Anal Bioanal Chem (2010) 397:3117–3125 DOI 10.1007/s00216-010-3880-8

ORIGINAL PAPER

Noninvasive depth profiling of walls by portable nuclear magnetic resonance Bernhard Blümich & Agnes Haber & Federico Casanova & Eleonora Del Federico & Victoria Boardman & Gerhard Wahl & Antonella Stilliano & Licio Isolani

Received: 14 January 2010 / Revised: 26 May 2010 / Accepted: 26 May 2010 / Published online: 19 June 2010 # Springer-Verlag 2010

Abstract A compact and mobile single-sided 1H NMR sensor, the NMR-MOUSE®, has been employed in the nondestructive characterization of the layer structure of historic walls and wall paintings. Following laboratory tests on a model hidden fresco, paint and mortar layers were studied at Villa Palagione and the Seminario Vescovile di Sant’ Andrea in Volterra, Italy. Different paint and mortar layers were identified, and further characterized by portable X-ray fluorescence spectroscopy where accessible. In the detached and restored fresco “La Madonna della Carcere” from the Fortezza Medicea in Volterra, paint and mortar layers were discriminated and differences in the moisture content of the adhesive that fixes the detached wall painting to its support were found in both restored and original B. Blümich (*) : A. Haber : F. Casanova Institute for Technical and Macromolecular Chemistry, RWTH Aachen University, Worringer Weg 1, 52056 Aachen, Germany e-mail: [email protected] E. Del Federico : V. Boardman Department of Mathematics and Science, Pratt Institute, 200 Willoughby Ave, New York, NY 11205, USA V. Boardman Department of History of Art and Design, Pratt Institute, 200 Willoughby Ave, New York, NY 11205, USA G. Wahl : A. Stilliano Centro Interculturale, Villa Palagione 56048 Volterra, Italy V. Boardman : L. Isolani Department of Fine Arts, Pratt Institute, 200 Willoughby Ave, New York, NY 11205, USA

sections. These investigations encourage the use of the portable and single-sided NMR technology for nondestructive studies of the layer structure and conservation state of historic walls. Keywords Nuclear magnetic resonance . X-ray fluorescence . Wall painting . Art conservation . Pigments . Nondestructive analysis

Introduction Nuclear magnetic resonance (NMR) [1] is a well-known technique in clinical diagnostics and chemical analysis. In these applications, large and stationary equipment is used with strong magnetic fields. But there are also miniaturized NMR instruments which can be carried to the site of the object [2]. Although most NMR investigations occur with the object inside the magnet, stray-field techniques use NMR magnets placed next to the object [3]. The first small device of this kind was the NMR Mobile Universal Surface Explorer, or NMR-MOUSE® [4], which was optimized recently to measure depth profiles through coplanar layers of a diverse range of objects with a depth resolution of higher than 10 µm [5]. There is great interest in using this Profile NMR-MOUSE® for noninvasive analysis of objects of cultural heritage such as to study the stratigraphy of paintings [6,7], the degradation of paper [8–10], wood, bones, and mummies, and other materials containing hydrogen [3,11], including consolidation treatments of porous building materials [12–14]. For the first time, depth profiles were measured through walls of historic buildings, and this work documents different applications of the Profile NMR-MOUSE® in measuring historical walls decorated with paint or fresco layers.

