Advanced Techniques in Art Conservation

GUEST EDITORIAL Advanced Techniques in Art Conservation It is well-known that historical artifacts and works of art det...

Author: Andrew J. Clark

248 downloads 2093 Views 83MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form

DOWNLOAD PDF

GUEST EDITORIAL Advanced Techniques in Art Conservation It is well-known that historical artifacts and works of art deteriorate

of decay. Owing to the advantages afforded by its high sensitivity,

because of physical and chemical interactions with the surround-

synchrotron radiation methods have been widely exploited in recent

ing environment. These processes can be thermodynamically and

years. Among the available tools, X-ray absorption spectroscopy

kinetically complex and can lead to discoloration, structural weak-

(XAS) offers a combination of features particularly well-suited for the

ening, corrosion, and other alterations that, in the absence of inter-

study of art. Chemical mapping at high spatial resolution, for

vention to stop or slow degradation, may result in a myriad of

instance, provides information on the distribution of local phase com-

negative consequences, ranging from the incorrect “reading” of a

position and chemical states. XAS is used to reveal techniques of

work of art to its complete loss. It follows that detailed knowledge

execution, to explain optical properties, and to visualize details of

of the composition and structure of the materials used by artists, to

alteration reactions. Since synchrotron radiation techniques offer

include the behavioral properties of these materials, must be a fun-

large depth of focus and fast acquisition, time-resolved measure-

damental component of successful and durable conservation and

ments are also possible in customized sample environments, as in

restoration methods.

the study of corrosion processes in metals as they occur.

Current scientific research is making appropriate contributions to

One research trend focuses on the development of methods for

the conservation of artistic heritage though projects that, for the most

the subsurface investigation of objects without recourse to invasive

part, are dedicated to the improvement of analytical techniques for

transversal cross sections. X-ray based tomography and laminogra-

diagnostic assessment, the identification of the causes and effects of

phy have been exploited to explore in three-dimensions the inner,

deterioration processes, and the development of new methods of

multilayered structure of paintings, demonstrating the possibility of

effective and secure restoration. Such research is especially challeng-

examining the in-depth structure, the state of conservation, and the

ing owing to the fact that historical objects are unique and irreplace-

technology of manufacture of other types of objects. Terahertz time-

able, which demands that analytical investigation must be

domain spectroscopy has been shown to be capable of highlight-

noninvasive or, at the very least, employ microsampling techniques.

ing interfaces in a stratigraphic structure, while macroscopic-scanning

In addition, conservation treatments must be completely reversible

X-ray fluorescence has been used to uncover the subsurface distri-

or executed with materials and methods compatible with the origi-

bution of pigments. Optical coherence tomography exemplifies an

nal object.

appropriate noncontact method of optically sectioning partially trans-

This special issue of Accounts of Chemical Research owes its gen-

parent objects, with a resolution in the micrometer axial range.

esis to Ivano Bertini, who brought together the guest editors at a

Stray-field nuclear magnetic resonance (NMR) also offers oppor-

meeting held in the quintessential city of art, Florence. The Accounts

tunities for research innovation and noncontact subsurface investi-

in this issue comprise a panorama of the most advanced ongoing

gations. Small portable stray-field sensors for NMR relaxation

research, yet, while offering as broad a view as possible, it cannot be

measurements have been developed that have a penetration depth

comprehensive because of the extraordinarily wide range of mate-

of a few millimeters in a variety of materials and a resolution of a

rials encountered in art and the complicated way in which they have

few micrometers. Multiband IR imaging permits better visualization

been used. The research activities described here focus on the devel-

of underdrawings, as well as the distribution of materials in paint-

opment and deployment of advanced analytical methodologies and

ings, by means of the processing of multispectral images as pixel by

procedures, the characterization of the origin and mechanisms of

pixel subtraction and ratio or as false color representations.

material decay, and the creation and testing of new cleaning and consolidation treatments.

Because works of art are especially vulnerable to degradation, the chemical characterization of their organic natural materials is of pri-

Today, a variety of analytical techniques are available that facil-

mary importance. The many significant advances made in the strat-

itate the investigation of the details of alteration processes, the iden-

egies and procedures employed in gas chromatography-mass

tification of degradation mechanisms, and the discovery of the onset

spectrometry (GC-MS) have greatly improved the ability to distin-

Published on the Web 06/15/2010 www.pubs.acs.org/acr 10.1021/ar100072f © 2010 American Chemical Society

Vol. 43, No. 6

June 2010

693-694

ACCOUNTS OF CHEMICAL RESEARCH

693

Guest Editorial

guish complex mixtures in degraded small samples. Of particular

in stoneware, and the use of ancient hybrid materials (inorganic

interest is the problem of determining the distribution of organic

and organic) in ancient buildings. Recent studies have been car-

materials in cross sections of paintings or alteration layers. Second-

ried out as well on materials and painting methods used in the

ary ion mass spectrometry (SIMS) and combined Fourier transform

production of Byzantine icons. These studies not only allow the

infrared and Raman (FT-IR/Raman) microspectroscopy techniques,

objects to be placed in their proper historical and cultural frame-

including mapping and imaging, are suitable for these purposes. To

work but also lead to the development of suitable long-term con-

obtain specific refinements in the identification of proteins, immun-

servation strategies.

ofluorescent probes can be used to detect targeted molecules with high sensitivity and specificity.

In the field of restoration intervention, today nanotechnology allows restorers to provide innovative cleaning and consolida-

Traditionally the identification of dyes has required relatively

tion methods, such as those that utilize water-based micelles,

large samples in order to be analyzed by high-performance liquid

microemulsions, or calcium hydroxide nanoparticles. The meth-

chromatography (HPLC). Recently, successful efforts have been made

ods are both effective and fully respectful of the physicochemi-

to identify dyes from samples as small as a few tens of microme-

cal properties of the original materials used by the artist.

ters in diameter exploiting surface-enhanced Raman spectroscopy

Substantial improvement in restoration is also afforded by the

(SERS). A noninvasive approach to the study of dyes in art is possi-

synthesis of innovative gels that are easily removed from the

ble using UV-vis fluorescence-based analysis. Macro- and microf-

painting surface after cleaning, owing to the fact that they are

luorimetry offer the great advantages of being noninvasive as well

rheoreversible or nonadhesive or because they contain magnetic-

as highly sensitive, despite having the disadvantage of the lack of

coated ferrite nanoparticles.

molecular fingerprints as disclosed by vibrational spectra. This limi-

Laser ablation, too, has a very important role among innova-

tation may be overcome, however, using pertinent databases com-

tive cleaning methodologies, since it offers the advantages of high

piled from measurements taken on accurate reproductions of

control, accuracy, material selectivity, and immediate feedback.

historical dyes.

Prior to its application, it is imperative to test the effectiveness and

Among the approaches discussed here, theoretical calcula-

gradualness of the ablation by means of choosing the appropri-

tions also play a significant role. They permit the structural, elec-

ate pulse duration, maximizing the selectivity, and minimizing the

tronic, and spectroscopic properties of materials to be modeled for

risk of photothermal and photomechanical effects. In this regard,

spectral assignment or modeling of the kinetics of chemical pro-

the potential use of femtosecond pulse lasers to overcome the

cesses in a changing environment, in order to simulate degrada-

limitations of nanosecond ablation methods has been recently

tion phenomena.

explored.

At present, considerable efforts are also under way to develop

The entirety of the work reported in this special issue clearly

portable instrumentation for noninvasive analyses to be carried

shows how, today, chemical research concerning the study and

out on-site, where the object is located or exhibited. Taking the

conservation of art is a very active and productive field. The

laboratory to the art, rather than vice versa, reduces the risks and

advances of recent years can be attributed to the enormous

costs associated with the transportation of precious and delicate

progress made in the development of analytical technologies, as

objects to the lab and opens the way to the scientific examina-

well as to the widespread and increased attention given by the

tion of a great many works of art. The noninvasive approach allows chemists to carry out multitechnique noncontact analyses on a virtually unlimited number of points and in many cases obtain a more thorough description than is possible with a limited number of specimens. Such advanced techniques have answered many questions about the analysis and weathering of glass, as well as elucidat-

public and by policy-makers to the preservation and enhancement of the historical heritage of all nations. The worldwide realization of the social and economic value of the conservation of our heritage, as well as the rapid invention of new technologies, will surely generate exciting perspectives for future research.

ing key points about the mechanisms responsible for damage to

Brunetto Giovanni Brunetti Antonio Sgamellotti

stone induced by in-pore crystallization. As a result, new meth-

Universita´ di Perugia

ods of treatment are being proposed that offer the possibility of attacking the actual cause of the problem, rather than merely

Andrew J. Clark

treating the symptoms. Other advances relevant to conservation

Guest Editors

concern studies of microstructure and color-formation processes

AR100072F

694


693-694

June 2010

Vol. 43, No. 6

New Methodologies for the Conservation of Cultural Heritage: Micellar Solutions, Microemulsions, and Hydroxide Nanoparticles RODORICO GIORGI, MICHELE BAGLIONI, DEBORA BERTI, AND PIERO BAGLIONI*,† Department of Chemistry and CSGI, University of Florence, via della Lastruccia 3 - Sesto Fiorentino, 50019 Florence, Italy RECEIVED ON JULY 4, 2009

CON SPECTUS

M

odern civilization’s inherited artworks have a powerful impact on society, from political, sociological, and anthropological points of view, so the conservation of our Cultural Heritage is fundamental for conveying to future generations our culture, traditions, and ways of thinking and behaving. In the conservation of cultural artifacts, scientists intervene in the degradation of often unique handcrafts, resulting from a delicate balance of aging, unpredicted events, environmental conditions, and sometimes incorrect previous restoration treatments, the details of which are often not precisely known. Nanoscience and nanotechnology are revolutionizing materials science in a pervasive way, in a manner similar to polymer chemistry’s revolution of materials science over the preceding century. The continuous development of novel nanoparticle-based materials and the study of physicochemical phenomena at the nanoscale are creating new approaches to conservation science, leading to new methodologies that can “revert” the degradation processes of the works of art, in most cases “restoring” them to their original magnificent appearance. Until recently, serendipity and experiment have been the most frequent design principles of formulations for either cleaning or consolidation of works of art. Accordingly, the past has witnessed a number of actively detrimental treatments, such as the application of acrylic and vinyl resins to wall paintings, which can irreversibly jeopardize the appearance (or even the continued existence) of irreplaceable works of art. Current research activity in conservation science is largely based on the paradigm that compatibility of materials is the most important prerequisite for obtaining excellent and durable results. The most advanced current methodologies are (i) the use of water-based micelles and microemulsions (neat or combined with gels) for the removal of accidental contaminants and polymers used in past restorations and (ii) the application of calcium hydroxide nanoparticles for the consolidation of works of art. In this Account, we highlight how conservation science can benefit from the conceptual and the methodological background derived from both soft (microemulsions and micelles for cleaning) and hard (nanoparticles for consolidation) nanoscience. A combination of different nanotechnologies allows today’s conservators to provide, in each restoration step, interventions respectful of the physicochemical characteristics of the materials used by artists. The “palette” of methods provided by nanoscience is continuously enriching the field, and the development of novel nanomaterials and the study of nanoscale physicochemical phenomena will further improve the performance of restoration formulations and our comprehension of degradation mechanisms.

Published on the Web 03/05/2010 www.pubs.acs.org/acr 10.1021/ar900193h © 2010 American Chemical Society

Vol. 43, No. 6

June 2010

695-704


695

Nanoscience for the Conservation of Cultural Heritage Giorgi et al.

Introduction It is well accepted that modern Conservation Science originated from the tragic floods that devastated Florence and Venice in 1966, imposing the search for new methodologies to restore and conserve the immense Cultural Heritage heavily damaged by the flood. A scientific method for conservation of carbonatic materials was proposed for the first time by Enzo Ferroni at the International ICOM Conference in Amsterdam (1969).1,2 This new method for the “in situ consolidation” of wall paintings was able to “restore”, in a two-step series of chemical reactions, the degradation of calcium carbonate, the principal constituent of wall paintings, drastically damaged by the flood. The new Science for Conservation has developed in a fast and healthy way following two main streams: (i) the analytical characterization of the materials constituting the works of art, the characterization of the pictorial technique used by the artists, and the chemical reactions involved in their degradation; (ii) the search for new scientific methods for the restoration/conservation, allowing the transfer to future generations of our Cultural Heritage. The last point will be the object of this Account, where we will review some of the most important methodologies developed, mainly using the cultural framework of soft-condensed matter and nanoscience. Generally speaking, the restoration of a work of art consists of two distinct aspects: (1) cleaning, which is a transient treatment, meant to remove the materials not originally belonging to the work of art. (2) consolidation, which is a durable intervention that should remediate, prevent, or slow down further degradation due to aging or external agents. All of the current research activity in Conservation Science is mainly based on the paradigm that compatibility of materials is the most important prerequisite in order to obtain excellent and durable results. In view of this seemingly intuitive, yet novel, awareness, it is clear that the use of polymers for the consolidation of inorganic materials should be avoided as much as possible in the restoration of the original features of the unaltered material. Currently this is mainly achieved thanks to the tremendous advancements witnessed by nanoscience and nanotechnology that provide theoretical and technical backgrounds to formulate innovative systems for restoration.3,4 We pioneered the application of nanoscience to Cultural Heritage conservation, and we devised several methods for the consolidation of wall paintings, for paper and wood deacidification, and for the removal of grime, dirt, polymers, and other organic materials soiling the original artifacts.5-12 In this 696


695-704

June 2010

Vol. 43, No. 6

Account, we will highlight (as an example of the potential of these new “nanotools”) the use of “soft” microemulsions and micelles for cleaning and “hard” nanoparticles for consolidation of works of art.

Degradation of Polymer Resins Used in Past Restoration Reversibility of the restoration treatment has been presented for decades as a milestone in conservation.13 This principle has supported several conservators in the usage of polymer resins, made available by the industry since 1960, which were considered the “panacea” for the restoration of works of art. Consolidation and protection of paintings, stone, wood, paper, glass, ceramic, and bones was made possible by simple application of organic solutions, through immersion, injection, or brushing, of acrylic, vinylic, and silicone-based polymers and (their) copolymers. Most of these polymers were considered “reversible”, that is, the solvent used for the application could be later used for the removal (if desired) of the applied resins. The large use of polymers was also favored by the “the brilliant water-white color of the polymers” that “makes it possible to secure masses of high light transmission and great optical clarity - because of the remarkable stability of the polymers to the action of heat and light, these properties are permanent” and by the “many striking and beautiful effects can be produced by the incorporation of dyes and pigments”.14 Unfortunately, the application of polymer coatings induces severe modifications of the main physicochemical properties of artifacts, particularly for inorganic matrices as wall paintings.15,16 Investigations performed on Paraloid B72 (or Acryloid) (EMA/MA 70:30 w/w; ethyl methacrylate/methyl acrylate; weight average molecular weight 88 000 g/mol; supplier Rohm and Haas (USA)), Elvacite 2046 (nBMA/iBMA; n-butyl methacrylate/i-butyl methacrylate), and Primal AC33 (or Rhoplex, EA/MMA 60:40 w/w; ethyl acrylate/methyl methacrylate; viscosity (Pa · s) 6, Tg (°C) 16; supplier Rohm and Haas (USA)) showed that water capillarity, water vapor permeability, and surface wettability of lime mortar change drastically. The application of polymers leads to a strong increase of surface hydrophobicity, to a decrease of about 70-85% of the capillary rise coefficients and 40-50% of the vapor permeability, and to an increase of contact angle from about 30° to 100°.17 When the surface is coated with a polymer, salt crystallization occurs within the pores of the wall paintings, generating strong mechanical stress beneath the painted layer (50-100 µm) and leading to the complete disruption of the pictorial layer in a period of time from about 5 to 30-40


FIGURE 1. Mural paintings in “Templo de los Nichos Pintados” in Mayapan (Yucatan). The pictures show the degradation that occurred to the paintings from the last restoration in 1999 due to the mowilith DM5 coating.

years, depending on the environmental conditions. The alteration of the natural “breathing” of the surface due to the presence of the polymer increases consistently the deleterious mechanical effect of salt crystallization from the interior of the wall with the complete detachment of the painting from its support, that is, the surface of the work of art. In addition to the degradation of the painted layer promoted by the drastic change of the physicochemical characteristics, polymers themselves degrade, and upon aging they are subjected to molecular weight changes due to cross-linking reactions and chain scissions. The main consequence of cross-linking is the loss of solubility, which makes the application irreversible.18-20 Photoaging under UV light was examined by some authors to detect the molecular weight distribution change due to chain scissions or coupling of macroradicals.18 The overall stability of the polymers was shown to be influenced by the alkyl side groups, particularly in the case of paraloid B66 (nBMA; poly(methyl,nbutyl methacrylate)) and paraloid B67 (iBMA; poly(isobutyl methacrylate)), where relatively longer ester chains are present. In these compounds, the oxidation is favored by hydrogen atoms, either on the methylenes of the n-butyl groups in B66 or on the tertiary carbon of the isobutyl groups in B67. Polymers containing n-butyl ester groups undergo fast and extensive cross-linking together with some chain scissioning. Paraloid B72 and paraloid B82 (MMA, poly(methyl methacrylate)) show a good stability toward oxidation due to the presence of the EMA and MMA methacrylic units19 and seem to be suitable as consolidants and protective agents for applications to wall paintings and other artifacts. However, it is important to notice that the composition of commercially available products contains almost 1% i-BMA, formed during the polymerization reaction. Even tiny amounts of i-BMA may initiate radical formation leading to extensive cross-linking reactions, similarly to paraloid B66 and B67. The difference in paraloid B72 with the other member of this polymer family is mainly related to the slower

degradation process but is quite common the evidence of enhanced degradation generated by paraloid B72 (as well as B66 and B67) used in past restorations. The degradation effects are particularly evident in locations where the environmental conditions are severe due to fluctuations of temperature and relative humidity or in very polluted urban areas where photochemical smog favors polymer degradation, as demonstrated in a recent paper by Favaro and co-workers.20 Conservation treatments based on resins have produced irreversible degradation of most of Mesoamerican paintings.21 Recent investigations in the archeological site of Mayapan (Mexico, Post-Classic Age 1200-1450 a.C.) have demonstrated the dramatic conditions of these paintings recently submitted to consolidation and protection treatments with mowilith (vinyl acetate/n-butyl acrylate 65:35 w/w copolymer) resins. A photographic documentation reveals that in less than 10 years a relevant portion of the original painting was lost (see Figure 1). This was due to the combination of the mowilith film with extensive salt crystallization processes that quickly destroyed the plaster beneath the paint layer. In addition to the mechanical degradation, the natural aging of polymer resins generates discoloration and loss of solubility, due to the increase of the polymer molecular weight. At the end of the degradation process, these polymers are very difficult or impossible to remove with conventional neat solvents. Considering the severe side effects generated by the use of polymers and the necessity of their removal, we proposed the use of nanostructured fluids; some examples are highlighted in the following.

Cleaning and Removal of Polymeric Coatings by Means of Amphiphile-Based Systems Historically, the first application of soft matter concepts was performed in Florence at the end of the 1980s during the restoration of the Renaissance masterpiece by Masaccio, Vol. 43, No. 6

June 2010

695-704


697


Masolino, and Lippi in the Brancacci Chapel in Florence (1984-1990).22 Investigations under UV light of the paintings revealed the presence of a multitude of wax spots due to the extinction of votive candles kept close to the wall painting. The removal of this hydrophobic material from the hydrophilic fresco painting was obtained using microemulsions.23-26 The microemulsion used in Brancacci’s paintings was very similar to the so-called “French microemulsion” and was formed by nanosized dodecane droplets, dispersed in an aqueous solution of ammonium27 dodecylsulfate (surfactant) and 1-butanol (cosurfactant). The oil in water microemulsion ensured low aggressiveness with respect to the original components of the painted layer, due to the presence of water as a dispersing medium that remained in direct contact with the hydrophilic surface of the wall painting.24 Since this pioneering example, several other surfactantbased micellar and microemulsive systems have been used by our group to clean wall paintings and canvases both from accidental contamination and from polymers that have been extensively used as consolidants and protectives in past restorations. Nanocompartmentalized systems specifically tailored for the removal of paraloid B72, primal AC33, and mowilith DM5 (vinyl acetate/n-butyl acrylate 65:35 w/w copolymer; viscosity (Pa · s) 3.5, Tg (°C) 2; supplier Hoechst, Germany) resins have been recently formulated by some of us. Both micelles and swollen micelles work as nanocontainers of solvents that can dissolve polymers to obtain their complete removal from the surface and from the porous texture of the work of art. The detergency capability mainly depends on the very large surface area of micelles and nanodroplets available for interaction with the polymer coating. The water-based system reduces the penetration into the artifact porous matrix that occurs with organic solvents and the associated polymer redissolution into the artifact but also reduces the toxicity of the formulation, offering better and faster performance than pure organic solvents. Table 1 reports the composition of some of the most successful formulations used for cleaning. Oil in water microemulsion made with sodium dodecyl sulfate (SDS), 1-pentanol (PeOH) as a cosurfactant, and a small amount of a mixture of p-xylene and nitro diluent (a mixture of 62% toluene, 15% butyl acetate, 15% ethyl acetate, 6% n-butyl alcohol, 2% cellosolve acetate) was shown to be effective for paraloid B72 removal (Classical system; see Table 1). Polar polymer resins such as poly(vinyl acetate), not soluble in the oil mixture above-reported, were successfully removed by using a four-component micellar solution with a high amount 698


695-704

June 2010

Vol. 43, No. 6

TABLE 1. Composition (% w/w) of Systems Used for Polymer Removal from Wall Paintingsa Classical component water SDS 1-PeOH p-xylene nitro diluent (ND)

Conegliano

composition component composition 86.2 3.9 6.5 1.8 1.6

water SDS 1-PeOH PC

69 5.1 3.9 22

Mayapan component

composition

water SDS 1-PeOH PC ethyl acetate (EA)

73.3 3.7 7 8 8

a

Classical refers to the first o/w microemulsion formulated for paraloid B72 removal; the other formulations are named after the places where they were used for the first time.

of propylene carbonate (PC), which replaced the nonpolar mixture of xylene and nitro diluent (Conegliano system). Previous investigations have shown that PC acts as a cosurfactant inducing a rod-to-sphere micelle transition reducing the size and aggregation number of SDS micelles by increasing the headgroup area.28,29 The Conegliano system could also be used for paraloid B72 removal since it favors the swelling of paraloid B72 films, which can be then removed by mechanical action for a complete cleaning of the surface. Very often paraloid B72 and mowilith DM5 have been used in mixtures or sequentially applied to wall paintings. An ideal system should remove simultaneously both resins. A major improvement in this direction has been recently obtained by adding ethyl acetate (EA) to the Conegliano formulation to obtain the Mayapan micelles, whose composition is reported in Table 1. The aromatic and aliphatic solvents of the Classical formulation are replaced by less toxic solvents. Pure ethyl acetate and propylene carbonate are known to solubilize paraloid B72. They are partially water-soluble (EA 8% w/w, PC 20% w/w), but a saturated aqueous solution of EA and PC provides only partial swelling of the polymer. The addition of SDS and 1-PeOH either to the EA or PC aqueous solutions provides a drastic improvement in the efficiency of the removal process, due to the presence of micelles that provide the large interfacial area necessary for polymer uptake. It is important to stress here that micellar solutions formed by SDS or SDS/PeOH alone are practically ineffective in polymer removal; therefore it is the synergistic action of the presence of micelles and PC that causes the polymer removal. Even better performance was obtained by adding simultaneously 8% EA and 8% PC to aqueous SDS/1-PeOH micelles (Mayapan system). This system allowed a complete cleaning of the surfaces treated with paraloid B72 in a few minutes. The Classical system is very efficient in removing acrylic polymers, but the polymer removal is quite slow compared with the Mayapan micelles. This is probably due to the lower amount of


organic solvents contained in the former leading to different microstructural features of the system. It is clear from the above examples that the design of efficient formulations for the tailored removal of polymers requires a deep comprehension of the solubilization mechanisms, which can be more complex than traditional detergency, due to the presence of partially water-soluble organic solvents (cosolvents). To better understand the polymer removal process, we performed swelling tests of polymer films deposited in glass vials and detergency tests on paraloid films on models of wall paintings. Quasi-elastic light scattering (QELS) is an efficient and noninvasive tool to characterize size and size distribution of complex systems in the nanometer domain. Figure 2 shows that FIGURE 2. Size distribution of the Mayapan micellar system obtained from QELS.

the intensity weighted size distribution of Mayapan micelles is centered around 4 nm with a relatively broad size distribution, while the average droplet diameter of the Classical sys-

FIGURE 3. Schematic representation (bottom panel) of the differences in the removal mechanisms of paraloid B72 performed by Conegliano and Mayapan formulations. Conegliano system only swells the polymer layer, while the application of Mayapan micelles to a paraloid B72-coated surface results in a liquid-liquid phase separation. SAXS spectra (upper panel) obtained after the application of Mayapan micelles to a paraloid B72-coated surface. The more dense phase (green line) still contains micelles, while the upper phase (red line) contains the polymer dissolved in a blend of EA/PC/PeOH.

Vol. 43, No. 6

June 2010

695-704


699


FIGURE 5. Experimental cleaning test on model fresco paintings treated with paraloid B72: (A) water/SDS/1-PeOH/(8%)EA; (B) water/ PC; (C) Classical system; (D) Conegliano system; (E) water/EA/PC; (F) Mayapan system.

FIGURE 4. Schematic representation of the hypothesized detergency mechanism of the Mayapan (upper panel) and Classical (lower panel) systems. While in the first case the partially watersoluble solvents (EA and PC) partition among the aqueous and the micellar phase, in the second case, we have a “traditional” microemulsion where the water-insoluble oil core is dispersed in water by the presence of a surfactant film. The “naked” micelles (SDS/PeOH) do not detatch paraloid B72 from the surfaces. In this cartoon, the Mayapan micelles act as solvent carriers and quickly mediate their interaction with the polymer, which “chooses” the right amounts of solvents to undergo a swelling process. It is worth noticing that the micelles of the Mayapan system, after the interaction with paraloid B72, change about 10% in size, while the microdroplets of the Classical system get sensibly smaller; for microemulsions, the kinetics is slower, due to the oil confinement in the micellar core. After the interaction, the micelles decrease in size because of the solvent release.

tem is about 17 nm with polydispersity of 25%, typical of a common o/w microemulsion. The smaller micellar size, together with the presence of specific organic solvents, results, for a given surfactant content in a very high interfacial area that is correlated to the kinetics of the cleaning process. The Conegliano “micelles” are characterized by very small aggregates with a hydrodynamic diameter of about 2 nm28,29 and refer to clusters formed by 7-10 SDS monomers and not to conventional micelles. For the sake of clarity, we indicated these clusters as “micelles”. Mayapan “micelles” are similar but present a higher effectiveness toward polymer removal that is nicely illustrated by the results of a semiquantitative test on 700


695-704

June 2010

Vol. 43, No. 6

model systems that we used to compare the removal efficiency of these two systems. Coating the bottom of two vials with a known amount of paraloid B72 (whose thickness is comparable to that spread on art handcraft) and putting the film in contact with Conegliano and Mayapan micellar solutions (see Figure 3, bottom), we found that the Conegliano micelles are able to swell the polymer, but the swollen polymer remains at the bottom of the vial. Mayapan micelles were also able to swell the polymer but in addition a spontaneous detaching of the polymer from the glass was observed with a liquid-liquid phase separation in the micellar phase. This behavior is ideal for polymer removal from artifacts. In order to understand this mechanism, both phases have been investigated with static and dynamic scattering techiniques. Figure 3 reports the SAXS spectra obtained from the polymer-rich and the surfactant-rich phases. The less dense phase contains the polymer and a mixture of H2O, PeOH, PC, and EA solvents, enriched in the less polar components. The more dense and abundant phase is still a micellar solution but is depleted of the organic phase. It is clear that the mixture of organic solvents, dispersed in the aqueous medium thanks to the presence of micelles, guarantees specific interactions with the polymer (see Figure 4) suggesting a mechanism where the Mayapan micelles work as a carrier of solvents during the phase separation process that is the final step of the cleaning procedure. Figure 5 shows a brick with a 1 cm thick lime plaster layer, painted with fresco technique, treated with paraloid B72. Visual


FIGURE 6. Mural painting sample treated, on the right side, with mowilith DM5 under glazing light (center), and SEM pictures of the uncoated (left) and coated (right) surface.

FIGURE 7. Details of the painting surface after removal of mowilith DM5 coating by using micellar solution (Mayapan system).

trative examples of a fresco painting treated with the Mayapan system.

Confinement of Micellar Solutions and Microemulsions in Chemical Gels

FIGURE 8. Linen canvas glued with mowilith DMC2 for a relining treatment. In the red box, the appearance of the treated area after micellar solution application.

analysis of this model painting after cleaning showed a complete recovery of the optical features of the surface in the area treated with the Mayapan system (see Figure 5F). The Classical and Conegliano systems gave slightly worse results, as is clearly observable from the polymer residues still present in the treated areas as shown in Figure 5, areas C and D, respectively. As already mentioned, the Mayapan micelles are also active for mowilith DM5 removal. Figures 6 and 7 report illus-

Microemulsions and micellar solutions are usually applied over a painted surface by using cellulose-pulp compresses or physical gels soaked with them.30 In many cases, as in the case of canvas or wood painting cleaning, a more controlled release of the microemulsion is required to avoid the swelling of inorganic and organic binding materials by water, that is, the most abundant component in an oil in water microemulsion. The advantage of using gels resides in the “easy” manipulation, the possibility of microemulsion and micellar solution confinement into the gel network, and the possibility to make them responsive to external stimuli (i.e., pH, magnetic fields, temperature, light).31-35 Gel structure may result from formation of relatively weak physical bonds (i.e., hydrogen bonds or van der Waals interactions) or covalent chemical bonds; both can be responsible for the cross-linking of particles or polymer chains to create the gelled state. In the first case, we refer to a physical gel, in the second to a chemical gel. Vol. 43, No. 6

June 2010

695-704


701


FIGURE 9. Wall paintings from Mayan Classic period in Calakmul (Mexico). Calcium hydroxide nanoparticles were used for consolidation.

Among the large number of polymers that can be used for gelation, cellulose derivatives, polyacrylic acids, and polyamines are the most used in the field of conservation science. Unfortunately, these physical gels present some difficulties in their removal and may release residues on the surface of the work of art after the application. Besides, the gels’ mesh size is difficult to control during gelation.36,37 Chemical gels are characterized by stronger interactions in the gel molecular network. This type of gel behaves like a soft solid, and the mesh size is tunable by controlling the crosslinking during the polymerization reaction. Chemical gels, based on an acrylamide/bis-acrylamide network and PEG or silane cross-linkers, loaded with the detergent solutions have been recently synthesized and applied to Conservation.11,38 The microemulsion release or uptake can be modulated by controlling the gel mesh size or applying a stimulus. In a previous paper,38 it was shown by SAXS that a microemulsion retains its structure when loaded into the gel. The uptake/ release capability of the gel was quantified by analysis of hydration/dehydration curves. Microemulsion-loaded gel was shown to be effective for the removal of paraloid B72 from limestone and wall paintings. Cleaning tests on canvases, previously treated with adhesives (i.e., mowilith DMC2 (vinylacetate/dibutylmaleate copolymers) and acrylate/methacrylate copolymers) that are normally used for the relining of canvas paintings, have been performed with a chemical gel, tailored in order to have good water retention. The gel was loaded with the Mayapan system specifically developed for the swelling and the removal of acrylate/methacrylate and vinylacetate/butylacrylate copolymers. After a 2-hour application, the capillary sorption of water through the ground and paint layer was restricted to the con702


695-704

June 2010

Vol. 43, No. 6

tact area of gel over the canvas. The swollen polymers were partly removed by the gel, and the complete cleaning was accomplished by gentle mechanical action with a blade (see Figure 8).39

Inorganic Nanoparticles for the Consolidation of Wall Paintings Very often the cleaning process (particularly for wall paintings) cannot be completely performed without a concomitant consolidation of the painted layer. Consolidation is often forced by the degradation of the painted layer enhanced by the polymer, as occurs in Mexico where the unfavorable climate conditions40,41 increase the rates of polymer degradation generating severe flaking of surfaces and paint detachments in a very short time. Consolidation of wall paintings and limestone by using compatible material is now achieved by using calcium hydroxide nanoparticle dispersions that are completely compatible with the original material constituting the works of art. They present increased reactivity and dramatically different transport properties in porous media.42 The high surface area of nanoparticles influences their chemical reactivity, producing just few days after the application a consistent consolidation of the treated surfaces, due to the reaction of hydroxide with CO2 from the air to give crystalline calcium carbonate (calcite). These positive effects are particularly important in archeological sites where the conservation in situ usually requires an immediate intervention after discovery. Nanoparticles have been used in several restoration workshops in Europe and are extensively used in Mexico for the in situ consolidation of paintings (see Figure 9) and limestone in “La Antigua Ciudad Maya de Calakmul”, a UNESCO World Heritage Site since 2002 (Campeche, Mexico).43,44 Nanoparticles


of calcium and barium hydroxide have been used in Calakmul for in situ consolidation of the wall paintings (from the Classic period (600-700 AD)) discovered during the excavations from 2004 to 2008 in the Structure I of the Acro´polis Chik Naab. Figure 9 reports an image of one of the main buildings in the Calakmul area, and some parts of the mural paintings were consolidated by using calcium hydroxide nanoparticles. The use of hydroxide nanoparticles is not restricted to consolidation of wall paintings but can be used to control the pH in cellulosic material (paper) or wood to avoid the catalytic cleavage of the β-glycosidic bond or the Fenton reaction usually occurring in the presence of transition metal ions, opening new perspective in the conservation of an enormous number of documents and artifacts.6,7

Conclusions and Outlook Conservation science has matured using plenty of methodological tools borrowed from nanoscience. Soft matter, surface science, detergency, and polymer and organic chemistry contribute to this new branch of science. Micelles, microemulsions, and nanoparticle systems are valuable systems for the optimal conservation of artifacts. These systems have many avenues for additional improvements at the hands of researchers. Improvements of these technologies will be possible through accurate investigations aiming at the knowledge of the following: • extraction capability and transport mechanisms in microemulsions and micellar solutions • transport mechanisms and structure of the physical and chemical gels • extraction capability and transport mechanisms of confined microemulsion and micellar systems • synthesis of new organic and inorganic nanoparticles and nano- or microcontainers for repairing or self-repairing These new smart materials add a new palette to the “classical” methodologies for the “cleaning” of works of art. Paraphrasing a well-known sentence we could say that “there is plenty of room” for nanoscience in Conservation. Thanks are due to Massimo Bonini, Doris Rengstl, Ilaria Lapini, Giacomo Pizzorusso, and David Chelazzi (CSGI) for the assistance in laboratory experiments, Michel Menu and Aurelia Chevalier (Centre de Recherche et de Restauration des Museés de France C2RMF) for the experiments on canvas paintings, Carla Giovannone, Tiziana Dell’Omo, and Lucia Di Paolo (ISCR, Rome) for assistance with the application tests on model wall

paintings, and Ramon Carrasco Vargas (director of the “Proyecto Arqueologico Calakmul”-Mexico), Diana Magaloni (UNAM-Mexico and director of the Museo Nacional de Antropologia), Maria Del Carmen Castro and Yareli Jaidar Benavides (CNCPC-Mexico), and Claudia Garcia Solis, Conservator of the “Proyecto Arqueologico Mayapan” (Mexico) for experiments in situ. Financial support from CSGI and Bilateral project ItalyMexico is acknowledged. BIOGRAPHICAL INFORMATION Rodorico Giorgi, Ph.D. in Science for Cultural Heritage Conservation at the University of Florence and master in Chemistry in 1996, is currently research fellow at the Department of Chemistry and CSGI. Giorgi’s main research activity is in physicochemical characterization of materials, investigation of degradation processes and the development of methodologies for the conservation of works of art materials (wall and canvas paintings, stone, paper, photographs, and wood). Giorgi is author of about 60 publications. Michele Baglioni is a doctoral candidate in Science for Cultural Heritage Conservation at the University of Florence and holds a Master in Technology for the Conservation of Cultural Heritage. He is currently working in the development of soft matter systems for conservation. Debora Berti obtained her Laurea Degree in Chemistry in 1993 and a Ph.D. in Physical Chemistry in 1997. She has been Visiting Scientist at the ETH (Zurich) during 1994-1995 and postdoctoral fellow at CSGI until 2000. In 2000, she joined the University of Florence, where she coordinates a research group working on phospholipid and nucleolipid self-assembly. She is the author of more than 60 publications and has presented more than 100 invited and contributed lectures to international conferences. Her background in physical chemistry of surfactant self-assembly has gradually shifted throughout the years to self-assembly of bioinspired and biorelevant functional amphiphiles, using scattering techniques. Piero Baglioni is a Full Professor of Physical Chemistry at the Department of Chemistry and CSGI of the University of Florence. He is the author of about 300 publications in the field of colloids and interfaces and a pioneer in the application of soft matter to the conservation of Cultural Heritage. He has produced several innovative methods applied worldwide. FOOTNOTES * To whom correspondence should be addressed. E-mail: [email protected]. Phone: +39 0555253033. Fax: +39 0555253032. † No kinship exists among the authors. REFERENCES 1 Ferroni, E.; Malaguzzi, V.; Rovida, G. Experimental Study by Diffraction of Heterogeneous Systems as a Preliminary to the Proposal of a Technique for the Restoration of Gypsum Polluted Murals. ICOM-CC Plenary Meeting - The International Council of Museums-Committee for Conservation, Amsterdam, Netherlands, 1969 [Paper read at the Amsterdam ICOM meeting in 1969; copy in ICCROM library, Rome].

