Non-destructive and micro-invasive testing techniques for characterizing materials, structures and restoration problems in mural paintings
Accepted Manuscript Title: Non-destructive and micro-invasive testing techniques for characterizing materials, structures and restoration problems in ...
Please cite this article as: {http://dx.doi.org/ This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Non-destructive and micro-invasive testing techniques for characterizing materials, structures and restoration problems in mural paintings Mariagrazia [email protected], Stefano Sfarrab* ##Email##[email protected] ##/Email## ##Email##[email protected]##/Email##, Marco Chiarinic ##Email##[email protected]##/Email##, Valeria [email protected], Giuliana Taglierid ##Email##[email protected]##/Email##, Giorgio Cerichellia ##Email##[email protected]##/Email## aUniversity of L’Aquila, Department of Physical and Chemical Sciences, Via Vetoio (Coppito 1), I-67100, Loc. Coppito, L’Aquila (AQ), Italy, bLas.E.R. Laboratory, University of L’Aquila, Department of Industrial and Information Engineering and Economics, Piazzale E. Pontieri 1, I-67100, Loc. Monteluco di Roio, Roio Poggio, L’Aquila (AQ), Italy, cUniversity of Teramo, Department of Bioscience and Technology for Food Agriculture and Environment, Via Carlo Lerici 1, I-64023, Mosciano Sant’Angelo, Teramo (TE), Italy, dUniversity of L’Aquila, Department of Industrial and Information Engineering and Economics, Piazzale E. Pontieri 1, I-67100, Loc. Monteluco di Roio, Roio Poggio, L’Aquila (AQ), Italy Tel.: +39 0862 434336, fax: +39 0862 431233.
Graphical abstract fx1Highlights► Infrared thermography allowed to identify structural damage and rising damp effect. ► The present approach provided insights on the used pigments and painting techniques. ► FT-IR, XRF and XRD analyses of the mortar sample showed the peculiar composition. ► 1D, 2D NMR analyses were useful for the identification of the restoration polymer. ► NMR technique also allowed to characterize the plasticizing agents. Abstract In this paper, chemical and structural studies of medieval wall paintings in Ocre (L’Aquila, Italy) are presented. During the latest restoration campaign, non-destructive (Near-Infrared Reflectography and Infrared Thermography) and micro-invasive (Nuclear Magnetic Resonance, Fourier Transform Infrared Spectroscopy, μ-Raman, Scanning Electron Microscopy with X-ray Microanalysis, X-Ray Diffraction, X-Ray Fluorescence, Optical Microscopy, Mass Spectrometry, Thermogravimetry) analyses were performed in order to determine the detachments of wall surfaces and the characterization of original and restoration materials. Data integration allowed to reconstruct the conservative history, the execution techniques and the conservation problems of the artefact, as well as to assess the effectiveness of restoration activities adopted. The combined use of physical and micro-chemical techniques proved to be effective for an in-depth study of materials stratification of paintings.
1
Keywords: Non-Destructive testing techniques; micro-chemical analyses; structural defects and rising
1. Introduction The monastery of Santo Spirito in Ocre (L’Aquila, Italy) is a complex of extreme historical and architectural importance as it is the sole example of coenobitic Cistercian architecture substantially intact in the Abruzzo region. The first nucleus of the monastery was built in 1222. Abandoned by the Benedictine order in 1652, it was used as a cemetery from the nineteenth century until 1939. The church stands between the north side of the monastery and the cloister; it consists of a single nave originally covered by an ogival vault, partially replaced by beams after the cave-in, likely occurred in the eighteenth century. The chapel-sacristy is decorated in the lower part with a velarium; above this, frescoes depicting scenes from the Beato Placido’s life are still visible in some areas of the walls. The cycle was supposed to start from the west wall and go further to the south wall. During the 70’s, the detachment of some sixteenth frescoes allowed both the discovery of older layers of confined paintings (indeed, a crucifixion attributed to the thirteenth century is still visible on the southern wall), and of large series of paintings on the eastern wall representing miracles performed by the Beato Placido, not belonging to the cycle and presumably superimposed a few years after the realization of the cycle itself (second half of XIV century) [1]. The historical complex has been the subject of a long and troubled history of interventions for its possible reuse. Unfortunately, an exhaustive outline of the situation regarding the restoration projects performed over the years is lacking to the knowledge of the authors because of the incompleteness of official documents. Since 1978, the site, still unguarded and unused, was the subject of vandalism and theft [2]. In the recent years, the state of preservation of the pictorial cycle appeared rather precarious because of graffiti and a noticeable fallen of plaster and color. In addition, the 2009 earthquake further worsened the situation causing the separation of the northern wall. During the recent restoration campaign completed in 2013 that provided the structural and aesthetic reinstatement of the vestry, diagnostic studies were performed both on the east and on the south walls. Non-destructive and micro-invasive testing techniques were employed with the aim to identify possible structural defects and, simultaneously, to characterize both the original and the subsequent materials applied for restoration purposes. For the sake of clarity of the readers, a summary of the performed analyses is shown in Figure 1.
2. Materials and methods
2
2.1 Non-destructive testing techniques Near-InfraRed Reflectography (NIRR) As an imaging technique, NIRR is a method that allows advanced types of visual analysis [3]. The NIRR method can concentrate its action into narrow-wavelength bands. Among the digital cameras currently in use in order to perform the NIRR method, it exists the Canon 40DH spectrum enhanced with the internal filter named EOS045, wideband spectral and EF55-200 lens of the same company, also equipped with a CMOS sensor 22.2 × 14.8 mm having 10 megapixels (5.7 μm pixel pitch). The latter was used in this work. In order to get the EOS 40DH, the substantial change consisted in the removal of the anti-IR filter, as well as in the subsequent positioning of the Idas Clear/AR filter. Adapting to the geometry of the sensor, this filter allows proper autofocus operation by avoiding, in the same time, the internal reflections effect. In addition, by transmitting all wavelengths between 380–1000 nm, the Idas Clear/AR filter makes the Canon EOS 40DH a suitable tool for the reflectographic acquisitions. This happens when a proper filter is placed in front of the camera lens by excluding the visible radiation. The possibility to perform infrared acquisitions is compatible with the manual setting of other functions, such as the white balance, the opening of the stop, and the variation of the exposure times. The experimental spectral response curve of the Canon 40DH along with its Signal-to-Noise-Ratio (SNR) in relation to the IR signal (i.e., the influence of the filtering phase and the exposure times) were studied before the implementation of the reflectographic campaign. For a sake of brevity, they are not reported herein, although a similar study (useful to the readers for a more in-depth understanding) can be found elsewhere [4]. The lighting of the mural paintings was carried out through two Philips IR 250 W lamps equipped with diffuser soft boxes and symmetrically positioned at 35° angle with respect to the focal axis of the camera. During the reflectographic measurements, a set of bandpass filters working at 680, 715, 760, 850, 950 and 1000 nm were mounted on the external lens of the camera in order to simulate a multispectral inspection. The idea follows a previous publication centred on semi-transparent materials [5], although only the reflectograms acquired at 1000 nm have been included in the present text.