3118

B. Blümich et al.

The original NMR-MOUSE® is a palm-sized NMR device with a time-invariant stray field positioned near an object to quantify the information from one pixel of a medical image from the volume of the object near the sensor. The signal is collected from a sensitive volume external to the sensor. Early iterations of this concept employed a simple U-shaped magnet with a radiofrequency (rf) coil positioned in the gap between the magnet poles (Fig. 1b). The coil is needed to excite and receive the NMR signal. The sensitive volume of the original NMR-MOUSE® is oddly shaped and prevents the measurement of depth profiles with high resolution [4]. The Profile NMR-MOUSE® [5] has a sensitive volume in the shape of a flat slice, coplanar to the sensor surface with a lateral extension of about 10 mm and a thickness adjustable between a few micrometers and about 100 µm. When measurements are performed, the sensor is placed close to the wall with its surface coplanar to the wall (Fig. 1a), so that the sensitive slice is located inside the wall (Fig. 1b). The distance between the sensor surface and the wall is increased using a stepper-motor-controlled displacement of the sensor. This makes the sensitive slice move outward toward the wall surface. The sensor position is varied in small steps, and at each step an NMR measurement is taken. In general, proton signals are measured because hydrogen nuclei give the best NMR signal and are abundant in many materials. Depending on the size of the sensor, maximum depths of 3, 10, and 25 mm can be accessed. The 10-mm sensor can be converted into a 5-mm sensor to provide better sensitivity at shallow depths. A typical measurement applies a string of rf impulses. The first impulse invokes an impulse response, which decays rapidly owing to the inhomogeneity of the stray field of the NMR magnet. Echoes of the original impulse are recalled many times with the subsequent rf impulses. The envelope of this echo train is recorded [1]. This echo-envelope function decays with time. Its initial amplitude provides the proton

density, which is proportional to the moisture content in wet walls. Its decay-time constant is called T2 in the NMR field and varies with the mobility of the protons. The protons in the dry binder of paint layers have a short T2, whereas liquid water has a long T2. Consequently, the NMR signal from paint and bound water in mortar decays quickly, often in less than 1 ms, whereas moist walls may provide NMR signals lasting tens of milliseconds. To record depth profiles, the acquired NMR signal s(t) may be reduced to one number w by integrating it from time zero to some time t1 and from t1 toRinfinity (Fig. 1c). The R ratio of the two resulting integrals 0 t1 s(t) dt and ti 1 s(t) dt defines the profile amplitude w at the given measurement depth. The value of t1 is adjusted beforehand to scale in the parameter w so that the relative contributions of signal amplitude, corresponding to proton density, and relaxation time T2, corresponding to molecular mobility, achieve maximum contrast to identify different features in the depth profile. Alternatively the amplitude sum of the first echos is plotted in arbitrary units that approximately scale with the proton density.

Fig. 1 Measurement of depth profiles with nuclear magnetic resonance (NMR). a NMR-MOUSE® with 25-mm depth access set up in Herculaneum to measure a depth profile. b Components of the

NMR-MOUSE® and principle of operation. c NMR signal and evaluation: Envelope of an echo train and definition of partial integrals for calculation of the profile amplitude w

Experimental NMR measurements in Volterra were taken using a Profile NMR-MOUSE® with 10 mm and 5 mm depth access. For some measurements the depth range of the 10 mm sensor was reduced to 5 mm by inserting a 5 mm spacer between the sensor and the rf coil to improve the sensitivity at shallow measurement depths. Sensors were mounted on computer-driven positioning devices. The NMR sensors and the positioning devices were manufactured by ACT (Aachen Germany). The spectrometer hardware was a Minispec spectrometer manufactured by Bruker Biospin (Rheinstetten, Germany). Table 1 summarizes the typical acquisition parameters.

Noninvasive depth profiling of walls by portable nuclear magnetic resonance Table 1 Acquisition parameters for nuclear magnetic resonance (NMR) depth profiles

3119

Parameter

Value (5-mm range)

Value (10-mm range)

Spectrometer

Bruker Minispec

Bruker Minispec

NMR-MOUSE® NMR frequency (MHz) Power amplifier (W) Pulse width (µs) Echo time (µs) Number of echoes Number of scans Time per profile (h)

PM 5 18.1 300 8 67 32 2,048 1.5

To complement the NMR measurements, different paint layers were analyzed using portable X-ray fluorescence (XRF) spectroscopy to determine the elemental composition of the paint and mortar layers and to monitor the presence of modern materials. A handheld Tracer III-V manufactured by Bruker AXS was used to perform the XRF measurements. Spectra were acquired without a filter, using a voltage of 40 kV, a current of 1.20 µA, and a vacuum pump attachment. These settings allowed detection of elements with an atomic number of 11 (Na) and higher.