Vol. 43, No. 6

June 2010

695-704


703


2 Ferroni, E.; Baglioni, P. Experiments About the Method for the Restoration of Sulfated Frescoes. In Proceedings of the Symposium ‘Scientific Methodologies Applied to works of Art’, Florence, May 2-5, 1984; Parrini, P. L., Ed.; Montedison Progetto Cultura: Milan, 1986, pp 108-109. 3 Baglioni, P.; Giorgi, R. Soft and Hard Nanomaterials for Restoration and Conservation of Cultural Heritage. Soft Matter 2006, 2, 293–303. 4 Baglioni, P.; Giorgi, R.; Dei, L. Soft Condensed Matter for the Conservation of Cultural Heritage. C. R. Chim. 2008, 12, 61–69. 5 Ambrosi, M.; Dei, L.; Giorgi, R.; Neto, C.; Baglioni, P. Colloidal Particles of Ca(OH)2: Properties and Application to Restoration of Frescoes. Langmuir 2001, 17, 4251– 4255. 6 Giorgi, R.; Dei, L.; Ceccato, M.; Schettino, C. V.; Baglioni, P. Nanotechnologies for Conservation of Cultural Heritage: Paper and Canvas Deacidification. Langmuir 2002, 18, 8198–8203. 7 Giorgi, R.; Bozzi, C.; Dei, L.; Gabbiani, C.; Ninham, B. W.; Baglioni, P. Nanoparticles of Mg(OH)2: Synthesis and Application to Paper Conservation. Langmuir 2005, 21, 8495–8501. 8 Giorgi, R.; Chelazzi, D.; Baglioni, P. Nanoparticles of Calcium Hydroxide for Wood Conservation. The Deacidification of the Vasa Warship. Langmuir 2005, 21, 10743– 10748. 9 Carretti, E.; Dei, L.; Baglioni, P. Solubilization of Acrylic and Vinyl Polymers in Nanocontainer Solutions. Application of Microemulsions and Micelles to Cultural Heritage. Langmuir 2003, 19, 7867–7872. 10 Carretti, E.; Giorgi, R.; Berti, D.; Baglioni, P. Oil-in-Water Nanocontainers as Low Environmental Impact Cleaning Tools for Works of Art: Two Case Studies. Langmuir 2007, 23, 6396–6403. 11 Bonini, M.; Lenz, S.; Giorgi, R.; Baglioni, P. Nanomagnetic Sponges for the Cleaning of Works of Art. Langmuir 2007, 23, 8681–8685. 12 Carretti, E.; Grassi, S.; Cossalter, M.; Natali, I.; Caminati, G.; Weiss, R. C.; Baglioni, P.; Dei, L. Poly(vinyl alcohol)-Borate Hydro/Cosolvent Gels: Viscoelastic Properties, Solubilizing Power, and Application to Art Conservation. Langmuir 2009, 25, 8656– 8662. 13 Mora, P.; Mora, L.; Philipot, P. Conservation of Wall Painting; Butterworths: London, 1984. 14 Neher, H. T. Acrylic Resins. Ind. Eng. Chem. 1936, 28, 267–271. 15 Resins in Conservation, Proceedings of the Symposium, Edinburgh, 21 and 22 May, 1982; Tate, J. O., Tennent, N. H., Eds.; Scottish Society for Conservation and Restoration: Edinburgh, 1983. 16 Horie, C. V. Materials for Conservation; Butterworths: London, 1987. 17 Carretti, E.; Dei, L.; Baglioni, P. Aqueous Polyacrylic Acid Based Gels: Physicochemical Properties and Applications in Cultural Heritage Conservation. Prog. Org. Coat. 2004, 49, 282–289. 18 Lazzari, M.; Chiantore, O. Thermal-Ageing of Paraloid Acrylic Protective Polymers. Polymer 2000, 41, 6447–6455. 19 Chiantore, O.; Lazzari, M. Photo-Oxidative Stability of Paraloid Acrylic Protective Polymers. Polymer 2001, 42, 17–27. 20 Favaro, M.; Mendichi, R.; Ossola, F.; Rosso, U.; Simon, S.; Tomasin, P.; Vigato, P. A. Evaluation of Polymers for Conservation Treatments of Outdoor Exposed Stone Monuments. Part I: Photo-Oxidative Weathering. Polym. Degrad. Stab. 2006, 91, 3083–3096. 21 Orea, H.; Magar, V. A Brief Review of the Conservation of Wall Paintings in Mexico. In Preprints of the 13th Triennial Meeting ICOM Committee for Conservation, ICOMCC, Rio de Janeiro, 22-27 September 2002; Vontobel, R., Ed.; James and James (Science Publishers): London, 2002; p 176. 22 Borgioli, L.; Caminati, G.; Gabrielli, G.; Baglioni, P. Removal of Hydrophobic Impurities from Pictorial Surfaces by Means of Heterogeneous Systems. Sci. Tech. Cult. Herit. 1995, 4, 67–74. 23 De Gennes, P. G.; Taupin, C. Microemulsions and the Flexibility of Oil/Water Interfaces. J. Phys. Chem. 1982, 86, 2294–2304. 24 Ferroni, E.; Gabrielli, G.; Caminati, G. Asportazione di Materiali Idrofobi da Superfici Pittoriche Murali Mediante Solubilizzazione in Sistemi Dispersi. In La Cappella Brancacci, la scienza per Masaccio, Masolino e Filippino Lippi, Quaderni del restauro; Olivetti: Milan, 1992; pp 162-171.

704


695-704

June 2010

Vol. 43, No. 6

25 Friberg, S. E.; Flaim, T. Surfactant Association Structures. In Inorganic Reactions In Organized Media; Holt, S. L., Ed.; ACS Symposium Series 177; American Chemical Society: Washington, DC, 1982; pp 1-17. 26 Fontell, K.; Ceglie, A.; Lindman, B.; Ninham, B. W. Some Observations on Phase Diagrams and Structure in Binary and Ternary Systems of Didodecyldimethylammonium Bromide. Acta Chem. Scand. 1986, A40, 247–256. 27 The ammonium counterion was chosen because the painting, after the cleaning from the wax, was consolidated using the Ferroni-Dini method, which uses the ammonium carbonate in the first step of the consolidation process. 28 Palazzo, G.; Fiorentino, D.; Colafemmina, G.; Ceglie, A.; Carretti, E.; Dei, L.; Baglioni, P. Nanostructured Fluids Based on Propylene Carbonate/Water Mixtures. Langmuir 2005, 21, 6717–6725. 29 Colafemmina, G.; Fiorentino, D.; Ceglie, A.; Carretti, E.; Fratini, E.; Dei, L.; Baglioni, P.; Palazzo, G. Structure of SDS Micelles with Propylene Carbonate as Cosolvent: A PGSE-NMR and SAXS Study. J. Phys. Chem. B 2007, 111, 7184–7193. 30 Carretti, E.; Fratini, E.; Berti, D.; Dei, L.; Baglioni, P. Nanoscience for Art Conservation: Oil-in-Water Microemulsions Embedded in a Polymeric Network for the Cleaning of Works of Art. Angew. Chem., Int. Ed. 2009, 48, 8966–8969. 31 Baglioni, P.; Dei, L.; Carretti, E.; Giorgi, R. Gels for the Conservation of Cultural Heritage. Langmuir 2009, 25, 8373–8374, and references therein. 32 Carretti, E.; Dei, L.; Weiss, R. G. Soft Matter and Art Conservation: Rheoreversible Gels and Beyond. Soft Matter 2005, 1, 17–22. 33 Peng, F.; Li, G.; Liu, X.; Wu, S.; Tong, Z. Redox-Responsive Gel-Sol/Sol-Gel Transition in Poly(acrylic acid) Aqueous Solution Containing Fe(III) Ions Switched by Light. J. Am. Chem. Soc. 2008, 130, 16166–16167. 34 Liu, Q.; Wang, Y.; Li, W.; Wu, L. Structural Characterization and Chemical Response of a Ag-Coordinated Supramolecular Gel. Langmuir 2007, 23, 8217–8223. 35 Carretti, E.; Dei, L.; Baglioni, P.; Weiss, R. G. Synthesis and Characterization of Gels from Polyallylamine and Carbon Dioxide as Gellant. J. Am. Chem. Soc. 2003, 125, 5121–5129. 36 Wolbers, R. C. Cleaning Painted Surfaces: Aqueous Methods; Archetype Publications: London, 2000. 37 Stulik, S.; Miller, D.; Khanjian, H.; Khandekar, N.; Wolbers, R.; Carlson, J.; Petersen, W. C. In Solvent Gels for the Cleaning of Works of Art: The Residue Question; Dorge, V., Ed.; The Getty Conservation Institute: Los Angeles, 2004; pp 18-83 and references therein. 38 Bonini, M.; Lenz, S.; Falletta, E.; Ridi, F.; Carretti, E.; Fratini, E.; Wiedenmann, A.; Baglioni, P. Acrylamide-Based Magnetic Nanosponges: A New Smart Nanocomposite Material. Langmuir 2008, 24, 12644–12650. 39 Chevalier, A.; Chelazzi, D.; Baglioni, P.; Giorgi, R.; Carretti, E.; Stuke, M.; Menu, M.; Duchamp, R. Extraction d’Adhe´sifs de Rentoilage en Peinture de Chevalet: Nouvelle Approche. In Proceedings of the 15th Triennal Conference ICOM Committee for Conservation, New Delhi, September 22-26; Bridgland, J., Ed.; Allied Publishers Pvt. Ltd: New Delhi, 2008; Vol. II, pp 581-589. 40 Riederer, J. The Restoration of Archaeological Monuments in the Tropical Climate. In Proceedings of 7th Triennial ICOM Meeting, Copenhagen, 10-14 September 1984; De Froment, D., Ed.; ICOM Committee for Conservation: Paris, 1984; pp 21-22. 41 Espinosa, A. Conservation and Restoration of the Murals of the Temple of the Paintings in Bonampak. In Proceedings of in situ Archaeological Conservation, 613 April 1986, Mexico City; Hodges, H. W. M., Ed.; INAH: Mexico City, 1987; pp 84-89. 42 Odom, T. W.; Pileni, M.-P. Nanoscience (guest editorial). Acc. Chem. Res. 2008, 41, 1565 and articles collected in this special issue. 43 Giorgi, R.; Chelazzi D.; Carrasco, R.; Colon, M.; Desprat, A.; Baglioni, P. The Maya Site of Calakmul: In situ Preservation of Wall Paintings and Limestone by Using Nanotechnologies. Proceedings of the IIC Congress 2006, Munich - The Object in Context: Crossing Conservation Boundaries; Saunders, D., Townsend, J. H., Woodcock, S., Eds.; James and James: London, 2006; pp 162-169. 44 Giorgi, R.; Dei, L.; Baglioni, P. A New Method for Consolidating Wall Paintings Based on Dispersions of Lime in Alcohol. Stud. Conserv. 2000, 45, 154–161.

Synchrotron-Based X-ray Absorption Spectroscopy for Art Conservation: Looking Back and Looking Forward MARINE COTTE,†,‡ JEAN SUSINI,*,‡ JORIS DIK,§ AND KOEN JANSSENS⊥ †

Laboratoire du Centre de Recherche et de Restauration des Musées de France (LC2RMF), CNRS UMR 171, Palais du Louvre, Porte des Lions, 14, Quai François Mitterrand, F-75001 Paris, France, ‡European Synchrotron Radiation Facility, Polygone Scientifique Louis Néel, 6, rue Jules Horowitz, F-38000 Grenoble, France, § Delft University of Technology, Department of Materials Science and Engineering, Mekelweg 2, NL-2628CD Delft, The Netherlands, and ⊥University of Antwerp, Department of Chemistry, Universiteitsplein 1, B-2610 Wilrijk, Belgium RECEIVED ON JULY 11, 2009

CON SPECTUS

A

variety of analytical techniques augmented by the use of synchrotron radiation (SR), such as X-ray fluorescence (SR-XRF) and X-ray diffraction (SRXRD), are now readily available, and they differ little, conceptually, from their common laboratory counterparts. Because of numerous advantages afforded by SR-based techniques over benchtop versions, however, SR methods have become popular with archaeologists, art historians, curators, and other researchers in the field of cultural heritage (CH). Although the CH community now commonly uses both SR-XRF and SR-XRD, the use of synchrotron-based X-ray absorption spectroscopy (SR-XAS) techniques remains marginal, mostly because CH specialists rarely interact with SR physicists. In this Account, we examine the basic principles and capabilities of XAS techniques in art preservation. XAS techniques offer a combination of features particularly well-suited for the chemical analysis of works of art. The methods are noninvasive, have low detection limits, afford high lateral resolution, and provide exceptional chemical sensitivity. These characteristics are highly desirable for the chemical characterization of precious, heterogeneous, and complex materials. In particular, the chemical mapping capability, with high spatial resolution that provides information about local composition and chemical states, even for trace elements, is a unique asset. The chemistry involved in both the object’s history (that is, during fabrication) and future (that is, during preservation and restoration treatments) can be addressed by XAS. On the one hand, many studies seek to explain optical effects occurring in historical glasses or ceramics by probing the molecular environment of relevant chromophores. Hence, XAS can provide insight into craft skills that were mastered years, decades, or centuries ago but were lost over the course of time. On the other hand, XAS can also be used to characterize unwanted reactions, which are then considered alteration phenomena and can dramatically alter the object’s original visual properties. In such cases, the bulk elemental composition is usually unchanged. Hence, monitoring oxidation state (or, more generally, other chemical modifications) can be of great importance. Recent applications of XAS in art conservation are reviewed and new trends are discussed, highlighting the value (and future possibilities) of XAS, which remains, given its potential, underutilized in the CH community.

1. Introduction

broken down into two main categories, a better

Whatever the technical and methodological approaches, questions commonly tackled by archeologists, art historians, and curators can be

understanding of the past and a well-founded pre-

Published on the Web 01/08/2010 www.pubs.acs.org/acr 10.1021/ar900199m © 2010 American Chemical Society

diction of the future. As regards the past, analyses are intended to reveal the manufacturing Vol. 43, No. 6

June 2010

705-714


705

Synchrotron-Based XAS for Art Conservation Cotte et al.

FIGURE 1. Evolution of the publications reporting XAS analyses on CH items over the past decade. Statistics are sorted by type of material (A) versus time and (B) versus elements and absorption edges.

secrets behind specific artifacts; choice of ingredients, manu-

Account. Here, we will focus mainly on the basic principles

facturing processes (extraction of the materials, purification,

and capabilities of X-ray absorption spectroscopy (XAS)

heat treatment, chemical synthesis, etc.), geographic prove-

techniques.

nance, and trade routes, as well as authentication, constitute

The use of SR in Cultural Heritage (CH) applications was first

the main information that art historians seek to uncover.

mentioned in 1986 by Harbottle et al. who anticipated that

Regarding the future, a great deal of research is being car-

SR-based techniques would “quickly take a prominent place in

ried out to understand alteration phenomena in a more pro-

archaeometric research”.1 As a natural extension of conven-

found manner with the final goal of improving methods of

tional laboratory techniques, SR-XRF and SR-XRD became rap-

restoration and conservation.

idly popular in the CH community. Conversely, use of XAS

Various traditional analytical tools are used in laboratories to perform chemical analyses on museum objects. How-

remained marginal, mostly because of a lack of interaction between CH specialists and SR physicists.

ever, such chemical characterizations are challenging for many

However, underpinned by the dramatic improvement of

reasons. First, the artifacts are precious, so sampling is usu-

both synchrotron sources and beamline instrumentation, XAS

ally forbidden or at least limited in number and size. Second,

became increasingly used in archaeometry. Indeed, in addi-

works of art are often made of complex (organic/inorganic,

tion to the above-mentioned advantages, XAS techniques are

crystallized/amorphous) heterogeneous, and multimaterial

appealing because they can be equally applied on amorphous

mixtures of simple compounds; hence high sensitivity and

or crystalline materials. When combined with the XRF detec-

high lateral resolution can be advantageous.

tion mode, XAS also benefits from low detection limits, thus

Several synchrotron radiation (SR)-based analytical tech-

permitting the analysis of both major and minor or even trace

niques are now available, such as X-ray fluorescence (XRF),

elements. Finally, it provides rather straightforward access to

X-ray diffraction (XRD), and Fourier-transform infrared spec-

oxidation states and more generally to chemical speciation

troscopy (FTIR). They are not conceptually different from their

information.

widespread laboratory counterparts. However, the higher sen-

As shown in Figure 1A, the applications of XAS cover an

sitivity, minimal sample preparation, and possibility of con-

array of materials, ranging from hard matter such as glass,

ducting analyses at the microscale (or submicroscale) provided

ceramics, and metal to softer materials such as paintings,

by SR offers significant advantages over laboratory methods.

bone, wood, or paper. In the current context, historical knowl-

A detailed review of such methods is beyond the scope of this

edge and art conservation studies rely on similar chemical

706


705-714

June 2010

Vol. 43, No. 6


FIGURE 2. Principles of XAS.

characterization methods for both groups of materials. However, they differ in their scientific scope: The main objective for the first group of materials is to understand their original fabrication process. This mostly concerns the craftsman’s control over synthetic reactions at high temperature resulting in various optical effects in glasses, glazes, lusters, ceramics, and pigments. Studies on the second group of materials focus on unintentional degradation reactions. In general, these occur over longer time scales and are either due to past conservation treatments or due to passive external circumstances or may even be inherent to the chemical compounding of the artwork itself. Both types of studies will be illustrated by different examples.

2. Basics of X-ray Absorption Spectroscopy XAS techniques are based on the measurement of the variation in the absorption coefficient while scanning the energy of the probing X-rays photons in a narrow region around an absorption edge (Figure 2). This variation is physically related to the excitation cross section of the core electrons into unoccupied electronic states or into the vacuum continuum. The spectral features observed close to the absorption edge,

referred to as the X-ray absorption near edge structure (XANES), reflect the molecular environment (oxidation state, coordination numbers, site symmetry, and distortion) of a given absorbing atom and provide the basic mechanism for imaging with chemical specificity.2 Information on different electronic states within systems that have the same elemental composition is therefore possible. A similar method is also used in electron microscopes by performing electron energy loss spectroscopy (EELS). However, the specificity of the interactions of X-rays with matter offers some advantages to XANES over EELS microscopy: micro-XANES has been proven to have a better spectral sensitivity, to produce less radiation damage, and to benefit from a larger penetration depth for bulk analysis.3 Extended X-ray absorption fine structure (EXAFS) is measured when the energy of the incoming X-rays is scanned far beyond the absorption edge of interest. EXAFS oscillations result from constructive and destructive interference between the outgoing and backscattered photoelectron waves. The physical processes giving rise to the EXAFS signal are generally well-understood and can be modeled using appropriate computer programs, which make it possible to assess, with Vol. 43, No. 6

June 2010

705-714


707


FIGURE 3. Sketch of a state-of-the-art hard X-ray microprobe on SR beamline. Several detectors allow a multimodal approach by simultaneous signal recording.

reasonable accuracy, the identity of the surrounding atoms, specific bond distances, and coordination numbers.4 As a result of the removal of core electrons due to the X-ray absorption processes described above, atom de-excitation leads to the emission of characteristic “secondary” X-rays. These X-ray fluorescence signals permit elemental identification, mapping, and often quantification of the elements present in a sample in a similar fashion as can be done with nonsynchrotron techniques such as scanning electron microscopy (SEM)-energy-dispersive X-ray analysis (EDXA) and particle-induced X-ray emission (PIXE). The advantage of SR-XRF lies in its sensitivity, due to the high photon flux available, the large production cross section for XRF signals, the possibility of tuning energy, and the low scattering cross sections. Quantification is comparatively straightforward because the physics of photon interaction with matter is relatively simple and well-understood. As discussed above, the absorption edge of an element is related to the chemical environment of the absorbing atom and its oxidation state. Therefore, by select-

minimizes radiation exposure without compromising the signal-to-noise and allows quantification with an accuracy of 3-5%.5 Most of the synchrotron set-ups can accommodate samples of different dimensions, and when using hard X-rays, they operate at atmospheric pressure. Therefore it is possible to perform local analyses on whole CH objects, as was done on Etruscan glass vessels and beads6 or Van Gogh paintings.7 However, today these examples remain marginal, mainly because of logistic limitations (historical value, insurance cover, travel range/physical size of scanning stages, etc.). In this respect, sampling is often favored not only because it simplifies sample handling but also because it can give access to cross-sectional information. It is noteworthy that “noninvasive analysis” does not always imply “nondestructive analysis” because X-ray irradiation may result (to some degree) in radiation damage. In addition to a possible local blackening (limited to the size of the probe), more subtle chemical changes due to photoreduction have been reported in cases of hybrid or organic materials such as iron gall ink8,9 and blood-containing African patinas.10 This photochemical effect usually decreases the oxidation state of the targeted element. Being directly proportional to the photon density, this process can occur very rapidly, in the range of a few minutes, when using a submicrometer probe. If underestimated, this artifact can lead to misleading conclusions and should be systematically addressed. Indeed, this problem is far from being specific to CH samples, and several ongoing developments on high-efficiency detectors aim to minimize time exposure and consequently the total absorbed dose at each examined position.

ing appropriate energies of the incoming X-ray photon, it is possible to generate chemical maps of an element in relation to its oxidation state and chemical bonding. As depicted in Figure 3, the most advanced hard X-ray microprobes usually offer a multimodal analysis platform where SR-XRF, SR-XAS, and SR-XRD can be combined. The sample is usually raster scanned, and data are collected at each pixel to generate corresponding elemental, structural, or chemical maps. Most SR-XRF instruments rely upon a specific 45°/45° geometry where the sample is canted 45° to the incident beam and X-ray detectors are placed in the plane of the storage ring at ∼90° to the incident beam. With the incoming beam being naturally polarized in the horizontal plane, this geometry minimizes the contribution of scattered primary X-rays. Furthermore, the full control of both tunability and spectral bandwidth of the incoming monochromatic radiation 708


705-714

June 2010

Vol. 43, No. 6

3. Historical Knowledge and Ancient Manufacturing Technologies XAS techniques were first applied to ancient glasses11,12 (Figure 1A,B). More recently, glazes and ceramics, based on similar vitreous matrices, have also benefitted from an increasing interest. Indeed, XAS techniques are particularly adapted for the study of these types of materials. Combined with the good match between method capabilities (short-range probe, sensitive) and material properties (amorphous and diluted state), XANES and EXAFS provide unique chemical information. Historical vitreous materials were highly valued because of their optical properties. The interplay between transparency, opacity, color, metallic shine, colored iridescence, and other properties are still appealing today but must have been highly appreciated in the object’s contemporary times. Such effects


can be induced by the presence of opacifying crystals, ionic chromophores, or metallic nanoparticles as well. The oxidation state of elements and more generally their chemical environment within the vitreous matrix is directly correlated to these optical effects. The historical production method therefore required adequate control of firing conditions (temperature, atmosphere, and time), as well as the introduction of oxidizing or reducing ingredients. Coloring variations are usually obtained in glass by modulating the oxidation states of transition elements such as Mn, Fe, Co, and Cu; these elements have characteristic absorption frequencies in the visible region as a result of d-d electronic transitions.11 This explains the high number of XAS studies at the K-edge of these elements (Figure 1B). For instance, XANES analyses of ancient glasses from the Patti roman villa (Messina, Sicily) at the Fe and Mn K-edges supported the hypothesis that pyrolusite (MnO2) could have been added intentionally as a decolorant during the melting procedure.13 However, many of the above-mentioned optical effects (such as transparency) can be affected by long-term corrosion. For instance, the oxidation of Mn2+ into Mn4+ has been observed in medieval glass windows exposed to progressive weathering in Cathedral du Bosc, Normandy, France, (14th Century).14 This oxidation results in the precipitation of manganese oxi-hydroxides, which in turn lead to opacification and a change in color (brown) of the glass panes. In the case of lusters, the skills of many craftsmen were honed toward producing a metallic shine and a colored iridescence on the surface of ceramic objects. These optical properties result from the presence of an inhomogeneous dispersion of metallic (usually silver or copper) nanoparticles in the first outermost micrometer of the glaze. Following pioneering transmission electron microscopy studies, a number of projects dedicated to luster characterization by EXAFS and XANES were initiated aiming mainly at establishing a correlation between color (red, gold, or green), chemical composition, and copper or silver oxidation states.15,16 Beyond the family of vitreous-based artworks, XAS techniques can also be exploited to track possible heat-induced color transformations in various pigments. This has proven to be particularly relevant for manganese-based pigments. For instance, Mn K-edge XANES and EXAFS were used to understand the thermal effect on the structural environment of Mn in fossilized mastodon ivory or bone. The heat-induced oxidation of Mn2+/3+/4+ into Mn5+ and further substitution for P5+ in the apatite matrix is responsible for the blue rendering of these pigments, usually called odontolite, or bone turquoise, since it perfectly imitates the color of mineral turquoise.17

In a different context, Mn XANES was employed for the characterization of black pigments found in Spanish and French prehistoric caves. Several atypical manganese oxide minerals such as manganite, groutite, todorokite, and birnessite were identified. TEM and XANES analyses revealed that, instead of synthesizing these compounds through heat-induced reactions, natural pigments were favored. Thus, these raw pigments, rare in nature, are likely to originate from nonlocal geological sites and may have had to be imported from a distant location.14,18

4. Alteration, Restoration and Preservation Most of the alteration processes, for example, metal corrosion, involve modification of the redox states of the original material, while the average elemental composition of the bulk material remains unchanged. Alteration is then limited to a highly superficial surface gradient, which may have a thickness in the micrometer or even submicrometer range. Yet, such pure surface changes can have effects on the objects visual appearance. Probing the oxidation of specific elements, and more generally their chemical environments, is therefore of prime importance when studying alteration mechanisms. Degradations may be detected with greater significance with XAS than the change in the macroscopic, visual appearance of the object would suggest as reported, for instance, in the study of degradation of cadmium pigment.19 While the conservators of the painting noted a slight discoloration of the yellow paint, the degradation layer itself could be better identified by XAS. Metal corrosion is a major concern in art conservation.20 XAS offers a direct observation (either as point, profile, or map acquisition) of metal speciation. Such measurements have mainly been performed on iron21 or bronze22,23 artifacts. In all of these works, mention is made of the interest of combining several microanalytical techniques such as SEM-EDX, XRF, XRD, Raman, and XAS. The XRF/XAS combination is rather natural as all the XRF emission lines below the XANES edge being investigated are recorded simultaneously. For instance, Reguer et al. demonstrated a clear correlation between the iron valence distribution in multilayered corrosion products and the evolution of chloride concentrations.21 XAS techniques are also increasingly used to study chemical reactions involved during pigment alteration processes. As an example, sulfur-based pigments, such as HgS (cinnabar when natural, vermilion when synthetic), as well as CdS, may suffer from discolorations. While red HgS tends toward shades of gray or black, bright yellow CdS may evolve into a white Vol. 43, No. 6

June 2010

705-714


709


FIGURE 4. Examples of XANES spectra (K-edge) of sulfur (left) and chlorine compounds (right) occurring in art materials.

transparent matter. XANES at the sulfur K-edge revealed that both HgS, in Pompeian paintings,24 and CdS, in paintings by the 19th Century artist James Ensor,19 are subject to oxidation. The XANES features probe the sulfur oxidation states directly (Figure 4) and can be used straightforwardly to map the different sulfur species. Besides sulfur oxidation, chlorine was identified as an additional potential player in cinnabar degradation. In some Pompeian paintings,24 as well as Gothic paintings from Pedralbes, Barcelona25 (Figure 5), XANES revealed the formation of mercury-chloride chemical bonds, specifically in grayaltered regions. Interestingly, the presence of the element chlorine was not systematically related to alteration. It was, for example, observed as the nonreactive pollutant NaCl on a surface of a red Pompeian painting.24 Chlorine XANES spectra exhibit features that make it possible to distinguish mercury chloride compounds from alkaline and alkaline earth chlorides (Figure 4). Thus it was possible to selectively map the distribution of mercury chloride compounds in cross sections of a Gothic painting, showing the blackening of cinnabar (Figure 5). This figure shows that mercury chloride compounds are limited to the gray region, on top of the cinnabar layer, while other chlorides, presumably CaCl2, are present beneath the pigment (in the mortar). More generally, chlorine is frequently involved in artwork degradation, either in paint layers, metals, or glass. It can affect not only materials previously immersed in salt water but 710


705-714

June 2010

Vol. 43, No. 6

also artifacts that are kept in the vicinity of the sea. Furthermore, chlorinated cinnabars have already been reported in paintings stored in museums, such as London National Gallery26 (Raman analyses) and the Louvre27 (XANES analyses). Accordingly, XANES at the Cl K-edge will certainly find many future applications in art conservation, as complementary methods to laboratory analyses. It is noteworthy that significant aesthetic damage is not necessarily associated with alterations throughout the bulk of the material and are limited to the objects outmost superficial layer. As observed in the previous examples, pigment alteration can be limited to the few outmost micrometers of the painting. Therefore, the characterization of such phenomena is challenging because the material (often the original pigment) present underneath the alteration can significantly contribute to the chemical signal detected. Submicrometer probes are particularly well-suited to the study of transversal cross sections, because they usually enable a discriminative analysis of both safe and altered regions. A multiscale approach is equally important as exemplified in various studies that combine in-plane millimetric analysis over large painting surfaces with in-depth micrometric analysis of transversal polished sections.7,24 It should be pointed out that spatial resolution is determined not only by lateral resolution but also by penetration depth. This aspect clearly discriminates X-ray microscopy from electron microscopy. It can be an inconvenience as well an asset. Different strategies can be used to restrict the volume analyzed by the X-ray methods (see below, current trends). Conversely, the relatively high penetration depth of X-rays (in particular at high energy) can be fully exploited for 3D imaging.28 At low energies, probe depth is limited by self-absorption. For instance, in HgS, 80% of the sulfur fluorescence lines are reabsorbed within the first 0.5 µm layer. XAS analyses are also employed to gain a better understanding of the alteration of soft materials such as wood or paper. Indeed, these reactions can be affected by surrounding metallic compounds (such as iron bolts in wooden war ships or iron-gall ink in paper), which can act as catalysts in oxidation processes. For example, measurements of the oxidation state of iron can be used to characterize ink diffusion in paper and related chemical reactions leading to the disintegration of ancient manuscripts.8,9,29 Similar to painting characterization, a dual macro-/microstrategy was followed by Sandstrom et al. while studying marine wood degradation in two major historical war ships, the Mary Rose and the Vasa. Series of wood fragments were sampled at different depths, starting from the surface and cov-


FIGURE 5. XANES spectra and chemical mapping of a blackening cinnabar from Pedralbes monastery.25 Optical light microscope picture of the multilayer structure and regions of interest. Two chlorine XRF mappings, recorded with two peculiar energies of the incoming beam, 2.8231 keV, specific to Hg-Cl bonding (noted as Cl*), and 2.85 keV, which excite unselectively all chlorine species (noted as Cl).

ering several centimeters. Samples were then ground to powder and analyzed with a large beam in order to obtain an average measure of the sulfur oxidation states within the wood. Meanwhile, thin transverse wood sections were studied with a microprobe to gain information on the subcellular distribution of reduced sulfur within the individual cells of the wood.30