3
The distance between the camera and the mural painting, as well as the distance between each lamp and the mural painting, have been imposed in equal manner during the acquisitions, i.e., ~1.50 m the first one, and ~1.60 m the second one. Infrared Thermography (IRT) The IRT method can be divided into the active, passive and hybrid approaches [6–7]. Recently, the hybrid approach has been introduced in the scientific panorama. In this case, the first one and the second one were used, by considering the purpose of the work centred on the evaluation of the presence of water that afflicts the mural paintings under analysis [8] through a bottom-up effect [9] caused by a porous behaviour of the materials [10]. The present work traces the HOS mode of operation (i.e., Kurtosis) [11–13], by adding the correlation technique [14] since in both cases only one final image summarizes the best features of the entire thermal sequences emerging along the heating up and cooling down phases. In particular, the IRT measurements were conducted by using a long wave (LW) thermal camera made by FLIR (7.5– 13 μm), S65 HS series. The frame rate was set at 1 Hz. The mural paintings were heated by using one 2 kW IR lamp made by STAR Progetti®, and the heating up phase lasted 446 s, followed by a cooling down phase of 543 s. The heating up phase was preceded by the acquisition of one cold image. In practice, a total of 990 thermal images were acquired and processed in Matlab® environment. In two cases, a comparison with the thermal images acquired thank to the passive approach is also added. At this point, it is important to remark: a) how the kurtogram will provide an indication of, first, the location of eventual subsurface defects, and second, their thermal diffusivity [15], b) how the surface temperature of the wall painting reached during the heating up phase a ∆T = 9.5 °C, the surface temperature during the heating up phase was continuously monitored by using virtual spots, i.e., a function available into the software of the thermal camera; they shown the value of the temperatures on the screen of the device, c) right after the heating up phase, any damage was noticed through a visual inspection,
4
how color measurements were not necessary immediately before the use of the active IR thermography method. Indeed, ∆T applied was very low. The information taken from [16, 17] were useful in order to understand how, in this condition, the thermographic measurement was completely nondestructive. In particular, these studies gave us very helpful hints on the reaction of medieval pigments to extended heat conditions in which a laser, instead of a lamp, was used in order to provide a thermal stress, how both a visual inspection and a finger-tapping test preceded the thermographic campaign [18]. The first one was very useful in order to exclude emissivity variations linked to the most important thermal anomalies discussed in section 3, while the second one was used in order to detect possible and very shallow subsuperficial splittings. The thermal camera was focusing thanks to the help of a lukewarm piece of aluminum placed near to the surface to be investigated. As a whole, the examined parts of the mural painting doesn’t have a specific emissivity value due to the surface heterogeneity, therefore, according to the literature, a mean value of 0.75 was used [19]. The environmental conditions were monitored by using a thermo-hygrometer just before the thermographic campaign. The main output parameters recorded were 20 °C and 70% as relative humidity (RH). These values, along with both the distance between the thermal camera and the inspected wall, and the reflected radiation from the surroundings – calculated via the Standard Test Methods for Measuring and Compensating for Reflected Temperature using Infrared Imaging Radiometers (E 1862 – 97), i.e., 20 °C, – were manually inserted into the software of the thermal camera, the technical characteristics of the IR camera, i.e., field-of-view = 20° × 15°, instantaneous field of view = 1.1 mrad, measurement distance = 30 cm to infinity, detector = focal plane array (FPA) uncooled microbolometer VOx, displayed pixel number = 320 × 240, measurement temperature range = - 40 ~ + 1500 °C, accuracy (% of reading) = ± 2 °C or ± 2 %, minimum detection temperature difference = < 0.05 °C with respect to 30 °C blackbody temperature, how the correlation technique was used in this work for the first time to the best of our knowledge, in order to detect moisture in a wall painting; therefore, the basic principles are described in the following.
2.1.1 Correlation Thermography The coefficient of linear thermal expansion (CTE) is a material property that is indicative of the extent to which a material expands upon heating. Different substances expand by different amounts, therefore, the defects inside a mural paintings, such as splitting areas, inclusions, pockets of water, etc. expand autonomously. The correlation coefficients are a solution to characterize the linearity between two signals. In the thermographic context, correlation involves two variables, i.e., the reference (i.e., the sound area) and the variable to be compared with the reference. The effect of offsets is canceled in correlation computation because offsets do not change the linear relationship between two variables. An interesting way to proceed is to compute the reference from a semi-infinite body assuming that the heat transfer can be described with sufficient accuracy using the semi-infinite body model [20, 21]. Therefore, the equation governing the sample surface temperature evolution in time after an energy impulse Q, is:
Tsemi inf inite ( z 0, t ) body
Q e t
1)
where, z is the distance from surface (thus, z = 0 corresponds to the sample surface), t is the time, e is the material effusivity and Q is the input energy. Taking the thermogram obtained at the time t’ before the first defect
5
becomes visible on the surface, the temperature of the sound area at location [i, j], ∆Ts[i,j](t’) can be obtained from Eq. (2):
Q[ i , j ] e[ i , j ]
t ' T[ i , j ] t '
2)
Therefore, sound area temperature can be calculated combining Eqs. (1) and (2) into Eq. (3):
Tsound i , j t
Qi , j ei , j t
t' Ti , j t ' t
3)
Applying Eq. (3) on the whole surface of the region of interest (ROI) of the mural painting, the time evolution of the surface temperature corresponds to the temperature evolution that would be obtained if no defect was present into the ROI. The latter method was used in this work in order to minimize: a) the discretional contribution of the operator, b) the non-uniform heating caused by the applied heat source, and c) the emissivity variations. Finally, to digitize the difference in trend, a defect-free point was selected in order to determine the correlation coefficient r with other pixels of chronological temperature change during the heating process. The correlation coefficient r was calculated by Equation (4): _ _ x x y y i1 i i n
r
n i 1
xi x _
2
n i 1
4)
yi y _
2
where, xi represents the chronological temperature change of the defect-free point, and yi represents the chronological temperature change of each pixel. Further, ̅ and
represent the average values of xi and yi, respec-
tively. The correlation coefficient r means normalizing xi and yi so as to make their averages zero and variances 1, to determine the covariance. The influence of reflection is removed by normalizing the temperature change. The signal processing ignored the time zones right before the beginning and after the cessation of heating, during which the temperature changed stepwise at the defect-free point, but instead used the gradual temperature change during the step heating [22, 23].
2.2 Mineralogical and chemical analyses
6
In order to characterize paint layers that, according to the stylistic analysis were referring to different historical periods, plaster and restoration products, it was necessary to take small samples. According to their typologies, samples were divided into three groups: pigments, original decorated plaster, restoration material (Table 1). Transmission spectra in the IR region of plaster and pigments samples were collected with a PerkinElmer 683 spectrophotometer. Few milligrams of each sample and about 100 mg of anhydrous KBr (IR grade, Merck) were powdered in an agate mortar in order to obtain 13 mm diameter pellets. Twenty-five signal averaged scans were acquired. SEM analyses of pigments were carried out by means of PHILIPS XL30CP microscope. EDS analyses were performed by using OXFORD ENERGY 250 equipped with INCAx-act LN2-free detector. Because of the vacuum conditions, the carbon coating of samples was not required. Mass measurement was performed using a MALDI-TOF spectrometer AB SCIEX TOF/TOF 5800 System. Unless otherwise stated, all starting materials, catalysts, and solvents were commercially available and were used as purchased. For lipids/proteins extraction the Bligh–Dyer [24] method was followed. The surface of sample M was scraped into a fine powder. 500 µL of CHCl3 /MeOH (1:2) were added to 31 mg of sample followed by vigorous vortex-mixing and ultrasonication (20 min). Then, 170 µL of CHCl3 followed by 170 µL of H2O were added and the mixture was vortexed and ultra-sonicated again (20 min) after each addition. Finally, the mixture was centrifuged (10 min at 3000 × g) to facilitate phase separation. The lower organic layer was collected, concentrated under a nitrogen stream, reconstituted into 100 µL of a CHCl3 /MeOH (1:1) solution and prepared for lipid MALDI analysis (matrix used DHB 10 mg mL−1 in MeOH:TFA 0.1%). The upper CH3OH/H2O phase was concentrated in a Speed-Vac apparatus and subjected to enzymatic hydrolysis for peptide analysis. The tryptic digest was obtained by following the digestion protocol. The protein pellet was dissolved in 20 µL of 400mM NH4HCO3 in 8 M urea. 5µL of 50mM dithiothreitol were added and the mixture was incubated for 30 min at 50 °C, chilled, added with 5 µL of 100 mM iodoacetamide
7