Laboratory study of a hidden painting Since antiquity, older wall paintings have been used as templates for newer, sometimes more fashionable, wall paintings. Chiseling the old painting in some areas allowed the preparation mortar (arriccio) to adhere. Then, by addition of a lime plaster layer, a mixture of calcium hydroxide and a support such as sand, clay, or marble dust, the surface was prepared for application of a new painting. The new painting layer is typically called the intonaco. A famous hidden and lost wall painting The Battle of Anghiari, 1505 CE, by Leonard da Vinci, is believed to be beneath a fresco by Giorgio Vasari in the Palazzo Vecchio in Florence, Italy [15]. To test if single-sided NMR lends itself to investigating hidden wall paintings, a covered painting (Fig. 2a) was fabricated in the secco and fresco techniques of medieval and Renaissance art (Fig. 2b) and analyzed using NMR depth profiles. A measurable signal was generated by spraying water on the surface and measuring the moisture profiles immediately after spraying and again 3 h later. The paint layer thickness was estimated from the position of the dips at about 5-mm depth in the profiles for red and gray to about 0.4 mm (Fig. 2c). The combined action of binder and pigments inhibited the moisture transport through the hidden secco in ways that vary for different paint formulations. This demonstrated that not only the position

PM 10 10.65 300 15 67 32 2,048 3

PM 10 10.65 300 15 67 32 8,192 10

and paint layer thickness can be accessed using NMR depth profiles, but also some information about the type of paint or the color [3].

Stratigraphy of walls in Villa Palagione Villa Palagione was built under the Medici dynasty in 1598. It was used by nobility and wealthy families as a summer retreat and hunting residence. At the end of the last century it was left to decay for several decades. Painted ceilings and walls were among the damaged areas when a vaulted ceiling collapsed. The villa was restored by the current owners Gerhard Wahl and Antonella Stillitano beginning in 1986. Photographs and oral accounts describe several rooms covered with wall and ceiling paintings. The original paintings are still present within the walls, which were repaired in the last decade with modern materials. Restorers left small untreated areas so the earlier paintings could still be seen. The paint layers were measured with a 5 mm spacer inserted between the magnet and the rf coil of the NMRMOUSE®. This spacer enhanced the instrument sensitivity by moving the coil closer to the sensitive plane while reducing the accessible depth from 10 to 5 mm. Three depth profiles were acquired across 5 mm depth on one wall at positions where several layers of old paint were preserved (Fig. 3a). An echo train was acquired every 100 µm. Two depth profiles were measured in dry conditions at adjacent points. Another profile was measured 90 min after moistening the wall by spraying it lightly with water. When dry, the paint layers and bound water in the mortar provided only a weak NMR signal owing to a low proton content. Nevertheless, the upper paint layer of 0.5 mm thickness was clearly identified with paint layers underneath, particularly at the position where the measurement was repeated after surface moistening. The profile measured after moistening mirrored the profiles measured while the wall was dry. This confirms that the profile amplitude variations are not dominated by noise but rather by the paint layer properties and

3120

B. Blümich et al.

Fig. 2 Hidden painting. a Phantom image of hidden wall paintings above the Profile NMR-MOUSE® for analysis of depth profiles. b Fresco and secco wall paintings created by medieval and Renaissance art techniques that were subsequently covered by a 5 mm layer of mortar. c Timedependent moisture profiles through different sections of the hidden secco. They reveal the position of the hidden painting and information about different colors. R red, Y yellow, G gray

some unevenness of the mortar interface. Owing to the higher content of mobile protons from the uptake of water, the profile amplitude from the moistened section is higher. Underneath the outer paint layer is one that absorbs water well. By Fig. 3 Profiles of paint layers. a Profiles across 3 mm depth into a wall with several paint layers and photograph of a nearby section of the wall where different paint layers have been removed one by one. b X-ray fluorescence (XRF) spectra of exposed paint layers. 1 surface paint layer containing modern materials; 2 exposed white paint layer; 3 exposed red paint layer; 4 exposed gray mortar

examining a nearby part of the wall, where paint layers were partially uncovered, we assigned this layer to whitewash that covers even older layers of paint, which include a reddish pattern.