5. Current Trends Recent methodological developments aim at pushing classical XAS experiments toward better control of additional experimental parameters, such as time or probing depth, or by combining XAS with other analytical techniques. Toward Time Resolution. Alteration mechanisms, as for any chemical reaction, must be ultimately addressed as kinetic processes. The specific nature of works of art causes the time scale of these processes to be rather broad, ranging from hours to centuries. An additional degree of complexity is the multifactor character of the alterations, which are determined not only by environmental conditions and their variations but also by the intrinsic composition of the object. Reproducing such chemical reactions in the laboratory, with or without artificial aging, can be particularly rewarding for conservation purposes. Such studies attempt to understand alteration phenomena, while assessing the effects and the efficiency of conservation and restoration treatments. When such model processes are performed, physical and chemical conditions must be carefully monitored, from near-reality to extreme conditions. In this context, SR techniques offer two main advantages: the large depth of focus, which allows scientists to use cus-

tomized sample environments (electrochemical cells, furnace, etc.) during experiments, and the high flux, permitting fast data acquisition, which is essential for precise observation of the very first reaction stages. As an example, Adriaens et al. developed an electrochemical cell dedicated to the in situ monitoring of metal corrosion in marine conditions. Time-resolved XAS and XRD were used to monitor copper corrosion at the surface-solution interface in a sodium sesquicarbonate solution and to evaluate the effectiveness of protective coatings of lead objects.31 Similarly, in situ temperature monitoring is relevant for a better understanding of manufacturing processes of various artworks (glass, ceramics, lusters, etc.). Although several experiments combining off-line heat treatment and XAS measurements have been reported, such as on different manganesebased pigments,17,18 online temperature-resolved experiments are still underexploited and so far have only involved XRD analyses. Pradel et al. studied high temperature reactions in model luster paint.32 The authors could correlate the decomposition of metacinnabar and the subsequent creation of a sulfo-reducing atmosphere with the reduction of Cu2+-containing compounds to Cu+; an intermediate chemical state that leads to the formation of metallic Cu nanoparticles. Toward in-Depth Resolution. Another trend consists in developing methods to tune the probed depth in order to obtain “in-depth” information without invasive transversal cross sections. Penetration depth can be tweaked by tuning the energy of the excitation photon beam. Gliozzo et al. measured the XANES spectra of Fe at both K- and L-edges on black gloss pottery from Northern Etruria to explain visual differences among different samples.33 To probe the gloss at difVol. 43, No. 6

June 2010

705-714


711


ferent depths, the authors relied on the difference in beam penetration between two energies, 7.1 keV (Fe K-edge) and 0.7 keV (Fe L-edges). Despite more complex experimental conditions (lower fluorescence yield, experiments in vacuum to prevent air absorption), XANES at the L-edges, with a penetration depth of about 0.1 µm, was essential to selectively probe the surface of the gloss. These low-energy acquisitions revealed a direct correlation between the Fe oxidation state and the visual appearance of the gloss. Working at several edges of the same element usually requires distinct experiments, and often, as in the above example, on two different instruments. Conversely, other strategies rely on multidetection systems capable of simultaneous measurements at two different depths. Cartechini et al. performed XAS analysis on a XVth century Italian painting to better understand copper resinate blackening and to establish a correlation between the color and the chemical environment of copper.34 XAS measurements were conducted both in X-ray fluorescence yield mode (XFY) and in total electron yield mode (TEY) based on the collection of secondary electrons emitted. While XFY probes the first tens of micrometers, TEY accesses only the first few hundred nanometers at the surface of the paint layer. In this example, the strong analogy of spectral features in both modes led to the assumption that the copper local environment did not change across the depth of the blackened layer. Similar to the combination XFY/TEY, combining XAS with optically detected X-ray absorption spectroscopy (ODXAS) offers identical modulation of probed depth. ODXAS relies on the measurements of near-optical photons (i.e., with wavelengths in the 200-1000 nm range), generated by secondary fluorescence processes. This method was exploited to perform high-resolution analysis of copper corrosion.35 Model corrosion systems deposited on a copper substrate were analyzed in air and in a sodium sesquicarbonate solution, simulating typical conservation methods for copper-based objects recovered from marine environments. In such a configuration, the ODXAS is significantly more surface specific by probing the first 0.5-1.5 µm layer, while the XRF data are dominated by signals originating from the substrate. A similar ODXAS experiment was reported by Roque´ et al. on the study of model lusters (metallic copper nanoparticles embedded in a glass matrix). The objective was primarily to correlate optical luminescence properties with chemical composition of metallic nanoparticles.36 Finally, the use of polycapillary-based set-ups provides a more versatile control of in-depth measurements. The so-called confocal geometry is based on the use of two X-ray 712


705-714

June 2010

Vol. 43, No. 6

lenses. One lens is used to focus the incoming beam while the second restricts the XRF detector angular acceptance to a small volume. The intersection volume of the incoming microbeam and the coinciding focus of the second lens determines the voxel size.37 Such strategy has already been employed for depth-selective elemental analyses.38 Extension to XAS studies is possible as well but is yet to find applications in the cultural heritage field. Toward Multimodal Microspectroscopy. Combining XAS with other methods, in particular spectroscopic methods, is another way of enhancing the method. Beyond XRF, which is a natural and indeed essential companion method to XAS, vibrational techniques can be coupled to XAS in a fruitful manner. Raman and FTIR spectroscopy are complementary to XAS as all these methods share several characteristics: They all probe molecular groups while being quasi-noninvasive, can be used on similar samples, and can all generate medium- to high-resolution chemical maps. While XAS is element-specific, vibrational spectroscopies simultaneously access a large number of molecular groups. Consequently, XAS is more sensitive and more adapted to the study of a single element of interest, whereas FTIR and Raman microscopy provide a global picture of the overall molecular composition. Such a combination can be performed off-line, as was done on paintings34 or on patina sampled from African statues.10 The new potential of combining vibrational and electronic microspectroscopies is increasingly addressed on various facilities.3

6. Conclusions As already stated by Nakai et al. in 1997,11 XAS techniques are particularly relevant for the analysis of artifacts. Experiments can be performed in air, directly onto the entire object, without any sample preparation. Conversely, focusing the beam down to submicrometric size enables precise and selective mapping. Analyses can be carried out on any kind of material (from hard to soft), in particular on vitreous matrices. They can also be applied to many elements, as opposed to Mo¨ssbauer spectroscopy, which is limited to a small number of elements.39 When XAS spectra are acquired in XRF detection mode, the sensitivity is very high. Thus, minor or trace elements, which may have major impacts on the appearance of works of art (e.g., chromophoric elements in glass or chlorine in paintings) can be probed. Similarly, the chemical sensitivity provided by XANES and EXAFS is very high and offered particularly significant information in many cases. On the one hand, many analyses are carried out to study histor-


ical production techniques used to create appealing visual effects in vitreous materials. This includes color, translucence, opacity, iridescence, etc. in glasses and ceramics. Many of these effects result from electronic transitions that are directly related to the chemical/electronic state of some transition metals. On the other hand, phenomena of unintentional, postproduction degradation are often directly connected with modification of the molecular surroundings (change of the oxidation state, reaction with exogenous chemicals, etc.) of these metals. In both cases, obtaining information about the chemical environment of specific elements is highly relevant. Furthermore, combining high lateral resolution with high chemical sensitivity (resulting from SR high flux and energy tunability) provide an unrivaled way of generating chemical maps. Finally, the possibility of combining these various analytical features in a single experiment is one of the unique assets of these methods. After one or two decades of exploratory and demonstrative studies, SR-XAS and SR-XRF are now recognized as appropriate methods for tackling challenging questions in art conservation that cannot be addressed using conventional laboratory methods. Both the SR and CH communities are now aware of the advantages associated with this synergy and several partnerships are being built, as exemplified by the IPANEMA institute at the SOLEIL40 synchrotron, as well as the creation of dedicated review panels in several other SR facilities. Finally, current technological developments such as time-resolved XAS measurements, the use of nanoprobes, and the possibility of performing ultrafast scans will undoubtedly open new avenues in the field of analytical chemistry applied to art conservation.

BIOGRAPHICAL INFORMATION After gaining the “agre´gation” of chemistry at the “Ecole Normale Supe´rieure” (Lyon), Marine Cotte obtained her Ph.D. for her research at the C2RMF on lead-based cosmetics and pharmaceutical compounds used in Antiquity. During her postdoctoral fellowship at the ESRF, she has expanded the application of X-ray and FTIR microspectroscopies to paintings. She has now a twofold position as a permanent CNRS scientist (C2RMF) and as a beamline scientist (ESRF). Jean Susini obtained his Ph.D. in Chemical Physics at the University Pierre et Marie Curie (Paris) and joined the ESRF in 1989. In 1994, he took responsibility for the design, construction, and operation of the X-ray microscopy beamline. He led the ESRF X-ray Microscopy and Microanalysis Group and coordinated the operation and scientific program of several instruments, including hard and soft X-ray microprobes and the infrared spectromicroscopy station. In 2009, he was appointed Head of the Instrumentation Services and Development Division of the ESRF.

Joris Dik trained as art historian at the University of Amsterdam under the supervision of Prof. E. Van de Wetering and then obtained a Ph.D. in X-ray crystallography at the Free University of Amsterdam, focusing on the synthesis and alteration behavior of the pigment lead-tin yellow. His research interest is to apply novel methods of scientific investigation to art-history problems. He currently is associate professor at the Technical University of Delft, The Netherlands. Koen Janssens obtained his Ph.D. in Analytical Chemistry from the University of Antwerp (Belgium) and became professor at this university in 2000. He has employed since 1990 intense beams of X-rays for nondestructive materials analysis, and he was, together with F. Adams and A. Rindby, editor of Microscopy X-ray Fluorescence Analysis (Wiley, 2000) and, together with R. Van Grieken, of the Non-destructive microanalysis of Cultural Heritage materials (Elsevier, 2004). His main fields of interest are X-ray based microanalysis of materials with special attention for local speciation of metals in (altered) environmental and cultural heritage materials such as glass and inorganic painters’ pigments. FOOTNOTES * To whom correspondence should be addressed. Phone: +33 4 76 88 21 24. Fax: +33 4 76 88 20 66. E-mail: [email protected].

REFERENCES 1 Harbottle, G.; Gordon, B. M.; Jones, K. W. Use of synchrotron radiation in archaeometry. Nucl. Instrum. Methods Phys. Res. B 1986, 14 (1), 116–122. 2 Sto¨hr, J., NEXAFS spectroscopy; Springer-Verlag: Heidelberg, Germany, 1992. 3 Special Issue: Workshop on simultaneous Raman-X-ray diffraction/absorption studies for the in situ investigation of solid state transformations and reactions at non-ambient conditions. Phase Transitions 2009, 82 (4), 291-355. 4 Koningsberger, D. C.; Prins, R., X-ray Absorption, Principles, techniques of EXAFS, SEXAFS and XANES; John Wiley & Sons: New York, 1988. 5 Vincze, L.; Janssens, K.; Vekemans, B.; Adams, F., Monte Carlo simulation for x-ray fluorescence spectroscopy. In X-ray Spectrometry Based on Recent Technological Advances; Tsuji K., Injuk J., Van Grieken R., Eds; Wiley: Chichester, U.K., 2004; pp 435-462. 6 Arletti, R.; Vezzalini, G.; Quartieri, S.; Ferrari, D.; Merlini, M.; Cotte, M. Polychrome glass from Etruscan sites: First non-destructive characterization with synchrotron µ-XRF, µ-XANES and XRPD. Appl. Phys. A 2008, 92 (1), 127–135. 7 Dik, J.; Janssens, K.; Van Der Snickt, G.; Van Der Loeff, L.; Rickers, K.; Cotte, M. Visualization of a lost painting by Vincent van Gogh using synchrotron radiation based X-ray fluorescence elemental mapping. Anal. Chem. 2008, 80 (16), 6436– 6442. 8 Kanngiesser, B.; Hahn, O.; Wilke, M.; Nekat, B.; Malzer, W.; Erko, A. Investigation of oxidation and migration processes of inorganic compounds in ink-corroded manuscripts. Spectrochim. Acta B 2004, 59, 1511–1516. 9 Wilke, M.; Hahn, O.; Woodland, A. B.; Rickers, K. The oxidation state of iron determined by Fe K-edge XANESsapplication to iron gall ink in historical manuscripts. J. Anal. At. Spectrom. 2009, 24, 1364–1372. 10 Mazel, V.; Richardin, P.; Debois, D.; Touboul, D.; Cotte, M.; Brunelle, A.; Walter, P.; Laprevote, O. Identification of ritual blood in African artifacts using TOF-SIMS and synchrotron radiation microspectroscopies. Anal. Chem. 2007, 79 (24), 9253–9260. 11 Nakai, I.; Matsunaga, M.; Adachi, M.; Hidaka, K. I. Application of XAFS in archaeology. J. Phys. IV 1997, 7 (2), 1033–1034. 12 Schofield, P. F.; Cressey, G.; Wren Howard, W.; Henderson, C. M. B. Origin of color in iron and manganese containing glasses investigated by synchrotron radiation. Glass Technol. 1995, 36 (3), 89–94. 13 Quartieri, S.; Triscari, M.; Sabatino, G.; Boscherini, F.; Sani, A. Fe and Mn K-edge XANES study of ancient Roman glasses. Eur. J. Mineral. 2002, 14 (4), 749–756. 14 Farges, F.; Chalmin, E.; Vignaud, C.; Pallot-Frossard, I.; Susini, J.; Bargar, J.; Brown Jr, G. E.; Menu, M. Archeological applications of XAFS: Prehistorical paintings and medieval glasses. Phys. Scr. 2005, T115, 885–887.

Vol. 43, No. 6

June 2010

705-714


713


15 Padovani, S.; Sada, C.; Mazzoldi, P.; Brunetti, B.; Borgia, I.; Sgamellotti, A.; Giulivi, A.; D’Acapito, F.; Battaglin, G. Copper in glazes of Renaissance luster pottery: Nanoparticles, ions, and local environment. J. Appl. Phys. 2003, 93 (12), 10058– 10063. 16 Smith, A. D.; Pradell, T.; Roque´, J.; Molera, J.; Vendrell-Saz, M.; Dent, A. J.; Pantos, E. Colour variations in 13th century hispanic lustre - an exafs study. J. Non-Cryst. Solids 2006, 352, 5353–5361. 17 Reiche, I.; Morin, G.; Brouder, C.; Sole´, V. A.; Petit, P. E.; Vignaud, C.; Calligaro, T.; Menu, M. Manganese accommodation in fossilised mastodon ivory and heatedinduced colour transformation: Evidence by EXAFS. Eur. J. Mineral. 2002, 14, 1069–1073. 18 Chalmin, E.; Vignaud, C.; Farges, F.; Menu, M. Heating effect on manganese oxihydroxides used as black Palaeolithic pigment. Phase Transitions 2008, 81 (2-3), 179–203. 19 Van der Snickt, G.; Dik, J.; Cotte, M.; Janssens, K.; Jaroszewicz, J.; De Nolf, W.; Groenewegen, J.; Van der Loeff, L. Characterisation of a degraded cadmium yellow (CdS) pigment in an oil painting by means of synchrotron radiation based X-ray techniques. Anal. Chem. 2009, 81 (7), 2600–2610. 20 Dillmann, P.; Beranger, G.; Piccardo, P.; Matthiesen, H. Corrosion of metallic heritage artefacts. Eur. Fed. Corros. Publ. 2007, xxx, 48. 21 Reguer, S.; Dillmann, P.; Mirambet, F.; Susini, J.; Lagarde, P. Investigation of Cl corrosion products of iron archaeological artefacts using micro-focused synchrotron X-ray absorption spectroscopy. Appl. Phys. A 2006, 83 (2), 189–193. 22 De Ryck, I.; Adriaens, A.; Pantos, E.; Adams, F. A comparison of microbeam techniques for the analysis of corroded ancient bronze objects. Analyst 2003, 128 (8), 1104–1109. 23 Northover, P.; Crossley, A.; Grazioli, C.; Zema, N.; La Rosa, S.; Lozzi, L.; Picozzi, P.; Paparazzo, E. A multitechnique study of archeological bronzes. Surf. Interface Anal. 2008, 40 (3-4), 464. 24 Cotte, M.; Susini, J.; Metrich, N.; Moscato, A.; Gratziu, C.; Bertagnini, A.; Pagano, M. Blackening of Pompeian cinnabar paintings: X-ray microspectroscopy analysis. Anal. Chem. 2006, 78 (21), 7484–7492. 25 Cotte, M.; Susini, J.; Sole, V. A.; Taniguchi, Y.; Chillida, J.; Checroun, E.; Walter, P. Applications of synchrotron-based micro-imaging techniques to the chemical analysis of ancient paintings. J. Anal. At. Spectrom. 2008, 23, 820–828. 26 Spring, M.; Grout, R. The blackening of vermilion: An analytical study of the process in paintings. Natl. Gallery Technol. Bull. 2002, 23, 50–61. 27 Cotte, M.; Susini, J. Watching ancient paintings through synchrotron-based X-ray microscopes. MRS Bull. 2009, 34, 403–405.

714


705-714

June 2010

Vol. 43, No. 6

28 Janssens, K.; Dik, J.; Cotte, M.; Susini, J. New techniques for in-depth analysis of cultural heritage artefacts. Acc. Chem. Res., 2010, 43, 10.1021/ar900248e. 29 Proost, K.; Janssens, K.; Wagner, B.; Bulska, E.; Schreiner, M. Determination of localized Fe2+/Fe3+ ratios in inks of historic documents by means of µ-XANES. Nucl. Instrum. Methods Phys Rev B 2004, 213, 723–728. 30 Sandstro¨m, M.; Jalilehvand, F.; Damian, E.; Fors, Y.; Gelius, U.; Jones, M.; Salome´, M. Sulfur accumulation in the timbers of King Henry VIII’s warship Mary Rose: A pathway in the sulfur cycle of conservation concern. Proc. Natl. Acad. Sci. U.S.A. 2005, 102 (40), 14165–14170. 31 Adriaens, A.; Dowsett, M.; Jones, G. K. C.; Leyssens, K.; Nikitenko, S. An in-situ X-ray absorption spectroelectrochemistry study of the response of artificial chloride corrosion layers on copper to remedial treatment. J. Anal. At. Spectrom. 2009, 24, 62–68. 32 Pradell, T.; Molera, J.; Pantos, E.; Smith, A. D.; Martin, C. M.; Labrador, A. Temperature resolved reproduction of medieval luster. Appl. Phys. A: Mater. Sci. Process. 2008, 90 (1), 81–88. 33 Gliozzo, E.; Kirkman, I. W.; Pantos, E.; Memmi Turbanti, I. Black gloss pottery: Production sites and technology in Northern Etruria. Part II: Gloss technology. Archaeometry 2004, 46 (2), 227–246. 34 Cartechini, L.; Miliani, C.; Brunetti, B. G.; Sgamellotti, A.; Altavilla, C.; Ciliberto, E.; D’Acapito, F. X-ray absorption investigations of copper resinate blackening in a XV century Italian painting. Appl. Phys. A 2008, 92 (1), 243–250. 35 Dowsett, M. G.; Adriaens, A.; Jones, G. K. C.; Poolton, N.; Fiddy, S.; Nikitenko, S. Optically detected X-ray absorption spectroscopy measurements as a means of monitoring corrosion layers on copper. Anal. Chem. 2008, 80 (22), 8717– 8724. 36 Roque´, J.; Poolton, N. R. J.; Molera, J.; Smith, A. D.; Pantos, E.; Vendrell-Saz, M. X-ray absorption and luminescence properties of metallic copper nanoparticles embedded in a glass matrix. Phys. Status Solidi B 2006, 243 (6), 1337–1346. 37 Vincze, L.; Vekemans, B.; Brenker, F. E.; Falkenberg, G.; Rickers, K.; Somogyi, A.; Kersten, M.; Adams, F. Three-dimensional trace element analysis by confocal X-ray micro-fluorescence imaging. Anal. Chem. 2004, 76, 6786–6791. 38 Kanngiesser, B.; Malzer, W.; Reiche, I. A new 3D micro X-ray fluorescence analysis set-up - First archaeometric applications. Nucl. Instrum. Methods Phys. Res. B 2003, 211 (2), 259–264. 39 Keisch, B. Mossbauer effect studies of fine arts. J. Phys. (Paris) 1974, 12 (35), C6151–C6-164. 40 Bertrand, L.; Vantelon, D.; Pantos, E. Novel interface for cultural heritage at SOLEIL. Appl. Phys. A 2006, 83 (2), 225–228.

Analytical Strategies for Characterizing Organic Paint Media Using Gas Chromatography/Mass Spectrometry MARIA PERLA COLOMBINI,* ALESSIA ANDREOTTI,† ILARIA BONADUCE,† FRANCESCA MODUGNO,† AND ERIKA RIBECHINI† Chemical Science for the Safeguard of the Cultural Heritage Group, Dipartimento di Chimica e Chimica Industriale, Universita` di Pisa, via Risorgimento 35, 56126 Pisa, Italy RECEIVED ON JUNE 27, 2009

CON SPECTUS

T

hroughout history, artists have experimented with a variety of organic-based natural materials, using them as paint binders, varnishes, and ingredients for mordants in gildings. Artists often use many layers of paint to produce particular effects. How we see a painting is thus the final result of how this complex, highly heterogeneous, multimaterial, and multilayered structure interacts with light. The chemical characterization of the organic substances in paint materials is of great importance for artwork conservation because the organic components of the paint layers are particularly subject to degradation. In addition, understanding the organic content and makeup of paint materials allows us to differentiate between the painting techniques that have been used over history. Applying gas chromatography/mass spectrometry (GC/MS) analysis to microsamples of paint layers is widely recognized as the best approach for identifying organic materials, such as proteins, drying oils, waxes, terpenic resins, and polysaccharide gums. The method provides essential information for reconstructing artistic techniques, assessing the best conditions for long-term preservation, and planning restoration. In this Account, we summarize the more common approaches adopted in the study of the organic components of paint materials. Our progress in developing GC/MS analytical procedures in the field of cultural heritage is presented, focusing on problems that arise from (i) the presence of mixtures of many chemically complex and degraded materials, (ii) the interference of inorganic species, (iii) the small size of the samples, and (iv) the risk of contamination. We outline some critical aspects of the analytical strategy, such as the need to optimize specific wet-chemical sample pretreatments in order to separate the various components, hydrolyze macromolecular analytes, clean-up inorganic ions, and derivatize polar molecules for subsequent GC/MS separation and identification. We also discuss how to interpret the chromatographic data so as to be able to identify the materials. This identification is based on the presence of specific biomarkers (chemotaxonomy), on the evaluation of the overall chromatographic profile, or on the quantitative analysis of significant compounds. GC/MS-based analytical procedures have for 20 years provided important contributions to conservation science, but challenges and opportunities still coexist in the field of organic-based paint materials. We give selected examples and provide case studies showing how a better understanding of the chemical composition of organic paint materials and of their degradation pathways contribute to a better knowledge our cultural heritage, and to its preservation for future generations.


Vol. 43, No. 6

June 2010

715-727


715

Strategies for Characterizing Organic Paint Media Colombini et al.

1. Introduction Paints are generally made up of the same fundamental components: a pigment, which is most typically a fine powder of inorganic or organic colored material, and, with the exception of frescos, a fluid binder, which enables the pigment to be dispersed and applied with a brush. The binder may be a proteinaceous material such as egg or casein, a vegetable gum, a drying oil, a natural wax, or a mixture of two or more of these materials. After drying or curing, a solid paint film is produced. The surface on which the paint is applied may be cotton or linen canvas, a wood panel, stone, metal, glass, or plaster on a wall. Such surfaces generally need to be prepared with a ground layer. For instance, a mixture of animal glue and gypsum was used for centuries as a ground for both wooden panels and canvases. Paintings were also often coated with a varnish containing plant resins and/or oils to produce saturated, deep-toned colors and protection against environmental agents.1 Painters were not only artists but also “material scientists”, since they had to be able to select the best paint materials, process them, and apply them in order to suit their needs and achieve the desired aesthetic results. They experimented with a wide range of natural materials1 and often used many layers of paint to produce particular effects. To our eyes, the appearance of a painting is thus the final result of the interaction of this complex, highly heterogeneous, multimaterial, and multilayered structure with light. This Account focuses on the chemical characterization of organic components, which is of great interest because the different organic paint materials used help us to differentiate between the various painting techniques and because the organic component of the paint layer is particularly subject to degradation. An analysis of organic paint materials is essential for their long-term preservation, to assess the best conservation and display conditions, to prevent and slow the decay processes, and to plan the best restoration. Macroscopic degradation phenomena, such as the yellowing of the varnish layers and the loss of cohesion and craquelures in the paint layers, are related to the chemical alterations of the organic media, such as depolymerization, oxidation, hydrolysis, crosslinking, and biological attack. Chemical reactions between organic materials and pigments lead to discoloration or color alteration. Organic paint constituents are very challenging from an analytical point of view, and the following critical factors always need to be considered when planning analytical procedures: 716


715-727

June 2010

Vol. 43, No. 6

• several organic natural and synthetic substances are often simultaneously present in the layered structure; • nonoriginal compounds, formed as a result of aging or introduced by restoration treatments and pollution, are generally also present; • a very low amount of organic matter (a few percentage points in the overall weight or even lower) is generally encountered in a minute heterogeneous paint sample (,1 mg). For a complete understanding of the composition of paint layers, several techniques need to be used, including scanning electron microscopy/energy dispersive X-ray (SEM-EDX), X-ray diffraction (XRD), micro-Fourier transform infrared (micro-FTIR), micro-Raman, secondary ion mass spectrometry (SIMS), and many others.2 Nevertheless, at present, the coupling of gas chromatography with mass spectrometry (GC/MS) is the preferred analytical approach to characterize organic paint materials. The versatility of GC in the investigation of a very broad set of natural organic materials that can be found in artwork was pioneered by Mills and White1 and confirmed by a number of successful applications and case studies (refs 3 and 4 and references therein). The choice of GC is driven by the fact that natural organic substances are complex mixtures of many chemical species that are very similar to each other; the resolution and determination of the molecular profile is essential in order to identify the materials present and the aging pathways. Consequently, in this specific field, the coupling of GC with mass spectrometry is necessary due to the high number of compounds with similar retention times. In addition, because the most significant compounds are not available as commercial standards, identification cannot be based only on retention times and requires the confirmation by mass spectra. This Account focuses on the critical evaluation and troubleshooting in the various steps of the analytical procedures used in our laboratory for the GC/MS analysis of organic paint media, including the choice of analytical strategies and reference materials, sample pretreatment and purification to avoid possible interferences, data interpretation, and the use of reference materials and databases. Possible future developments are also discussed.

2. Analytical Strategies Most organic materials in a painting are macromolecular.1 In some cases, they are natural polymers, such as proteins or plant gums. Others undergo polymerization or cross-linking as an effect of exposure to light and air, such as natural resins


and drying oils. For this reason, it is not possible to analyze them directly by GC/MS, and a preliminary step to reduce the macromolecular analytes to low molecular weight molecules is needed. This can be achieved by coupling analytical pyrolysis with gas chromatography/mass spectrometry (Py-GC/MS) or by a wet-chemical treatment of the samples prior to GC/MS.3 When Py-GC/MS is used, the chemical composition of the sample is reconstructed on the basis of the interpretation of the molecular profile of the thermal degradation products of the original components. The technique involves minimal sample manipulation and no sample pretreatment, thus reducing the problems of contamination and sample loss associated with wet-chemical procedures. With the exception of some synthetic polymers (such as acrylic polymers), paint materials under pyrolysis produce polar, low volatile molecules, which need thermally assisted in situ derivatization, with either silylating or methylating agents, in order for them to be efficiently revealed through GC/MS (refs 5 and 6 and references therein). Even though Py-GC/MS is a fast and efficient technique to obtain the fingerprint of the organic materials in paint samples, GC/MS after wet chemical sample pretreatment is unsurpassed in its capacity to unravel the composition of the samples at a molecular level. The wet chemical pretreatment often includes chemolyses and derivatization reactions. The first step is needed in order to free small molecules from macromolecules or polymers (such as amino acids from proteins, fatty acids from triglycerides, or sugars from polysaccharides) thus providing molecules that are suitable for GC/MS analysis. Derivatization reactions transform molecules containing polar functional groups, such as carboxylic or alcoholic moieties, into less polar compounds, such as the corresponding silyl esters or ethers, thus increasing their volatility. As a result, in the case of heterogeneous multimaterial paint samples, when more than one class of compound, with very different chemical properties, need to be investigated, the complexity of the GC/MS procedure is increased by the necessity to separate and specifically treat the various types of analytes:4,7 • Proteins as polysaccharides need to undergo acidic hydrolysis in order to free the amino acids and sugars, respectively. However, to ensure the completeness of the reaction, by minimizing any loss of the most labile components, hydrolysis conditions must be specifically optimized, that is, milder for polysaccharides and harsher for proteins. • Glycerolipids and natural waxes require alkaline hydrolysis, but the complete hydrolysis of wax esters is much more difficult than that for glycerides.

FIGURE 1. Chromatograms obtained by GC/MS analysis of amino acids after acidic hydrolysis followed by silylation of (a) a sample from an Italian wall painting from the 13th century (Crypt of the Cathedral of Siena) containing high amounts of azurite (a coppercontaining blue pigment),8 (b) the same sample after the cleanup of the inorganic species, (c) a sample from an Italian panel painting from the 14th century containing high amounts gypsum in the preparation layer, and (d) the same sample after the clean-up of inorganic species.

• Polyesters of shellac components require alkaline hydrolysis before subsequent GC/MS analysis, but the choice of reagents affects the resulting molecular profile. In addition, the analysis of some paint binders may involve some interference in terms of the presence of inorganic compounds deriving from the support or from pigments: the characterization of proteinaceous and polysaccharide materials can Vol. 43, No. 6

June 2010

715-727


717


FIGURE 2. Flowchart showing a combined analytical procedure for the GC/MS analysis of glycerolipid, waxy, resinous, proteinaceous, and polysaccharide materials in one individual paint microsample.7

be affected by the occurrence of high amounts of inorganic species. Metal cations, including Hg2+, Fe3+, Cu2+, Pb2+, Cd2+, Zn2+, and Ca2+, and anions give rise to analytical interferences in the characterization of proteinaceous materials.8 To overcome these problems, a purification step based on the purification of the extract using a miniaturized sorbent tip (C18 or C4 stationary phase) can be included in the analytical procedure.7-9 As an example, Figure 1 shows the results of the amino acid analysis of two paint samples, one from the wall paintings in the Crypt of the Cathedral of Siena (Italy, 13th century) and the other from an Italian panel painting from the 14th century, before and after a cleanup aimed at eliminating cations and anions.8 GC/MS analysis was performed by hydrolysis and silylation of the amino acids. The chromatogram in Figure 1a refers to a wall paint sample containing high amounts of azurite (a copper-containing pigment) analyzed without the cleanup step. It clearly shows the hexadecane peak (I.S.1, the injection internal standard), but the peaks relative to amino acids, including norleucine (I.S.2, the derivatization internal standard) are absent. This is due to the analytical interference caused by the Cu2+ cations from azurite, which form complexes with amino acids, thus removing them from the silylation reaction. Applying a cleanup procedure to the same sample to remove inorganic species (chromatogram 718


715-727

June 2010

Vol. 43, No. 6

in Figure 1b) solves the problem and reveals the presence of a proteinaceous binder, which in this case was recognized as egg.8 High amounts of sulfates (the major components of gypsum paint grounds) can also be very problematic, as highlighted in Figure 1c, relating to a sample collected from an Italian painting from the 14th century containing a portion of the gypsum ground layer. In this case, in addition to the hexadecane peak, there is a huge peak relative to the trimethylsilyl-ester of sulfuric acid. The alanine and glycine peaks are visible, but there are no other amino acids with a higher retention time than sulfuric acid, thus highlighting the chromatographic contamination caused by the sulfates. Once again, purification using a C4 tip (chromatogram in Figure 1d) removed the interference, and the proteinaceous binder could thus be identified as egg. Taking into account the issues mentioned above and the necessity to identify as many materials as possible in just one paint microsample, a combined analytical procedure was developed for the simultaneous characterization of proteinaceous, polysaccharide, lipid, and resinous paint materials.7 The procedure (Figure 2) is based on a multistep chemical treatment of the sample, where solvent and solid phase extractions are used to separate materials with different physicochemical properties, thus obtaining three purified fractions for GC/MS analysis: a lipid-resinous, an amino acidic,


FIGURE 3. Sample collected from the gilding of the Orvetari Chapel: (a) cross-sectional image in dark field (white light); (b) same image in vis-fluorescence (UV light A filter); (c) sample composition, layer by layer.10

and a saccharide fraction. The method is microinvasive and can also be used to characterize very small heterogeneous samples whose organic content is only a few tenths of micrograms. The GC/MS analysis of a paint sample as a whole allows us to identify the organic materials contained in it but gives no information on their spatial distribution in the various layers that make up the paint. Assessing the composition of each layer is essential in order to fully understand how the artist worked. One approach is to try to physically separate the layers under an optical microscope, obtaining subsamples to be separately analyzed by GC/MS. Another possibility is to combine the results of the GC/MS analysis of the whole sample with the information obtained applying imaging techniques to the cross section of a duplicate sample by specular reflection FTIR, SIMS, and SEM-EDX. This approach was used in the characterization of the gilding technique in a sample collected from a fragment of the mural paintings in the Orvetari Chapel (15th century) that remained after the collapse as a result of

the bombing in 1944 of the Eremitani Church in Padua, Italy.10 The results are summarized in Figure 3. A double metallic leaf was revealed, made up of a gold leaf and a tin leaf glued together with an oily mordant. Double metallic leaf decorations were prepared by the artist in advance, cut into the desired shape and then applied onto the wall: a mordant layer was applied on the tin side of the decoration, and then pressed onto a colored mordant previously positioned on the wall. The thin layer of proteinaceous material directly applied on the plaster support guaranteed adhesion between the plaster and the superimposed layers.

3. Data Interpretation In works of art, the identification of organic materials characterized by GC/MS techniques is generally based on three main principles: chemotaxonomy, that is, the identification of the presence of one or more specific compounds (biomarkers), evaluation of the chromatographic pattern, and quantitative analysis. Vol. 43, No. 6

June 2010

715-727


719


FIGURE 4. Decisional scheme for the identification of polysaccharide gums.

Chemotaxonomy. Biomarkers are considered as stable

Natural waxes are typically identified on the basis of their

and diagnostic molecules present in a material, which have

molecular profile: beeswax, after hydrolysis, is character-

either been preserved intact or formed over the centuries

ized by the presence of long chain fatty acids with an even

due to aging. Identifying the composition at a molecular

number of carbons (from palmitic to dotriacontanoic acid),

level and thus the presence of biomarkers is crucial in order

(ω-1)-hydroxy acids with an even number of carbons (from

to obtain information on the kinds of organic substances

15-hydroxyhexadecanoic acid to 23-hydroxytetracosanoic

that were originally present in the sample and to under-

acid), long chain linear alcohols with an even number of

stand the natural or anthropogenic alteration processes that

carbons (from tetracosanol to tetratriacontanol), long chain

have modified the original composition of the samples.

(R,ω-1)-diols with an even number of carbons (from 1,23-

Biomarkers are mainly related to the source (animal/vege-

tetracosandiol to 1,27-octacosandiol), and long chain lin-

table or botanical origin) from which the material was obtained. For example, cholesterol is the most abundant sterol in animal fats, while phytosterols, mainly sitosterol, indicate a vegetable origin. Biomarkers can also shed light on the degradation processes undergone by the material, since they can be linked to the transformations caused by environmental factors and aging. The assessment of the presence and of the origin of terpenic resins, in both plants and animals, is based on the recognition of molecular biomarkers. Sandarac, colophony, dammar, and mastic resins are the most common natural resins found in paint samples, mainly as varnishes or as gilding mordants. The biomarkers used for the recognition of resins have been reported in the literature.