and kept in the dark at room temperature for 30 min. Finally, 140 µL of H2O and 1.4 µg of sequence grade trypsin were added and the mixture was incubated overnight at 37 °C. The digests were then stored at −20 °C. Aliquots (10 µL) of sample prepared for peptide MALDI analysis were mixed 1:1 with the matrix (CHCA) solution (10 mg mL−1 in ACN:TFA 0.1%, 50:50). After drying, the spot was washed with 2 µL of water and analyzed. Cross section of sample M was embedded in Buehler Epoxycure bi-component epoxy resin and then polished with diamond suspensions. Microscope observation under UV light (365 nm) was performed using a ZEISS AXIO IMAGER A2 equipped with a LEICA DFC 320 camera. The chemical composition of plaster and pigments portions of sample M were determined using an X-ray fluorescence set up (XRF - X SPECTRO mod. XEPOS III). The sample was ground and sieved (<100 mm); 1.3791 gr of powder were weighed and analyzed using the semi-quantitative method. Mineralogical analysis of sample M were performed on both binder-enriched fraction, aggregate and pigments. Binder and aggregate fractions were separated according to UNI 11305; in particular, the sample was subjected to a dry and gentle manual disintegration, using an agate mortar, taking care not to damage the aggregate fraction. The sample of crushed mortar was then sieved, using a stack of sieves (UNI 2332-1) according to sequence 1-6-10-20-25-29-33-38, until 63 µm mesh size. As far as the binder: aggregate ratio is concerned, although the amount of material available did not meet the standards requirements, the authors proceeded with its determination taking into account sample’s peculiarities related to its cultural and historical value. The aggregate fraction was also characterized by visual analysis (UNI 11176). Pigment and mortar fractions were analyzed by Xray diffraction (XRD) analysis using a PANalytical X’Pert X-ray diffractometer and CuKα radiation. Data were collected in the angular range of 5-80 ° 2theta, characterized by a step size 0,026 °, and a time for step 200 seconds. The experimental diffraction patterns were analyzed using PANalytical Highscore Plus software. The qualitative analysis of the phases was carried out by comparing the experimental pattern with ICDD and international ICSD reference files supplied by the software. Semi-
8
quantitative results were performed too, by means of the Reference Intensity Ratio (RIR) method, which is based upon scaling all diffraction data to the diffraction of standard reference materials [25]. As regards thermo-gravimetric (TG/DTA) analyses, valuable in estimating the water content in a selected ROI as representative of the inspected walls, a LINSEIS L81 apparatus was used. Mortar samples, previously conditioned at different RH values (50%, 70% and 100%), were ground and 100 mg were considered for each measurement. Dynamic experiments were carried out, with a temperature gradient of 20°C/min. from ambient temperature up to 200°C, in a static air atmosphere. Sample A taken from the south wall, in correspondence with the additions of lacunae of paint layer made with mortars, was analyzed after a pre-treatment in a 10% HCl solution for removing residual CaCO3. Micro-Raman spectra of sample A were obtained using a LABRAM spectrometer (HoribaJobin Yvon, λ = 633 nm, 1 μm spatial resolution and spectral resolution of 2 cm−1). The microRaman spectrometer is equipped with a confocal optical microscope (100 × MPLAN objective with a numerical aperture of 0.9 and a measured work distance of 0.15 mm). Laser radiation was focused directly onto the samples, without a preliminary preparation. Infrared spectrum of sample A was collected using horizontal attenuated total reflection (HATR) technique on a PerkinElmer Spectrum One spectrometer equipped with the HATR accessory and a flat ZnSe crystal. The reflection variant of IR analysis was chosen as it allows little or no sample preparation and it proves to be very useful in studying polymers surfaces. Sample A was also characterized by means of NMR spectroscopy. Almost 100 mg of sample was left in 700 µL of deuterated chloroform for 4 days at 303 K, afterwards the filtered CDCl3 was analyzed. 1H NMR and 13C NMR spectra of chloroform extracts were acquired employing a Bruker Avance III operating at 400.13 and 100.6 MHz, respectively. Chemical shifts are reported in ppm (δ) relative to tetramethylsilane (TMS, δ = 0.00) for 1H and referenced to the chemical shifts of residual solvent resonances CDCl3 at 77.0 ppm for 13C. Gradient enhanced heteronuclear experiments (HSQC) have been performed by acquiring a bi-dimensional matrix consisting of 512 × 2048 points
9
with a window of 29000 and 4500 Hz, respectively, for 13C and 1H nuclei; a repetition time of 5 s and 32 scans for each experiment have been used. Diffusion ordered spectroscopy (DOSY) experiments were carried out by carefully choosing a stimulated radio frequency pulse sequence. The measurements were made using a 5 mm direct probe with a z-gradient coil capable of producing magnetic field pulsed gradients of 5.35 G cm−1; DOSY experiments were performed using the bipolar pulse pair longitudinal eddy current delay pulse sequence and 20 spectra (32 transients each) were collected. The values of δ and ∆ were 2.0 ms and 250 ms duration, the eddy current delay (Te) was set to 5 ms in all experiments. The pulse gradients (δ) were incremented from 2 to 95% of the maximum gradient strength in a linear ramp. The temperature was set and controlled to 298 K. After Fourier transformation and baseline correction the diffusion dimension was processed with the Bruker TopSpin 3.2 software package.
3 Results and discussion
3.1 NDT techniques Fig. 2a shows the right wall where micro-samplings were collected while the influence of the rising damp can be observed by looking at the bottom part of the wall into the long wave infrared spectrum in Fig. 2b. Indeed, the temperature of the wall surface is affected by the heat exchange (by convection and radiation) with the surrounding environment and by water mass transfer and evaporation. The energy associate with the mass transfer within the wall (water and salts) is less significant than the energy lost by evaporation. As well known, the amount of the heat carried by a certain quantity of water from a point to another of the wall is approximately two orders of magnitude smaller than the energy required to the same quantity of water to change its state on the surface. In order to obtain a clear and unequivocal dependence between the measured temperature and the evaporative rate, the wall surface was studied at the environmental conditions since it increases the transpiration process. An evaporation causes a cooling of some degrees. This is partially compensated by the increase of convection and conduction from the inner parts of the wall [26, 27].
10
The passive approach also indicates the configuration of the masonry, as well as the presence of a sub-superficial wood beam just below the ceiling. The latter is obviously shown with a higher temperature, by considering the low conductivity of this material. Unfortunately, P3 was not analyzed from the point-of-view of the thermographer due to the presence of a stair which obstructed the field-of-view. Later, reflectographic and active infrared thermographic inspections were performed, by capturing the point of interest as well as an area around it. The first one was preparatory for the micro-chemical analysis described in section 3.2, since it clearly marks some pigments, such as the green and the dark, in an unmistakable way (Fig. 2d) with respect to the visible image (Fig. 2c). The second one confirmed the presence of moisture due to the rising damp phenomenon at least in the mortar layer, previously signaled by the passive approach. In particular, by comparing Fig. 2c with Figs. 2e and 2f, two uncovered zones can be detected on the top and on the bottom of the figures. They are surrounded by irregular dotted lines. A possible explanation could be due to the presence of soluble salts. When the water evaporates from the surface, the soluble salts crystallize in the natural pores of the wall, growing in volume and making fractures in the microstructure of the wall painting materials. This leads to breakage of the superficial layers into flakes, that eventually cause a complete defoliation of the paint layers. In particular, regarding the correlation result (Fig. 2e), the red dotted lines mark the confines between the mortar layer and the painted layer, while the yellow dotted lines mark the confines between the mortar zones affected by moisture and the healthy zones. In addition, Kurtogram result points out four detached zones indicated by arrows. Finally, the cold image gathered before the starting of the heating up phase is added in Fig. 2g. In this way, it is possible for the reader the understanding of the improvement made at the state of the scientific research with the use of HOST and correlation techniques. As explained in [28], it is possible to graphically schematize the different behaviour along the time of two representative zones (Fig. 3). Since evaporative flux was inhibited due to the critical ambient conditions of the frescoed room, it starts when a thermal stimulus is applied. Fig. 3 shows
11
the thermal curves obtained during the heating up phase; they manifest clear differences because the square represent the dry area, while the crosses indicate the damp area (Fig. 2e). Unfortunately, the dependence of effusivity from the water content is not so simple to be singled out. The main problems in the comparison among angular coefficients are due to two contrasting effects. The first one concerns the evaporation flux of the damp area, that introduces a negative term into the energy balance. The second one is the different absorption coefficient due to the water presence that increases the amount of absorbed heating [29]. However, the first insight provided by the combination of advanced algorithms (Figs. 2e,f) and the graph in Fig. 3a was confirmed by thermogravimetric analyses. In particular, mortar samples were firstly conditioned for one week at different relative humidity (RH) values in order to simulate dry, intermediate and moist conditions (RH= 50%, 70% and 100%, respectively) and, subsequently, analyzed. The value equal to 70% was chosen by considering the ambient RH of the vestry at the time in which the sample was collected. Results point out that mortar samples show a slight but appreciable hygroscopic behavior; e.g., the sample maintained at RH condition equal to 100% showed a weight loss lower than 1%. This joined approach can be considered as an independent variable with respect to the presence or not of the painted layer, by considering the recognized ability of the described algorithms to minimize the emissivity variation’s effect [30]. Similarly, the P4_P5 and P6 area, corresponding to the thirteenth century painting on the southern wall (dotted dark rectangle in Fig. 4a), was investigated by using the same approach focused on the non-thermal (Fig. 4b) and thermal parts of the infrared spectrum (Figs. 4c-f). The marks in Fig 4a exhibit the ancient practice of renewing the painted decoration simply superimposing a layer of plaster on the original painting, partially perforated to ensure a better cohesion. Once again the NIRR approach retrieves some particulars, which are difficult to notice in the visible spectrum. Above all, the simplicity to distinguish the dark color with respect to the brown color. Readers can notice the shape of the symbol above the left hand of the character which is indicated by
