582, 623 nm). green trees at the top of the painting are mainly painted The reflectance map for cobalt blue (Fig. 7e) shows a with a mixture of cobalt blue and viridian (red and black good match to the cobalt elemental map indicating the spectra in Fig. 8c). Delaney et al. Herit Sci (2018) 6:31 Page 12 of 12 References Conclusions 1. Delaney JK, Ricciardi P, Glinsman LD, Facini M, Thoury M, Palmer M, de la This paper presents the design and construction of an Rie ER. Use of imaging spectroscopy, fiber optic reflectance spectroscopy, imaging spectroscopic scanner capable of collecting XRF and X-ray fluorescence to map and identify pigments in illuminated manuscripts. Stud Cons. 2014;59:91–101. and FORS image cubes. Because the illumination sources 2. Alfeld M, Janssens K, Dik J, de Nolf W, van der Snickt G. Optimization of and detectors remain immobile while instead the paint- mobile scanning macro-XRF systems or the in situ investigation of histori- ing is moved, multiple point-based spectroscopy modali- cal paintings. J Anal At Spectrom. 2011;26:899–908. 3. Van der Snickt G, Legrand S, Caen J, Vanmeert F, Alfeld M, Janssens K. ties can easily be added or tested. While higher spatial Chemical imaging of stained-glass windows by means of macro X-ray resolution and shorter collection times can be obtained fluorescence (MA-XRF) scanning. Microchem J. 2016;124:615–22. with hyperspectral reflectance imaging cameras or high 4. Alfeld M, Pedroso JV, van Eikema Hommes M, van. der Snickt G, Tauber G, Blass Haschke JM, Erler K, Dik J, Janssens K. A mobile instrument for flux X-ray sources, this paper shows the utility of a mod - in situ scanning macro-XRF investigation of historical paintings. J Anal At erate spatial resolution imaging system to map the chem- Spectrom. 2013;28:760–7. ical composition of painted works of art in a museum 5. Dooley KA, Conover DM, Glinsman LD, Delaney JD. Complementary standoff chemical imaging to map and identify artist materials in an early environment. Moreover, this approach eliminates the Italian renaissance panel painting. Angew Chem Int Ed. 2014;53:13775–9. need to purchase two hyperspectral reflectance cameras 6. Delaney JK, Zeibel JG, Thoury M, Littleton R, Palmer M, Morales KM, de that operate together to cover the spectral range from la Rie ER, Hoenigswald A. Visible and infrared imaging spectroscopy of Picasso’s Harlequin musician: mapping and identification of artist materi- 400 to 2500 nm since the FORS instrument used here is als in situ. Appl Spectrosc. 2010;64:584–94. sensitive over this entire spectral range. Although hyper- 7. Dooley KA, Coddington J, Krueger J, Conover DM, Loew M, Delaney JK. spectral reflectance cameras can provide < 0.2 mm spatial Standoff chemical imaging finds evidence for Jackson Pollock’s selective use of alkyd and oil binding media in a famous ‘drip’ painting. Anal Meth- sampling with 500× to 1000× faster collection speeds, ods. 2017;9:28–37. the relatively higher cost of the imaging cameras may be 8. Conover DM. Fusion of reflectance and X-ray fluorescence imaging spec- prohibitive for many institutions. Thus using the FORS troscopy data for the improved identification of artists’ materials. Ph.D. thesis. The George Washington University (Summer 2015). instrument as the reflectance modality makes for a rela - 9. Jackall Y, Delaney JK, Swicklik M. ‘Portrait of a Woman with a Book’: a tively lower cost, dual imaging modality XRF and FORS ‘Newly Discovered Fantasy Figure’ by Fragonard at the National Gallery of scanner. Art, Washington. Burlingt Mag. 2015;157(1345):248–54. 