1,4,6,11,12

ear saturated hydrocarbons with the prevalence of an odd number of carbons (from tricosane to tritriacontane).1,4 The characterization of plant gums, which are naturally occurring polysaccharides exuded by several species of plants or extracted from some seeds, is also based on the evaluation of molecular profiles. In the Mediterranean basin, the gums traditionally used were Arabic gum, tragacanth gum, fruit tree gum, and locust bean. Ghatti and karaya gum were important materials in the Indian subcontinent. They are high molecular weight polymers consisting of aldopentoses, aldohexoses, and uronic acids joined together by a glycoside bond. Their identification by GC/MS requires hydrolysis in order to free sugars, followed by derivatization.4 Since each

Evaluation of the Chromatographic Pattern. In some

gum shows a different composition, their identification is

cases, identification is not based on a few very specific

based on the use of a decisional scheme,13 as reported in Fig-

biomarkers but on the simultaneous presence in the chro-

ure 4.

matogram of a series of significant molecules. These are

The presence of mixtures of saccharide materials and the

nonspecific if considered alone but become extremely indic-

effect of aging and biological attack make it difficult to iden-

ative when the overall chromatographic pattern is evalu-

tify a plant gum in a paint sample. Figure 5 shows the chro-

ated such as the entire series of alkanes, alcohols, or

matograms of six samples collected from mural paintings of

carboxylic acids.

Macedonian tombs (4th to 3rd centuries B.C.; AGAT and AZ

720


715-727

June 2010

Vol. 43, No. 6


FIGURE 5. Chromatograms of six samples collected from the mural paintings of Macedonian tombs (4th to 3rd centuries B.C.; AGAT and AZ samples) and from the Nestor palace (13th century B.C.; P sample), Greece.

samples) and from the Mycenaean Nestor palace (13th cen-

tions of the chromatographic profiles of plant gums

tury B.C.; P samples), Greece.

observed in paint samples still needs to be fully under-

All samples showed the presence of saccharide materi-

stood.

als: sample AGAT 30 contained fruit tree gum, samples 4

Quantitative Analysis. The specific identification of paint

P9 and 17 P11 contained a mixture of tragacanth and fruit

binders entails a quantitative determination using GC/MS of

tree gums, but samples AZ 3, AZ 18, and AGAT 4 were not

the following compounds:

classified. It is fundamental to stress that most of the chro-

• amino acids, for proteinaceous binders (egg, animal glue,

matographic profiles obtained, from a quantitative point of view, do not correspond to any of the reference gums analyzed.2,13 The reason for these quantitative modifica-

casein, or milk); • monosaccharides and uronic acids for polysaccharide gums; Vol. 43, No. 6

June 2010

715-727


721


FIGURE 6. Amino acid percentage content of a sample collected from an Italian mural painting from the 16th century attacked by fungi and of reference casein, egg, and collagen samples.

• monocarboxylic and R,ω-dicarboxylic acids for drying oils (linseed, walnut, poppy seed, or tung oil). Proteins used as paint media are essentially limited to three kinds, egg, casein, and collagen, so they can be distinguished on the basis of their different amino acid profiles.1,3 A comparison of the amino acidic profile of an unknown sample with reference materials can be efficiently and rapidly achieved by principal component analysis (PCA). The main limitation of an analytical approach to proteinaceous paint media identification based on amino acid quantitative determination is the presence of mixtures of different proteins in the same sample. Moreover, if unexpected proteinaceous materials are present, such as blood (used as binder in African wooden sculptures14) or garlic (an ingredient in adhesives for the application of gilding on paintings15), they will give rise to a different amino acid profile. One important aspect that must be considered when evaluating the protein content in a paint sample is the effect of biological agents such as fungi or bacteria, which can induce changes in the resulting amino acidic profile: glycine is increased, since it is the metabolism product of many bacteria and fungi, and some other amino acids are significantly decreased. Figure 6 shows the amino acid percentage content of a sample from an Italian mural painting by Niccolo` D’Abbate (16th century) attacked by fungi, compared with egg, collagen, and casein reference samples. The glycine content is quite high, but because hydroxyproline is absent, the presence of collagen can be ruled out. Moreover, the contents of aspartic acid and glutamic acid are very low, revealing a profile that cannot be ascribed to egg or to casein. With the increased availability of advanced instrumentation, proteomics may soon become the preferred approach for protein determination in paintings. Peptide mass mapping by 722


715-727

June 2010

Vol. 43, No. 6

MALDI-TOF and HPLC-MS/MS, when optimized and validated, will provide not only an identification of the kinds of proteinaceous binders but also additional information that is not accessible by amino acid analysis: for example, the distinction between egg yolk and egg glair, and the determination of the animal species from which collagen or casein have been obtained.16 A quantitation of monosaccharides and uronic acids must also be performed. Since saccharides are common environmental contaminants, an analysis of polysaccharide binders requires the quantitative evaluation of each sugar in order to assess whether it is present above or below the detection limit and can or cannot thus be used to identify the gum. Blank evaluations will be discussed later in this section. Moreover the identification of polysaccharide gums can be supported by the multivariate pattern analysis.13 The identification of lipids is based on the quantitation of mono- and dicarboxylic acids after saponification. Polyunsaturated acylic chains contained in fresh, nonaged drying oils (linseed oil, walnut oil, poppy seed oil, or tung oil) are not observed in aged oil paint layers because polyunsaturated acylic chains are subject to oxidative cleavage thus forming R,ω-dicarboxylic fatty acids, such as pimelic acid, suberic acid, sebacic acid, and azelaic acid being the most abundant, which can then be considered as markers for the presence of an aged drying oil. The amount of dicarboxylic acids formed in the curing and aging of egg lipids is substantially smaller than that in drying oils.17 Therefore the ratio between the amounts of azelaic and palmitic acid (A/P) is taken as a parameter for differentiating drying oils from egg lipids in paint samples (values of A/P > 1 indicate a drying oil, values of A/P < 0.3 are typical of egg lipids, while intermediate A/P ratio values are observed for “tempera grassa” in which egg and a drying oil are mixed).1 The presence of egg must be confirmed by amino acid analysis of the proteinaceous matter. Obviously the amount of dicarboxylic acids formed (sum of the percentage content of dicarboxylic acids, ∑D) is strictly dependent on the degree of oxidation, so the values observed may vary considerably and may be influenced by many factors such as the preheating of the oil media before use, the age of the paint, the conservation environment, and the effects of radical reactions initiated by pigments. The ratio between the amounts of palmitic and stearic acids (P/S) has been proposed as a possible index for differentiating between drying oils.1 The ratio is considered constant over time since these two saturated monocarboxylic acids are less subject to degradation during curing and aging. Typical P/S ratio values are reported as 1.4-2.4 for linseed oil, 2-4.5 for


als, as in “tempera grassa”, and the contribution of fatty acids from other sources, such as natural waxes. When ready-made oil colors arrived on the market, several vegetable oils were introduced into commercial formulations, to obtain some specific physical properties of the binder or as cheap adulterating agents. One of these oils, for example, is castor oil, a common ingredient in alkyd paint media. Very little is known about modern oily media, and a systematic investigation is needed. Moreover, fatty acids, and especially palmitic and stearic acids, are abundant in the environment and may conFIGURE 7. PCA score plot of a paint sample collected from an Afghan clay sculpture of the 6th century, obtained with and without performing the correction for the daily recovery.

walnut oil, 3-8 poppy seed oil, and 2.5-3.5 for egg. However, evaluating this parameter is particularly delicate because of the possible presence of mixtures of different lipid materi-

taminate the paint layer (hand contact, residues of burning vegetable oils and animal fats for lighting, etc.). Microorganisms can also alter the P/S values of those expected for pure materials. Another important aspect is that fatty acids react with metal cations in paint films thus forming metal soaps (ref 18 and references therein). Different reactivity, migration

FIGURE 8. Sample collected from Greek Byzantine icon from the 15th century: (a) TIC chromatogram relative to the lipid-resinous fraction where the compounds deriving from an acyl-lipid material (A), shellac (S), a Pinaceae resin (P), and beeswax (B) are indicated.

Vol. 43, No. 6

June 2010

715-727


723


speed, and solubility of the metal carboxylates in the paint media and in the cleaning solvents could alter the amounts of acids that are recovered in the analysis. Lastly, it is known that fatty acids can sublimate over time, altering the ratios between the saturated acids with a different number of carbons, that is, palmitic and stearic acids.19 As a result of all these factors, the widely used P/S parameter should be very carefully considered, and in our opinion, further research is needed to define the criteria for reliably identifying the botanical source of a drying oil in a paint sample. A quantitative analysis of compounds that have been previously subjected to derivatization should take several factors into account. The first is the fundamental need to use a derivatization internal standard and calibration curves. Modifications of GC/MS performances will affect compounds with different physical-chemical properties in different ways over time. Running daily standards enables us to evaluate the daily response of each analyte. Figure 7 explains the effect due to an uncalibrated response. It shows the PCA score plot of a paint sample collected from an Afghan clay sculpture from the 6th century: the sample not corrected for the daily recovery is located as an outlier, and its profile remains unassigned. After the correction, the sample is located perfectly in the casein cluster, thus indicating that this was the binder used. Lastly, dealing with proteins, lipids, and saccharides could involve a high level of environmental contamination. Blanks must thus be run frequently and the corresponding LOD (limit of detection) and LOQ (limit of quantitation) calculated, in order to avoid any misleading interpretation when the amount of analytes determined is not significant. For example, a painting sample taken from a Chinese clay sculpture from the 16th century showed the following amino acidic profile: Ala 6.7; Gly 12.6; Val 6.8; Leu 11.1; Ile 6.0; Ser 6.2; Pro 8.4; Phe 3.9; Asp 18.7; Glu 19.6; Hyp 0.0. This profile is similar to that of egg, and in fact, when this sample is processed with the PCA method, it falls very close to the egg cluster. In actual fact, the protein content, evaluated as the sum of the quantified amino acids, was 0.29 µg, and the LOD (limit of detection) was 0.30 µg; this indicated that the sample did not contain significant level of proteinaceous binder, and the amino acids detected were likely due to environmental contamination and not to the presence of egg. Case Study. Figure 8 shows three portions of the TIC chromatogram relating to the lipid-resinous fraction of a paint sample collected from a Greek Byzantine icon from the 15th century,9 in order to show how the marker recognition, evaluation of the chromatographic pattern, and quantitative analysis can be used to identify the organic materials present. The 724


715-727

June 2010

Vol. 43, No. 6

FIGURE 9. Sample collected from Greek Byzantine icon from the 15th century: (a) TIC chromatogram of the amino acid fraction; (b) PCA score plot.

characteristic biomarkers highlight the presence of shellac and pine resin. The profile of fatty acids, hydroxy acids, alcohols, diols, and alkanes points to beeswax and another acyl-lipid material. Figure 9a shows the TIC chromatogram of the amino acid fraction, and Figure 9b shows the PCA score plot obtained by comparing the amino acid profile with reference tempera paint layers, thus indicating that the sample contained animal glue.

4. Reference Materials and Databases Reference materials play an important role in analytical chemistry. Their basic purpose is to calibrate instruments and validate analytical procedures. In Conservation Science, they are an essential tool for laboratories involved in the development of analytical strategies for the analysis and characterization of materials from works of art. They can provide corroborating evidence for the presence of specific materials, allowing us to build up databases of chromatographic profiles and to expand mass spectral libraries. This aspect is particularly important in the case of natural organic materials. In fact, only a few compounds present both in fresh and in aged natural materials are commercially available as standards, and the analysis of ref-


FIGURE 10. (a) Average amino acidic composition of reference samples of garlic, dry and artificially aged. (b) PCA score plot of gilding and reference samples.

erence materials of a known origin is an essential step in the set up of analytical procedures and to support data interpretation. Reference materials can be divided into different kinds: raw materials and painting replicas of a known composition made according to recipes reported in historical treatises. Degradation due to aging from environmental parameters leads to changes in the composition of the original materials. Thus the study of painting replicas submitted to artificial aging treatments (using UV radiation, temperature, moisture, and environmental contaminants such as SO3, NOx, or O3) plays an important role by simulating the degradation processes and providing similar reference materials to naturally aged paintings.4 Even though there is currently no internationally accepted artificial aging protocol, it is clear that aging tests and the use of raw materials and replicas are the only possible approaches to study and understand degradation processes. Over the last 20 years, several collections of raw materials and painting replicas have been set up whose preparation has been made according to documented recipes, and there is an increasing interest on historical documentary source research and reconstruction on art materials and techniques.20 Studying old technical treatises is a very useful tool for the selection and collection of reference materials and to shed

light on old recipes. For example, the study of technical treatises from the 8th to the 15th century highlighted that garlic was commonly used as an ingredient in the manufacture of adhesives in gilding. The garlic protein content was then determined in several fresh, dry, and thermally aged replicas,15 and the average amino acidic composition obtained is reported in Figure 10a. Sixteen gilding samples collected from Italian mural paintings (14-17th century) were analyzed, and garlic was identified in four of these samples. Figure 10b shows the score plot obtained. An artist may also have used a very unusual material, and the reference materials may not yet be available or studied. The corresponding molecular markers are thus unknown. In these cases, analytical data may suggest that an unexpected material is present, and the study of selected reference materials, together with the interpretation of historical documents, may enable us to confirm this initial hypothesis and finally to improve the databases and mass spectral libraries.

5. Conclusions For the last 20 years, developing procedures based on GC/MS for analyzing organic paint materials has been a fundamenVol. 43, No. 6

June 2010

715-727


725


tal field of research in Conservation Science. Such procedures have enabled us to unravel these complex, structured mixtures of aged natural materials that constitute paint layers, to expand our knowledge of artists’ techniques, and to contribute to the preservation of paintings. Research on organic paint materials is still an open issue, where challenges and opportunities coexist. Our knowledge of the degradation processes during aging is still far from complete; much effort needs to be made to model and understand the relations between the macroscopic alteration and degradation of the paint layers, the chemical modification undergone by the constituent materials, and the influence of the interaction with the environment. Understanding these processes and interactions is fundamental for planning an efficient conservation and to preserve Cultural Heritage for future generations. These two aims can only be achieved if model paint replicas are used correctly, appropriate artificial aging protocols are developed, and collaborations continue to be set up between research groups in order to share reference materials and databases and to perform interlaboratory exercises and round-robin analyses on shared samples so as to validate and compare analytical procedures.

BIOGRAPHICAL INFORMATION Maria Perla Colombini currently holds the post of Full Professor of Analytical Chemistry in the Department of Chemistry (Faculty of Science) at the University of Pisa. She holds courses on Analytical Chemistry and the Chemistry of Cultural Heritage. She is Director of the Masters Course on “Materials and Diagnostic Techniques in the Cultural Heritage field”. Her research work includes developing analytical procedures based on spectroscopic and chromatographic techniques for characterizing micropollutants in the environment and, especially, organic materials and their degradation products in works of art and archaeological objects. She is head of the Chemical Sciences for the Safeguard of Cultural Heritage research group and specializes in the characterization of binders, organic dyes, and resins using chromatographic and mass-spectrometric techniques. Alessia Andreotti graduated in Chemistry in 2002 at the University of Pisa with a thesis on laser cleaning applied to the restoration of paintings. Since 2004, she has been working as a technician at the Department of Chemistry and Industrial Chemistry in the technical-scientific and data evaluation areas. Her research focuses on the characterization of natural and synthetic organic materials collected from samples in the field of Cultural Heritage using instrumental analytical techniques such as HPLC, GC/MS, Py-GC/MS, and direct exposure mass spectrometry (DEMS). She also specializes in using lasers and other state-of-theart techniques for cleaning of easel paintings, mural paintings, and other artifacts. 726


715-727

June 2010

Vol. 43, No. 6

Ilaria Bonaduce is a lecturer and permanent researcher in the Department of Chemistry and Industrial Chemistry at the University of Pisa; she received her Ph.D. in Chemical Science from the University of Pisa, Italy, in 2006. Her research focuses on the characterization of natural and synthetic organic materials used in works of art and the study of how they degrade during aging. Another major research interest is the development of analytical procedures for the identification of organic materials in paint samples, using mass spectrometric techniques, such as GC/MS, Py-GC/ MS, and DE-MS. Francesca Modugno received a Ph.D. in Analytical Chemistry in 2002 from the University of Pisa (Italy) and is currently a lecturer and permanent researcher in the Department of Chemistry and Industrial Chemistry at the University of Pisa. Her research deals with using analytical techniques such as GC/MS and PYGC/MS in the diagnosis and conservation of historical, artistic, and archaeological objects. Her research involves studying organic materials, such as terpenic resins, protein, lipids, and wood, and their degradation processes. Erika Ribechini received a Ph.D. in Chemical Sciences in 2006 from the University of Pisa. Since January 2009, she is a lecturer and permanent researcher in Analytical Chemistry in the Department of Chemistry and Industrial Chemistry at the University of Pisa. Her research mainly involves the development of analytical procedures based on chromatography and mass spectrometry for the study and characterization of organic natural substances from archaeological findings and works of art, with specific focus on vegetable oils, plant resins, and natural waxes. Her current research also involves the chemical characterization of archaeological waterlogged wooden artifacts using mass spectrometric methods. FOOTNOTES * To whom correspondence should be addressed. E-mail: [email protected]. Fax: 00390502219260. Phone: 00390502219305. † For A.A., e-mail: [email protected], fax: 00390502219260, phone: 00390502219252. For I.B., e-mail: [email protected], fax: 00390502219260, phone: 00390502219252. For F.M., e-mail: [email protected], fax: 00390502219260, phone: 00390502219405. For E.R., e-mail: [email protected], fax: 00390502219260, phone: 00390502219312. REFERENCES 1 Mills, J. S.; White, R. The Organic Chemistry of Museum Objects; Butterworth Heinemann: Oxford, U.K., 1994. 2 Scientific Examination for the Investigation of Paintings. A Handbook for Conservator-restorers; Pinna, D., Galeotti, M., Mazzeo, R., Eds.; Centro Di della Edifimi srl: Firenze, Italy, 2009. 3 Organic Mass Spectrometry in Art and Archaeology; Colombini, M. P., Modugno; F., Eds.; John Wiley & Sons: Chichester, U.K., 2009. 4 Andreotti, A.; Bonaduce, I.; Colombini, M. P.; Modugno, F.; Ribechini, E. Organic paint materials and their characterization by GC-MS analytical procedures. In New Trends in Analytical, Environmental and Cultural Heritage Chemistry; Tassi, L., Colombini, M. P., Eds.; Transworld Research Network: Kerala, India, 2008, pp 389-423. 5 Bonaduce, I.; Andreotti, A. Py-GC/MS of Organic Paint Binders. In Organic Mass Spectrometry in Art and Archaeology; Colombini, M. P., Modugno, F., Eds.; John Wiley & Sons: Chichester, U.K., 2009; pp 303-325. 6 Scalarone, D.; Chiantore, O. Py-GC/MS of Natural and Synthetic Resins. In Organic Mass Spectrometry in Art and Archaeology; Colombini, M. P., Modugno, F., Eds.; John Wiley & Sons: Chichester, U.K., 2009; pp 327-361. 7 Lluveras, A.; Bonaduce, I.; Andreotti, A.; Colombini, M. P. A GC/MS analytical procedure for the characterization of glycerolipids, natural waxes, terpenoid resins,


8

9

10

11 12 13

proteinaceous and polysaccharide materials in the same paint micro sample avoiding interferences from inorganic media. Anal. Chem. 2010, 82, 376–386. Gautier, G.; Colombini, M. P. GC-MS identification of proteins in wall painting samples: A Fast clean-up procedure to remove copper-based pigment interferences. Talanta 2007, 73, 95–102. Bonaduce, I.; Cito, M.; Colombini, M. P. The development of a gas chromatographic-mass spectrometric analytical procedure for the determination of lipids, proteins and resins in the same paint micro-sample avoiding interferences from inorganic media. J. Chromatogr. A 2009, 1216, 5931–5939. Bonaduce I.; Boon, J. J. An integrated mass spectrometric and molecular imaging analytical approach to identify and localise constituents in paintings applied to gilded multilayer structures from 14th to 16th C works of art. In New Trends in Analytical, Environmental and Cultural Heritage Chemistry; Tassi, L., Colombini, M. P., Eds.; Transworld Research Network: Kerala, India, 2008; pp 345-388. van der Doelen, G. A. Molecular studies of fresh and aged triterpenoid varnishes, PhD thesis, University of Amsterdam, 1999. De la Rie, E. R. Old master paintings: A study of the varnish problem. Anal. Chem. 1989, 61, 1228A-1233A, 1237A-1240A. Bonaduce, I.; Brecoulaki, H.; Colombini, M. P.; Lluveras, A.; Restivo, V.; Ribechini, E. Gas chromatographic-mass spectrometric characterisation of plant gums in samples from painted works of art. J. Chromatogr. A 2007, 1175, 275–282.

14 Mazel, V.; Richardin, P.; Debois, D.; Touboul, D.; Cotte, M.; Brunelle, A.; Walter, P.; Lapre’vote, O. Identification of ritual blood in African artifacts using ToF-SIMS and synchrotron radiation microspectroscopies. Anal. Chem. 2007, 79, 9253–9260. 15 Bonaduce, I.; Colombini, M. P.; Diring, S. Identification of garlic in old gildings by gas chromatography-mass spectrometry. J. Chrom. A 2006, 1107, 226–232. 16 Kuckova, S.; Hynek R.; Kodicek, M. Matrix-assisted Laser Desorption Ionisation Mass Spectrometry, Applied to the Analysis of Protein Paint Binders. In Organic Mass Spectrometry in Art and Archaeology; Colombini, M. P., Modugno, F., Eds.; John Wiley & Sons: Chichester, U.K., 2009; pp 165-187. 17 van den Berg, J. D. J.; van den Berg, K. J.; Boon, J. J. Identification of non cross-linked compounds in methanolic extracts of cured and aged linseed oil-based paint films using gas chromatography-mass spectrometry. J. Chromatogr. A 2002, 950, 195–211. 18 Keune, K. Binding media, pigments and metal soaps characterised and localised in paint cross-sections, Ph.D. thesis, University of Amsterdam, 2005. 19 Schilling, M.; Carson, D.; Khanjian. H. Gas chromatographic determination of the fatty acid and glycerol content of lipids IV. Evaporation of fatty acids and the formation of ghost images by framed oil paintings. Bridgland, J.; Brown, J. In Preprints of 12th Triennial Meeting of ICOM Committee for Conservation, James and James: London, 1999. LYON, pp 242-247. 20 www.clericus.org/astr/index.htm.

Vol. 43, No. 6

June 2010

715-727


727

In Situ Noninvasive Study of Artworks: The MOLAB Multitechnique Approach COSTANZA MILIANI,*,† FRANCESCA ROSI,‡ BRUNETTO GIOVANNI BRUNETTI,†,‡ AND ANTONIO SGAMELLOTTI†,‡ †

Istituto CNR di Scienze e Tecnologie Molecolari (CNR-ISTM), c/o Dipartimento di Chimica, Universita` degli Studi di Perugia, via Elce di Sotto 8, I-06123 Perugia, Italy, and ‡Dipartimento di Chimica, Universita` degli Studi di Perugia, via Elce di Sotto 8, I-06123 Perugia, Italy RECEIVED ON JANUARY 9, 2010

CON SPECTUS

D

riven by the need to study precious and irreplaceable artworks without compromising their integrity, researchers have undertaken numerous efforts to develop noninvasive analytical tools and methodologies that can provide a chemical description of cultural heritage materials without any contact with the object. The challenge is that artworks are made of complex mixtures, often with heterogeneous and unknown layered materials. Their components must be identified over a range of size scales, from the molecular identification of constituent compounds to the mapping of alteration phases. In this Account, we review recent research in spectroscopic techniques accessible from the mobile laboratory (MOLAB). The lab is equipped with an array of state-of-the-art, portable, and noninvasive instruments specifically tailored to tackle the different issues confronted by archaeologists, curators, and conservators. The MOLAB approach is suitable for studying a variety of objects, from ceramics to manuscripts or from historical wall paintings to contemporary canvases. We begin by discussing issues related to the acquisition and interpretation of reflectance or backscattered spectra from the surface of heterogeneous materials. Then we show how the selectivity needed for the noninvasive identification of pigments in paintings, even in mixtures or in layered matrices, can be acquired by combining elemental information from X-ray fluorescence with molecular and structural insights from electronic and vibrational spectroscopies. Discriminating between original pigments and restoration retouches is possible, even when both comprise similar chromophores, as highlighted in the study of paintings by Jordaens and Raphael. The noninvasive approach permits the examination of a very large number of artworks with a virtually limitless number of measurements. Thus, unexpected and uncommon features may be uncovered, as in the case of a lead pyroantimonate yellow doped with zinc that was discovered by micro-Raman and X-ray fluorescence on an Italian Renaissance majolica. For characterizing binding media, we discuss the strengths and limitations of using mid- and near-FTIR (Fourier transform infrared) spectroscopies supported by a multivariate statistical analysis, detailing the study of organic materials in a wall painting by Perugino and a survey of the painting technique on 18 contemporary paintings by Burri. In Michelangelo’s David, we show how the noninvasive mapping of contaminants and alteration phases might inform decisions on preventive conservation plans. The multitechnique MOLAB approach overcomes the intrinsic limitation of individual spectroscopic methods. Moreover, the ability to analyze artworks without the need to move them is an invaluable asset in the study and preservation of cultural heritage.

728


728-738

June 2010

Vol. 43, No. 6

Published on the Web 05/07/2010 www.pubs.acs.org/acr 10.1021/ar100010t © 2010 American Chemical Society

The MOLAB Multitechnique Approach Miliani et al.

1. Introduction In recent decades, several advanced analytical methods have been successfully employed to investigate cultural heritage materials working on both minute bulk samples or cross sections. However, the use of microdestructive techniques reveals two main limitations. The first is the requirement of sampling that, although minimal, is not advisible when dealing with irreplaceable or precious items. Second, even in instances where microspecimens may even be available (typically from the edges or from damages or lacunae), the sampling sites may not be representative of the entire artwork. To overcome these shortcomings, there has been a growing interest in the use of noninvasive analytical techniques able to yield information by remote examination of the entire artwork. In this context, we carry out our research activity aimed at setting up new spectroscopic tools and integrated methods for analyzing cultural heritage materials directly, without any contact with the surface of the object. In particular, in order to carry out noninvasive measurements in situ (i.e., in the same places where the artworks are exhibited or conserved), we moved toward the exploitation of portable equipment. Bringing analytical equipment from the laboratory to the artwork, rather than the reverse, conveys particular advantages. Any risk (and costs) connected with the transportation of a high-value and fragile object into a lab is avoided, paving the way for the investigation of a very large number of artworks. The results are obtained practically in real time, creating a new form of relationship between scientists, conservators, and curators, based on immediate discussion of results. The interest of the cultural heritage community for the noninvasive in situ approach is proved by the fact that research

groups based in conservation institutes, such us the Getty Conservation Institute,1 the Canadian Conservation Institute,3 and the Centre de Recherche et de Restauration de Museés de France,2 have assembled some analytical instruments to form portable laboratories in order to facilitate their research in artwork conservation. In the present Account, we report on recent developments in optimizing instruments and methods of a mobile laboratory (MOLAB) made of an integrated suite of several spectroscopic techniques ranging from X-ray to near-infrared. We initially discuss relevant issues concerning the interpretation of spectral signals collected in reflectance or backscattering mode from heterogeneous matrices. Afterward, selected experimental results are detailed with the aim of highlighting actual performances of the noninvasive approach in the study of the chemical composition of different types of objects. Remarks concerning the current shortcomings, as well as possible developments, are also given in the conclusion.

2. Noninvasive Portable Equipment and the Multitechnique Approach MOLAB is equipped with an array of state-of-the-art portable and noninvasive instruments for both point and imaging analyses. The spectroscopic point techniques, discussed in this Account, range from the X-ray to the IR region of the electromagnetic spectrum; their main characteristics in terms of portability, spectral parameters, spatial resolution, and probeartwork distance are provided in Table 1. Most of the systems have been intentionally set up, using commercial optical and electronic components, selected by achieving the best compromise between efficiency and portability. They all are designed to be compactly transported as separate kits that can be assembled and aligned directly onsite (Figure 1).

TABLE 1. Main Technical Parameters of MOLAB Equipment for Spectroscopic Point Analysisa technique

spectral range

*XRF4

spectral resolution

1.7-30 keV 5

250-2000 cm

micro-Raman 6

-1

mid-FTIR

7000-900 cm 7

7

UV-vis reflectance

8

UV-vis fluorescence TCSPT

12 0.1

dimension (cm3)/weight (kg)

probe-artwork distance (mm) 20

W anode X-ray source/Si drift detector

probe head: 20 × 10 × 10/0.5

15

Nd:YAG (532 nm) and diode (785 nm) laser/CCD/quartz fiber optic globar source/MCT detector/chalcogenide glass fiber optic halogen lamp/InGaAs detector/quartz fiber optic deuterium-halogen lamp/CCD/quartz fiber optic xenon lamp/CCD/quartz fiber optic LED (455 nm) and diode (375, 650 nm) laser/photocathode detector/silica fused fiber optic

12

50 × 50 × 50/35

4-8

-1

12500-4000 cm

4 cm

12

50 × 50 × 50/35

4-8

250-850 nm

2 nm

10

30 × 20 × 20/10

2

200-1100 nm

25 nm

2

30 × 20 × 20/10

4

12

60 × 70 × 50/35

4

350-850 nm

c

100 ps

experimental setup (excitation source/ detection system/sampling probe)

probe head: 40 × 40 × 30/3

-1

4 cm -1

*near-FTIR

9b

150 eV (FWHM) at 5.9 keV 10 cm-1

-1

spatial resolution (mm2)

a

Further experimental details are reported in the cited references. Asterisk indicates commercial equipment. fluorescence lifetime measurements. c Lifetime resolution.

Vol. 43, No. 6

June 2010

b

Time correlation single photon counting for

728-738


729


FIGURE 1. On-site installation of the mobile facility MOLAB at the Statens Museum for Kunst, Copenhagen.10

The challenge of on-site noninvasive measurements arises from matrix effects, due to the fact that the spectral signals are generally collected in backscattering or reflectance mode from the surface of heterogeneous and often layered systems. Matrix effects convey different consequences for each spectroscopic method and must be carefully evaluated to obtain reliable information from spectral features. Regarding X-ray fluorescence (XRF), the matrix effect leads to the loss of proportionality between intensity and concentration since the emission intensity of chemical elements is affected by the overall composition of the area under investigation.11,12 Some attempts have been recently reported on the quantitative interpretation of XRF spectra of favorable cases.13,14 Micro-Raman measurements from painting surfaces lead to a large interfering background generated by the luminescence of binders and varnishes.15 To minimize this problem, a timeresolved Raman portable spectrometer has been recently developed,16 while others have applied the subtracted shifted Raman spectroscopy method.17 Concerning UV-vis fluorescence, it has been recently shown that the luminescence, traveling through the paint, can be affected by several physical factors (i.e., self-absorption, multiple scattering, inner filter effects) resulting in shifts or even distortions of emission bands.15,16 In the attempt to avoid such pitfalls, a correction method based on KubelkaMunk theory has been recently optimized on both mock-ups and genuine artworks.18,19 Reflectance fiber optic mid-FTIR (Fourier transform infrared) measurements generally imply a normal optical layout (0°/0° geometry) and therefore the collection of both diffuse and specular reflection with a ratio that basically depends on 730


728-738

June 2010

Vol. 43, No. 6

FIGURE 2. Reflectance mid-FTIR spectra collected on a smooth (red lines) and rough (blue lines) surface of (a) an acrylic film, Paraloid B72 and (b) CaCO3. For comparative purposes, the respective transmission spectra are reported.

the roughness of the object surface. The consequences of competitive components of reflectance lead to challenges in the spectral interpretation, as well illustrated in Figure 2. In particular, the specular reflection, governed by Fresnel’s law, depends on both the absorption index (k) and refractive index


(n). Accordingly, organic compounds show derivative-like spectral features (Figure 2a), while spectra of minerals are often

TABLE 2. List of the Most Used Organic and Inorganic Blue Pigments from Antiquity to 19th Centurya

distorted by the inversion of those fundamental bands that show k . n (reststrahlen effect, Figure 2b).20 The diffuse reflection is ruled by Kubelka-Munk’s law and depends on both the absorption index and scattering coefficient. The main distortion generated by diffuse reflection is the change of relative band intensities with respect to transmission mode. Weak absorption bands strongly increase in their intensity; thus, in reflectance mode, absorptions relative to minor components or combination and overtone bands of major components become visible (Figure 2). The MOLAB approach for the integrated noninvasive study of a polychromatic artwork initially provides for a general survey using imaging methods (UV-vis fluorescence imaging and near-IR reflectography, not discussed in this Account) aimed at selecting specific areas of interest to tackle the issues posed by archeologists, curators, or conservators. The second step consists of a wide campaign of XRF measurements, followed by molecular spectroscopies: first, mid- and nearFTIR (for inorganic pigments and binders) and then UV-vis absorption and emission (for organic pigments). Afterward, micro-Raman measurements (for inorganic pigments) are usually attempted in the case of low fluorescing objects (i.e., ceramics and manuscripts).