12
a red arrow (Fig. 4b). Into this field-of-view, two different thermographic acquisitions have been performed. The first one is linked to Figs. 4c-d and puts in evidence some integrated details. In Fig. 4c sub-superficial stones (not observable in Fig. 4d) were detected and surrounded by irregular yellow dotted lines. In addition, by comparing the figures, it is possible to notice two white (in Fig. 4c) or dark (in Fig. 4d) areas surrounded by irregular red dotted lines. The respective scales assure the same peak of intensity into each figure. The reverse behaviour in term of intensity between the techniques establishes the presence of moisture in each sub-superficial cavity [31]. Instead, this is not true for the splitting surrounded by an irregular white dotted line in Fig. 4d. For comparison purposes, a finger-tapping test has been conducted above the three areas and the reverberation is identical (i.e., alive) for the areas surrounded by irregular red dotted lines, while it is different (i.e., dead) for the area surrounded by an irregular white dotted line [32]. It is possible to assume that the splitting is located in the upper layers, since the presence of a sub-superficial stone in the same area was detected in Fig. 4c by using the correlation technique. At the same time, the behaviour of the splitting surrounded by an irregular red dotted line in Figs. 4e-f appears identical to the previous one illustrated in Figs. 4c-d. Also in this case, the correlation algorithm was able to signal the presence of subsuperficial stones, surrounded by irregular yellow dotted lines (Fig. 4e). Surprisingly, a white spot in which the intensity decreases from the inside out, is present in Fig. 4f. Its external area is surrounded by an irregular azure dotted line. Also in this case, it is located above a sub-superficial stone (Fig. 4e) although its sound is ``alive'' after a finger-tapping technique. Therefore, the presence of moisture in the inner part of the splitting can be ascertained. The degradation behaviour of the intensity can be explained taking into account the fact that the surface underwent an active approach by lamps; Fig. 4f summarizes the thermal contribution coming from several thermograms linked to both the heating up phase and the cooling down phase. It is possible to assume that the evaporation phase started from the peripheral part of the splitting by considering the ``thermal imprint'' left at the end of the image processing [33]. Also in this case, the cold images acquired before the starting of the heating up
13
phase are added in Figs. 4g,h. The motivation is the same discussed for the Fig. 2. In addition, by comparing Figs. 4c,d to Fig. 4g, and Figs. 4e,f to Fig. 4h, it is possible to see how the instrumental noise was greatly reduced after the image processing. Fig. 5. North wall: (a) ROI inherent to the reflectographic and thermographic acquisitions, (b) Thermogram recorded by means of the passive approach, (c) Visible image of the ROI, (d) NIRR at 1000 nm of the ROI, (e) Correlation result, (f) Kurtogram result, and (g) the cold image (i.e., the thermal image before the starting of the heating up phase).
For the sake of completeness on the state of conservation of the building structure, the north wall was also inspected (Fig. 5a). Using the passive approach, sub-superficial cracks on the top of the wall which surround the window and cross the right corner, the stonework and the rising damp phenomenon can be detected (Fig. 5b). Taking into account the distance between the inspected wall (Fig. 5a) and the opposite one, it is not possible to apply the HOST approach nor the correlation technique to Fig. 5b. Indeed, Fig. 5b, i.e., a thermogram, was built through a mosaicking procedure consisting of 4 × 3 thermal images forming a matrix, in order to ``explore'' the whole surface. This is a routine procedure in which the time component is lost, although the details of the final image are improved, thank to the radiometric characteristic of the instrument. It was also necessary its application, by considering the field-of-view of the thermal camera that is maintained in the hands of the operator (i.e., the thermographer) [34]. The same can be said regarding Fig. 2b, also reconstructed via Image Builder® software, made by FLIR System. Following the same method used for the previous walls, the inspected part is indicated by a dotted dark rectangle and its magnification is reported in Fig. 5c. A selection of the pigments enhanced by the NIRR technique, working at 1000 nm, was signaled by dotted lines in Fig. 5d. Even in this case, the comparative results permitted to distinguish the zones affected by the rising damp effect. They are signaled by yellow arrows in Figs. 5e-f. On one hand, they reflect the behaviour previously explained, although in a reverse way in term of color-intensity. On the other hand, the rising damp ef-
14
fect can also be detect by using the passive approach (Fig. 5b). In the first case, the presence of moisture should be superficial. Finally, the defect surrounded by an irregular azure dotted line in Fig. 5f, just like that in Fig. 4f, can be explained by the forced convection provided by lamps on the wall painting, which stimulated the phase change from liquid to aeriform [35] during the heating up phase. In this second case, the presence of moisture should be sub-superficial. This assumption can be reinforced by looking at the cold image recorded before the starting of the heating up phase (Fig. 5g), that is inherent to the same ROI of Figs. 5e,f. In the early stage, the presence of superficial humidity is evident; however, it is surrounded by an irregular red dotted line for a sake of clarity (Fig. 5g). The visual behaviour dramatically change after the image processing (Figs. 5e,f) according to [36] which explained how materials having high thermal effusivity values cannot hold heat in the long period of time, because heat quickly dissipates itself from the surface as soon as the surrounding temperature drops down.
3.2 Micro chemical analyses
3.2.1 Pigments SEM-EDS analyses were realized to determine the elemental composition of pigments of different colors. For all samples, EDS results show the presence of C, Ca and O, ascribable to the mortar of the support and to the binder, as well as that of S, maybe linked to the presence of gypsum. Besides these elements, also Al, Mg, K, Fe, Na, Cl and Si are present, compatible with earth-based pigments (ochres) [37], presumably containing different percentages of aluminous-silicates and iron oxides. Therefore FTIR transmission spectra of P3, P4, P5 and P6 were collected (Fig. 6). For all samples the bands associated with the presence of calcite (713, 874, 1396 (Br), 1798 cm-1) and quartz (1166, 798, 694 e 510 cm-1) are detected. The characteristic peaks of gypsum (596, 669, 1105, 1624, 1684 cm-1) were also identified in samples P3, P5 and P6. While we can identify the presence of iron oxides in the form of goethite (FeOOH or Fe2O3 · H2O), the bands due to hematite, Fe2O3, the main responsible for the reddish hue [38] in samples P5 and P2, are not detectable by reason of the overlap of
15
its peaks at 3436, 3150, 800, 535, 470 cm-1, with those of quartz and with the stretching vibrations of O-H bonds. The dark color of sample P3 can be attributed to pyrolusite [39], MnO2, because of the very weak bands at 505, 1086 and 1155, 1952 and 2138 cm-1 or to an organic black (peaks at 1088 and 1059 cm-1). Unfortunately, it was not possible to uniquely identify the presence of organic pigments as the amorphous C, main element of vegetable blacks, is not detectable with this type of analysis, while the characteristic stretching signals of the PO43- group of animal blacks is not easily recognizable in the region between 1250-900 cm-1. It is possible to exclude the presence of kaolinite in absence of the peaks at 1032, 1009, 938 and 914 cm-1. We can conclude that east and south walls decorations, although related to different historical periods, were realized using ochre pigments mixed with calcite. Further confirmations to this assumption came from XRF and XRD analyses of the painted portion of sample M. As we can see in Table 2 the mineralogical and elemental semi-quantitative results show that the main phases were constituted by quartz and calcite, together with lower percentages of chalk, weddellite and, in the case of red pigment, a 2% of hematite too. The analysis of FTIR spectra recorded for all pigments didn’t show bands attributable specifically to oxalates, perhaps because of the overlap with peaks of gypsum, quartz and calcite (1616, 1318, 783, 660, 517 cm-1). According to literature, oxalates on murals may be due to degradation of natural organic compounds used as binders and/or protective products [40–42], or to the activity of biodeteriogens such as bacteria [43, 44] and fungi [45, 46]. In this light, the rising damp phenomenon shown in Fig. 2a could have promoted the biological growth and the subsequent formation of oxalates. Finally, all FTIR spectra recorded show transmission bands in the window region 2800-1800 cm-1 typical of adjacent double and triple bonds of polymer compounds containing isocyanate or nitrile [47]. These bands may be related to acrylonitrile or polyurethane [48] products used during an old consolidation activity of mural paintings.