Noninvasive depth profiling of walls by portable nuclear magnetic resonance

XRF analysis made it possible to determine that the wall surface was coated with modern materials owing to high signal intensity for titanium. Titanium white pigment (TiO2) was not widely used before 1940, so its presence dated the outer layer to after 1940 [16]. The XRF studies on the partially exposed areas of the wall revealed that the whitewash found as a second layer was most likely a lime wash (possibly CaCO3) and not plaster (gypsum, CaSO4·2H2O), owing to the high amounts of calcium and the absence of a sulfur signal. The third layer Fig. 4 Profiles through mortar layers of walls. a Profile across 10 mm depth of a wall after wetting before beginning the 10 h acquisition and section of the wall measured after opening up the mortar for inspection. b Profiles across 5 mm at a similar position of another wall measured before and after moistening the wall: The photograph shows the state of the wall after finishing the measurements. The measured spot is still wet, giving rise to a high-amplitude profile of the outer paint layer. Behind the measured spot, an older wall painting was discovered after opening up the wall. c XRF spectra of the wall layers. 1 surface wall and paint layer; 2 gray mortar layer; 3 more absorbent ochre mortar layer

3121

detected consisted of iron pigments and calcium compounds, possibly a lime mortar matrix owing to the absence of sulfur and the presence of silicon. The fourth and last layer detected also consisted of an iron-rich lime mortar but without any pigments. A depth profile was measured across a larger depth of 9 mm after removing the 5 mm spacer from the NMRMOUSE® on the opposite wall near the balcony door (Fig. 4a). This reduced the sensitivity, and the signal from the dry wall was very weak. To enhance the signal, this

3122

section of the wall was sprayed with water several times, and the signal was scanned 8,192 times and accumulated. With a step width of 30 µm, it took 10 h to measure a profile across 9 mm depth. During this time the water sprayed on the wall surface before the start of the measurement penetrated further into the wall and evaporated from the surface. Therefore, the resultant depth profile reflects the moisture profile with a maximum at about 5 mm into the wall. A sharp increase in the moisture content at a higher depth is observed in the acquired profile at about 3-mm depth. This indicates a change in the mortar properties. With permission of the owner, the wall was opened up at the measured position and the mortar inspected. At the 3 mm depth the mortar consistency changed, which could easily be seen by the color change from gray in the outer layer to ochre in the more absorbent inner layer. Similar depth profiles were measured at another spot on a different wall in the same room, dry and after moistening (Fig. 4b). Best visible in the profile of the dry wall are indications of at least two layers of mortar in the outer 3 mm of the wall and the paint layer. The paint layer shows a lowamplitude profile when dry and a high-amplitude profile when wet. This part of the wall was also opened up after measuring. When analyzing the mortar pieces removed from the hole, we found a surface paint layer, a very thin grayish layer, and a 3 mm-thick gray layer. The gray portion was found to consist of at least two phases, one more compact and closer to the surface, the other more porous and deeper in the wall. There were also small pebbles and white fragments visible within the deeper layer. XRF analysis of the top layer and the two mortars (Fig. 4c) confirmed the NMR results and visual observations that the chemical compositions of the layers were different. The top layer consisted of lead and iron pigments and possibly gypsum (Fig. 4c, surface). The second and third layers displayed similar compositions which consisted of mortars containing aluminosilicates, potassium, calcium, iron, and traces of manganese, titanium, and copper (Fig. 4c, gray mortar, ochre mortar). Differences in peak intensities for potassium may suggest that the mortars although chemically similar contained different concentrations of clay or other aluminosilicates.