3. In Situ Experimental Results In the past few years, we have been involved in several research projects studying over two hundred artworks: from marble sculptures21 (by Michelangelo, Antelami, Canova, etc.) to wall paintings22-24 (by Perugino, Gaddi, Lippi, etc.), from ancient easel paintings10,19,25,26 (by Raphael, Mantegna, Bronzino, Vasari, Jordaens, etc.) to modern and contemporary canvases4,7,27 (by Cezanne, Renoir, Rothko, Munch, Burri, etc.), and from ceramics5,28 (Renaissance Italian and early Meissen lusterwares) to manuscripts9 (Book of Kells, Codice Cospi, etc.). The activity has been executed in Italy, within national projects, as well as in Europe, within the transnational access service offered to European researchers through the EuARTECH and CHARISMA projects.29 3.1. Noninvasive Identification of Pigments. The identification of pigments used to paint works of art is fundamental to further the understanding of an object’s history or an artist’s technique, and may provide evidence for dating or attribution. Characterization of the artist’s original colored materials as well as materials applied later (by artists or restor-

name

formula

period32

techniqueb

indigo egyptian blue azurite lapis lazuli smalt Prussian blue synthetic ultramarine Thenard’s blue cerulean blue

C16H10N2O2 CaCuSi4O10 2Cu(CO3) · Cu(OH)2 Na7.5[Al6Si6O24]S4.5 SiO2(vit)Cox M(I)Fe(III)[Fe(II)(CN)6 · nH2O] Na7.5[Al6Si6O24]S4.5

antiquity antiquity antiquity antiquity antiquity from ∼1720 from 1824

UV-vis fluo8,9 and FTIR UV-vis fluo33 and FTIR XRF and mid-FTIR22 XRF and mid-FTIR34 XRF and near-FTIR35 XRF and mid-FTIR22,28 XRF and mid-FTIR34

CoAl2O4 CoO · n[SnO2]

from ∼1800 from ∼1870

XRF and near-FTIR4,7,30 XRF and near-FTIR30

a

MOLAB techniques used for their onsite noninvasive identification in paintings are provided. b Raman spectroscopy is not reported, since in easel paintings very often the scattering is hampered by fluorescence background from binders and varnishes.

ers) is important for providing criteria for an optimal conservation. Conversely, from other materials, there is rather ample literature on the exercising of a noninvasive and in situ approach to gather an insight into the chemical composition of colored compounds by means of a single analytical method, with XRF11-14 and UV-vis reflectance30,31 being the most commonly applied. On this basis, we developed a multitechnique strategy able to provide a description of molecular compositions comparable to that achievable from microdestructive conventional analyses.7 Some examples of the MOLAB multitechnique approach for the pigment identification in easel paintings and ceramics are here discussed. Blue Colors in Easel Paintings. The painting The Ferry Boat from Antwerp by Jacob Jordaens (c. 1623) was studied by MOLAB at the Statens Museum for Kunst in Copenhagen (Figure 1).10 One of the questions addressed in the project concerned the distribution of blue pigments over the painting surface, since several different tonalities were visible. By the multitechnique approach, properly combining XRF data with those from UV-vis fluorescence, mid-FTIR, and/or near-FTIR, it is currently possible to noninvasively discriminate between the nine different compounds listed in Table 2. Regarding Jordaens painting, five different compounds have been pinpointed alone, in mixtures or even layered. Smalt and indigo have been found on original areas such as on the woman’s skirt (Figure 3). Jordaens used indigo as a dark underpaint highlighted by partially discolored smalt brushstrokes. The presence of smalt has been inferred on the basis of cobalt signals in the XRF spectra and near-FTIR evidencing the typical shape of the electronic absorption due to d-d transition of Co(II) in pseudotetrahedral coordination35 (Figure 3a). Indigo has been ascertained on the basis of mid-FTIR rich features and the typical UV-vis fluorescence emission at about 730 nm (Figure 3b). These results have been confirmed Vol. 43, No. 6

June 2010

728-738


731


FIGURE 3. Study of blue pigments of The Ferry Boat from Antwerp (c. 1623) by Jordaens (detail of the painting is shown in the middle). In situ spectra (black lines) used for the identification of (a) smalt (left XRF, right near-FTIR); (b) indigo (left mid-FTIR, right UV-vis fluorescence); (c) azurite (left XRF, right mid-FTIR); (d) cerulean blue (left XRF, right near-FTIR), and (e) Prussian blue (left XRF, right mid-FTIR). Mid-FTIR spectra of references are reported for comparison (blue lines). Inset in (f) shows the optical image of a cross-section taken from the skirt of the woman on the boat.10

by the analyses of selected microsamples carried out by the museum conservators.10 As an example, the optical image of a cross section from the woman’s skirt is reported in Figure 3f, showing an upper thin layer of discolored smalt over a base of indigo. Jordaens also used azurite to color both the sky and the sea, often mixed with malachite and lead white. Azurite has been identified through the presence of the copper signal in XRF and infrared combination bands of both the copper carbonate (at about 2500 cm-1) and hydroxide (at about 4248 cm-1) moieties (Figure 3c). The latter features permit the identification of azurite also when in mixture with malachite.22 Beyond the original pigments, a further two blue compounds (cerulean blue and Prussian blue) have been revealed, assigned to previous retouches as these are not temporally compatible with Jordaens’s period (Table 2). Cerulean blue has 732


728-738

June 2010

Vol. 43, No. 6

been distinguished from smalt by the signal of tin in XRF and by the different position and shape of the Co(II) electronic transition in the near-infrared (Figure 3d).30,35 Prussian blue has been identified by the strong CN asymmetric stretching at 2090 cm-1 in the mid-FTIR spectrum22,28 (Figure 3e) in small traces on the woman’s skirt. The residual blue retouches are ascribable to the early restorations dating back to 1797-1884,10 probably carried out to disguise the discoloration of smalt. A similar issue has been addressed by MOLAB during the recent restoration of the Madonna of the Goldfinch by Raphael (c. 1506, Galleria degli Uffizi, Firenze). Conservators posed the problem to distinguish between early restoration materials and 19th century overpaints. In fact, the panel sustained severe damage only 40 years following its creation, when the owner’s house was unexpectedly ruined; consequently, it was


FIGURE 4. Study of the blue pigments featured in the Madonna of the Goldfinch (c. 1506) by Raphael. On the right, an image of the painting during the restoration process is shown (points refer to the measurement locations). On the left, mid-FTIR spectra recorded on the upper left part of the painting are shown (spectrum a corresponds to point 34, and spectrum b to point 35). Inset depicts the lapis lazuli chromophore structure made of sodalite β-cages containing CO2 as well as S3•-.34

restored for the first time by the painter Ridolfo del Ghirlandaio. Interestingly, noninvasive measurements proved that, for the reconstruction of damaged areas, Ridolfo used the same blue pigments originally employed by Raphael: lapis lazuli painted over a layer of azurite. In Figure 4, two mid-FTIR spectra are shown collected from areas of the sky, one painted by Raphael (spectrum a, point 34) and the other, at the extreme left-hand corner of the panel, painted by Ridolfo (spectrum b, point 35). Beyond the typical combination bands of azurite, both spectra have a sharp band at 2342 cm-1 that we proved to be due to the asymmetric stretching of CO2 entrapped in the sodalite β-cages of lapis lazuli, from its geological genesis.34 Diversely, the later retouches resulted to be made of smalt, Prussian blue, and synthetic ultramarine. This last one is the synthetic analogoue of lapis lazuli obtained by the furnacing of a mixture of kaolin, sulfur, and sodium carbonate. The two pigments share the same chromophore molecular structure (a sodalite framework containing the blue trisulfur radical anion, Figure 4) but can be differentiated since only the natural one retains traces of CO2.34 Encapsulated carbon dioxide can be straightforwardly identified by noninvasive midFTIR measurements through diffuse minor absorption band enhancement.34 Naples Yellows in Binary and Ternary Composition. In ceramics, pigments are applied as dispersed cations or crystalline phases embedded in transparent or opacified glazes. In

this fashion, they can be conveniently analyzed on-site by exploiting portable micro-Raman.36 In this regard, we have been recently engaged in a wide study on Naples yellow and its modified forms in glaze Renaissance ceramics.37 Naples yellow is one of the oldest known synthetic pigments, since its production goes back to about 3500 years ago; in Western European art, it has been used since the 16th century in Italian majolica and later in paintings. It consists of a lead antimonate oxide (Pb2Sb2O7) displaying the cubic pyrochlore structure as depicted in Figure 5a. The interest for modified versions of Naples yellow arises from a study by Roy and Berrie that in 1998 discovered the ternary form Pb2Sb2-xSnxO7-x/238 on Italian paintings of the 17th century. The existence of the modified pigment was later confirmed on a number of paintings from the 17th to the 19th century.26,39 Very recently, during a MOLAB project on 11 Italian Renaissance istoriato wares belonging to the Victoria and Albert Museum, we disclosed a further ternary form of Naples yellow on an original plate from Castel Durante (1537) where tin was substituted with zinc.5 The plate is shown in Figure 5b. The zinc modified pigment was used to color the bright orange stole of the boy, while the light yellow decorations on the background were depicted with the conventional lead pyrochlore (Figure 5c). In Figure 5d and e, XRF and Raman spectra collected on yellow and orange areas are compared. XRF graphs highlight the presence of tin (ascribable to cassiterite), lead, and antimony on both areas, while zinc is evidenced only on the orange one. The yellow decoration displays the typical Raman feature of Naples yellow, that is, a strong scattering at about 510 cm-1 related to the symmetric stretching of the SbO6 octahedra.37 The same Raman scattering mode in the orange decoration appears split in two bands, one again at 510 cm-1 and the other, less intense, at 450 cm-1. On the basis of structural study on standards of lead antimony yellows, we proved this spectral feature distinctive of a doped pyrochlore with antimony partially substituted by zinc or tin.37 This experimental observation finds an explanation in historical records, considering what Piccolpasso suggested about producing the antimonate yellow: “...to make the most beautiful zalulino (i.e., Naples yellow) many add some tutia allesandrina, which is very good”, where very likely tutia refers to zinc oxide as indicated by a long series of quotations in Western historical sources.40 3.2. Study of the Painting Technique: Binding Media Characterization. The binding media (basically proteins, glycosides, and lipids in ancient art, yet a wide range of synthetic polymers in contemporary art) with their own chemical-physVol. 43, No. 6

June 2010

728-738


733


FIGURE 5. (a) Crystal structure of pyrochlore represented as two interpenetrating networks. The A2O′ network corresponds to fourcoordinate O′ ions and two-coordinate Pb cations. The B2O6 framework consists of SbO6 octahedra sharing all vertices to form large cavities.37 (b) Image of plate 7688-61 from the V&A collection (1537). (c) Detail of the plate showing the orange stole and the yellow decorations on the back. (d) XRF spectra collected on the yellow (black line) and orange (gray line) decorations. (e) Raman spectra collected on the yellow (black line) and orange (gray line) decorations.

ical properties are the components that mostly affect the modus operandi of an artist. Even if there are benchtop analytical techniques that, on microsamples, may give detailed specification on a binder up to its biological origin, the value and benefit of a methodology able to noninvasively produce information on the binder’s chemical class (proteins, lipids, etc.) are self-evident. To this purpose, we have recently set up a methodology for the on-site noninvasive binder characterization exploiting reflectance vibrational spectroscopy in the near-27,41 and mid-6,22,27 infrared region. Organic Binders in Wall Paintings. The characterization of organic materials in wall paintings is particularly challenging due to their small amount with respect to the inorganic matrix (basically composed of lime-based plasters and pigments). To tackle this issue, we explored strengths and limitations of using the reflectance mid-FTIR technique supported by multivariate statistical analysis.6 Specifically, we have constructed a model based on the principal component analysis (PCA) of reflectance spectra collected on hundreds of mocksups (painted with different binders and pigments) and then tested on a number of wall paintings by Perugino, Gaddi, and Lippi. As an example, the t1 vs t4 score plot resulting from the PCA of a data matrix composed of spectra collected on mockups painted a secco with three different binders (casein, animal glue, whole egg) and a fresco with water is shown in Figure 6a. Application with water and casein are distinguished 734


728-738

June 2010

Vol. 43, No. 6

from applications with whole egg and rabbit skin glue. The possibility to differentiate between whole egg and animal glue depends on the amount of the lipid component in egg, whose carbonyl bond at 1740 cm-1 is not always well resolved from the reststrahlen band of calcium carbonate. This PCA model has been tested on spectral data acquired on a wall painting by Perugino (Adoration of the Magi, 1521).22 The painting is characterized by the presence of residues from an original secco overpaint that is appreciable in the UV-fluorescence image (see the damask decoration of the Wise King’s gown, Figure 6c), while not visible in normal light. The univariate analysis of the spectra from the Wise King’s gown highlighted the following: a proteinaceous binder showing strong signals of νCH, amide I and amide II, Figure 6d, most probably for a gold application on the fluorescent damask decorations, and weak signals of an organic compound in the spectra recorded on nonfluorescing areas. Projecting all these data on the t1 vs t4 space generated by the PCA model, all the spectra collected on fluorescent damask decorations are located in the upper part of the diagram (full red stars, Figure 6a) corresponding to the a secco application by glue or egg. Conversely, the spectra collected on nonfluorescing areas fall in the area of pure fresco (open red stars, Figure 6a), because the weak signals of the organic component are not considered relevant by the model. Organic Binders in Contemporary Paintings. In the present section, a mid- and near-FTIR study for the noninva-


FIGURE 6. Characterization of the organic binders in the wall painting Adoration of the Magi by Perugino (1521). (a) t1 vs t4 score plot resulting from the PCA calculated on a matrix data set of 150 spectra of replicas painted with water (black square), glue (gray triangle), casein (black circle), and egg (light gray inverted triangle). (b) Images of the replicas. (c) UV-fluorescence imaging of a detail of the painting. (d) Comparison of reflectance mid-FTIR spectra collected on fluorescing and nonfluorescing areas of the Wise King’s dress. The stars in (a) refer to the projection of spectral data from fluorescing (full stars) and not fluorescing (open stars) areas of the Wise King’s dress.

sive characterization of organic binders on 18 paintings by Alberto Burri dating from 1948 to 1975 and belonging to the Collezione Albizzini (Citta` di Castello, Italy) is detailed.27 The survey highlighted that Burri widely used poly(vinyl acetate) (PVA) starting from the early works dated to 1948, along with more traditional media such as proteins and lipids. In Figure 7a, two spectra collected on a smooth (B52_a) and rough (B52_b) area of the painting Bianco (1952) are reported and compared with the spectra of vinyl, alkyl, and acrylic resins. The combined presence of the νsym(C-O-C) (at 1240 cm-1) with the intense peak of δsym(CH3) (at 1370 cm-1) and the weaker signal at 1440 cm-1 has been used for the identification of PVA. In the near-infrared region, the ν+δ(CH) combination bands and the ν(CH) first overtone result in a characteristic profile, distinct for each synthetic resin, thus confirming the presence of PVA. In Figure 7b, two mid-FTIR spectra collected on two paintings are compared with an animal glue standard. The combined presence of signals at 1660 cm-1 (amide I), 3070 cm-1 (2δNH), and 3309 cm-1 (νNH) sug-

FIGURE 7. Study of Burri’s painting technique. Upper: (a) mid-FTIR spectra collected on smooth (B52_a) and rough (B52_b) areas of Bianco; for comparative purposes the spectra of standards are reported (gray lines). (b) Mid-FTIR spectra collected on Bianco (B52) and Gobbo (G52) compared with a glue standard (light gray line). (c) Reflectance near-FTIR spectra acquired on Rosso (R50), SZ1 (Sz49), Bianco (B51), and Nero 1 (N48) compared with the near-FTIR spectra of standards. Lower: Detail of Bianco; the compositional distribution is detailed.27

gests the presence of a proteinaceous binder. Near-FTIR measurements (Figure 7c) have also revealed the presence of lipidic components, detected by the characteristic doublet at 4340-4250 cm-1 ascribed to the ν+δ(CH) of aliphatic chains and the first overtone of ν(CH) at about 5800-5680 cm-1.41 Burri employed synthetic and natural binders to obtain different morphologies and optical effects as observed in the white monochrome painting Bianco (1952) shown in Figure 7. A multivariate statistical approach applied to the combined vibrational mid-FTIR and elemental XRF data has permitted chemical composition of the white monochrome surface to be mapped, thus correlating the visual morphological differences to specific materials and techniques. Specifically, the multivariate analysis revealed that the “cold” white areas were painted with ZnO mixed with PVA, while the opaque and “warm” white areas with CaCO3, BaSO4, and a minor amount of ZnO mixed with proteins and lipids (Figure 7d). Notably, in Vol. 43, No. 6

June 2010

728-738


735


statue probably due to rain that could have affected the higher part of the sculpture when in its original open air location, leading to the enriched formation of oxalates in the lower part. Spectral features of beeswax have been recognized on specific areas of the sculpture, likely a residue of the detachment additives used during the casting of the David in 1847 or from an old protective substance.21

4. Breakdowns and Perspectives

FIGURE 8. Mapping of the surface of Michelangelo’s David.21 Reflectance mid-FTIR spectra collected from (a) a sulfate incrustation (on the left arm), (b) an oxalate patina (on the right foot), and (c) a wax contamination (on the right thigh). For comparative purposes, the spectra of standards gypsum, wewhellite, and beeswax are reported in red.

correspondence of the “warm” white areas, Zn carboxylates have also been recognized, likely originating from the interaction of free fatty acids of the lipidic binding medium with zinc oxide.27 3.3. Molecular Mapping of Alterations and Contaminants. The value and benefit of applying on-site noninvasive methodologies is not only limited to the characterization of the constituent original materials of a work of art but may also be extended to identifying alteration products and contaminants as well as to monitor online the efficacy and harmfulness of cleaning treatments aimed to their removal. In this context, the MOLAB facility has been successfully exploited to monitor the conservation state and cleaning methods of ancient paintings (by Antonello da Messina and Raphael), modern artworks (by Zolla42 and Munch), and carbonate sculptures (by Antelami and Michelangelo21). In particular, Michelangelo’s David was studied by portable mid-FTIR spectroscopy within a wide diagnostic program aimed at designing a preventive conservation plan. On-site infrared point measurements, carried out on scaffoldings, allowed for a noninvasive examination of the David’s entire surface21 (Figure 8). Sulfates were clearly detected on residual gray incrustations but also in areas that did not appear macroscopically altered, and these were attributed to dry indoor depositions. Oxalates were also identified mainly localized in the lower part of the 736


728-738

June 2010

Vol. 43, No. 6

The results outlined in this Account demonstrate that noninvasive on-site methodologies have opened up a new way for conservation scientists to analyze artworks giving valuable information for their understanding and long-term preservation, while fully respecting their aesthetical and historical value. The MOLAB multitechnique approach allows one to overcome the intrinsic limitation of each single spectroscopic method, thus gaining a high selectivity in the identification of materials. Studying easel paintings, the current main limitation concerns the pigments composed of oxides or sulfides of heavy metals (such as lead oxides, arseniate, antimonate and stannate yellows, etc.) that generally are not analyzable by Raman spectroscopy due to the large interference from binder and varnish fluorescence. The recent development of portable equipment for noninvasive X-ray diffraction measurements is opening up the possibility of overcoming these shortcomings.43,44 The characterization of organic binders is intrinsically more challenging. However, employing mid- and near-infrared spectroscopy, encouraging results have been achieved, especially in contemporary canvas paintings that are generally less affected by constraints due to the small quantity of organic binder (with respect to wall paintings) or to the interferences from added restoration materials (with respect to ancient paintings). Possible improvements could arise from the combined exploitation of other techniques such as time-resolved fluorescence spectroscopy45 and unilateral NMR.46,47 Finally, it is worth underlining that the currently available portable noninvasive methods have strong limitations in fully resolving complex layer stratigraphy. In this regard, it is of interest to mention that the recent efforts to bring the three-dimensional micro-XRF from the synchrotron to the laboratory48 are paving the way for on-site measurements of three-dimensionally resolved X-ray fluorescence of painting layers. The work has been carried out through the support of the EU within the sixth FP (Eu-ARTECH, RII3-CT-2004-506171) and the seventh FP (CHARISMA Project No. 228330) and through the support of the Italian MIUR (PRIN 2006035484, FIRB RBNE03SML9). The authors are grateful to several researchers


that have contributed to the MOLAB activity: C. Anselmi, L. Cartechini, C. Clementi, A. Cosentino, A. Daveri, B. Doherty, K. Kahrim, V. Manuali, F. Presciutti, A. Romani, and M. Vagnini.

BIOGRAPHICAL INFORMATION Costanza Miliani is Researcher at the CNR-ISTM (Istituto di Scienze e Tecnologie Molecolari) in Perugia. She has authored 45 articles concerning structural, electronic, and vibrational properties of materials of interest for cultural heritage. Francesca Rosi is post-Doc at the Chemistry Department of the University of Perugia. Her research interests include the application and development of noninvasive and portable spectroscopic techniques for studying materials of interest in the field of cultural heritage. Brunetto Giovanni Brunetti is Professor of General Chemistry at the University of Perugia and author of 120 scientific publications on chemical reaction dynamics and spectroscopy applied to cultural heritage. He has been coordinator of the EU supported project EU-ARTECH (sixth FP) and currently coordinates CHARISMA (seventh FP). Antonio Sgamellotti is Professor of Inorganic Chemistry at the University of Perugia, President of the Center of Excellence SMAArt (Scientific Methodologies applied to Archaeology and Art), and author of more than 300 scientific publications on advanced computational chemistry and on spectroscopic investigations of artwork materials.

REFERENCES 1 http://www.getty.edu/conservation/science/about/portablelab.html (last accessed on 30 March 2010). 2 http://www.cci-icc.gc.ca/about-apropos/nb/nb37/analy-eng.aspx (last accessed on 30 March 2010). 3 http://www.c2rmf.fr/pages/page_id18170_u1l2.htm (last accessed on 30 March 2010). 4 Rosi, F.; Burnstock, A.; Van den Berg, K. J.; Miliani, C.; Brunetti, B. G.; Sgamellotti, A. A non-invasive XRF study supported by multivariate statistical analysis and reflectance FTIR to assess the composition of modern painting materials. Spectrochim. Acta, Part A 2009, 71, 1655–1662. 5 Rosi, F.; Manuali, V.; Miliani, C.; Grygar, T.; Bezdicka, P.; Burgio, L.; Seccaroni, C.; Sgamellotti, A.; Brunetti, B. G. Raman scattering features of lead pyroantimonate compounds: implication for the non-invasive identification of yellow pigments on ancient ceramics. Part II. In-situ characterization of Renaissance plates by portable micro-Raman and XRF. J. Raman Spectrosc. 2010, in press. 6 Rosi, F.; Daveri, A.; Miliani, C.; Verri, G.; Benedetti, P.; Pique´, F.; Brunetti, B. G.; Sgamellotti, A. Non-invasive identification of organic materials in wall paintings by fiber optic reflectance infrared spectroscopy: a statistical multivariate approach. Anal. Bioanal. Chem. 2009, 395, 2097–2106. 7 Rosi, F.; Miliani, C.; Burnstock, A.; Brunetti, B. G.; Sgamellotti, A. Non-invasive insitu investigations versus micro-sampling: a comparative study on a Renoirs painting. Appl. Phys. A: Mater. Sci. Process. 2007, 89, 849–856. 8 Clementi, C.; Miliani, C.; Romani, A.; Santamaria, U.; Morresi, F.; Mlynarska, K.; Favaro, G. In-situ fluorimetry: a powerful non-invasive diagnostic technique for natural dyes used in artefacts. Part II Identification of orcein and indigo in Renaissance tapestries. Spectrochim. Acta. Part A 2009, 71 (5), 2057–2062. 9 Romani, A.; Clementi, C.; Miliani, C.; Brunetti, B. G.; Sgamellotti, A.; Favaro, G. Portable Equipment for Luminescence Lifetime Measurements on Surfaces. Appl. Spectrosc. 2008, 62, 1395–1399. 10 Filtenborg, T.; Hendrikman, L.; Noldus, B.; De la Fuente Pedersen, E.; Schlotter, A.; Verhave, J.; Wadum, J. Jordaens The Making of a Masterpiece; Statens Museum for Kunst: Copenhagen, 2008.

11 Janssens, K.; Vekemans, B.; Adams, F.; Oost, A. A compact small-beam XRF instrument for in-situ analysis of objects of historical and/or artistic value. Spectrochim. Acta, Part B 1999, 54, 1697–1710. 12 Pessanha, S.; Guilherme, A.; Carvalho, M. L. ; Comparison of matrix effects on portable and stationary XRF spectrometers for cultural heritage samples. Appl. Phys. A: Mater. Sci. Process. 2009, 97, 497–505. 13 De Viguerie, L.; Sole, V. A.; Walter, P. Multilayers quantitative X-ray fluorescence analysis applied to easel paintings. Anal. Bioanal. Chem. 2009, 395, 2015–2020. 14 Bonizzoni, L.; Galli, A.; Poldi, G.; Milazzo, M. In situ non-invasive EDXRF analysis to reconstruct stratigraphy and thickness of renaissance pictorial multilayers. X-Ray Spectrom. 2007, 36, 55–61. 15 Vandenabeele, P.; Tate, J.; Moens, L. Non-destructive analysis of museum objects by fibre-optic Raman spectroscopy. Anal. Bioanal. Chem. 2007, 387, 813–819. 16 Osticioli, I.; Mendes, N. F. C.; Porcinai, S.; Cagnini, A.; Castellucci, E. Spectroscopic analysis of works of art using a single LIBS and pulsed Raman setup. Anal. Bioanal. Chem. 2009, 394, 1033–1041. 17 Rosi, F.; Paolantoni, M.; Clementi, C.; Doherty, B.; Miliani, C.; Brunetti, B. G.; Sgamellotti, A. Subtracted shifted Raman spectroscopy of organic dyes and lakes. J. Raman Spectrosc. 2010, 41, 452–458. 18 Verri, G.; Clementi, C.; Comelli, D.; Cather, S.; Piquee´, F. Correction of UltravioletInduced Fluorescence Spectra for the Examination of Polychromy. Appl. Spectrosc. 2008, 62, 1295–1302. 19 Clementi, C.; Miliani, C.; Verri, G.; Sotiropoulou, S.; Romani, A.; Brunetti, B. G.; Sgamellotti, A. Application of the Kubelka-Munk Correction for Self-Absorption of Fluorescence Emission in Carmine Lake Paint Layers. Appl. Spectrosc. 2009, 63, 1323–1330. 20 Korte, H.; Roseler, A. Infrared reststrahlen revisited: commonly disregarded optical details related to n < 1. Anal. Bioanal. Chem. 2005, 382, 1987–1992. 21 Ricci, C.; Miliani, C.; Brunetti, B. G.; Sgamellotti, A. Non-invasive identification of surface materials on marble artifacts with fiber optic mid-FTIR reflectance spectroscopy. Talanta 2006, 69, 1221–1226. 22 Miliani, C.; Rosi, F.; Borgia, I.; Benedetti, P.; Brunetti, B. G.; Sgamellotti, A. FiberOptic Fourier Transform Mid-Infrared Reflectance Spectroscopy: A Suitable Technique for in Situ Studies of Mural Paintings. Appl. Spectrosc. 2007, 61, 293– 299. 23 Sotiropoulou, S.; Sister Daniilia, Miliani, C.; Rosi, F.; Cartechini, L.; PapanikolaBbakirtzis, D. Microanalytical investigation of degradation issues in Byzantine wall paintings. Appl. Phys. A: Mater. Sci. Process 2008, 92, 143–150. 24 Rosi, F.; Miliani, C.; Borgia, I.; Brunetti, B.; Sgamellotti, A. Identification of nineteenth century blue and green pigments by in situ XRF and micro-Raman spectroscopy. J. Raman Spectrosc. 2004, 35, 610–615. 25 Miliani, C.; Daveri, A.; Spaabaek, L.; Romani, A.; Manuali, V.; Sgamellotti, A.; Brunetti, B. G. Bleaching of red lake paints in encaustic mummy portraits. Appl. Phys. A: Mater. Sci. Process., DOI: 10.1007/s00339-010-5748-3. 26 Grygar, T.; Hradil, D.; Hradilova´, J.; Bezdie`ka, P.; Gruˆnwaldova´, V.; Fogasˇ, I.; Miliani, C. Microanalytical identification of Pb-Sb-Sn yellow pigment in historical European paintings and its differentiation from lead tin and Naples yellows. J. Cult. Heritage 2007, 8, 377–386. 27 Rosi, F.; Miliani, C.; Clementi, C.; Kahrim, K.; Presciutti, F.; Vagnini, M.; Manuali, V.; Daveri, A.; Cartechini, L.; Brunetti, B. G.; Sgamellotti, A. An integrated spectroscopic approach for the non invasive study of modern art materials and techniques. Appl. Phys. A: Mater. Sci. Process., in press. 28 Miliani, C.; Doherty, B.; Daveri, A.; Loesch, A.; Ulbricht, H.; Brunetti, B. G.; Sgamellotti, A. In situ non-invasive investigation on the painting techniques of early Meissen Stoneware. Spectrochim. Acta, Part A 2009, 73, 587–592. 29 www.eu-artech.org; www.charismaproject.eu (last accessed on 30 March 2010). 30 Bacci, M.; Magrini, D.; Picollo, M.; Vervat, M. A study of the blue colors used by Telemaco Signorini (1835-1901). J. Cult. Heritage 2009, 10, 275–280. 31 Dupuis, G.; Menu, M. Quantitative characterization of pigment misture used in art by fibre-optics diffuse-reflectance spectroscopy. Appl. Phys. A: Mater. Sci. Process. 2006, 83, 469–474. 32 Eastaugh, N.; Walsh, V.; Chaplin, T.; Siddal, R. Pigment Compendium A Dictionary of Historical Pigments; Elsevier: Oxford, 2004. 33 Accorsi, G.; Verri, G.; Bolognesi, M.; Armaroli, N.; Clementi, C.; Miliani, C.; Romani, A. The exceptional near-infrared luminescence properties of cuprorivaite (Egyptian blue). Chem. Commun. 2009, 3392–3394. 34 Miliani, C.; Daveri, A.; Brunetti, B. G.; Sgamellotti, A. CO2 entrapment in natural ultramarine blue. Chem. Phys. Lett. 2008, 466, 148–151. 35 Bacci, M.; Picollo, M. Non-destructive spectroscopic detection of cobalt (II) in painting and glass. Stud. Conserv. 1996, 41, 136–144.

Vol. 43, No. 6

June 2010

728-738


737


36 Colomban, P.; Milande, W. On-site Raman analysis of the earliest known Meissen porcelain and stoneware. J. Raman Spectrosc. 2006, 37, 606–613; and references therein. 37 Rosi, F.; Manuali, V.; Miliani, C.; Brunetti, B. G.; Sgamellotti, A.; Grygar, T.; Hradil, D. Raman scattering features of lead pyroantimonate compounds. Part I: XRD and Raman characterization of Pb2Sb2O7 doped with tin and zinc. J. Raman Spectrosc. 2009, 40, 107–111. 38 Roy, A.; Berrie, B. H. A New Lead-based Yellow in the Seventeenth Century in Painting Techniques History, Materials and Studio Practice; Roy, A., Smith, P., Eds.; The International Institute for Conservation of Historic and Artistic Works: London, 1998; pp 160-165. 39 Sandalinas, C.; Ruiz-Moreno, S. Lead-tin-antimony yellow historical manufacture, molecular characterization and identification in seventeenth-century Italian paintings. Stud. Conserv. 2004, 49, 41–52. 40 Seccaroni, C. Giallorino: storia dei pigmenti gialli di natura sintetica; De Luca Eds D’Arte: Rome, 2006. 41 Vagnini, M.; Miliani, C.; Cartechini, L.; Rocchi, P.; Brunetti, B. G.; Sgamellotti, A. FT-NIR spectroscopy for non-invasive identification of natural polymers and resins in easel paintings. Anal. Bioanal. Chem. 2009, 395, 2107–1118. 42 Kahrim, K.; Daveri, A.; Rocchi, P.; de Cesare, G.; Cartechini, L.; Miliani, C.; Brunetti, B. G.; Sgamellotti, A. The application of in situ mid-FTIR fibre-optic reflectance

738


728-738

June 2010

Vol. 43, No. 6

43

44 45

46

47

48

spectroscopy and GC-MS analysis to monitor and evaluate painting cleaning. Spectrochim. Acta, Part A 2009, 74, 1182–1188. Gianoncelli, A.; Castaing, J.; Ortega, L.; Dooryheé, E.; Salomon, L.; Walter, P.; Hodeau, J.-L.; Bordet, P. A portable instrument for in situ determination of the chemical and phase compositions of cultural heritage objects. X-Ray Spectrom. 2008, 37, 418–423. Chiari, G. Saving Art in situ. Nature 2008, 453, 159–159. Nevin, A.; Comelli, D.; Valentini, G.; Anglos, D.; Burnstock, A.; Cather, S.; Cubeddu, R. Time-resolved fluorescence spectroscopy and imaging of proteinaceous binders used in paintings. Anal. Bioanal. Chem. 2007, 388, 1897–1905. Presciutti, F.; Perlo, J.; Casanova, F.; Glo¨ggler, S.; Miliani, C.; Blu¨mich, B.; Brunetti, B. G.; Sgamellotti, A. Non-invasive NMR profiling of painting layers. Appl. Phys. Let. 2008, 93, 033505. Del Federico, E.; Centeno, S. A.; Kehlet, C.; Currier, P.; Stockman, D.; Jerschow, A. Unilateral NMR applied to the conservation of works of art. Anal. Bioanal. Chem. 2010, 396, 213–220. Mantouvalou, I.; Lange, K.; Wolff, T.; Gro¨tzsch, D.; L¨; uhl, L.; Haschke, M.; Hahn, O.; Kanngieβer, B. A compact 3D Micro X-Ray fluorescence spectrometer with X-ray tube excitation for archaeometric applications. J. Anal. At. Spectrom. 2010, 25, 554–561.

Advances in Laser Cleaning of Artwork and Objects of Historical Interest: The Optimized Pulse Duration Approach SALVATORE SIANO* AND RENZO SALIMBENI Istituto di Fisica Applicata “Nello Carrara”, Consiglio Nazionale delle Ricerche, Sesto Fiorentino (Florence), Italy RECEIVED ON JULY 8, 2009

CON SPECTUS

L

aser ablation has found numerous applications in biomedical and industrial settings but has not spread as quickly as a means of cleaning artwork. In this Account, we report recent advances in the study and application of laser cleaning to the conservation of cultural heritage. We focus on the solution of representative cleaning problems of encrusted stones, metals, and wall paintings that were achieved through the optimization of laser pulse duration. We begin by introducing the basic mechanisms involved in the laser ablation of stratified materials and the criteria for preventing undesired side effects to the substrate and then briefly present case studies for each of these materials. Laser interaction effects are reviewed in a schematic way, with a concise overview of the physical models needed to support intuitive interpretations of the phenomenology observed, both in laboratory tests and in practical applications on important artifacts. This approach aims to provide keys of generalization that will favor the rigorous application of laser cleaning, repeatability of the successful results reported in this work, and further dissemination and acceptance of the technique. The topics treated examine the ablation mechanisms along with the efficiency, gradualness, selectivity, and effectiveness of the technique as a function of the pulse duration of neodymium laser systems and the operating conditions. Physical modeling and experimental evidence support the selection of pulse durations of between several tens of nanoseconds and several tens of microseconds, making it possible to minimize the risk of photothermal and photomechanical effects and maximize the selectivity of the ablation process. The sections dedicated to stones and metals also deal with the important problem of discoloration, which has significantly slowed the spread of the laser cleaning technique. The well-known problem of a yellowish appearance after laser cleaning is shown to be closely related to the ablation process; it can therefore be prevented by a suitable selection of irradiation parameters. The metal surfaces investigated are amalgam gilding, gold leaf gilding, and, for the first time, silver artifacts. We also describe the criteria used for applying laser ablation techniques to restoring unique masterpieces, such as Lorenzo Ghiberti’s Porta del Paradiso and Donatello’s David. Furthermore, a novel and unusual cleaning approach for archaeological silver is reported. Based on underwater laser irradiation, it provides a way to prevent oxidative effects and amplify the photomechanical coupling to the hard, thick concretions that usually accompany archaeological pieces. Finally, the experimental extension of the laser cleaning approach to wall painting and its practical use in important restoration works is presented. The practical examples reveal a significant advance in perspective for the application, which was unthinkable until recently. In sum, this Account describes novel technological and methodological contributions of laser cleaning that are having a significant impact in the field of cultural heritage conservation.