3.2.2 Painting technique The observation of cross section of sample M via Scanning Electron Microscopy allowed the authors to recognize the presence of only two layers: the mortar and the pigment. As it is possible to see in Fig 7 (up), the pigment layer, having a thickness equal to 40 µm or less, was applied without adding
16
more finely ground lime (intonaco) nor limewash (intonachino) layers [49]. The EDX maps shown in Fig 7 (middle), substantially confirm the results obtained by using XRF and XRD analyses on the red and white pigments (see section 3.2.1); they also show that the presence of calcite can only refers to the thin coloured layer and to the mortar binder fraction. On account of these data, it is possible to deduce that the technical execution of the paintings was very simple, a kind of secco on the rough wall, using single painting layers. Secco generally relates to a painting technique in which pigments are applied on the dry plaster using inorganic (limewater) or organic binders [50]. Looking at FTIR spectra of pigment samples, it is also possible to note absorption bands that could suggest the presence of an organic binder, in particular between 3000 and 2830 cm-1, around 1740 and 1630 cm-1, characteristic of lipid and protein binders [51]. Nevertheless, these bands can also be found in reference spectra of calcium carbonate and gypsum, respectively [52]. For an in-depth study of the execution technique of the paintings, the cross section of sample M was also observed under UV illumination: images (Fig. 7 bottom) show a brown color, most noticeable at the red pigment, possibly due to an egg tempera [53]. This hypothesis was proved by MALDI TOF MS analysis of the pigment portion of sample M. As to the lipid fraction, intact lipids as well as their degradation products were identified. In mass spectrum (Fig. 8) the representative marker ions for intact eggs phospholipids (m/z = 760.58, 774.56, 788.58) and their degradation products (m/z = 496.19, 524.23, 536.22) [54] are clearly visible. Referring to the peptide fraction, many eggs peptides were identified (m/z = 1003.39, 1193.82, 1307.46, 1308.44 chicken ovotransferrin, 1033.70, 1233.10, 1234.09, 1277.83, 1451.01 chicken vitellogenin, 1065.50 Lysozyme), as well as some unattributed signals (m/z = 1320.65, 1340.60, 1765.34) linked to egg binder fingerprinting in old painting [54]. In light of these results, oxalates might be explained as deterioration products formed during breakdown of the organic binder, possibly mediated by micro-organisms.
3.2.3 Plaster
17
Sample M plaster portion was first analyzed by XRD that revealed the presence of calcite, gypsum and quartz, and then characterized from a physical-chemical and mineralogical-petrographic point of view. The mortar consists of about 35% binder fraction, while 45% of the total material has a particle size distribution comprised between 106 and 2360 mm. The remaining 20% consists of aggregates with diameters greater than or equal to 4mm. The particle size distribution enabled us to estimate a 1/3 binder: aggregate ratio. In particular, the white /cream aggregate mostly consists of arenaceous materials (aggregates of less than 2 mm), but also of conglomerates (between 2 and 4 mm) and coarse-grained inerts (over 4 mm). The aggregate fraction is quite selected (σ = 1,00 Ø), with mainly sub-angular and sub-rounded clasts with low-sphericity. Lime mortar were prepared by crushing local calcareous sedimentary stones. XRF results (Table 3) of binder and aggregate fractions are slightly different. Based on total amount oxides values (not shown), Hydraulic and Cementation Indexes [55, 56] were calculated (0,09 and 0,24 respectively) which confirmed the quicklime nature of the sample. The XRD study underlines that the major components are calcite and quartz. Also bernalite and diopside are present in the aggregate fraction, maybe intentionally added to confer little hydraulic properties to the mortar [57, 58]. Despite the rising damp shown in Fig. 2a, just at the sampling zone, and the closeness of cement mortars, according to the analytical results and macroscopic observation, the mortar sample is not particularly affected by degradation phenomena related to the crystallization of soluble salts. All things considered, sample M shows fairly chemical/physical and mechanical properties and good conservation conditions.
3.2.4. Restoration product Micro Raman and HATR-FTIR spectra of sample A were collected (Figs. 9, 10). Table 4 shows the assignments of the signals according to literature data [59–63]. Even if Raman and IR spectra are very similar to those of acrylates [64, 65], the assignments of frequencies allow us to characterize sample A as polyvinyl acetate based product. Many products used in the restoration field are referred to as vinyl derivatives: the EVA and VAE copolymers for example. But even PVAc homo-polymers actually differ according to their additives. In order to better understand the real composition of sample A, 1D and 2D NMR experiments were performed.
18
13
C NMR data of sample A in CDCl3, compared with those in literature [66], verify the characterization of the
resin as a polyvinyl acetate product. Signals at δ = 170.4-170.3, 67.9-66.17, 39.9-38.7 and 21.0 ppm are assigned to carbonyl (A), methine (B), methylene (C) and methyl (D) carbons respectively. Moreover, signals unrelated to the polymer suggest the presence of an aromatic structure. As regards the production of PVAc emulsions, two plasticizers are normally used: Di-butyl phthalate (DBP) and Di-butyl Maleate (DBM) [67]. Signals at δ 130.9 and 128 ppm confirmed the presence of DBP and were assigned to A' and B', δ 65 ppm in C', δ 30.6 ppm in D', δ19.2 ppm E' and δ13.7 ppm F'. The quaternary and carbonyl carbons’ signals are in the background noise. The characterization of the sample was also supported by a 1H NMR spectrum in CDCl3. Peaks at δ = 7.72-7.70 ppm and 7.53-7.52 were assigned to the aromatic protons of DBP in B' and A', at δ ppm 4.30 to the four protons in C'. D' signals fall below those of the polymer, peaks at 1.44 δ ppm belongs to the protons in E' and terminal methyl protons F’ resonated at 0.97 ppm, partially overlapped with another methyl signal at 0.98 ppm, not assigned. Signals at δ 4.87 ppm and in the region δ 1.9-1.6 ppm (protons in B), at δ 2.01 (D) were assigned to protons from PVAc. These assignments were confirmed by 1H-13C correlations of peaks in the 2D 1H-13C HSQC spectrum (Fig. 11) that correlates the chemical shift of proton with the chemical shift of the directly bonded carbon. Finally, the DOSY NMR experiment allowed us to distinguish the components of the sample according to the value of their diffusion coefficients and, therefore, to their different sizes and shapes. In the case of linear polymer chains, the diffusion coefficient is linked to the molecular mass [68]. The axis of the first dimension (F1) (Fig. 12) represents the logarithm of the diffusion coefficient while the second dimension (F2) represents the chemical shift. At higher diffusion coefficients, the lower part of the spectrum, we can find the slight residual solvent (CDCl3) at 7.25 ppm, TMS (tetramethylsilane) at 0.00 ppm and the signals of DBP, while lowest coefficients belong to the polymer, whose enlargement of the signals is due to its poly-dispersity. Sample A was recognized as a polyvinyl acetate added with di-butyl phthalate. Currently, commercial products that correspond to these compositional characteristics are formulations of trademark Vinamul (Vinamul 9146 or 9910 according to [48]) usually largely used as adhesives in restoration operations, but also as binders and protective.