10. Ravaud E, Pichon L, Laval E, Gonzalez V, Eveno M, Calligaro T. Develop- Authors’ contributions ment of a versatile XRF scanner for the elemental imaging of paintworks. DMC, JKD and ML designed and implemented the FORS scanner. DMC, LDG, Appl Phys A. 2016;122:17. JKD and KJ designed and built the XRF scanner. DMC developed and wrote 11. Romano FP, Caliri C, Nicotra P, Di Martino S, Pappalardo L, Rizzo F, Santos the GUIs. JKD, KAD, DMC conducted the experiments, and JKD, DMC, KAD, HC. Real-time elemental imaging of large dimension paintings with a LDG did the data analysis. JKD and DMC, with help from the coauthors, wrote novel mobile macro X-ray fluorescence (MA-XRF) scanning technique. J the paper. All authors read and approved the final manuscript. Anal At Spectrom. 2017;2:773–81. 12. Alfeld M, Janssens K. Strategies for processing mega-pixel X-ray fluo - Author details rescence hyperspectral data: a case study on a version of Caravaggio’s Department of Scientific Research, National Gallery of Art, Washington, DC painting Supper at Emmaus. J Anal At Spectrom. 2015;30:777–89. 20565, USA. School of Engineering and Applied Science, George Washington 13. Vekemans B, Janssens K, Vincze L, Adams F, Van Espen P. Analysis of X-ray University, Washington, DC 20052, USA. Present Address: Sensors and Elec- spectra by iterative least squares (AXIL)—new developments. X-Ray tron Devices Directorate, U.S. Army Research Laboratory, Adelphi, MD 20783, Spectrom. 1994;23:278–85. USA. Department of Chemistry, University of Antwerp, Antwerp, Belgium. 14. Solé VA, Papillon E, Cotte M, Walter P, Susini J. A multiplatform code for the analysis of energy-dispersive X-ray fluorescence spectra. Spectrochim Acknowledgements Acta Part B. 2007;62:63–8. The authors acknowledge funding from the National Science Foundation 15. Conover DM, Delaney JK, Loew MH. Automatic registration and (Award 1041827). J.K.D. and D.M.C. acknowledge funding from the Andrew mosaicking of technical images of Old Master paintings. Appl Phys A. W. Mellon and Samuel H. Kress Foundations. The authors are grateful to 2015;119:1567–75. David Martin and Dennis Murphy of SmartDrive Ltd., Gary Fager of Malvern 16. Bayer A, Gallagher M, Centeno SA, Delaney JK, Read E. Andrea del Sarto’s PANalytical, and Gao Ning of XOS for advice. KJ acknowledges support from Borgherini holy family and charity: two intertwined late works. Metrop EU-InterReg project SmartLight and from GOA Project SolarPaint (University of Mus J. 2017;52(1):34–55. Antwerp Research Council). 17. Bacci M, Magrini D, Picollo M, Vervat M. A study of the blue colors used by Telemaco Signorini (1835–1901). J Cult Heritage. 2009;10:275–80. Competing interests 18. Bacci M, Picollo M, Trumpy G, Tsukada M, Kunzelman D. Non-invasive The authors declare that they have no competing interests. identification of white pigments on 20th-century oil paintings by using fiber optic reflectance spectroscopy. JAIC. 2007;46:27–37. Ethics approval and consent to participate 19. Clark RN. Spectroscopy of rocks and minerals, and principles of spectros- Not applicable. copy. In: Rencz AN, editor. Manual of remote sensing, remote sensing for the earth sciences, vol. 3. New York: Wiley; 1999. p. 3–58. 20. National Gallery of Art. In-house reflectance spectral database. Washing- Publisher’s Note ton, D.C.: National Gallery of Art. Springer Nature remains neutral with regard to jurisdictional claims in pub- lished maps and institutional affiliations. Received: 23 January 2018 Accepted: 9 May 2018
Heritage Science – Springer Journals
Published: May 30, 2018
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.
All for just $49/month
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.
Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.
All the latest content is available, no embargo periods.
“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”Daniel C.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud
“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”@deepthiw
“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”@JoseServera