1. Introduction Despite the laser cleaning of artworks being introduced about 10 years before industrial and bioPublished on the Web 01/28/2010 www.pubs.acs.org/acr 10.1021/ar900190f © 2010 American Chemical Society

medical applications of laser ablation1 and many successful studies having been reported,2 the dissemination of this technique is proceeding relatively more slowly than the latter applications. The Vol. 43, No. 6

June 2010

739-750


739

Advances in Laser Cleaning of Artwork Siano and Salimbeni

main obstacles to a rapid spread arise from the difficulty in suitably addressing a variety of complex cleaning problems, which involve material stratifications that have variable physical and chemical properties. In principle, the optimization of the laser cleaning of artwork is much more complex than any other application of laser ablation. Furthermore, the degree of cleaning to be achieved can be univocally defined only in some cases, whereas it is often the result of intertwined multidisciplinary factors including objective material assessments, fruition expectations, intrinsic limits of the methodology, the skilfulness of the restorer, and others. During the past decade, we have provided substantial evidence that the laser pulse duration is the most crucial parameter. A careful optimization allows achieving high gradualness and selectivity in stone and metal cleaning, which is producing a significant application impact. During the 1990s, a novel fiber-coupled Nd:YAG (1064 nm) emitting pulses of 20 µs, the so-called short free running (SFR) temporal regime, was introduced.3 This pulse duration domain can provide unique gradualness and self-termination performances. As a result of its successful experimentation, two versions emitting pulses with a duration between 40 and 150 µs (1-2 J/pulse) were marketed several years ago, as alternatives to the most commonly

2. Stones Extensive applications of laser cleaning were carried out on marble and limestone.6 Up until recent years, mostly QS lasers were used for this purpose. There were essentially two limitations that provided room for proposing longer pulse duration: the occurrence of surface damage to the substrate at relatively low operative fluences, especially in cases of strong decohesion, and the recurrent yellowish appearance of the surface uncovered. The longer pulse duration allows overcoming these problems, although it reduces the efficiency. 2.1. Ablating with Different Laser Pulse Durations. The dependence of the ablation rate, zab (µm/pulse), on the laser pulse duration (tL) was systematically investigated through ablation tests carried out on laboratory samples that simulated typical black crusts on stones. The samples were prepared by applying the following mixture on sandstone substrates: 60% gypsum, 30% quartz powder, 8% carbon black, and 2% burnt sienna. Cross-section examinations showed fairly homogeneous distributions of the inert component and pigment load within the gypsum matrix. Reflectance and transmittance measurements using an integrating sphere enabled us to estimate the optical penetration

used Q-switching (QS) Nd:YAG laser emitting pulses of 5-20

at 1064 nm (δ = 27 µm), through the fitting of the transmis-

ns (0.1-1 J/pulse).

sion fluence, F(z) ) Fae-z/δ, where Fa is the absorption fluence.

More recently, a contribution for extending the application

It is much lower than the whole thickness of the crust, which

of laser cleaning to metal surfaces has been provided. The

makes the following considerations on ablation rates of gen-

state of the art on this topic was in its very early stages until

eral valence at least in stone cleaning.

the beginning of the past decade, when we started investigat-

Ablation tests were carried out using three Nd:YAG lasers

ing the conservation problems of Ghiberti’s Porta del Parad-

systems: QS, LQS, and SFR at a pulse repetition frequency of

iso.4 The study proved the need and advantages of using

2 Hz. The surface under treatment was humidified with neb-

pulse durations of several tens of nanoseconds and supported

ulized water at the same repetition frequency. Lastly, zab val-

the overall application of laser cleaning to the friezes of the

ues associated with different fluences and pulse durations

wings. A special Nd:YAG laser with a pulse duration of 70 ns

have been determined through the measurement of the depth

known as long Q-switching (LQS) was developed for perform-

(dab) produced by a number of laser pulses (np) using a contact microprofilometer (zab ) dab/np).

5

ing this important conservation treatment.

As one of the main consequent follow-ups, two years ago

The experimental ablation plots achieved are shown in Fig-

a novel, LQS Nd:YAG laser (120 ns) was marketed. Mean-

ure 1. They can be interpreted with the help of the so-called

while, the effectiveness of the SFR has been proven also for

blow-off model. This was formerly introduced for describing

gold leaf gilding. Finally, both temporal regimes began to be

vaporization-mediated ablation of homogeneous materials,

tested to clean wall paintings.

but it can easily be adapted to the present heterogeneous

Evidence of the apropriateness of the approach have

crust. The model assumes that the material removal begins

emerged from new systematic investigations and case stud-

above a characteristic threshold, Fth ) δεcr, where εcr is the

ies. In this Account, we present the most recent methodolog-

average critical energy density (J/cm3), which provides the fol-

ical advances achieved in the laser cleaning of stones, metals,

lowing scaling law: zab ) δ ln(F/Fth). The saturation of the abla-

and wall paintings, based on suitable selections of the laser

tion rate is defined by the equation zab ) δ, which occurs at

pulse duration and irradiation conditions.

F ) Fs ) eFth.

740


739-750

June 2010

Vol. 43, No. 6


FIGURE 1. Ablation rates measured and fitting using the blow-off model for QS, LQS, and SFR.

Despite the previous expression of zab being referred to a vaporization-mediated ablation of homogeneous materials, its application to fit the data achieved for the present stratification provides useful information on the basic mechanisms involved. For short pulses (QS), the logarithmic law apparently reproduces the rate increase only up to about 1 J/cm2 (Figure 1), which allows derivation of Fth ) 0.28 J/cm2, Fs ) 0.76 J/cm2, and δ ) 16.5 µm. Above 1 J/cm2, the data show a saturation behavior more pronounced than that estimated by the model. Furthermore, the latter significantly underestimate δ with respect to the measured value (δ ) 27 µm). This accelerated saturation is likely due to the occurrence of nonlinear optical absorption phenomena produced by local ionization of highly absorbing particles, which increase the typical effect of saturation due to the shielding of the expanding ablation plume.

A better agreement was achieved for the LQS and SFR. The fitting parameters of the former were Fth ) 0.56 J/cm2, Fs ) 1.51 J/cm2, and δ ) 25.8 µm. Those of the latter depended on tL: Fth ) 2.2-4.7 J/cm2 and Fs ) 5.9-12.3 J/cm2, when tL ) 50-150 µs, respectively, which showed that the linear approximation holds over a large temporal range. In all these cases, the ablation process is driven by water vaporization around absorbing particles, which are overheated because of their high opacity and act as photothermal converters. For black crusts, water occupies at least 60% of the irradiated volume, and it is certainly the first component to be partially vaporized, since the critical temperature of the others is much higher. Some indications about the amount of water vaporized is provided by the estimated critical energies (εcr ) Fth/δ): 170, 217, and 863-1844 J/cm3 for QS, LQS, and SFR, respectively. Among these, only the latter is the same order of magnitude as the critical energy of water (εw ) Cw∆T + Qw ) 2591 J/cm3, where Cw is specific heat and Qw is latent heat of vaporization), whereas the others are much lower. Ablation begins with vaporization of a variable amount of water around absorbing components. An estimation of the typical radial size of the microvolume vaporized is derivable from the thermal diffusion length, lth ) 2(DwtL)1/2 (Dw ) thermal diffusivity of water): 59 nm, 186 nm, and 5.4-9.3 µm for QS, LQS, and SFR, respectively. The comparison of operative energy densities with εcr estimated enables us to make the following considerations. SFR energy densities (103-104 J/cm3) are compatible with the vaporization of a relevant fraction of the water that imbibes the irradiated volume, which is often noticeable to the naked eye. Conversely, the corresponding amount for short pulses (QS and LQS) can be almost negligible, but a large increase in volume is associated with the phase explosion because the heating is much more localized. For QS, the local peak pressures generated in proximity of the absorbing particles is dependent on the laser pulse duration (Pmax ≈ tL-1/2) and can easily reach several hundred bars. This represents a limitation for short pulses whenever uncovering very fragile or absorbing layers to be safeguarded. The thermal conduction within the irradiated volume determines the differences between the ablation thresholds but is not the only factor. The present parametrization also shows the presence of another relevant contribution to the material removal. Both QS and LQS generate coherent pressure waves, which propagate through the irradiated volume at the speed of sound. By assuming roughly an average speed of 3000 m/s, the propagation length corresponding to 6 and 60 ns is 18 and 180 µm, respectively, which produces coherent superVol. 43, No. 6

June 2010

739-750


741


FIGURE 2. Representative stratigraphy of encrusted stones: BC, black crust; PS, pigmented scialbatura; CO, Ca-oxalates film; MS, sulfated marble substrate.

position of the point source waves and then the formation of two wavefronts propagating in inner and outer directions. The total reflection of the latter at the sample-air interface produces a rarefaction wave. This stretches the material, thus taking part in the material removal. This contribution to the ablation is called spallation. For the present samples, it is more pronounced for QS laser around Fth, according with the relatively low value of the latter and the high slope of the curve, but more generally the estimation of the pressure transient wavelength suggests spallation could became relevant also for LQS laser. 2.2. Gradualness and Appearance. The highest efficiency was achieved when using QS at low fluences (e1 J/cm2), but it rapidly saturates because of the nonlinear absorption and shielding. The optical linearity of LQS makes it more efficient than QS above 1.5 J/cm2. Conversely, the ablation rates of SFR were decidedly lower, though the extension of the linear ablation produces higher ablation depths. The efficiency can be exploited for increasing the productivity, but in several cases, a low rate would be preferable with respect to single-shot deep ablation, which can be uncontrollable. The main materials composing natural black crusts due to urban pollution are gypsum, quartz, black carbon, iron oxides, oily residues, other earthy materials, and minor components. However, cases of pure black crusts are rare. Art objects have undergone several treatments over the centuries, including the application of coats, such as organic binder-patinations, whitewashes, pigmented scialbaturas, and others, which nowadays must be removed for conservation and fruition issues. Figure 2 displays a representative example of multiple layers on stone, but obviously many different situations can be encountered. 742


739-750

June 2010

Vol. 43, No. 6

FIGURE 3. Comparative tests using QS and SFR: differences between the operative ranges. The detail shows the occurrence of visible damage induced by QS at relatively low fluence.

The strong inhomogeneity of the optical properties along the cross section of the irradiated stratification has a crucial importance in laser cleaning. In the example of Figure 2, there is an increasing diffusivity moving from the outer black crust to the inner Ca-oxalates film, and then to the sulfated stone underneath. The fluence needed to ablate the inner Ca-oxalates layer may be significantly higher than that of the black crust also due to the significant ochre content, which implies a fundamental difference between the QS and longer pulse lasers. As already described, the fluence increase of the former is limited by nonlinearity and the consequent photomechanical side effects. Conversely, the larger operative domains of LQS and especially SFR can enable a more precise control of the final degree of cleaning. This is shown by the comparative cleaning tests of Figure 3, which were carried out on a stone fragment from the façade of Florence’s Cathedral whose stratigraphy is that of Figure 2. The higher controllability associated with longer pulses concerns the final phase of the cleaning, whereas the gradualness of the removal of the different temporal regimens through the stratification strongly depends on composition and microstructure of the strata. Most problems of yellow-orange appearances after the laser cleaning of marbles are due to residues of iron oxides, Ca-oxalate films with pigment loads, organic substances, and other pigmented components of the stratification. The attribu-


tion of yellow appearance to residues of materials coming from the stratification has been pointed out also by other authors.7 However, deeper degrees of cleaning using higher fluences typically provide lighter appearances and colder color hues. Often, the usual operative fluences of QS lasers (0.1-1 J/cm2) are insufficient for removing the last pigmented film in proximity to the whitish marble substrate, whereas this can be achieved at the operative fluences of longer pulses (2-8 J/cm2). The damage thresholds of LQS and SFR are significantly higher than that of QS. Thus for example, for aged white marbles, we have typically measured 3-4, 20-30, and 1-1.5 J/cm2, respectively. In addition to pigmented residues, the case of yellow hues well beneath the stone surface are also encountered. This condition is usually observed for gypsum, stucco, plaster, and porous stone treated in the past with oily substances. Comparisons between different lasers and mechanical cleaning have shown very similar chromatic results. The final appearance does not depend on the cleaning method in such cases. To complete the picture, we also investigated the possible occurrence of staining associated with laser ablation. Cleaning tests on samples prepared by applying pure black carbon or black crust scraped from genuine marble artifacts were carried out. For all three lasers used, the ablation of black carbon in dry conditions produced grayish staining, which was more pronounced at the lowest fluences and gradually disappeared at the higher fluences. No relevant staining effects were observed in water-assisted conditions, apart from very light grayish discoloration at low fluences. Similar effects were noted for the genuine crust, with the difference that the color of the stains for QS and LQS was pale yellow rather than grayish, as for SFR. However, fluence increase and water assists effectively prevented any discoloration effect. 2.3. Stone Case Study. The Architrave of San Ranieri is a bass relief of Roman origins (reused and integrated during the Late Middle Ages) that decorates the north portal of the Cathedral of San Ranieri in Pisa. The entire cleaning of the carved parts (about 6 m2) was carried out using laser ablation. The stratification was similar to that reported in Figure 2, whereas the present substrates are Greek marble (main central relief) and local marble of San Giuliano (top and bottom integrations). The state of conservation of the artwork was very serious. The hard stratification to be removed lay on sulfated marble substrate presenting deep decohesion effects. This difficult situation, along with some preliminary tests, led us to select the laser technique and to optimize an overall treatment using the SFR. The operative fluences were relatively high (up to 7-9

FIGURE 4. Architrave of San Ranieri, Pisa. Examples of selectivity and gradualness of the SFR: (0) before cleaning; (1) early degree of laser cleaning; (2) final degree of laser cleaning. The detail (bottom) shows the safeguard of the marble surface even in cases of strong decohesion.

J/cm2, 80 µs) due to the high reflectance and hardness of the scialbaturas and Ca-oxalates layers. The QS was found to be too invasive for the complete removal of the sulfated surface coats. Details of the relief during the laser cleaning are shown in Figure 4. The early degree of cleaning (1), which was achieved with fluences around 5 J/cm2, was considered unsatisfactory, since a significant amount of sulfated residues of the stratification was still present. The laser was therefore set at a higher fluence (about 8 J/cm2), in order to achieve a deeper cleaning. Stratigraphic diagnostics and colorimetric controls allowed assessment of a deeper removal and the safeguard of the most internal Ca-oxalate film (whewellite). Vol. 43, No. 6

June 2010

739-750


743


A particular situation was found on the genuine Roman part, where the SFR was not able to remove some very dark spots that were identified as residues of biological attacks. The associated ramified and tenacious microstructures did not permit a sufficient degree of removal by means of the slow vaporization-mediated ablation of the SFR. The problem was solved using the most intensive photomechanical coupling of the QS.

3. Metals As already mentioned, the milestone in metal cleaning using laser ablation was the study of Ghiberti’s Porta del Paradiso,4 the restoration of which will be completed at the end of this year (Figure 5). Here, we summarize the general physical criteria and extend the laser approach to other metal surfaces. 3.1. “Semi-infinite” and “Negligible” Metal Thickness. The high reflectance of metals (Rm) is associated with an extremely high absorption coefficient. The light is dissipated in a nanometer-scale depth, which makes the thermal conduction and hence pulse duration of critical importance. For this reason, it is appropriate to approach the cleaning of metals by starting from the thermal features of laser-metal interaction. The one-dimensional thermal conduction theory provides estimations of the temperature rise ∆T(z,t) in a metal slab with a thickness l. By consideration of the conditions in which the thermal conduction of the adjacent materials is negligible, the temperature rise produced by the absorption intensity Ia(t) ) (1 - Rm)IL(t), where IL(t) is the incident laser intensity, is the following:8

∆T(z, t) ) ∞

1 Km

t Ia(t

∑ ∫0

n)1

Dm π

- τ)

√τ

{

∫0t Ia(t - τ) e

[e-(2nl - z)

2⁄(4D

-z2⁄(4Dmτ)

mτ)

√τ

dτ +

+ e-(2nl + z)

2⁄(4D τ) m

}

] dτ (1)

where Km and Dm are the thermal conductivity and diffusivity of the metal, respectively. Dm ) Km/(FmCm), where Fm and Cm are density and specific heat of the metal, respectively. The first term in eq 1 represents the solution to the semiinfinite medium, while the sum accounts for the reflection of the thermal wave at the interfaces. If lth < l, the first term represents a good approximation, whereas the second increases for lth > l, that is, tL > l2/(4Dm), until it produces a homogenization of the temperature inside the slab when lth . l. Conversely, if the heat generated by the irradiation of the metal film is massively transferred to the underlying low conductivity (Ki) substrate, temperature estimations can be 744


739-750

June 2010

Vol. 43, No. 6

achieved by considering the surface-layer approximation:

∆T(z, t) )

[

Ia 2b bKi

Dit -η2 e - (1 + bz)erfc η + π

]

eb(z+Dibt) erfc(η + √Dit) (2) where η ) z/[2(Dit)1/2], b ) riCi/Cm1, and Cm1 is the thermal capacity per unit area of the gold film. In practical applications, eq 1 can be used to estimate the temperature rise in metal films with thickness above some micrometers, such as for amalgam gilding, gold and silver inlays, and silver objects. In contrast, eq 2 is suitable for gold leaf gilding, of which the typical thickness is 100-500 nm. These considerations derive from the estimation of lth. Thus, for example, if we assume a gold film (Km ) 286 W/(m °C)), Dm ) 1.23 cm2/s) with an average thickness of about 6 µm, as is that of the Porta del Paradiso, lth ) 1.7, 5.9, and 140 µm for QS (6 ns), LQS (70 ns), and SFR (40 µs), respectively. For both the QS and LQS, the metal film behaves as a semi-infinite medium since the reflection at the internal interface is negligible. Conversely, for SFR laser heating, the thermal wave undergoes many reflections, which produce an almost constant temperature along the thickness. The assumption of negligible conduction outside the film aims to include the most delicate cases where the film is detached from the substrate. Conversely, disregarding the conduction of the layer beneath the gold leaf is too unrealistic since lthi . l in any case. Equation 1 shows that the temperature of a metal significantly increases when the laser pulse duration decreases: ∆T(0,tL) ≈ tL-1/2, as is also evident from the decrease of lth. Thus, the cleaning of low-melting point metals with a QS is very harmful. At the same time, eq 1 indicates significant temperature rises at the typical operative fluences of the SFR due to the contribution of the reflection term. Thus, for example, by assuming a constant Fa ) 150 mJ/cm2, eq 1 estimates a peak temperature of about 450 and 170 °C for the QS and the LQS, respectively. Instead a realistic absorption fluence for the SFR is around 1 J/cm2, which would involve a homogeneous temperature rise in the film up to 240 °C lasting for a longer time. These are the basic reasons that led us to propose LQS for cleaning amalgam gilding, gold and silver inlays or laminas, and silver alloy objects. A general optimization rule for metal films with a thickness on the order of micrometers is provided by the condition l ) lth (tL ) l2/(4Dm)). By using operative fluences of 0.5-1 J/cm2, one can achieve safe laser ablation without any relevant risk of micromelting.


FIGURE 6. Atomic force microscopy of amalgam gilding (a) before treatment (10 × 10 µm2) and (b) after ablation of malachite in dry conditions (3 × 3 µm2).

tion is so effective to produce satisfactory degrees of cleaning only in rare cases. 3.2. Ablation Dynamics and Degree of Cleaning of Metals. Let us focus on “thick” gilding, silver, and other cases FIGURE 5. Lorenzo Ghiberti’s Porta del Paradiso: a tondo before and after LQS laser cleaning.

Conversely, the thermal analysis (eq 2) for gold leaf suggested a limited application perspective for the QS and LQS (though not ruled out), because the presence of the gold layer produces a strong concentration of the laser heating within the leaf and immediately underneath, which could easily destroy the gilding. The SFR was found to be more suitable in this case. It can also be used on thick metals; however the abla-

in which l > lth, whereas laser ablation involved in the cleaning of gold leaf using SFR is mainly driven by slow vaporization, as described above. In metal cleaning, water assists are even more important than for stones, since gold and silver surfaces can undergo serious staining effects. Figure 6 shows nanoscale structures following laser ablation in dry conditions of malachite on a synthetic amalgam gilding sample. They were generated by redeposition of ablation products. For silver also the direct oxiVol. 43, No. 6

June 2010

739-750


745


dation associated with temperature gradients of some hundreds of degrees can occur. Laser cleaning using the LQS can be described in accordance with the inhomogeneous vaporization and spallation dynamics mentioned above with the addition of a further contribution provided by the transient heating of the metal surface. The latter easily exceeds the vaporization threshold of water, thus generating expansion and a pressure wave at the bottom of the irradiated volume, which can significantly amplify the efficiency of the material spallation. Especially in the present case of relatively low operative fluences, the thickness of the water-assists layer is very important. In principle, water should simply imbibe the material to be removed, since thicker layers reduce the intensity of the rarefaction wave and increase the acoustic impedance against material ejection. On the other hand, the case of underwater irradiation can also have certain practical applications in metal cleaning. Underwater irradiation of an absorbing stratification generates an intensive microexplosion that releases a strong recoil stress against the target. Above a given threshold, the latter can mechanically fragment or erode the stratification. Furthermore, as is well-known, an intense underwater explosion also generates the expansion of a cavitation bubble, the subsequent collapse of which produces a water jet, which releases a second mechanical stress against the target. At sufficiently high fluences, a sequence of two to three cavitation cycles could be triggered by the single laser pulse. The peak pressure associated with the early microexplosion is closely dependent on the pulse duration. As for black particles in gypsum crusts, Pmax ≈ tL-1/2, whereas the water hammer pressure of the collapse depends on the speed of the water jet (vj): Pwh = Fwcwvj, where Fw is the density of water and cw is its sound speed. Typical collapse speeds vj ) 10-100 m/s produce a water hammer pressure Pwh = 750-1500 bar, which certainly contributes to the material removal. 3.3. Metal Case Studies. To provide some example applications, we selected the recent restoration work of a small silver relief (15 × 25 cm2) by Guglielmo della Porta (crafted around 1550), the Roman silver coin Treasure from Rimigliano (Livorno), and Donatello’s David. The former represents the very general problem of tarnished silver, in which the thin sulfur-rich layer is intimately bound to the metal, which makes this case more difficult than the removal of deposits or varnishes. For the restoration of the coin Treasure, we applied underwater laser cleaning for the first time. Lastly, 746


739-750

June 2010

Vol. 43, No. 6

FIGURE 7. Detail (10 cm vertical size) of a silver relief by Guglielmo della Porta (XVIth century) during laser cleaning. Reproduced by permission of Nardini Editore, Florence.

the David represents the most recent conservation problem of oil gilding solved using laser ablation. As for gilded bronzes, the irradiation of tarnished silver in dry conditions produces unacceptable discoloration effects. For this reason, cleaning tests using the LQS (0.5-1 J/cm2) were carried out under water-assisted conditions, with the addition of air, nitrogen, or argon flux. None of them enabled us to completely solve the problem of redeposition, even though argon assists minimized such a side effect. However, it was observed that a minimal mechanical action with wet cotton or other material easily removes the redeposition film, which makes gas assists redundant. However, a finishing of this type must be performed soon after the laser treatment, since the residues rapidly react with the substrate in a way that makes the removal very difficult after a few hours. Laser cleaning of tarnished silver is quicker than traditional methods, including preliminary treatments with acetone and distilled water and then chemical and mechanical actions using sodium carbonate solutions. The LQS allows self-terminated ablation within a sufficiently wide operative range, before the occurrence of direct oxidation and micromelting. The final appearance (Figure 7) is similar to that provided by traditional methods with the difference that the strong abrasion involved in the latter is more invasive. The Treasure from Rimigliano was an agglomerate of about 3500 silver-alloy Roman coins (230-260 AC). The archeologists decided to detach about 300 coins from the block in order to clean them for a thorough classification. Furthermore, they decided to clean the coins at the top of the agglomerate for exhibition purposes.


FIGURE 9. Time evolution of the cavitation bubble (LQS, 120 ns, 1.5 J/cm2): radius measured using time-resolved laser shadowgraph (images at three different delays are also shown).

FIGURE 8. The Treasure from Rimigliano: detail (top) and general view (bottom) during underwater laser cleaning using the LQS and the SFR.

After assessing that it was practically impossible to remove safely the hard and thick mineral stratifications using laser irradiation in air, we decided to test underwater cleaning. The result were better than expected. Both the LQS and SFR provided very satisfactory results without any alteration of the natural surface texture and, more importantly, much better than those achieved with traditional mechanical cleaning using a piezoelectric ablator. Figure 8 shows the first cleaning test (top) and a general view of the Treasure after being partially cleaned (bottom). Figure 9 reports the cavitation cycles that were imaged in operative conditions (LQS, 120 ns, 1.5 J/cm) using a timeresolved laser shadowgraph setup. This confirmed the previous description of underwater cleaning. Furthermore, pressure measurements using a needle hydrophone were conducted, which showed that both lasers induced phase explosion and water hammer pressure peaks of comparable amplitude. For the LQS, the former was more intense than the latter, whereas the situation was inverted with the SFR. In both cases, the ablation process was self-terminated at the reflecting metal surface.

The restoration work recently performed on Donatello’s David was aimed at removing a brown coat applied in the past. Thorough XRD, FT-IR, and gas-chromatography analyses enabled us to assess that this coat was composed of a mixture of quartz, gypsum calcite, Ca- and Cu-oxalates, Alsilicates, a pigment load of Fe-oxides, and carbon black in organic binder (linseed oil and colophony). The thickness was quite variable up to 100 µm. Similar intentional layers have been found in previous restoration work on other Florentine artworks. The coat strongly altered the fruition and was considered unsafe for the future conservation of this unique masterpiece. Furthermore, it completely covered the original oil gilding decorations (Figure 10). In accordance with basic concepts already discussed, the cleaning of the latter was carried out using the SFR laser at maximal irradiation fluences of around 2 J/cm2, which were sufficient to produce thermal alteration and a safe removal of the organic binder-patination. This allowed recovery of the relicts of the original gilding, which were abundant on the hair (Figure 10) and minimal in the lower part (Goliath’s beard). Furthermore, laser irradiation was also used in order to effect a preliminary disaggregation of the patination in nongilded zones before sodium carbonate poultices and then mechanical removal.

4. Wall Paintings The cleaning of painted surfaces represents a big challenge for laser cleaning. Studies were conducted on easel paintings for which excimer (mainly KrF*, 248 nm) and free running Er:YAG (2.94 µm) lasers have been proposed. Most of the materials, in particular varnishes, strongly absorb at the wavelengths of these two lasers (δ e 1 µm), which can therefore be used for lightening varnishes darkened by aging. However, 248 nm and 2.94 µm are also strongly absorbed by paint layers and Vol. 43, No. 6

June 2010

739-750


747


FIGURE 10. Donatello’s David: SEM of oil gilding coated by patination applied in the past (top) and a view of the hair after laser cleaning (bottom). The former is reproduced by permission of Giunti Editore, Florence.

hence require a careful control of the irradiation fluence and ablation depth. This is a different approach with respect to the idea of developing self-terminated processes. A systematic laboratory investigation was carried out on frescoed samples in order to assess the potential of QS (6 ns), LQS (120 ns), and SFR (40-50 µs) for the solution of specific cleaning problems. Several pigments were used, including red and yellow ochre, raw sienna, green earth, azurite, ultramarine and Egyptian blue, malachite, verdigris, and zinc white. We determined the damage thresholds of all these pigments for the three pulse durations available; various clean748


739-750

June 2010

Vol. 43, No. 6

ing problems were then simulated by applying a layer of black carbon, white and dark limewashes, or dark Paraloid to the paint samples. Moreover, we have learned from practical experience that limewashes were often applied on existing dark deposits. Thus, samples with whitewash on black carbon were realized in order to simulate a real situation of this type. None of the pigment investigated underwent discoloration in water-assisted conditions. The damage was essentially represented by partial or total ablation of the paint layer. This is very important, since prepared paint layers were weaker than genuine ones because of the weak carbonatization. Thus, the damage thresholds are expected to be higher in practical cases. As for stones and metals, also for wall paintings, we found that the LQS and SFR provide larger operative ranges and more discrimination potential than the QS. The damage thresholds were fairly high (between 0.7-0.9 J/cm2 and 3.5-5 J/cm2 for LQS and SFR, respectively) and therefore permit the safe removal of black carbon, dark limewash, and dark Paraloid. The removal of pure whitewash is possible only in a few cases. Conversely, whitewash on black carbon represents the best situation. A very effective ablation of the stratification was achieve with LQS at only 0.2 J/cm2. The SFR irradiation provided the selective removal of whitewash at 2 J/cm2, while the black layer underneath was ablated at a higher fluence (3 J/cm2). The process involved in the removal of whitewash, which is a not absorbing layer, is always a spallation that is induce by heating the absorbing layer underneath. It can be referred to as “secondary spallation” in order to distinguish it from the primary spallation introduced above. Secondary spallation is not a laboratory exercise. We are finding that it represents a very effective process, which plays a fundamental role in practical cases of limewash removal, as well as in other situations such as those of the following cases. 4.1. Wall Painting Case Studies. Laser cleaning has been successfully tested and consequently extensively applied to uncover the wall paintings of the Sagrestia Vecchia (1446-1449) and the Cappella del Manto (1370) in Santa Maria della Scala, Siena, those of the donjon of the Castle of Quart (Late Middle Age), in Aosta Valley. The stratifications to be removed were aged Paraloid with residues of a previous limewash for the Sagrestia Vecchia, whitewash for the Cappella del Manto, and limewash with the addition of animal glue for the paintings of the Castle of Quart. In all these cases, we found a black layer of carbon deposits (Figure 11) under


FIGURE 13. Cappella del Manto, Siena: two-step laser cleaning treatment. FIGURE 11. Stratigraphies of the wall paintings in the Sagrestia Vecchia (top) and the Cappella del Manto (bottom). From top to bottom: limewash (with Paraloid for the former), carbon deposits, iron-based paint layers.

FIGURE 14. Donjon of the Castle of Quart: SFR laser cleaning on red and black pigments. FIGURE 12. Sagrestia Vecchia, Siena: comparison between traditional cleaning (two angels on the left) and LQS cleaning (angel on the right partially cleaned).

the whitish layers to be removed, which favored ablation by secondary spallation. The cleaning results achieved are surprisingly good; indeed they are better than those of any other alternative technique tested (Figure 12-14). The dark stratification on the face of the angel in Figure 12 (Sagrestia Vecchia) was removed using the LQS at 0.7 J/cm2. The ablation process self-terminated, thanks to the high reflectance of the iron pigments and calcite of the paint layers uncovered. The early damages were observed at 1.5 J/cm2. As expected, this value is somewhat higher than those of the laboratory samples. Two different cleaning levels are shown in Figure 13 (Cappella del Manto). They were achieved by using the SFR (bot-

tom) and the SFR followed by LQS (top), respectively. A residual veil of the ancient carbon deposits was visible after the former treatment, which was completely removed with a further irradiation with LQS. The cleaning of the Gothic Linear Style paintings of the Castle of Quart was also approached by a combined use of the LQS and SFR. The former was effective on lead white-rich paint layers, whereas SFR was used on ochre and black carbon overpaint (Figure 14). This result is either really impressive or a kind of paradox, since laser ablation is typically used for removing “black” from “white”. The SFR allows removal of a whitish layer (lime scialbatura) from a black pigment.

5. Conclusions In this work, we have shown that a suitable selection of Nd:YAG laser pulse duration and irradiation conditions can Vol. 43, No. 6

June 2010

739-750


749


significantly extend the potential of laser cleaning. A set of successful results have been presented for stones, metals, and wall paintings that evidence the achievement of improved degrees of selectivity and control of the ablation processes. The methodological criteria introduced for the different typology of artifacts are intended as starting points for optimizing the laser cleaning procedures rather than as general recipes. In practical applications, preliminary tests should always be carried out in order to assess the phenomenology and therefore adapt what is stated here to the specific problem under study, taking into account all its unique features. Furthermore, since laser cleaning is very rarely applied as stand alone, also the combination with chemical and mechanical techniques should be suitably pondered in order to minimize the invasiveness and improve the result of the cleaning treatments.

BIOGRAPHICAL INFORMATION Salvatore Siano graduated in physics from the University of Florence. He leads a research group at the Applied Physics InstituteCNR of Florence that works on the development and application of laser and optoelectronic techniques for the study and conservation of cultural heritage. He is the person in charge of several research projects on the topic and is involved in the application

750


739-750

June 2010

Vol. 43, No. 6

of the laser techniques to important restoration works of famous masterpieces. Renzo Salimbeni graduated in physics from the University of Florence and is Director of the Applied Physics Institute-CNR of Florence. His scientific activity has been dedicated to laser technologies and their application in various fields. He has been in charge of several research projects devoted to the conservation of the cultural heritage. FOOTNOTES * E-mail: [email protected]. REFERENCES 1 Asmus, J. F.; Murphy, C. G.; Munk, W. H. Studies on the interaction of laser radiation with art artifacts. In Developments in laser Technology II; Weurker, R., Ed.; SPIE: Bellingham, WA, 1973; Vol. 41, pp 19-30. 2 Georgiou, S.; Anglos, D.; Fotakis, C. Photons in the service of our past: lasers in the preservation of cultural heritage. Contemp. Phys. 2008, 49, 1–27. 3 Margheri, F.; Modi, S.; Masotti, L.; Mazzinghi, P.; Pini, R.; Siano, S.; Salimbeni, R. Smart Clean: a new laser system with improved emission characteristics and transmission through long optical fibres. J. Cult. Heritage 2000, 1, S119-S123. 4 Siano, S.; Salimbeni, R. The gate of Paradise: Physical optimization of the laser clearing approach. Stud. Conserv. 2001, 46, 269–281. 5 Salimbeni, R.; Pini, R.; Siano, S. A variable pulse width Nd:YAG lasser for conservation. J. Cult. Heritage 2000, 4, 72s–76s. 6 Bromblet, P.; Laboure´, M.; Orial, G. Diversity of the cleaning procedures including laser for the restoration of carved portals in France over the last 10 years. J. Cult. Heritage 2003, 4, 17s–26s. 7 Pouli, P.; Fotakis, C.; Hermosin, B.; Saiz-Jimenez, C.; Domingo, C.; Oujja, M.; Castillejo, M. The laser-induced discoloration of stonework: A comparative study on its origins and remedies. Spectrochim, Acta, Part A 2008, 71, 932–945. 8 Siano, S.; Grazzi, F.; Parfenov, V. A. Laser cleaning of gilded bronze surfaces. J. Opt. Techn. 2008, 75, 419–427.