19
Conclusions The diagnostic project of mural paintings was realized combining non-destructive and micro-invasive analytical techniques. In particular, IRT technique identified structural problems, rising damp phenomena, as well as splitting of the painted layer. Instead, NIRR technique improved the visualization of particular pigments. As regards micro-analytical techniques, EDS, FTIR and XRF analyses allowed the identification of pigments as ochres, natural earths used for their relative stability in painting which were applied ``dry'', according to the analysis of the cross-section of sample M. MALDI-TOF-MS analysis applied on lipid and protein fractions of painted sample suggests that painting binder is organic one, probably egg, although effect of pigments on binder components ageing leave many MS picks unattributed. Tempera painting technique, along with rising dump and biodegradation, could explain the presence of oxalates. FT-IR, XRF and XRD analyses of the mortar sample showed that is essentially composed by calcite, quartz, and minimum percentages of bernalite and diopside observed in the aggregate fraction. The lime mortar exhibits good chemical and structural characteristics and no obvious degradation products connected to water capillarity are present. According to UNI 11087, the study of the latter sample and other ones, will permit to accurately determine the presence and the influence of soluble salts for the conservation status of mortars. Finally, 1D and 2D NMR analyses were useful for the identification of the restoration polymer named sample A. The technique allowed characterizing not only the main polymer component, but also the plasticizing agents and it permitted to uniquely identify the trade name of the product used in the previous restoration, probably employed in incorrect quantities and/or manners. Indeed, it was applied both as a binder of the restoration mortar and as protective product in high concentrations, such as to facilitate its degradation and detachment. In conclusion, it is possible to assert that the conservation problems of the vestry were mainly caused by vandalism and structural failures, due to the earthquake and aggravated by rising damp phenomena. The previous restoration activities, even if made with wrong materials, guaranteed a quite good conservation of the painting cycle. The analytical results were crucial for the development of a detailed plan of restoration of the mural paintings [69–71].
Acknowledgements
20
The authors are grateful to Dr. Michele Nardone, Dr. Lorenzo Arrizza and Ms. Fabiola Ferrante of the University of L'Aquila for their valuable assistance in performing the µ-Raman, SEM-EDS and TG/DTA analyses, respectively. The authors want also to thank the continuous support of the Soprintendenza per i Beni Ambientali, Architettonici, Artistici e Storici per l’Abruzzo (Italy), as well as Ms. Jenny Rolo (restorer) for the kind help during the sample taking procedure. References
[1] D. Piccirilli,;1; Gli affreschi del Beato Placido nel monastero di Santo Spirito di Ocre, Iconographica 2 (2003) 82-107. [2];1; Cultural Heritage Office Historical Archive, L’Aquila (accessed 02.13). [3] A. Bendada, S. Sfarra, C. Ibarra-Castanedo, M. Akhloufi, J.-P. Caumes, C. Pradere, J.-C. Batsale, X. Maldague,;1; Subsurface imaging for panel paintings inspection: a comparative study of the ultraviolet, the visible, the infrared and the terahertz spectra, Opto-Electron. Rev 23 (2015) 88–99. 10.1515/oere-2015-0013 [4] D. Ambrosini, C. Daffara, R. Di Biase, D. Paoletti, L. Pezzati, R. Bellucci, F. Bettini,;1; Integrated reflectography and thermography for wooden paintings diagnostics, J Cult Herit 11 (2010) 196– 204. 10.1016/j.culher.2009.05.001 [5] S. Sfarra, C. Ibarra-Castanedo, C. Santulli, F. Sarasini, D. Ambrosini, D. Paoletti, X. Maldague.;1; Eco-friendly laminates: from the indentation to non-destructive evaluation by optical and infrared monitoring techniques, Strain 49 (2013) 175–189. 10.1111/str.12026 [6] X.P.V. Maldague,;1; Theory and practice of infrared technology for nondestructive testing, John Wiley & Sons, New York, 2001. [7] S. Sfarra, E. Marcucci, D. Ambrosini, D. Paoletti,;1; Architectural heritage at infrared: from the passive infrared thermography to the hybrid infrared thermography (HIRT) approach, Mater Construcc (2016) Accepted for publication, in press [8] E. Grinzato, N. Ludwig, G. Cadelano, M. Bertucci, M. Gargano, P. Bison,;1; Infrared thermography for moisture detection: a laboratory study and in-situ test, Mater Eval 69 (2011) 97–104. [9] E. Barreira, R.M.S.F. Almeida, J.M.P.Q. Delgado,;1; Infrared thermography for assessing moisture related phenomena in building components, Constr Build Mater (2016) Accepted for publication, in press10.1016/j.conbuildmat.2016.02.026 [10] P. Vásquez, C. Thomachot-Schneider, K. Mouhoubi, G. Fronteau, M. Gommeaux, D. Benavente, V. Barbin, J.-L. Bodnar,;1; Infrared thermography monitoring of the NaCl crystallization process, Infrared Phys Techn 71 (2015) 198–207. 10.1016/j.infrared.2015.03.013 [11] F.J. Madruga, C. Ibarra-Castanedo, O.M. Conde, J.M. López-Higuera, X. Maldague,;1; Infrared thermography processing based on higher-order statistics, NDT&E Int. 43 (2010) 661–666. doi 10.1016/j.ndteint.2010.07.002 [12] S. Sfarra, P. Theodorakeas, C. Ibarra-Castanedo, N.P. Avdelidis, D. Ambrosini, E. Cheilakou, D. Paoletti, M. Koui, A. Bendada, X. Maldague,;1; How to retrieve information inherent to old resto-
21
rations made on frescoes of particular artistic value using infrared vision?, Int J Thermophys 36 (2015) 3051–3070. 10.1007/s10765-015-1962-8 [13] R. Hidalgo-Gato, J. Andrés, J.M. López-Higuera, F. Madruga,;1; Quantification by signal to noise ratio of active infrared thermography data processing techniques, Optics and Photonics Journal 3 (2013) 20–26. 10.4236/opj.2013.34A004 [14] M. Klein, C. Ibarra-Castanedo, X. Maldague, A. Bendada,;1; A straightforward graphical user interface for basic and advanced signal processing of thermographic infrared sequences, Proc. of the SPIE 6939 – Thermosense XXX (Orlando, FL, USA) Proceedings 693914-1–693914-9, VP Vavilov and DD Burleigh, Ed.s, 2008a. 10.1117/12.776781 [15] F.J. Madruga, C. Ibarra-Castanedo, O.M. Conde, X.P. Maldague, J.M. Lopez-Higuera,;1; Enhanced contrast detection of subsurface defects by pulsed thermography based on fourth order statistic moment, Kurtosis. Proc. of the SPIE 7299 – Thermosense XXXI (San Diego, CA, USA) Proceedings 72990U-1–72990U-8. DD Burleigh and RB Dinwiddie, Eds., 2009. 10.1117/12.818684 [16] P. Pouli, D.C. Emmony,;1; The effect of Nd:YAG laser radiation on medieval pigments, J Cult Herit 1 (2000) S181–S188. 10.1016/S1296-2074(00)00143-6 [17] S. Rickerby,;1; Heat alterations to pigments painted in the fresco technique, The Conservator 15 (1991) 39–44. 10.1080/01410096.1991.9995063 [18] S. Sfarra, C. Ibarra-Castanedo, D. Ambrosini, D. Paoletti, A. Bendada, X. Maldague,;1; Nondestructive testing techniques to help the restoration of frescoes, Arab J Sci Eng 39 (2014) 3461– 3480. 10.1007/s13369-014-0992-z [19] E.Z. Kordatos, D.A. Exarchos, C. Stavrakos, A. Moropoulou, T.E. Matikas,;1; Infrared thermographic inspection of murals and characterization of degradation in historic monuments, Constr Build Mater 48 (2013) 1261–1265. 10.1016/j.conbuildmat.2012.06.062 [20] M. Susa, H.D. Benítez, C. Ibarra-Castanedo, H. Loaiza, H. Bendada, X. Maldague,;1; Phase contrast using a differentiated absolute contrast method, QIRT J 3 (2006) 219–230. 10.3166/qirt.3.219-230 [21] M. Klein, C. Ibarra-Castanedo, A. Bendada, X. Maldague,;1; Thermographic signal processing through correlation operators in pulsed thermography. Proc. of the SPIE 6939 – Thermosense XXX (Orlando, FL, USA) Proceedings 693915-1–693915-6. VP Vavilov and DD Burleigh, Eds., 2008b 10.1117/12.777002 [22] T.C. Chu, W.F. Ranson, M.A. Sutton, W.H. Peters,;1; Applications of digital-image correlation techniques to experimental mechanics, Exp Mech 25 (1985) 232–244. 10.1007/BF02325092 [23] H. Endo, T. Kusaka,;1; Efficient inspection for gas pipes by infrared thermography, Kobelko Technology Review 33 (2015) 50–55. http://www.kobelco.co.jp/english/ktr/pdf/ktr_33/050-055. pdf, (accessed 06.01.16). [24] E. G. Bligh, W. J. Dyer,;1; A rapid method of total lipid extraction and purification, Can. J. Biochem. Physiol. (1959), 37, 911–917.