Paintings at the Seminario Vescovile di Sant’Andrea The church complex at Sant’Andrea in Volterra originates from 1170, when Ugolino della Gherardesca, Count Donoratico in Pisa (and infamous Inferno character of Dante’s Divine Comedy [17]), and his mother sold the property to the Benedictine Order (Sant’Andrea library staff, personal communication). The church was originally simple in design and has undergone constant renovations

B. Blümich et al.

since its beginnings. By 1336 the building had been established as a seminary that also housed and aided the ill. Renovations from the seventeenth century favored Rococo decorations, which were covered at some point after 1730, most likely in the early nineteenth century when plain-colored stucco walls came into fashion. Historical accounts of the Seminario describe an exterior courtyard decorated in a program of illusionistic frescoes, which had been covered with contemporary building materials and paints. During a 2006 restoration, caretakers rediscovered these hidden wall paintings and carefully removed portions of the contemporary paint, revealing the modified layer structure. This is a unique example of known hidden wall paintings, which can be measured both at the surface and through layers of typical contemporary construction materials. Figure 5 displays a courtyard wall with exposed strata including a formerly hidden wall painting consisting of stripes of different colors, and its preparation layer. It is still possible to see areas in the strata beneath the hidden wall painting that were chiseled to prepare for application of the newer painting. Two depth profiles (Fig. 5a) were recorded with the 5 mm sensor set up at the same point from the wall painting, one in the original dry state and one after spraying the wall with water. The moisture absorbed by the wall provided better signals, and steps in the moisture profile identified different interfaces between different layers. In the dry state, the profile amplitude is very low but does show some variation. These variations are far more pronounced in the profile measured after moistening the wall. During the 90 min required to measure the depth profile, the water diffused into the wall, so the profile amplitude decreased with increasing depth and merged with the profile of the dry wall at larger depth. Four layers can be distinguished in the upper 3 mm: an outer paint layer about 0.2 mm thick, over a 1.0-mm-thick layer of primer (layer 1), and a 0.4 mm-thick paint layer (layer 2) followed by another layer (layer 3) of the same thickness between 1.4 and 2.2 mm depth (layers 2 and 3 are likely to be the wall painting and its preparation layer as their depths agree with the layers that were uncovered); layer 4 is mortar. The NMR measurements were complemented by XRF analysis of the surface pigments of uncovered parts of the painting (Fig. 5b). The analysis revealed a historically accurate palette of pigments, containing mostly iron and manganese oxides. These earth pigments are consistent with the traditional medieval and Renaissance painting palette. However, relatively high concentrations of arsenic were present in the green areas of the hidden painting (Fig. 5b, layer 3) consisting of orange, red, green, and gray stripes, allowing it to be dated to not earlier than the mid1780s when green arsenic pigments such as Scheele’s green (CuHAsO3) became available [18] (Fig. 5b). Owing to its high toxicity, use of the pigment was discontinued by 1950.

Noninvasive depth profiling of walls by portable nuclear magnetic resonance

3123

Fig. 5 a Depth profiles of a wall painting acquired with the 5-mm sensor with and without wetting the surface, photograph of the wall with parts of the painting uncovered and the position for the XRF measurements marked, and enlarged view of the measurement position. b XRF spectra of the different paint layers of the formerly hidden wall painting. 1 layer 1 of modern paint and building materials covering the wall painting; 2 layer 2, a beige preparation layer; 3 layer 3, a light-orange paint layer belonging to the wall painting displaying stripes of different colors—red, orange, gray, and green paint shows up; 4 layer 4, exposed mortar layer beneath wall painting

Further XRF measurements were taken on sites of the walls where the deeper paint layers were exposed. The first layer from the surface contained significant amounts of titanium, showing that at least the wall surface had been treated with modern painting materials (Fig. 5b, layer 1). The second layer showed high signal intensity for sulfur and calcium and the absence of titanium, indicating that this layer was probably composed of gypsum (Fig. 5b, layer 2). The third layer consisted of stripes of different colors as mentioned above. The red colors were determined to contain iron compounds, and low sulfur concentration, whereas the green displayed a high arsenic concentration. This painting layer was possibly applied directly on lime mortar either with the secco or with the fresco technique (Fig. 5b, layer 3). The last layer showed a typical composition of a lime mortar with iron impurities (Fig. 5b, layer 4).