New Frontiers in Materials Science for Art Conservation: Responsive Gels and Beyond EMILIANO CARRETTI,† MASSIMO BONINI,† LUIGI DEI,† BARBARA H. BERRIE,‡ LORA V. ANGELOVA,§ PIERO BAGLIONI,*,† AND RICHARD G. WEISS*,§ †

Department of Chemistry & CSGI Consortium, University of Florence, via della Lastruccia 3, I-50019 Sesto Fiorentino (Florence), Italy, ‡Conservation Division, National Gallery of Art, Washington, D.C. 20565, §Department of Chemistry, Georgetown University, Washington, D.C. 20057-1227 RECEIVED ON NOVEMBER 30, 2009

CON SPECTUS

T

he works of art and artifacts that constitute our cultural heritage are subject to deterioration, both from internal and from external factors. Surfaces that interact with the environment are the most prone to aging and decay; accordingly, soiling is a prime factor in the degradation of surfaces and the attendant disfigurement of a piece. Coatings that were originally intended to protect or contribute aesthetically to an artwork should be removed if they begin to have a destructive impact on its appearance or surface chemistry. Since the mid-19th century, organic solvents have been the method of choice for cleaning painted surfaces and removing degraded coatings. Care must be taken to choose a solvent mixture that minimizes swelling of or leaching from the original paint films, which would damage and compromise the physical integrity of all the layers of paint. The use of gels and poultices, first advocated in the 1980s, helps by localizing the solvent and, in some cases, by reducing solvent permeation into underlying paint layers. Unfortunately, it is not always easy to remove gels and their residues from a paint surface. In this Account, we address the removal problem by examining the properties of three classes of innovative gels for use on artwork-rheoreversible gels, magnetic gels, and “peelable” gels. Their rheological properties and efficacies for treating the surfaces of works have been studied, demonstrating uniquely useful characteristics in each class: (1) Rheoreversible gels become free-flowing on application of a chemical or thermal “switch”. For art conservation, a chemical trigger is preferred. Stable gels formed by bubbling CO2 through solutions of polyallylamine or polyethylenimines (thereby producing ammonium carbamates, which act as chain cross-links) can be prepared with a wide range of solvent mixtures. After solubilization of varnish and dirt, addition of a weak acid (mineral or organic) displaces the CO2, and the resulting free-flowing liquid can be removed gently. (2) Incorporation of magnetic, coated-ferrite nanoparticles into polyacrylamide gels adds functionality to a versatile system comprising oil-in-water microemulsions, aqueous micellar solutions, or xerogels that act as sponges. The ferrite particles allow the use of magnets both to place the gels precisely on a surface and to lift them from it after cleaning. (3) Novel formulations of poly(vinyl alcohol)-borate gels, which accept a range of organic cosolvents, show promise for swelling and dissolving organic coatings. This family of gels can be quite stiff but can be spread. They are non-sticky and have sufficient strength to be removed by peeling or lifting them from a sensitive surface. These three classes of gels are potentially very important soft materials to augment and improve the range of options available for conserving cultural heritage, and their interesting chemical-physical properties open a rich area for future scientific investigation.

Published on the Web 04/13/2010 www.pubs.acs.org/acr 10.1021/ar900282h © 2010 American Chemical Society

Vol. 43, No. 6

June 2010

751-760


751

Frontiers in Materials Science for Art Conservation Carretti et al.

1. Introduction

appropriate solvent system to solubilize the substance(s) to be

Pictorial surfaces can be viewed as solid/gas interfaces where the solid phase is intrinsically microheterogeneous along the surface and in the dimension orthogonal to it. Both easel and mural paintings consist of a macroscopic support onto which a thin layer of a paint material has been applied. This pictorial layer, with or without a varnish coating, is the preferential site of degradation because it is where the artwork exchanges matter and energy with the surrounding environment (i.e., the atmosphere).1 This Account highlights some new approaches to cleaning easel paintings using gels. 1.1. A Scientific Approach to Cleaning Artwork. A varnish coating can even the appearance of a paint surface, provide saturation of color, add gloss over a picture surface, or increase contrast in matte and glossy areas. Coatings also provide barriers to dirt and (if formulated appropriately) ultraviolet radiation; protective varnishes may be considered sacrificial. Varnishes are water-based (such as egg white,2 gum, or glue) or solvent-based (such as “spirit varnishes” from tree resins or oil-resin mixtures3 or synthetic organic polymers4). They deteriorate as a result of cross-linking and chain scissions, becoming brittle, yellow, and (sometimes) increasingly insoluble at rates that depend on their chemical compositions and ambient conditions.5 Thus, paint surfaces can become disfigured by deteriorated coatings and other deposits such as old consolidants and overpaint. Ideal cleaning agents selectively remove deposits, additions, and deteriorated varnish without affecting underlying paint layers and are removed completely. In practice, ideal cleaning is unattainable. The approach to ideality depends on the materials to be removed, the layers beneath, and (of course) the cleaning agent: the more similar are the properties of the materials to be removed and those to remain (the paint), the greater the challenge. Solvent cleaning has limitations: How to remove varnishes without damaging oil-based paint films, which are sometimes compounded with resins to make “meguilps”?6 Unfortunately, even contemporary materials and methods, cleaning gels, enzyme-based solutions,7 and laser ablation,8 present unanswered issues regarding their use. If materials to be removed are identified, the solvent blend that maximizes selectivity of the cleaning action can be chosen from Teas plots.9 These diagrams plot the bulk properties of the solvents as a function of the intensity of their intermolecular, noncovalent interactions,10 hydrogen bonding, dispersion forces, and dipolar interactions. Using this or related tools, the conservator can select what should be an 752


751-760

June 2010

Vol. 43, No. 6

removed while avoiding swelling or solubilization of the underlying paint layer.11 Removal of the polymeric substances involves the following steps: (1) penetration of solvent orthogonal to the exterior layer (at ∼10 µm/s through a polymer network12); (2) swelling of the polymer layer in contact with the solvent; (3) transformation of the polymer layer into a gel-like fluid; (4) solubilization or softening of the polymer. Concurrently, solvent can penetrate the paint layer, swelling it and altering interactions between pigments and their binders (usually an oil or protein-based material).13 Consequently, the paint layers can be softened, embrittled, or leached and the pigment-medium binding can be disrupted. Additionally, the increase of volume of the paint layer (a manifestation of swelling) and subsequent decrease in volume (after solvent evaporation or leaching) can induce mechanical stresses that lead to long-term damage such as microfractures, alteration of the surface morphology with local whitening effects, and decreased structural stability of the paint layer.14,15 For these reasons, total control of the cleaning action is extremely difficult. Michalski modeled swelling in terms of Hansen parameters16 using a 3D representation that includes a detailed distinction among different solvent properties to identify the one most effective in solubilizing a substance.12 Leaching of small binder molecules, induced when solvent swells a paint layer, also changes the stiffness of surface materials.15 Several innovative approaches that combine the properties of solvents and soft matter17 will be discussed in section 3.

2. Gels for Cleaning Works of Art Cleaning artwork with gels and poultices has increased enormously during the last decades. Aqueous, nonaqueous, and mixed gels have been devised to remove varnish and overpaint from paint surfaces18 and to remove stains from stone19 and stains and adhesives from paper.20 Sequestration of solvents in gel matrices minimizes the deleterious effects of using liquids for cleaning paint surfaces and introduces several advantages: (1) Slow release of the active solvent across the gel/coating interface reduces the risk of swelling of the paint layers for a broad range of cleaning-action rates. (2) The high viscosity of the gel reduces bulk diffusion of solubilized molecules within the gel liquid and, consequently, slows the kinetics of solubilization. Thus, in many cases,

Frontiers in Materials Science for Art Conservation Carretti et al.

gelled solvents work more slowly and are more easily controlled on a paint surface than neat liquids. (3) The varied nature of materials to be removed, especially old varnishes, requires gels with solvents of different polarities. “Solvent gels”18 and gels with poly(vinyl alcohol)borax formulations (PVA-B) as the gellant21 satisfy this exigency. (4) The gel-based cleaning technique accommodates different solvents22 and highly selective cleaning agents such as enzymes,23 chelating molecules, and microemulsions.24–26 Because of their diversity, gels have been used to clean frescoes,27 glasses,28 metals,29 and even feathers30 besides paint layers on wood or canvas. (5) The surface area exposed to a gelled solvent corresponds much more closely to the area of application; gelled solvents spread much less than neat liquids. Solvent gels developed by Wolbers have been the most frequently used to remove varnishes.18 They employ partially deprotonated (carboxylate/carboxylic acid forms) poly(acrylic acid)s (PAcAs) as the gellants. Gels with polar solvents or aqueous liquids have been made also with a cellulose derivative23 or agar agar as the gellant. In the presence of a weak base, usually a nonionic surfactant (Ethomeen C12 or C25), PAcA chains adopt extended conformations due to electrostatic repulsion by the negatively charged carboxylate groups. Chain entanglement and other interchain interactions then establish the 3D networks that are necessary to immobilize the liquid components on a macroscopic scale; at the micro and smaller distance scales, the vast majority of liquid molecules diffuse at rates comparable to those in the neat bulk.31 Solvent gels are applied onto a paint surface directly (or using a spatula or other instrument), sometimes followed by mechanical action with a swab to increase gel-paint surface contact. Transfer of the liquid component at the gel/painted interface is controlled by thermodynamics (the chemical potential of the liquid in the two layers) and kinetics (the ease with which the liquid crosses the interface). A serious potential problem with the use of these gels as cleaning agents, even though they are often very effective, has been ensuring their complete removal (after the cleaning action) in a manner that does not damage the paint surface.32 Solvents applied neat may have other problems as noted, but they can be removed with small mechanical force, and residual solvent evaporates with time. The viscoelasticity of gels requires application of some type of force to remove them from a paint surface, and some of their gellant components may stay on the surface or in cracks and fissures after removal of the bulk. Burnstock and Kieslish33 demon-

strated by GC-MS and SEM analyses that residues of Ethomeen from PAcA-based gels were present after a gel was removed mechanically by a swab roll and even after applying clearing solvents.

3. Responsive Gels for Cleaning Works of Art Gels that react or respond to an external stimulus (especially pH, temperature, and magnetic fields discussed here) increase the options available to conservators. 3.1. Rheoreversible Gels. Our efforts to clean and remove degraded varnishes from surfaces of artwork using gels began with the development of gels with low molecularmass organogelators (LMOGs)31,34–36 that can be transformed reversibly and without heating into free-flowing liquids. Carbon dioxide gas is passed through a solution of a solvent and a low concentration (1000, >1d >1000, >20d (B) Infrared Tetrahertz Radiation-Based Methods absorption absorbing species 2D absorption absorbing species 3D optical density interfaces 3Da

tube/SR tube/SR tube/SR

21-23 22, 23 12, 21, 23 30-32 26-29 12 tube/SR

X-ray SR X-ray X-ray X-ray SR >0.1b >0.1b >0.1b 250-1000 >10c >0.3b 2D 2D 2D 2D 3Da 3Da (A2) Scanned Beam Methods elemental composition species composition phase composition elemental composition elemental composition element/phase composition

2D 3D 3D 3D 3D (A) X-ray-Based Methods

(A1) Full-Field Imaging Methods (electron) density (electron) density (electron) density interfaces (electron) density absorption absorption absorption refraction absorption

elemental chemical state crystal structure elemental elemental elemental/crystal strusture

>500 >600 0.3-10 0.3-10 >1

resolution (µm) dimensionality information imaged contrast type acronym imaging method

TABLE 1. Overview of Imaging Methods Suitable for Nondestructive Subsurface Examination of Paintings and Related Works of Art

photon source

X-ray tube/SR X-ray tube X-ray tube/SR SR SR

refs

4 5-7 8-10 10, 13-15, 18, 19 18-20

Subsurface Analysis of Cultural Heritage Artifacts Janssens et al.

a canvas or panel.2 Many painterly effects critically depend on the layer buildup; e.g., the translucent shine of colorful tissues, the suggestion of shadow in fleshtones, or the illusion of an object’s texture may be realized by deliberately including the optical contribution of lower layers. Knowledge about the stratigraphy of a painting often is highly relevant in conservation when stability problems such as paint discoloration or delamination are considered. Thus, the study of a painting’s stratigraphy is a research theme common to curators, conservators, and conservation scientists. Traditionally, the visualization of the inner structure of painted cultural heritage (CH) artifacts relies on penetrative, two-dimensional imaging techniques such as infrared reflectography (IRR) and X-ray radiography (XRR), optionally complemented with microscopic analysis of cross-sectioned samples.3 However, there are significant limitations to this approach. (a) The imaging techniques are sensitive to a limited number of materials and provide only flat, two-dimensional (2D) images of complex, three-dimensional (3D) structured systems. (b) Cross sections require destructive sampling that (c) provides only local information. Recently, a number of new approaches of imaging the integral, three-dimensional buildup of hidden paint layers systems were proposed and put into practice. They can be considered to be modernized extensions of IRR and XRR. Terahertz time-domain spectroscopy (THz-TDS) can be used to map interfaces inside a paint multilayer. Laminography, a variant of X-ray absorption tomography, is a method that allows visualization in three dimensions of the density variations inside a small subvolume a much larger painting. Macroscopic scanning X-ray fluorescence (MA-XRF) has been employed to visualize the distribution of pigments within some of these layers. Combined use of these methods provides topographic and color information about the micrometer depth scale, essential parameters for looking “into” paintings on an entirely new level. After outlining their principles (see Table 1), we discuss the first CH cases studied with these new methods, highlighting their possibilities of extracting information about the inner structure of painted artworks.

2. X-ray-Based Methods Because X-rays can penetrate almost any object and yield information about the interior of specimens opaque to visible light, since their discovery they have been used as a powerful tool for nondestructive inspection of materials. While the art and museum world heavily relies on XRR for the visualization of the invisible inner structure of paintings, convenVol. 43, No. 6

June 2010

814-825


815


FIGURE 1. Principle of different variants of X-ray absorption-computed tomography: (a) fan-beam tomography (X-ray tube source), (b) parallel beam tomography (synchrotron source), and (c) laminography.

tional (X-ray tube-based) XRR4 has a number of important limitations. Since the observed X-ray absorbance is a summation of all element-specific absorbances, the contributions to the overall image contrast from (small quantities of) weakly absorbing elements will frequently be obscured by those of heavier elements that are present in higher concentrations. Thus, the absorption contrast in XRR images is mostly caused by the heavy metal paint components (e.g., lead in lead white, mercury in vermillion). Moreover, since canvases usually are primed with a homogeneous lead white layer, an overall absorbance background is often present. Finally, the polychromatic character of an X-ray tube also reduces the contrast. Thus, conventional XRR images of paintings frequently provide only a fragmentary view of their substructure, hampering the readability of hidden compositions. 2.1. Tomographic Imaging. 2.1.1. Principle. The images produced by XRR may also be blurred because the entire 3D shape of the irradiated object becomes projected onto a 2D screen or camera. To circumvent this, X-ray tomography makes use of an extended series of projection images recorded under many different angles between the object and primary beam; the object hereby rotates around an axis perpendicular to the source-detector axis (Figure 1a). Mathematical reconstruction then allows creation of a virtual, 3D rendition of the object’s shape and (inner) density variations and visualization of its inner parts without physically sectioning or otherwise destroying it. This can be done both at the macroscopic (decimeter to meter) level, e.g., in medical computed tomography (CT), and at the microscopic level (MCT). 2.1.2. Macro-CT and Micro-CT. Macro-CT has been widely used for 3D visualization, for both medical and nonmedical purposes.5 There are far less stringent limitations with regard to X-ray exposure and scanning time for the examination of CH artifacts in wood, paper, glass, or metals instead of patients. A recent report describes the X-ray CT inspection of the sculpture of the Egyptian pharaoh-queen Nefertiti (18th dynasty), one of 816


814-825

June 2010

Vol. 43, No. 6

the treasures of ancient Egyptian art,6 with the aim of assessing its conservation status and of gaining information about its creation. Multisection CT with a section thickness of 0.6 mm was performed, resulting in the observation that the stucco layer on the face and the ears was very thin (1-2 mm) while the rear part of the reconstructed crown showed two thick stucco layers of different attenuation values, indicating the use of a multistep manufacturing process. The discovery of air voids in the stucco and of filamentous fissures parallel to the surface led to the conclusion that very careful handling of this artifact is mandatory to avoid any pressure and shearing forces in the crown and the shoulders. Another example involved the study of Roman glass fragments contained within large clumps of soil, excavated from a burial site.7 High-resolution tomography allowed “virtual removal” of the soil, making it possible to determine the shape and morphological class of five of the 14 excavated objects. In addition, CT was useful for the reconstruction of the layout of the burial chamber, the compilation of a list of grave contents, and the positioning of these contents within the chamber. Next to medical style macro-CT equipment, several manufacturers offer table top MCT instruments with an effective spatial resolution typically situated in the 1-10 µm range; “nanotomographic” equipment (resolution between 0.2 and 1 µm) is also available. MCT has been successfully applied in many fields such as archeology, soil science, and biology. Bugani et al. recently used MCT for quantitatively estimating the changes in the porosity of sandstone8 and of archeological waterlogged wood9 after consolidation treatments. 2.1.3. Full-Field 3D Imaging by Means of Synchrotron Radiation (SR). The improving availability and performance of synchrotron radiation sources has significantly boosted the 3D imaging possibilities of paintings and paint layer stacks. Synchrotrons are large particle accelerators that produce intense, quasi-parallel beams of X-rays, at flux levels many orders of magnitude higher than those of X-ray tubes. Monochromatic beams of tunable energy with very small cross sec-


FIGURE 3. Visualization of a spherical lead soap protrusion by high-resolution absorption tomography (ESRF-ID19): (a) mechanism of protrusion formation and (b and c) views of a large protrusion (ca. 100 µm in diameter). FIGURE 2. (a) Photograph of a 16th Century prayer nut and (b) (top row) volume reconstructions with a vertical cut through the middle of the nut, revealing the shell structure, and (bottom row) volume reconstructions of the outer shell with a Gothic motif.

tions (in the micrometer to tens of nanometers range)10 can be generated. The results of CT investigation of an early 16th Century prayer nut,11 a spherical wooden object ca. 4 cm in diameter, are shown in Figure 2; its interior holds highly finished miniature wood carvings with scenes from the life of Christ with carving details well beyond the millimeter scale. CT revealed that the central part of the relief was cut from a single piece of wood, rather than assembled from multiple components, underlining the extraordinary manual dexterity of its maker. Next to making use of absorption contrast, where a transmission detector records the amount of radiation that disappears inside the irradiated object, other types of CT, based on fluorescence, diffraction/scattering,12 and refraction13 of X-rays, were developed. Phase-contrast CT exploits enhanced edge contrast caused by interference between the original X-ray beam and its refracted equivalent. In slightly absorbing materials, this significantly improves the clarity with which the interfaces between various material phases may be visualized. Via measurement of the thickness of the growth layers inside teeth of key Neanderthal fossils, information about

the age, food habits, and health of these prehistoric men could be revealed.14 Also extinct insects, trapped in opaque amber, could be visualized in this manner.15 MCT (Figure 1b) is also a very promising technique for the inspection of stratigraphic paint layer samples without the need of physically embedding them into a resin and crosssectioning them, a procedure after which only the exposed surface is available for microscopic inspection and analysis. Already embedded paint fragments can also be studied. The possibilities of MCT for obtaining virtual cross sections of small, submillimeter-sized, paint layer samples with the traditional, semidestructive embedding and polishing method were recently compared, leading to the conclusion that nearly equivalent information can be obtained.16 Figure 3 shows 3D renditions of a spherical lead-soap protrusion that formed below the surface of a 15th Century painting as a result of the saponification reaction of Pb2+ ions (from lead white) with fatty acid residues in oil paint; their gradual increase in size can lead to the puncture of the originally closed paint surface.17 MCT allows observation of the in situ growth of such bodies, e.g., during accelerated aging treatments. 2.1.4. Laminography. The types of tomography mentioned above require that the lateral dimensions of artifacts being examined are all roughly the same so that under all Vol. 43, No. 6

June 2010

814-825


817


FIGURE 4. Laminographic images obtained a various depths below the surface of a test painting. The texture of the various layers can be inspected in detail.

observation and irradiation angles, the total path length through the material that the transmitted radiation must follow does not vary more than, say, one order of magnitude. In case of paintings and other objects that are much more extended along two dimensions (length and width) than along the third (depth), conventional CT cannot be employed; during the rotation of the painting relative to the source-detector axis, in a particular orientation, its entire length or width would be in the radiation path, blocking all transmission. The related method of laminography, originally developed for the inspection of complex, flat, multilayered objects such as printed circuit boards,18 does not suffer this limitation; the rotation takes place around an axis that is tilted relative to the radiation source-detector axis (Figure 1c). Experiments on mock-up paintings19 showed that voids and hidden compartments can be visualized in a nondestructive manner via this technique as well as the texture of the different layers in a stratigraphy20 (Figure 4). 2.2. Scanned Beam Imaging. While the full-field forms of tomography are excellent for 3D imaging of density variations or of interfaces between materials, they are not sensitive to changes in elemental makeup, speciation, or crystallographic composition. Imaging by means of a scanning X-ray micro- or nanobeam while various material-specific signals are recorded can provide more detailed information of this type. Both monochromatic and polychromatic beams can be used, and in some cases, the energy of the monochromatic beams is varied, allowing for spectroscopic imaging. Often the total amount of transmitted, reflected, and/or scattered photons is recorded while energydispersive photon detectors are frequently employed. According to Bertrand et al.,21 in the period from 1986 to 2005, the most important (scanning) method of investi818


814-825

June 2010

Vol. 43, No. 6

gation was XRF (yielding information about the local elemental composition), followed by X-ray diffraction (XRD) (local crystallographic phases), X-ray absorption spectroscopy (XAS) (chemical state/electronic environment contrast), and Fourier-transform infrared (FTIR) spectroscopy (presence and distribution of specific types of molecules and/or chemical bonds). Cotte et al.22 have reviewed the use of (combinations) of these methods for the investigation of various types of cultural heritage materials, among which are (partially) altered paint layer stratigraphies. The distribution of Cd compounds at or just below the surface of altered cadmium yellow (CdS) samples derived from a painting of the avant-garde painter James Ensor (1860-1949)23 was studied in this manner. As demonstrated by µ-XRD and µ-XANES (X-ray absorption near-edge spectroscopy), the alteration involved the oxidation of CdS to CdSO4 · 2H2O under the influence of light, oxygen, and moisture (Figure 5). Sulfur species-specific imaging versus depth showed that during the last century, the sulfur oxidation front progressed ca. 10 µm below the surface. These observations are relevant for painting conservators when they are establishing optimal conditions (relative humidity and level and type of illumination) for long-term preservation of works of art. 2.3. Micro- and Macro-XRF. In case energetic primary radiation is employed to induce the emission of characteristic radiation, XRF signals not only are generated superficially but also emerge from extended depths (several tens to hundreds of micrometers) below the surface of the material. This phenomenon was exploited in two different manners for oil painting imaging (Figure 6): either by employing confocal µ-XRF, a depth-selective variant of µ-XRF, or by employing


FIGURE 6. Irradiation-detection geometry (a) of MA-XRF using multiple detectors D1--D4 (M denotes the video camera/microscope and P the primary beam). (b) Top view, showing different takeoff angles (R) associated with different detectors (c) of confocal XRF.

FIGURE 5. (A) Optical micrograph of a cross section of the partially degraded paint surface. (B) RG composite S chemical state map of panels C and D (red for sulfides and green for sulfates). Distribution maps of (C) sulfides (2.4730 keV) and (D) sulfates (2.4820 keV). Map size of 184 µm × 50 µm; step size of 1 µm × 1 µm. (E) Pb distribution. Reprinted with permission from ref 23. Copyright 2009 American Chemical Society.

highly energetic millibeams of primary X-rays that are scanned over large areas of a painting (MA-XRF). 2.3.1. Confocal µ-XRF. Scanning electron microscopy supplemented with energy-dispersive X-ray spectroscopy (SEM-EDS)24 is often employed for paint layer characterization. While powerful, this technique requires a sample to be extracted from the painting. Typically, only a limited number of samples may be taken, especially from the areas of greatest interest, such as the facial area in a portrait. During the past several years, improvements to X-ray optics based on

hollow glass capillary tubes have led to the development of depth selective or confocal µ-XRF (CXRF)25 employing two X-ray lenses with coinciding foci (Figure 6c). One optic focuses the incident beam, while the other, perpendicular to the first, gathers X-ray fluorescence only from the microregion of the sample where the focal cone of the second optic intersects that of the first. By scanning materials through this volume, one can obtain intensity profiles reflecting the local composition variation versus depth or versus a lateral coordinate with a resolution in the range of 10-50 µm. While for primary beam focusing, several types of X-ray lenses may be used, as collecting optics for the fluorescent radiation, only polycapillary lenses,26 consisting of a tapered bundle of many thousands of hollow glass tubes, are employed. A promising feature of CXRF is the possibility of providing information not only about the 3D distribution of different elements in a sample but also about their local concentrations. To do so, one must first determine the elemental composition of the outermost layers in a multilayered material and then calculate the effect of absorption by those layers on the fluorescence intensity from layers buried below.27,28 When the Vol. 43, No. 6

June 2010

814-825


819


FIGURE 7. (A) Teniers’ ‘Armourers Workshop’ and (B) in-depth distribution of Pb, as obtained by CXRF, revealing the discontinuity between panel 1 and panel 2 in the paint layers at a depth of ca. 100 µm below the surface.

energy-dependent resolution of the confocal microscope is explicitly taken into account, the position and thickness of buried layers can be more precisely determined than the nominal, instrumental resolution. Woll et al.29 have employed CXRF to examine the continuity of the paint layers across a joint between two panels present in Breugel’s ‘The Armorers Shop’. The resulting depth distribution of Pb (Figure 7B) clearly shows that only the upper paint layers are continuous and that the original paint layers were applied separately on both panels, prior to their merger. 2.3.2. High-Energy MA-XRF. Upon irradiation with an energetic X-ray beam (see Figure 6a,b), the covering surface layers will not significantly attenuate the high-energy fluorescence signals from heavy elements in the deeper layers; in this manner, the distribution of selected minor and major components in the painting may be visualized. The use of highintensity X-ray beams leads to sufficiently small data acquisition dwell times per pixel so that large, decimeter-sized areas can be scanned. Using lower-energy X-ray beams, Bergmann et al.30,31 used the same method to reveal the original (Fe rich) writing in the Archimedes’ palimpsest. Conversely, in oil paintings, to penetrate through an overcoat of lead white of, for example, 50 µm without significant losses, the energy of the fluorescent radiation must be higher than 10 keV. To demonstrate the suitability of MA-XRF for visualization of hidden paint layers, V. van Gogh’s canvas ‘Patch of Grass’ was examined at HASYLAB.32 XRR of this painting suggested that below the multicolored landscape, a painted portrait was present (Figure 8). A pencil beam (0.5 mm × 0.5 mm) of quasi-monochromatic SR with an energy of 38.5 keV was used for scanning an area of approximately 17.5 cm × 17.5 cm, corresponding to the position of the covered head. A dwell time of 2 s per pixel was employed, making the total scan time approximately 2 days. Next to chemical elements correspond820


814-825

June 2010

Vol. 43, No. 6

ing to the landscape in the upper layer, such as Cr (chromium oxide, green), Fe (prussian blue, ochre), and Zn (zinc white), the element Sb (antimony) was present, due to the use of the pigment Naples yellow [lead antimonate, Pb(SbO3)2 · Pb3(Sb3O4)2] in the covered portrait. The Sb map revealed the details of a female portrait (Figure 9); the maxima in the Sb distribution corresponded to the lighter tones of the portrait, while in the Hg map, van Gogh’s use of red accents (via the use of the red pigment vermillion, HgS) was reflected. The approximate reconstruction of the portrait based on the Sb and Hg maps presents a significantly clearer and more detailed image of the hidden composition than the XRR and IRR images (Figure 8b,c). Individual brushstrokes and all physiognomic details could be visualized. The reddish intensity of the flesh tones of the lips, cheek, and forehead adds to the readability of the portrait. Several overpainted works by Rembrandt were successfully examined by means of the same method.33 Because of the requirement to transport paintings to SR facilities and the scarcity of irradiation time represent significant limitations, activities have been started to construct an optimized X-ray tube-based MA-XRF scanner that may be employed in musea.

3. Infrared Radiation-Based Methods While X-ray-based techniques are very suitable for visualizing the internal distribution of inorganic, high-Z (Z being the atomic number) components in artworks such as lead white, vermillion, and Naples yellow, they do yield information about the distribution of low-Z components in the substructure of paintings. The heavy metal compounds (and especially the omnipresent lead white) tend to obscure or completely absorb the signals of lower-Z materials that they cover or are mixed with, yet some of these components (e.g., chalk, graphite, and ochre) are frequently employed for underdrawing or underpainting.


FIGURE 8. ‘Patch of Grass’ by V. Van Gogh: (a) optical photograph and (b) XRR and (c) IRR images of the area indicated by the red square in panel a. Reprinted with permission from ref 32. Copyright 2008 American Chemical Society.

Over the past several decades, IRR has become a routine form of analysis in many painting collections, almost exclusively for the study of carbon-based underdrawings. Recently, new infrared methods such as OCT and THz-TDS have emerged, adding the possibility of incorporating depth profiling (OCT), as well as material-specific classification (THz-TDS imaging) in the infrared imaging domain. 3.1. Infrared Reflectography. IRR was introduced in the 1960s34,35 and uses an IR source of ∼1200 nm to illuminate objects. The radiation (900-1700 nm) reflected by the object is recorded with a GaAs-array sensor yielding images with a resolution of up to 0.1 mm, covering areas of typically 0.5-0.5 m2. IRR is most suitable for the study of underdrawings that consist of infrared absorbing carbon black on reflective chalk or gyp-

sum grounds, as found in paintings from the 16th Century and earlier. Note the detailed and sharp rendering of the contour lines delineating the figures in Figure 10. Examination with IRR of 17th or 18th Century paintings tends to be less rewarding because these later paintings often were set up in sketchy touches of earth pigments or underdrawn in white chalk. These pigment are very poor infrared absorbers. Furthermore, many 17th Century paintings were done on non-infrared-reflective colored grounds. Many of the paints contain infrared absorbing pigments, such as carbon black, that make it hard to distinguish the underdrawings from the covering paint layers. 3.2. Optical Coherence Tomography (OCT). A promising new development is OCT,36,37 a point scanning method based on the use of a near-IR source coupled to a Michelson Vol. 43, No. 6

June 2010

814-825


821


FIGURE 9. (a) Result of macro-XRF scanning of the central 15 cm × 15 cm area of ‘Patch of Grass’, V. Van Gogh, revealing the portrait of a peasant woman inside the read square indicated in Figure 8a. (b and c) Some comparable portraits of peasant women from the same period by V. Van Gogh. Reprinted with permission from ref 32. Copyright 2008 American Chemical Society.

FIGURE 10. Detail of ‘Adoration of the Name of Jesus’ (1578-1580) by El Greco. (a) Photograph and (b) IRR image showing the contour lines in the underdrawing of the figures adjacent to Philip II (in the black coat).

interferometer. The source, similar to those used for conventional IRR, illuminates both a reference mirror and the object under examination. Constructive interference occurs when the length of 822


814-825

June 2010

Vol. 43, No. 6

the optical path of the light that is backscattered within the object matches, within the coherence length, the length of the optical path of the radiation reflected by the mirror. The interference


FIGURE 11. (a) Francesco Francia, ’The Virgin and Child with an Angel’ (NG 3927), detail of the angel’s eye. Copyright The National Gallery, London. (b) Infrared image of the same region using the OSIRIS camera. (c) En-face OCT image at 930 nm. The size of the area on the painting is 10 mm × 15 mm.

measurement permits the determination of the depth at which the reflection took place within the object. This adds depth resolution to the infrared investigation of paintings, allowing mapping of the distribution of specific materials and material interfaces throughout the paint stratigraphy. This can be done in the form of cross sections that are perpendicular or parallel (enface OCT) to the paint surface. Series of such maps can be combined to yield three-dimensional information about entire volumes within a painting. The depth resolution of this method is in the single-micrometer range, while the maximum probing depth depends on the thickness of the paint layers and the opacity of the composing materials in the IR domain. Because the technique is based on the collection of single-point measurements, the acquisition is slower than in the case of IRR; in a few hours, it is possible to examine decimeter-sized areas. For the study of thinly painted layers, as found in works from the 16th Century and earlier, the technique proves to be a powerful imaging tool, in particular for the study of near-surface features, notably translucent layers such as glazes and varnish. However, OCT is unable to penetrate thick and opaque paint layers, as mostly present in post-16th Century paintings. Figure 11 compares an en-face OCT image to a conventional IRR map. The depth selectivity renders the OCT image more sharp, allowing individual strokes in the underdrawing to be visualized. 3.3. Terahertz Imaging. Far-infrared or terahertz pulses can also be used to detect the presence of hidden paint layers in paintings. The technique works by using an extremely short, focused THz pulse to illuminate a painting, either from the canvas side or from the front side. The back-reflected THz pulse is analyzed for reflections originating from the various interfaces present in the painting, such as the canvas-ground interface and the interfaces between the various paint layers. The temporal spacing between the reflections is a measure of the optical thickness of the paint layers, whereas their sign and

amplitude provide information about the THz refractive index contrast between the different layers. In this sense, the method works in a manner similar to that of “acoustic echo” equipment (medical imaging). THz spectroscopy is a noncontact technique: THz pulses are focused on the painting from a distance of ∼10 cm. T-rays do not ionize the material, and the extremely low intensities employed rule out adverse effects on the painting. Moreover, the refractive-index contrast in the THz range is larger than in the visible and near-infrared (VIS/NIR) region, making it easier to observe contrast between the paint layers even if they are very thin (8 keV) characteristic lines, i.e., Z g 29 (Cu). On the other hand, by means of the recently developed variants of IRR, low-Z materials such as charcoal, chalk, and ochre that are often used for underpainting or underdrawing can be visualized in greater detail and with greater depth selectivity. Thus, it becomes possible to employ the different state-ofthe-art variants of XRR and IRR in a complementary fashion to obtain a more complete picture of a painting and many of its superimposed paint layers. This research was supported by the Interuniversity Attraction Poles Programme-Belgian Science Policy (IUAP VI/16). The text also presents results of FWO (Brussels, Belgium) projects nr. G.0704.08 and G.0179.09 and from the UA-BOF GOA programme. BIOGRAPHICAL INFORMATION Koen Janssens obtained his Ph.D. in Analytical Chemistry from the University of Antwerp and became a professor at this university in 2000. He has employed since 1990 intense beams of X-rays for nondestructive materials analysis. His main field of interest is X-ray-based microanalysis of materials, with special attention paid to local speciation of metals in (altered) environmental and cultural heritage materials such as glass and inorganic painters’ pigments. Joris Dik trained as art historian at the University of Amsterdam and then obtained a Ph.D. in X-ray crystallography at the Free University of Amsterdam (Amsterdam, The Netherlands), focussing on the synthesis and alteration behavior of the pigment lead-tin yellow. His research interest is the application of novel methods of scientific investigation to art historical problems. He currently is an associate professor at the Delft University of Technology. After gaining the “agre´gation” of chemistry at the “Ecole Normale Supe´rieure” of Lyon, Marine Cotte obtained a Ph.D. in lead824


814-825

June 2010

Vol. 43, No. 6

based cosmetics and pharmaceutical compounds used in Antiquity at C2RMF (Centre of Research and Restoration of French Museums, Paris, France). During her postdoc at the European Synchrotron Radiation Facility (ESRF), she has broadened the application of micro X-ray and FTIR spectroscopies to paintings. She currently has a twofold position as a permanent CNRS scientist at C2RMF and as a beamline scientist at ESRF. Jean Susini obtained his Ph.D. in Chemical Physics at the University Pierre et Marie Curie (Paris, France) and joined the ESRF in 1989, where he became responsible for the research and development of X-ray mirrors for synchrotron beamlines. In 1994, he took responsibility for the design, construction, and operation of the X-ray microscopy beamline. In 2009, he was appointed Head of the Instrumentation Services and Development Division of ESRF. He is also a regular lecturer at the University Joseph Fourier (Grenoble, France) and at the University Pierre et Marie Curie. His main fields of interest are X-ray optics, X-ray imaging techniques, and their applications.