22
[25] R. L. Snyder, D. L. Bish,;1; Quantitative Analysis, in: D.L. Bish, J.E. Post (Eds.), Powder Diffraction, Reviews in Mineralogy 20, Mineralogical Society of America, Washington D.C., 1989, pp. 101-144. [26] N.P. Avdelidis, A. Moropoulou, P. Theoulakis,;1; Detection of water deposits and movement in porous materials by infrared imaging, Infrared Phys Techn 44 (2003) 183–190. [27] E. Grinzato, G. Cadelano, P. Bison,;1; Moisture map by IR thermography, J Mod Optic 57 (2010) 1770–1778. [28] N. Ludwig, E. Rosina,;1; Detection of damp surfaces by active and passive thermography, in Atti della Fondazione Giorgio Ronchi no. 3, pp. 377-390, L’arcobaleno, Florence, 2001. [29] N. Ludwig, V. Redaelli, E. Rosina, F. Augelli,;1; Moisture detection in wood and plaster by IR thermography, Infrared Phys Techn 46 (2004) 161–166. 10.1016/j.infrared.2004.03.020 [30] C. Ibarra-Castanedo, A. Bendada, X. Maldague,;1; Image and signal processing techniques in pulsed thermography, GESTS Int’l Trans. Computer Science and Eng. 22 (2005) 89–100. [31] S. Sfarra, C. Ibarra-Castanedo, F. Lambiase, D. Paoletti, A. Di Ilio, X. Maldague,;1; From the experimental simulation to integrated non-destructive analysis by means of optical and infrared techniques: results compared, Meas Sci Technol 23 (2012) 115601 (14 p). 10.1088/09570233/23/11/115601 [32] S. Sfarra, A. Bendada, C. Ibarra-Castanedo, D. Ambrosini, D. Paoletti, X. Maldague,;1; Santa Maria di Collemaggio Church (L’Aquila, Italy): historical reconstruction by non-destructive testing techniques, Int J Arch Herit 9 (2015) 367–390. 10.1080/15583058.2013.794376 [33] F. Bisegna, D. Ambrosini, D. Paoletti, S. Sfarra, F. Gugliermetti,;1; A qualitative method for combining thermal imprints to emerging weak points of ancient wall structures by passive infrared thermography – A case study, J Cult Herit 15 (2014) 199–202. 10.1016/j.culher.2013.03.006 [34] D. Paoletti, D. Ambrosini, S. Sfarra, F. Bisegna,;1; Preventive thermographic diagnosis for consolidation, J Cult Herit 14 (2013) 116–121. 10.1016/j.culher.2012.05.005 [35] S. Sfarra, C. Ibarra Castanedo, M. Tortora, L. Arrizza, G. Cerichelli, I. Nardi, X. Maldague,;1; Diagnostics of wall paintings: a smart and reliable approach, J Cult Herit 18 (2015) 229–241. 10.1016/j.culher.2015.07.011 [36] J.A. Balderas-López, J. Díaz-Reyes, O. Zelaya-Angel,;1; Photoacoustic technique for simultaneous measurements of thermal effusivity and absorptivity of pigments in liquid solution, Rev Sci Instrum 82 (2011) 124901. 10.1063/1.3666863 [37] C. Genestar, C. Pons,;1; Earth pigments in painting: characterization and differentiation by means of FTIR spectroscopy and SEM-EDS microanalysis, Anal Bioanal Chem 382 (2005) 269– 274. doi 10.1007/s00216-005-3085-8 [38] G.A. Mazzocchin, F. Agnoli, S. Mazzocchin, I. Colpo,;1; Analysis of pigments from Roman wall paintings found in Vicenza, Talanta 61 (2003) 565–572. 10.1016/S0039-9140(03)003230 [39] D. Hradil, T. Grygar, J. Hradilova, P. Bezdicka,;1; Clay and iron oxide pigments in the history of painting, Appl Clay Sci 22 (2003) 223–236. 10.1016/S0169-1317(03)00076-0
23
[40] M. Alvarez de Buergo, R. Fort Gonzalez,;1; Protective patinas applied on stony facades of historical buildings in the past, Constr Build Mat 17 (2003) 83–89. 10.1016/S09500618(02)00107-1 [41] M. Monte,;1; Oxalate film formation on marble specimens caused by fungus, J Cul Herit 4 (2003) 255–258. 10.1016/S1296-2074(03)00051-7 [42] R. Quaresima, E. Di Giuseppe,;1; Analysis of oxalate films by means of HPLC. International symposium (II): the oxalate films in the conservation of works of art (Milan, March 25-27), Proc. 391-406 EDITEAM, Cento, 1996. [43] S. Capodicasa, S. Fedi, A.M. Porcelli, D. Zannoni,;1; The microbial community dwelling on a biodeteriorated 16th century painting, Int Biodeter Biodegr 64 (2010) 727-733. 10.1016/j.ibiod.2010.08.006 [44] C. Milanesi, F. Baldi, S. Borin, R. Vignani, F. Ciampolini, C. Faleri, M. Cresti,;1; Biodeterioration of a fresco by biofilm forming bacteria, Int Biodeter Biodegr 57 (2006) 168–173. 10.1016/j.ibiod.2006.02.005 [45] M. Guggiari, R. Bloque, M. Aragno, E. Verrecchia, D. Job, P. Junier,;1; Experimental calciumoxalate crystal production and dissolution by selected wood-rot fungi, Int Biodeter Biodegr 65(2011)803–809. 10.1016/j.ibiod.2011.02.012 [46] M.V. Dutton, C.S. Evans, P.T. Atkey, D.A. Wood,;1; Oxalate production by Basidiomycetes, including the white-rot species Coriolus versicolor and Phanerochaete chrysosporium, Appl Microbiol Biotechnol 39 (1993) 5–10. doi 10.1007/BF00166839 [47] M.R. Derrick, D. Stulik, J.M. Landry,;1; Infrared spectroscopy in conservation science, The Getty Conservation Institute, Los Angeles, 2000. [48] V. Horie,;1; Materials for conservation Organic consolidants, adhesive and coatings, Rotledge, New York, 2011. [49] E. Fiorin, P. A. Vigato,;1; Teodelinda’s tales at Monza Cathedral: A physico-chemical diagnosis of the pictorial cycle, J Cult Herit 8 (2007) 13-25. 10.1016/j.culher.2006.05.003 [50] M. Gil, T. Rosado, I. Ribeiro, J. A. Pestana, A. T. Caldeira, M. L. Carvalho et al,;1; Are they fresco paintings? Technical and material study of Casas Pintadas of Vasco da Gama house in Évora (Southern Portugal), X-Ray Spectrom. 44 (2015) 154–162. DOI 10.1002/xrs.2593 [51] S. Daniilia, E. Minopoulou, Fr. D. Demosthenous, G. Karagiannis,;1; A comparative study of wall paintings at the Cypriot monastery of Christ Antiphonitis: one artist or two?, J Archaeol Sci 35 (2008) 1695-1707. 10.1016/j.jas.2007.11.011 [52] S. Sotiropoulou, Z. E. Papliaka, L. Vaccari,;1; Micro FTIR imaging for the investigation of deteriorated organic binders in wall painting stratigraphies of different techniques and periods, Microchem J 124 (2016) 559–567. http://dx.doi.org/10.1016/j.microc.2015.10.002 [53] R. Mazzeo, E. Joseph, S. Prati, A. Millemaggi,;1; Attenuated Total Reflection–Fourier transform infrared microspectroscopic mapping for the characterisation of paint cross-sections, Anal Chim Acta 599 (2007) 107–117. 10.1016/j.aca.2007.07.076
24