Detached wall painting “La Madonna della Carcere” A further study concerned a Madonna and child wall painting dating to about 1500 CE from the Fortezza Medicea at Volterra which was undergoing conservation.

The Fortezza was built in 1266 and received numerous additions after Lorenzo de’ Medici’s 1472 conquest of Volterra. Today, the fortress serves as a high-security prison and the Madonna and child painting is housed in its chapel. The painting had been detached from the wall by a previous restoration procedure using the “strappo” technique. This technique involves the removal of the paint layer (or intonaco) by the application of linen fabric embedded in animal glue or gum arabic binders. The painting is then gently tapped to detach the intonaco layer from the arriccio layer (a rougher layer underneath the intonaco layer present in wall paintings as a preparation layer). The intonaco layer is then applied to canvas on a wooden support using an organic adhesive. Owing to careful record-keeping and special brush strokes which distinguished restored areas from the original painting, it was possible to compare the NMR and XRF signals of side-by-side points. NMR depth profiles were acquired with the 5 mm setup at three different sections of the fresco. In two of these areas, original and restored parts were measured and compared, in terms of the weight function w. The painting had a consistent layer structure, evident in a characteristic increase in profile amplitude between 4 and 4.5 mm that is attributed to the interface

3124

B. Blümich et al.

between mortar with low proton content and the organic adhesive holding the painting to its support. The adhesive layer behind the restored points on the neck and red garment gave a stronger signal than that behind the original points, indicating a more elastic texture of the adhesive. Moreover, original and restored areas of the neck and the red garment, and the halo show a weak signal decrease at about 1.4 mm depth which can be attributed to different

levels of bound water in mortar with and without pigments. The first three data points on the right of each profile correspond to zero signal from the gap between the sensor and the fresco. XRF measurements of the gilding in this wall painting showed the presence of lead and tin, suggesting the use of lead tin yellow, Pb2SnO4, as a substitute or enhancement for gold in the restored areas (Fig. 6c, gilding). Traces of

Fig. 6 a Restored wall painting “La Madonna della Carcere” from the Fortezza Medicea in Volterra. b Detailed photographs of the positions where measurements were taken and depth profiles at the marked positions of adjoining original and restored sections and comparison

of profiles with the original parts of the wall painting. c XRF spectra. 1 gilding of Christ child’s halo; 2 red garment worn by the Madonna; 3 neck area as originally painted; 4 neck area after restoration using modern materials

Noninvasive depth profiling of walls by portable nuclear magnetic resonance

gold were present even in the restored areas, suggesting either that this pigment may have been used in a mixture with gold or that the lead pigment was used in the restoration procedure to replace the gold that had flaked off. The NMR depth profile of the gilded sections closely matches the profile measured in the neck region. No particular adhesive or organic material was detected underneath the gilded layer. Moreover, the rf NMR signal easily penetrated the layer, suggesting that the gilding is either extremely thin or consists of gold powder pigments. XRF analysis of the original portions of the Madonna’s red mantle displayed high intensities for both mercury and sulfur, suggesting the presence of a precious mercuric sulfide pigment such as natural cinnabar, also known as vermillion (HgS) (Fig. 6b, red pigment). The presence of cinnabar, an expensive pigment, speaks to the significance of the subject and suits a painting owned by the wealthy and powerful Medici family. In the restored areas, only iron oxides are found, mixed to match the original red color. XRF measurements of the restored and original regions of the Madonna’s neck (Fig. 6b, original, restored) revealed the use of a modern pigment such as zinc white, ZnO, possibly as a substitute for lead white, 2Pb(CO3)2 Pb(OH)2, present in the original painting.