[54] I. D. van der Werf, C. D. Calvano, F. Palmisano, L. Sabbatini,;1; A simple protocol for Matrix Assisted Laser Desorption Ionization- time of flight-mass spectrometry (MALDI-TOF-MS) analysis of lipids and proteins in single microsamples of paintingsAnal. Chim. Acta, (2012), 718, 1-10. 10.1016/j.aca.2011.12.056 [55] H. Boke, O. Cizer, B. Ipekoglu, E. Ugurlu, K. Serifaki, G. Toprak,;1; Characteristics of lime produced from limestone containing diatoms, Constr Build Mater 22 (2008) 866–874. 10.1016/j.conbuildmat.2006.12.010 [56] K. Callebaut, J. Elsen, K. Van Balen, W. Viaene,;1; Nineteenth century hydraulic restoration mortars in the Saint Michael's Church (Leuven, Belgium) Natural hydraulic lime or cement?, Cement Concrete Res 31 (2001) 397–403. 10.1016/S0008-8846(00)00499-3 [57] C. Sabbioni, G. Zappia, C. Riontino, M.T. Blanco-Varela, J. Aguilera, F. Puertas, K. Van Balen, E.E.;1; Toumbakari Atmospheric deterioration of ancient and modern hydraulic mortars, Atmospheric Environment 35 (2001) 539-548. 10.1016/S1352-2310(00)00310-1 [58] N. Bianco, A. Calia, G. Denotarpietro, P. Negro,;1; Hydraulic mortar and problems related to the suitability for restoration, Periodico di Mineralogia 82 (2013) 529-542. 10.2451/2013PM0031 [59] L. Costa, M. Avataneo, P. Bracco, V. Brunella,;1; Char formation in polyvinyl polymers I. Polyvinyl acetate, Polym Degrad Stab 77 (2002) 503–510. 10.1016/S0141-3910(02)001088 [60] M. Shimoyama, H. Maeda, K. Matsukawa, H. Inoue, T. Ninomiya, Y. Ozaki,;1; Discrimination of ethylene/vinyl acetate copolymers with different composition and prediction of the vinyl acetate content in the copolymers using Fourier-transform Raman spectroscopy and multivariate data analysis, Vib Spectrosc 14 (1997) 253–259. 10.1016/S0924-2031(97)00010-6 [61] D. Chelazzi, A. Chevalier, G. Pizzorusso, R. Giorgi, M. Menu, P. Baglioni,;1; Characterization and degradation of poly(vinyl acetate)-based adhesives for canvas paintings, Polym Degrad Stabil 107 (2014) 314–320. 10.1016/j.polymdegradstab.2013.12.028 [62] M. Cocca, L. D'Arienzo, L. D'Orazio, G. Gentile, C. Mancarella, E. Martuscelli, C. Polcaro,;1; Water dispersed polymers for textile conservation: a molecular, thermal, structural, mechanical and optical characterization, J Cult Herit 7 (2006) 236–243. 10.1016/j.culher.2005.11.002 [63] S. Wei, V. Pintus, M. Schreiner,;1; Photochemical degradation study of polyvinyl acetate paints used in artworks by Py–GC/MS, J Anal Appl Pyrol 97 (2012) 158–163. 10.1016/j.jaap.2012.05.004 [64] C. Miliani, M. Ombelli, A. Morresi, A. Romani,;1; Spectroscopic study of acrylic resins in solid matrices, Surf Coat Tech 151–152 (2002) 276–280. 10.1016/S0257-8972(01)01606-1 [65] M. Cocca, L. D’Arienzo, L. D’Orazio, G. Gentile, E. Martuscelli,;1; Polyacrylates for conservation: chemico-physical properties and durability of different commercial products, Polymer Testing 23 (2004) 333–342. 10.1016/S0142-9418(03)00105-3
25
[66] K. Dutta, M. Mukherjee, A.S. Brar,;1; Spectral Assignment of Poly(vinyl acetate) by One and Two Dimensional NMR Spectroscopy: Revisited, J Polym Sci A1 37 (1999) 551–556. 10.1002/(SICI)1099-0518(19990301)37:5<551::AID-POLA5>3.0. CO;2-R J POLYM SCI A1 [67] F. Toja, D. Saviello, A. Nevin, D. Comelli, M. Lazzari, M. Levi, L. Toniolo,;1; The degradation of poly(vinyl acetate) as a material for design objects: A multi-analytical study of the effect of dibutyl phthalate plasticizer. Part 1, Polym Degrad Stabil 97 (2012) 2441–2448. 10.1016/j.polymdegradstab.2012.07.018 [68] J. Viéville, M. Tanty, M.A. Delsuc,;1; Polydispersity index of polymers revealed by DOSY NMR, J Magn Reson 212 (2011) 169–173. 10.1016/j.jmr.2011.06.020 [69] F. Toschi, A. Paladini, F. Colosi, P. Cafarelli, V. Valentini, M. Falconieri, S. Gagliardi, P. Santoro,;1; A multi-technique approach for the characterization of Roman mural paintings, Appl Surf Sci, 284 (2013) 291-296. 10.1016/j.apsusc.2013.07.096 [70] V. Crupi, G. Galli, M.F. La Russa, F. Longo, G. Maisano, D. Majolino, M. Malagodi, A. Pezzino, M. Ricca, B. Rossi, S.A. Ruffolo, V. Venuti,;1; Multi-technique investigation of Roman decorated plasters from Villa dei Quintili (Rome, Italy), Appl Surf Sci, 349 (2015) 924-930. 10.1016/j.apsusc.2015.05.074 [71] M. Sawczak, A. Kamińska, G. Rabczuk, M. Ferretti, R. Jendrzejewski, G.;1; Śliwiński, Complementary use of the Raman and XRF techniques for non-destructive analysis of historical paint layers, Appl Surf Sci, 255 (2009) 5542-5545. 10.1016/j.apsusc.2008.07.138
Fig. 1 Summary scheme of the procedure followed for the diagnostic study of Ocre paintings.
Fig. 2. Right wall: (a) area of interest with mapped positions of the samples taking, along with the ROI inherent to the reflectographic and thermographic acquisitions (P3), (b) Thermogram recorded by means of the passive approach, (c) Visible image of the ROI, (d) NIRR at 1000 nm of the ROI, (e) Correlation result – P3, (f) Kurtogram – P3, and (g) the cold image (i.e., the thermal image before the starting of the heating up phase).
Fig. 3. Temperature increasing in active thermographic test on two representative areas marked in Fig. 2e (dry area: violet square; damp area: red crosses).
Fig. 4. South wall: (a) area of interest with mapped positions of the sample taking, along with the ROI inherent to the reflectographic and thermographic acquisitions (P4_P5 and P6), (b) Visible image on the top, and NIRR at 1000 nm on the bottom of the ROI, (c) Correlation result – P4_P6, (d) Kurtogram – P4_P6, (e) Correlation result – P5, (f), Kurtogram – P5, (g) the cold image (i.e., the thermal image before the starting of the heating up phase) inherent to P4_P6, and (h) the cold image (i.e., the thermal image before the starting of the heating up phase) inherent to P5.
Fig. 5. North wall: (a) ROI inherent to the reflectographic and thermographic acquisitions, (b) Thermogram recorded by means of the passive approach, (c) Visible image of the ROI, (d) NIRR at 1000 nm of the ROI, (e) Correla-
26
tion result, (f) Kurtogram result, and (g) the cold image (i.e., the thermal image before the starting of the heating up phase).
Fig. 6. FT-IR spectra of sample P3 and P4.
Fig. 7. Up: SEM images of white (A) and red (B) pigment portion of sample M cross-section; middle: EDS elemental map distributions; bottom: fluorescent images after UV illumination.
Fig. 8. MALDI-TOF mass spectrum of the lipid fractions obtained from BD extraction of sample M pigment portion.
Fig. 9. μ-FTRaman spectrum of sample A.
Fig. 10. HATR-FTIR spectrum of sample A.
Fig. 11. 2D 1H-13C HSQC spectrum of sample A.
Fig. 12. DOSY NMR spectrum of sample A
Tables
Table 1. Summary and location of samples analyzed via micro-chemical techniques. Sample
Location
Color/ Structure
P1 P2 P3 P4 P5 P6 M A
Right wall Right wall Right wall South wall/Crucifiction Palimpsest South wall/Crucifiction Palimpsest South wall/Crucifiction Palimpsest Right wall South wall/Crucifixion Palimpsest
Greenish Reddish Yellow Bluish Red Flesh-coloured Plaster-red/white pigment Restoration polymer
Table 2. Chemical and mineralogical composition of painted portion of sample M. XRF % MgO SiO2 P2O5 . SO3 Cl K2O CaO Fe2O3 TiO2 Other