Conclusions The cultural and historical value of wall paintings and their numerous conservation issues favors a nondestructive, minimally invasive approach to analysis and characterization. These studies demonstrate that NMR is a promising tool in analyzing and characterizing the layer structures of wall paintings and historical walls, and in the discovery of hidden wall paintings. By analyzing the moisture content in terms of proton density and mobility, the NMR-MOUSE® provides a method for distinguishing layer structure and properties. Combined with surface elemental composition from XRF data, the use of NMR allows distinction in the adhesive layer behind original and restored paint and mortar layers of detached wall paintings. Such a difference can provide the means to distinguish restored and original portions of detached wall paintings when an otherwise consistent layer structure is observed. Renaissance and medieval wall paintings that had been covered with modern materials during renovations at Villa Palagione and the Seminario Vescovile di Sant’Andrea exhibited NMR behavior similar to that of laboratory-prepared hidden frescoes, verifying that NMR is a viable nondestructive

3125

means for detecting such hidden wall paintings. Comparison of NMR measurements with destructive sampling at Villa Palagione confirmed the accuracy of the NMRMOUSE® in depth profiling over an interface a few millimeters deep with good reproducibility. Acknowledgements We would like to thank Bruker AXS for the loan of a Bruker Tracer III-V with which all XRF measurements in this study were performed; Pratt Institute’s faculty development grant; Cecilia and Sandro Sirigetti, restorers at the Conservation Laboratory in Volterra; Laura Capozzoli for information on the conservation of the Madonna della Carcere; Alessio Ciampini from Art Lab; and the staff of the Seminario Vescovile di Sant’Andrea for their assistance in making overnight measurements possible. We thank Pratt Institute’s alumni Lindsey Tyne, Ken Girard, Cyndi O’Hern, Megan Welchel, Penelope Currier, and Daria Souvorova for their assistance in this project.

References 1. Blümich B (2005) Essential NMR. Springer, Berlin 2. Blümich B, Casanova F, Appelt S (2009) Chem Phys Lett 477:231–240 3. Blümich B, Casanova F, Perlo J (2008) Prog Nucl Magn Reson Spectrosc 52:197–269 4. Eidmann G, Savelsberg R, Blümler P, Blümich B (1996) J Magn Reson A 122:104–109 5. Perlo J, Casanova F, Blümich B (2005) J Magn Reson 176:67– 70 6. Proietti N, Capitani D, Lamanna R, Presciutti F, Rossi E, Segre AL (2005) J Magn Reson 177:111–117 7. Proietti N, Capitani D, Rossi E, Cozzolino S, Segre AL (2007) J Magn Reson 186:311–318 8. Blümich B, Anferova S, Sharma S, Segre AL, Federici C (2003) J Magn Reson 161:204–209 9. Casieri C, Bubici S, Viola I, De Luca F (2004) Solid State Nucl Magn Reson 26:65–73 10. Proietti N, Capitani D, Pedemonte E, Blümich B, Segre AL (2004) J Magn Reson 170:113–120 11. Rühli FJ, Böni T, Perlo J, Casanova F, Baias M, Egarter E, Blümich B (2007) J Cult Herit 8:257–263 12. Proietti N, Capitani D, Cozzolino S, Valentini M, Pedemonte E, Princi E, Vicini S, Segre AL (2006) J Phys Chem B 110:23719– 23728 13. Sharma S, Casanova F, Wachen W, Segre AL, Blümich B (2003) Magn Reson Imag 21:249–255 14. Blümich B, Casanova F, Perlo J, Anferova S, Anferov V, Kremer K, Goga N, Kupferschläger K, Adams M (2005) Magn Reson Imag 23:197–201 15. Pieraccini M, Mecatti D, Luzi G, Seracini M, Pinelli G, Atzeni C (2005) Non Destr Test Eval Int 38:151–157 16. Gettens GR, Stout GL (1966) Painting materials: a short encyclopedia. Dover, New York, pp 160–161 17. Alighieri D (2008) Inferno cantos 32–33. Oxford University Press, Oxford 18. Fitzhugh EW (ed) (1997) Artist’s pigments. A handbook of their history and characteristics, vol 3. Oxford University Press, Oxford, pp 219–271