REFERENCES 1 van de Wetering, E. The Painter at Work; Amsterdam University Press: Amsterdam, 2000. 2 Hendriks, E. Van Gogh’s working Practice: A technical study. In New Views on Van Gogh’s Development in Antwerp and Paris: An integrated Art Historical and Technical study of his Paintings in the Van Gogh Museum; Hendriks, E., Van Tilborgh, L., Eds.; University of Amsterdam: Amsterdam, 2006; pp 231-245. 3 Khandekar, N. Preparation of cross sections from easel paintings. Rev. Conserv. 2003, 4, 52–64. 4 Van Heugten, S. Radiographic images of Vincent van Gogh’s painting in the Kro¨llerMu¨ller Museum, Otterlo, and the Van Gogh Museum, Amsterdam. In Van Gogh Museum Journal; Van Gogh Museum: Amsterdam, 1995; pp 63-85. 5 van Kaick, G.; Delorme, S. Computed tomography in various fields outside medicine. Eur. Radiol. 2005, 15, D74–D81. 6 Huppertz, A.; Wildung, D.; Kemp, B. J.; Nentwig, T.; Asbach, P.; Rasche, F. M.; Hamm, B. Nondestructive Insights into Composition of the Sculpture of Egyptian Queen Nefertiti with CT. Radiology 2009, 251, 233–240. 7 Janssen, R.; Poulus, M.; Kottman, J.; De Groot, T.; Huisman, D. J.; Stoker, J. CT: A new nondestructive method for vizualising and characterizing Ancient Roman glass fragments in situ in blocks of soil. Radiographics 2006, 26, 1837–1844. 8 Bugani, S.; Camaiti, M.; Morselli, L.; Van de Casteele, E.; Janssens, K. Investigating morphological changes in treated vs. untreated stone and building materials by Xray micro-CT. Anal. Bioanal. Chem. 2008, 391, 1343–1350. 9 Bugani, S.; Modugno, F.; Lucejko, J. J.; Giachi, G.; Cagno, S.; Cloetens, P.; Janssens, K.; Morselli, L. Study on the impregnation of archaeological waterlogged wood with consolidation treatments using synchrotron radiation microtomography. Anal. Bioanal. Chem. 2009, 395, 1977–1985.


10 Attwood, D. Microscopy: Nanotomography comes of age. Nature 2006, 442, 642– 643. 11 Reischig, P.; Blaas, J.; Botha, C.; Bravin, A.; Porra, L.; Nemoz, C.; Wallert, A.; Dik, J. A note on medieval microfabrication: The visualization of a prayer nut by synchrotron-based computer X-ray tomography. J. Synchrotron Radiat. 2009, 16, 310–313. 12 De Nolf, W.; Janssens, K. Micro X-ray diffraction and fluorescence tomography for the study of multilayered automotive paints. Surf. Interface Anal. 2010, 42, 411– 418. 13 Weitkamp, T.; David, Ch.; Bunk, O.; Bruder, J.; Cloetens, P.; Pfeiffer, F. X-ray phase radiography and tomography of soft tissue using grating interferometry. Eur. J. Radiol. 2008, 68S, S13-S17. 14 Tafforeau, P.; Smith, T. A. Nondestructive imaging of hominoid dental microstructure using phase contrast X-ray synchrotron microtomography. J. Hum. Evol. 2008, 54, 272–278. 15 Lak, M.; Fleck, G.; Azar, D.; Engel, M.; Kaddumi, H.; Neraudeau, D.; Tafforeau, P.; Nel, A. Phase contrast X-ray synchrotron microtomography and the oldest damselflies in amber (Odonata: Zygoptera: Hemiphlebiidae). Zool. J. Linnean Soc. 2009, 156, 913–923. 16 Boon, J. J.; Ferreira, E. S. B.; Van Der Horst, J.; Stampanoni, M.; Marone, F. X-ray tomographic microscopy compared to ion polished paint cross sections of 19th century paints with and without metal soap aggregates. In Book of Abstracts;TECHNART 2009 Conference, April 2009, Athens; p 39. 17 Keune, K.; Boon, J. J. Analytical imaging studies of cross-sections of paintings affected by lead soap aggregate formation. Stud. Conserv. 2007, 52, 161–176. 18 Helfen, L.; Baumbach, T.; Cloetens, P.; Baruchel, J. Phase-contrast and holographic computed laminography. Appl. Phys. Lett. 2009, 94, 104103. 19 Krug, K.; Porra, L.; Coan, P.; Wallert, A.; Dik, J.; Coerdt, A.; Bravin, A.; Reischig, P.; Elyyan, M.; Helfen, L.; Baumbach, T. Relics in Medieval Altarpieces? Combining X-ray Tomographic, Laminographic and Phase-Contrast Imaging to Visualize Thin Organic Objects in Paintings. J. Synchrotron Radiat. 2008, 15, 55–61. 20 Dik, J.; Reischig, P.; Krug, K.; Wallert, A.; Coedrt, A.; Helfen, L.; Baumbach, T. Three-dimensional Imaging of Paint Layers and Paint Substructures with Synchrotron Radiation Computed R-Laminography. J. Am. Inst. Conserv. 2009, 28, 185–197. 21 Bertrand, L.; Vantelon, B.; Pantos, E. Novel interface for cultural heritage at SOLEIL. Appl. Phys. A: Mater. Sci. Process. 2006, 83, 225–228 (see also http://www.synchrotron-soleil.fr/heritage). 22 Cotte, M.; Susini, J.; Dik, J.; Janssens, K. Synchrotron-Based X-ray Absorption Spectroscopy for Art Conservation: Looking Back and Looking Forward. Acc. Chem. Res. 2010, 43, 000 (in press). 23 Van der Snickt, G.; Dik, J.; Cotte, M.; Janssens, K.; Jaroszewicz, J.; De Nolf, W.; Groenewegen, J.; Van der Loeff, L. Characterization of a Degraded Cadmium Yellow

24

25

26

27

28

29

30 31 32

33 34

35 36

37 38

(CdS) Pigment in an Oil Painting by Means of Synchrotron Radiation Based X-ray Techniques. Anal. Chem. 2009, 81, 2600–2610. Derrick, M.; Souza, L.; Kieslich, T.; Florsheim, H.; Stulik, D. Embedding paint crosssection samples in polyester resins: Problems and solutions. J. Am. Inst. Conserv. 1994, 33, 227–245. Kanngiesser, B.; Mantouvalou, I.; Malzer, W. Non-destructive, depth resolved investigation of corrosion layers of historical glass objects by 3D Micro X-ray fluorescence analysis. J. Anal. At. Spectrom. 2008, 23, 14-819. Janssens, K.; Proost, K.; Falkenberg, G. Confocal microscopic X-ray fluorescence at the HASYLAB microfocus beamline: Characteristics and possibilities. Spectrochim. Acta, Part B 2004, 59, 1637–1642. Kanngiesser, B.; Malzer, W.; Reiche, I. A new 3D micro X-ray fluorescence analysis set-up: First archaeometric applications. Nucl. Instrum. Methods Phys. Res., Sect. B 2003, 211, 259–264. Kanngiesser, B.; Malzer, W.; Rodriguez, A. F.; Reiche, I. Three-dimensional microXRF investigations of paint layers with a tabletop setup. Spectrochim. Acta, Part B 2005, 60, 41–47. Woll, A. R.; Mass, J.; Bisulca, C.; Cushi-Nan, M.; Griggs, C.; Wanzy, T.; Ocon, N. The Unique History of The Armorer’s Shop, An Application of Confocal X-ray fluorescence microscopy. Stud. Conserv. 2008, 53, 93–109. Bergmann, U. Archimedes brought to light. Phys. World 2007, 20, 39–42. Service, R. F. Imaging: Brilliant X-rays reveal fruits of a brilliant mind. Science 2006, 313, 744–745. Dik, J.; Janssens, K.; van der Snickt, G.; van der Loeff, L.; Rickers, K.; Cotte, M. Visualization of a Lost Painting by Vincent van Gogh Using Synchrotron Radiation Based X-ray Fluorescence Elemental Mapping. Anal. Chem. 2008, 80, 6436–6442. van de Wetering, E. Rembrandt Laughing c. 1628: A painting resurfaces. In Kroniek van het Rembrandthuis; Rembrandthuis: Amsterdam, 2007; pp 18-40. Van Asperen de Boer, J. R. J. An introduction to the scientific examination of paintings. In Scientific examination of early Netherlandish painting; Filedt-Kok, J. P., Van Asperen-De Boer, J. R. J., Eds.; Fibula-van Dishoeck: Bussum, The Netherlands, 1976; pp 1-40. Saunders, D.; Billinge, R.; Cupitt, J.; Atkinson, N.; Liang, H. A new camera for highresolution infrared imaging of works of art. Stud. Conserv. 2006, 51, 277–290. Liang, H.; Cid, M. G.; Cucu, R. G.; Dobre, G. M.; Podoleanu, A. Gh.; Pedro, J.; Saunders, D. En-face Optical Coherence Tomography: A novel application of noninvasive imaging to art conservation. Opt. Express 2005, 13, 6133–6144. Hughes, M.; Spring, M.; Podoleanu, A. Speckle noise reduction in optical coherence tomography of paint layers. Appl. Opt. 2010, 49, 99–107. Adam, A. J. L.; Planken, P. C. M.; Meloni, S.; Dik, J. TeraHertz imaging of hidden paint layers on canvas. Opt. Express 2009, 17, 3407–3416.

Vol. 43, No. 6

June 2010

814-825


825

Structural Examination of Easel Paintings with Optical Coherence Tomography PIOTR TARGOWSKI,*,† MAGDALENA IWANICKA,† ‡ ´ LUDMIŁA TYMINSKA-WIDMER, MARCIN SYLWESTRZAK,† AND EWA A. KWIATKOWSKA† † Institute of Physics, Nicolaus Copernicus University, ul. Grudziadzka 5, ´ Poland, ‡Institute for the Study, Restoration and Conservation of 87-100 Torun, Cultural Heritage, Nicolaus Copernicus University, ul. Gagarina 7, ´ Poland 87-100 Torun,

RECEIVED ON JULY 6, 2009

嘷 w This paper contains enhanced objects available on the Internet at http://pubs.acs.org/acr.

CON SPECTUS

I

dentification of the order, thickness, composition, and possibly the origin of the paint layers forming the structure of a painting, that is, its stratigraphy, is important in confirming its attribution and history as well as planning conservation treatments. The most common method of examination is analysis of a sample collected from the art object, both visually with a microscope and instrumentally through a variety of sophisticated, modern analytical tools. Because of its invasiveness, however, sampling is less than ideally compatible with conservation ethics; it is severely restricted with respect to the amount of material extirpated from the artwork. Sampling is also rather limited in that it provides only very local information. There is, therefore, a great need for a noninvasive method with sufficient in-depth resolution for resolving the stratigraphy of works of art. Optical coherence tomography (OCT) is a noninvasive, noncontact method of optical sectioning of partially transparent objects, with micrometer-level axial resolution. The method utilizes near-infrared light of low intensity (a few milliwatts) to obtain cross-sectional images of various objects; it has been mostly used in medical diagnostics. Through the serial collection of many such images, volume information may be extracted. The application of OCT to the examination of art objects has been in development since 2003. In this Account, we present a short introduction to the technique, briefly discuss the apparatus we use, and provide a paradigm for reading OCT tomograms. Unlike the majority of papers published previously, this Account focuses on one, very specific, use of OCT. We then consider two examples of successful, practical application of the technique. At the request of a conservation studio, the characteristics of inscriptions on two oil paintings, originating from the 18th and 19th centuries, were analyzed. In the first case, it was possible to resolve some questions concerning the history of the work. From an analysis of the positions of the paint layers involved in three inscriptions in relation to other strata of the painting, the order of events in its history was resolved. It was evident that the original text had been overpainted and that the other inscriptions were added later, thus providing convincing evidence as to the painting’s true date of creation. In the second example, a painting was analyzed with the aim of confirming the possibility of forgery of the artist’s signature, and evidence strongly supporting this supposition is presented. These two specific examples of successful use of the technique on paintings further demonstrate how OCT may be readily adaptable to other similar tasks, such as in the fields of forensic or materials science. In a synergistic approach, in which information is obtained with a variety of noninvasive techniques, OCT is demonstrably effective and offers great potential for further development.

826


826-836

June 2010

Vol. 43, No. 6

Published on the Web 12/31/2009 www.pubs.acs.org/acr 10.1021/ar900195d © 2010 American Chemical Society

Structural Examination of Easel Paintings with OCT Targowski et al.

Introduction Easel paintings, especially when prepared according to traditional techniques, are multilayered structures. A support, usually stretched canvas or a wooden panel, is covered with glue size, and then primed in order to smooth its surface and to ensure proper adhesion of the consecutive layers. The composition may be outlined in an underdrawing executed by one of various techniques, and then the opaque paint layers are applied. To obtain a desirable visual effect, an additional cover of one or more layers of semitransparent glazes and finally a transparent varnish may be added. Apart from the obvious protective function of the varnish, it has a very important effect on the final appearance of the artwork.1,2 Additionally in many cases, this primary structure may be modified by alterations introduced by the artists themselves or commissioned by later owners. Among these, traces of previous renovation treatments, for example, overpaintings, are particularly important to detect. Another separate group of interventions in the structure of the artwork are various kinds of forgeries, sometimes committed many years or even one or more centuries ago and having their own history. For conservation or inventory purposes, it is very important to reveal the stratigraphy, that is, the order, thickness, composition, and possibly the origin, of the layers building up the painting’s structure. This knowledge is important in confirming the attribution and history of the artwork and is also essential for the planning of future conservation treatments. For these reasons, the attention of art historians and conservators has long been focused on methods of obtaining such information. The most common approach has been, and still largely is, to collect a small sample, embed it in resin, and then analyze its cross section under the microscope. At present, in addition to or replacing classical microchemical analysis,3 a wide range of microspectroscopic methods, for example, micro-Raman (µRaman), micro-infrared Fourier transform (µFTIR), micro-X-ray fluorescence (µXRF), scanning electron microscopic energy-dispersive X-ray (SEM-EDX), and particle-induced X-ray emission (PIXE) spectroscopies, is used to better understand artwork composition throughout examination of the sample.4 However, despite the fact that the results obtained in this way have been and to a large extent still are usually considered the most unequivocal, there are two disadvantages to this approach. First, sampling is rightly regarded as invasive, and so its application is strictly limited by conservation ethics. Usually it is allowed to be performed only within an area of former damage, where collecting material would have a

negligible impact on the final condition of the artwork. Therefore, information collected this way must be considered rather random and not necessarily representative of the object as a whole. In addition, the results obtained from sampling in the area of destruction must of necessity be interpreted with extreme care, since the structure of the object in this very place could have already been more or less heavily perturbed by the damage. The second disadvantage of sampling methodologies lies in their local character: the information gained is always specific to the site of sampling, since the structure of the painting, especially the thickness of particular layers, may vary rapidly5 across even fairly small regions. All of these limitations have long been known, and in response, alternative noninvasive investigations have gained significant attention. Paintings are X-rayed to detect alterations and traces of former compositions covered by later superimposed additions. The inspection of paintings by means of UVexcited fluorescence has also become very popular: not only may old paint and varnish layers (fluorescent) be distinguished from new ones (nonfluorescent),6 but the chemical composition of the varnish may be revealed due to the characteristic fluorescence exhibited by some of its ingredients.7 Visual inspection of UV-excited fluorescence may also indicate the presence of certain pigments in the paint layer.1 Another technique used for pigment identification is reflectance spectroscopy in visible light.8 The significant transparency of many pigments to infrared light permits utilization of IR radiation in the wavelength range 0.7 to 7 µm9,10 to reveal underdrawings. The methods described above allow general diagnoses of paintings for the purpose of resolving causes of their deterioration and to detect damaged regions as well as areas of former conservation treatments. Information on the location of underdrawings, on regions of previous restorations, on the thickness of paint layers, and on the presence of delaminations can also be obtained through various thermographic inspection techniques, including pulsed phase thermography (PPT).10 The identification of delaminations may also be achieved through air-coupled ultrasound echo detection.11 Holographic interferometry techniques provide information on bulk structural faults and mechanical discontinuities in a piece of artwork.12 All these noninvasive methods, as well as modern structurally resolving ones like XRF (see above) or CT (computed tomography) either are only sensitive to surface properties or do not have sufficient spatial resolution to be useful for structures with an overall thickness of only a fraction of a millimeter: the resulting information comprises an overlap of that from all different depths. Vol. 43, No. 6

June 2010

826-836


827


This Account is devoted to a novel technique, optical coherence tomography (OCT), for high-resolution structural imaging, operational in conservation science since 2003.13 OCT uses low-power light in the spectral range of 700 to 1500 nm to probe structure noninvasively from a distance of a few centimeters, in order to provide detailed micrometer-resolved information on its cross section, and thus stratigraphy, in a purely optical, noncontact, noninvasive, and nondestructive way. The major drawback to practical application of OCT in the examination of artwork arises from the limited transparency of the strata examined to the light used. However, layers like varnishes and glazes, usually semitransparent and thereby decisive for aesthetic impression, may readily be successfully tested in this way. In addition, the very high sensitivity of modern OCT modalities enables the collection of signals from deeper structures. Until now, OCT has been employed in the investigation and conservation of artwork to examine varnish and glaze layers of easel paintings,14-18 analyze underdrawings,15 examine historic jades,19 ceramics, and glass objects,20,21 inspect atmospheric corrosion of stained glass,22 and precisely image punchwork.23 Another group of applications is related to monitoring various dynamic processes such as the drying of a varnish layer,24,25 laser ablation of varnish,26-28 and the tracing of environmentally induced deformations of paintings on canvas.29 In this Account, the application of OCT to investigation of the structure of easel paintings is revisited and extended to more practical aspects. Using examples of oil paintings from a conservation studio, it will be shown how information obtained with OCT may be used to verify their history and authenticity.

Methodology The OCT technique originates from diagnostic medicine30 and at present is used mostly in ophthalmology for examination of the human retina, though OCT tomographs are also commercially available for examination of the anterior segment of the eye and of skin and other tissues. These instruments, as well as those designed for material science, may be directly used to examine works of art. In every OCT instrument, infrared radiation penetrates the object and is partially reflected at interfaces of layers of different refractive indices, or sometimes scattered from sites of inhomogeneity in its structure. Returning light is collected, and the time of propagation from the given depth of the structure is determined, thus providing a measurement of the opti828


826-836

June 2010

Vol. 43, No. 6

cal path to this structure. However, at the micrometer path resolution desired, this propagation time cannot be measured directly and is accomplished instead through a measurement of interference between the object and reference beams in an interferometer. Many specific configurations of OCT instrumentation are described in detail elsewhere,30 so only a brief summary of the principle and description of the instrument used by the authors will be given here. To explain the general idea of the OCT device, consider the interferometer (usually of the Michelson type) with the object to be investigated in one arm and a mirror in the second (reference) arm. An original beam from the broadband light source is split into two, often propagated in optic fibers, as depicted in Figure 1. The object is penetrated by a narrow beam of light from one of them, some of which is reflected or scattered back from various elements (1, 2, ..., n) of the structure and collected in the interferometer. The second beam is simultaneously back-reflected from a reference mirror. The beam splitter recombines these beams to allow interference. If the difference of round trip propagation times in the two arms of the interferometer is τn for a given structural element n, the interference signal may be described by the general formula

I(ω) ) Ir(ω) +

∑ In(ω) + 2 ∑ [√Ir(ω)In(ω) cos(ωτn)] n

{

) S(ω) Rr +

n

∑ Rn + 2 ∑ [√RrRn cos(ωτn)]} n

(1)

n

where the index r refers to the signal from the reference arm. In the first line expression, the common assumption that the (intensity) reflection coefficients R do not depend on the optical frequency ω is invoked. S(ω) denotes the spectrum of the light source. It can be seen in eq 1 that τn appears as its product with the optical frequency ω. As a result, two approaches to extracting structural information (Rn and τn) from the interferometric signal are possible. In the first-developed method, time-domain OCT (TdOCT), the interference signal is collected by means of a photodiode and thus integrated over its spectrum. In this case, the cosine term averages to zero everywhere except at τ ≈ 0. More precisely, the full width at half-maximum (FWHM) of the interference signal as a function of τ defines the axial resolution of the OCT instrument. For a Gaussian-shaped light source of central wavelength λcenter and a spectral bandwidth ∆λFWHM, this width, ∆τFWHM, and therefore the axial resolution, ∆x, are given by the formulas


∆τFWHM )

8 ln 2 ∆ωFWHM

and

∆x )

2 1 2 ln 2 λcenter (2) nR π ∆λFWHM

where both the forward and backward directions of light propagation in the object and the refractive index, nR, of the investigated region are taken into account in calculating ∆x. As can be seen from eq 2, the in-depth resolution increases with increasing bandwidth of the light source. Thus, to achieve the required axial resolution for these studies, broadband light sources with ∆λFWHM between 30 and 200 nm are utilized. Since ∆x quickly increases (i.e., deteriorates) for deeper infrared, radiation in the spectral range 800-1050 nm is usually employed. In time-domain OCT, the reference mirror is moved gradually, and the interference signal is simultaneously registered to complete the in-depth scan. When the optical paths match to within ∆x, an interference signal is detected. Knowing the position of the mirror, a profile of the internal structure of the object along the probing beam can be recovered. To recover complete cross sections, the probing beam is scanned over the object. By analogy with the usage in ultrasonography, such a cross-sectional image is called a B-scan, whereas the profile along a single vertical line is called an A-scan. Different images are obtained using the full-field modality of the OCT technique (FF-OCT),15 which produces cross sections in planes perpendicular to the direction of probing light (C-scans). Among the various FF-OCT systems available, microscopic devices (utilizing Linnik31 or Mirau17 configurations) have outstanding in-plane resolution, at the price, however, of a field of view limited to a fraction of a millimeter. Moreover, these systems very often utilize light from the visible range, which, according to eq 2, also provides better in-depth resolution.17 FF-OCT additionally holds promise for spectroscopic imaging.32 Fourier-domain OCT was developed later33 and differs from TdOCT in the way the structural information is extracted from the interferometric signal given by eq 1. In this case, the spectral signal is registered by means of a spectrograph (spectral OCT), or a very fast-tuned laser is used as a source with a single photodiode detector (swept source OCT). In both modalities, the spectrum of the interference signal is analyzed by means of Fourier transformation, and the components obtained have frequencies proportional to delay differences τn and amplitudes proportional to reflectivities Rn. The advantages of Fourier domain methods lie in the absence of movable parts in the reference arm, which increases the acquisition speed 100-fold, and in significantly higher sensitivity due to the multiplex advantage.34

FIGURE 1. OCT tomograph used in this study: LS, light source; OI, optical isolator; FC, fiber coupler; PC, polarization controller; NDF, neutral density filter; DC, dispersion compensator; RM, referencearm mirror; X-Y, transversal scanner; DG, diffraction grating; SL, spectrograph lens, CCD, linear CCD camera.

Another important consequence of eq 1 arises from the fact that in the interference term, (IrIn)1/2, a weak useful signal In is multiplied by a strong reference signal Ir. This leads, due to heterodyne gain, to the very high, shot-noise limited sensitivity of the method. In the case studies reported in this Account, a prototype laboratory-constructed high-resolution spectral OCT imaging system (tomograph) was employed (Figure 1). Its axial resolution is measured at 4 µm in air. It comprises a broadband infrared light source (LS, λcenter ) 845 nm, ∆λFWHM ) 107 nm) made up of two superluminescent diodes coupled together. The light output, of high spatial coherence (ensuring sensitivity), is launched into a single-mode optical fiber and passed through an optical isolator (OI), which protects the source from back-reflected light. This tomograph employs a fiber-optic Michelson interferometer, comprising a 50:50 fiber-optic coupler (FC) dividing the light into reference and object arms. The reference arm includes a polarization controller (PC), which provides optimal polarization conditions for interference, a neutral density filter (NDF) for adjusting the light power to achieve shot-noise-limited detection, a block of glass (DC) acting as a dispersion compensator, and the stationary reference mirror (RM). The object arm comprises a lens forming a narrow probing beam and a transversal scanner (X-Y) responsible for scanning the beam across the sample. The light beams, back-reflected from the reference mirror and reflected and or Vol. 43, No. 6

June 2010

826-836


829


backscattered from structural elements of the object, return to the coupler and interfere. To extract structural information from the interference signal, the tomograph is equipped with a custom-designed spectrograph employing a fast CCD camera as detector. The spectrograph comprises a volume-phase holographic grating (DG) with 1200 grooves/mm and an achromatic lens (SL), which focuses the spectrum onto a 12-bit single-line CCD camera (2048 pixels, 12-bit A/D conversion). Following Fourier transformation, the interference signal yields an A-scan, that is, one line of the cross-sectional image. Moving the beam across the sample in the X or the Y direction with the transversal scanner (X-Y) enables collection of a B-scan, that is, a 2-D cross-sectional image, while the addition of scanning in the perpendicular direction (Y or X, respectively) yields complete 3-D information on the spatial structure within seconds. Cross-sectional images obtained by OCT are usually presented in a false-color scale: black areas correspond to nonscattering media, such as air or clear, transparent varnish. Cold colors (blue and green) indicate low-scattering or reflecting areas, whereas warm colors (yellow and red) designate highscattering or -reflecting regions. More details on mapping the scattered or reflected intensities with false colors may be found elsewhere.35 In all tomograms shown here, the incident light path is from the top, through air. Thus, the first strong line corresponds to the air-painting (i.e., most usually the air-varnish) interface, while other layers accessible by the light utilized appear below this. The lowest layer depicted is always the first nontransparent one in the structure. It must be emphasized that all distances within an object are presented in the OCT tomograms as optical ones, with the assumption inherently made in processing that the photons are reflected or scattered only once at the structure sensed. Thus, if a gradual decay of the signal is observed below the boundary of an opaque layer (see Figure 4a in Results section), it means that many multiscattering events occur in this layer and create a virtual downward elongation of the imaged stratum. In contrast, if the OCT signal decays rapidly, creating a narrow trail in a tomogram (see Figure 4b), it indicates that the layer absorbs light strongly. In neither of these cases, therefore, is it possible to determine the layer thickness. Another consequence of the representation of distances as optical ones is that all of the internal structures of the object appear vertically stretched by a factor equal to the refractive index of the material. Consequently, as follows from eq 2, the axial resolution of the instrument is increased by the same factor. If the strata in the material examined are not perpendic830


826-836

June 2010

Vol. 43, No. 6

ular to the beam direction, refraction of the incident and returning light also leads to some additional distortion of the image. It is possible to correct this effect numerically. However, in the cases of the tomograms presented in this Account, while exact image corrections were not performed, the vertical scale bar is adjusted to present what would, assuming a refractive index of 1.5 and no distortion by refraction, be real distances. In addition, the vertical scale of all tomograms presented is expanded with respect to the lateral scale, because this improves appreciation of the axial structure of the object.

Practical Examples In this contribution, the results of OCT examinations made on two paintings on canvas of different style and time of origin are presented. The first depicts a Franciscan monk, Leonard of Porto Maurizio (1676-1751). In the lower right-hand corner, the painting bears an inscription (St. Leonard), and this is paired up with a date (1797) painted in the lower left-hand corner (Figure 2). It is known36 that Leonard was beatified by Pope Pius VI in 1796, and proclaimed Saint by Pope Pius IX in 1867. This history has raised doubts about the origin of the inscriptions in the painting. If it had been executed in 1797, Leonard would not yet have been canonized. Thus it was reasonable to suspect that the inscription St. Leonard was added later. An inspection of the picture in raking (obliquely incident) light revealed the existence of another inscription in the background, located above the date 1797 and hidden under a thick layer of varnishes and an overpainting. It was not possible to read particular letters with standard techniques; neither X-ray nor infrared examination had been helpful. Fortunately, the inscription was painted quite thickly, and the canvas and priming layers were thin enough to allow examination with white-light penetrating the picture from an intense source placed behind it (see Figure 6a in Results section). The two-line text could be deciphered as B. Leonardus d.[a?] Maurzio. However, it was not possible to determine the position of the inscription within the stratigraphy of the painting, which had been restored at least three times. It was therefore necessary more precisely to investigate the structure of the painting in order to verify the configuration of paint and varnish layers in the areas of all three inscriptions and thus provide unambiguous technical information with which to resolve its history. The second painting examined dates from the late nineteenth century and depicts an unknown woman. It is skillfully painted and signed GioRdi (Figure 3). Routine inspection of UV-excited fluorescence of the painting proved the existence of two kinds of


FIGURE 3. Upper panels, Portrait of an unknown woman, oil on canvas, 41.5 × 30 cm2, photography in front illumination (VIS) and of UV-excited fluorescence (UV); photo by W. Grzesik and used with permission. Lower panels: area of signature.

mine whether it lies on top of the remains of the original varnish (which had been removed), which would imply intentional forgery.

FIGURE 2. Saint Leonard of Porto Maurizio, picture from the Franciscan Church of St. Bonaventure in Pakosć´, Poland, oil on canvas, 121 × 84 cm2, photography in front illumination, taken by Magdalena Iwanicka and used with permission. Circles mark the regions in which the tomograms shown in Figure 4 were taken. Rectangles mark the regions of the inscriptions investigated: tomograms shown in figures indicated.

varnish on its surface: primary, fluorescing yellow, and secondary, of slightly blue fluorescence. Although it is certain that such a difference in the color of fluorescence of two varnishes may be due only to their different composition, visual color impression alone (i.e., without precise spectroscopic analysis7) is not reliable for unambiguous identification of varnishes. However, some rough indication may be given by conservation practice: old oilresin varnishes tend to fluoresce yellow-green, whereas modern synthetic varnishes develop rather blue fluorescence over time. UV examination did not reveal any differences (which usually lead to the assumption of forgery) between the region of the signature and its surroundings. It can be seen in Figure 3 that the signature is placed within a strip-like area where the primary varnish had been removed before the whole painting (including the area of the signature) was covered with a different varnish, of rather blue UV-excited fluorescence. The aim of the OCT examination of the signature was to deter-

Results a. Structural Imaging of Multilayer Paintings. The analysis of the painting Saint Leonard of Porto Maurizio by means of optical coherence tomography reveals quite a simple primary structure. In a sample tomogram (Figure 4a), no glaze layer is seen to be present, but up to four layers of varnish are revealed. They are distinguishable mostly due to particles of dirt originally present on the surface of the painting, which then became trapped under the next layer of varnish.37 The signal from light scattered by these particles forms a line of dots in the tomogram, marking the location of the interface between the surface of the painting and the subsequent layers of varnish. Had the interface line in the tomogram been more continuous and better pronounced, reflection due to the difference in refractive indices would have been indicated. The signal from scattering at the upper interface of the primary, most probably original, varnish layer (yellow arrows in the figure) is quite continuous, and the strongest of all those originating from interfaces between the varnish layers. This suggests that the painting was well dried, even weathered and not cleaned, before revarnishing. Such a structure is quite commonly present in OCT tomograms of varnish layers and may imply that such varnishes belong to different chronological phases. Vol. 43, No. 6

June 2010

826-836


831


FIGURE 4. OCT tomograms from Saint Leonard of Porto Maurizio: (a) multilayer varnish (image width 7 mm); (b) semitransparent overpainting (image width 7 mm); (c) opaque overpainting (image width 12.3 mm). Yellow arrows indicate the surface of the primary varnish layer; green arrows indicate the primary opaque paint layer; circles show boundaries between original and overpainted areas; rectangle indicates a region in which the overpainting is completely opaque; see text for details. The bars indicate real distances in media of refractive index 1.5.

The opaque paint layer (green arrows) in the area of this tomogram absorbs the probing light moderately, and some residual multiscattering is present. In the next cross section (Figure 4b), the original paint layer (green arrows) is more absorbing, due to its different pigment composition, and thus visible as a narrow line. On the right-hand side, it is apparently covered by another layer of paint. It is clear that this paint layer, which is partially permeable to IR light, lies on at least three layers of varnish, and the original paint layer continues underneath. Such an OCT image is characteristic for

overpainting onto the surface of the picture. The practice of this kind of renovation was quite common in the past. A similar situation is shown in Figure 4c, but the pigments in this overpainting absorb probing light strongly, and only traces of subsequent layers are visible. If only the part of this image marked with a green rectangle were analyzed separately, the image would suggest a nontransparent primary paint layer. However, detailed inspection in relation to the unaltered structures visible in Figure 4c outside the area of the overpainting, clearly indicates that, at the boundary between

FIGURE 5. OCT examination of inscriptions from Saint Leonard of Porto Maurizio: (a) first digit 7 from date 1797 (image width 9.3 mm); (b) letter S from the text St. Leonard. (image width: 7 mm). Letters are located between arrows; see text for details. Bars indicate real distances in media of refractive index 1.5. 嘷 w Flow-through OCT movies of the first digit 7 from date 1797 and letter S from the text St. Leonard are available.

832


826-836

June 2010

Vol. 43, No. 6


FIGURE 6. Analysis of the overpainted inscription B Leonardus a Maurzio: (a) letter D in intense penetrating light; (b) shape of the same letter reconstructed from the OCT data by extracting the signal from depths between 137 and 145 µm; (c) the OCT scan in the same lateral scale. Yellow solid lines in panels a and b indicate the location of the scan.

the two areas (circles), the paint layers do not match and that the original layer (green arrows) extends underneath, similarly to that in Figure 4b.

Analyses of the kind described above may provide conservators with helpful information, which is especially important for the prospective removal of secondary layers to be carried out with the intention of preserving the original ones. b. Analysis of Inscriptions and Artists’ Signatures. A similar analysis applies to signatures and other inscriptions. The order of layers provides significant information on the painting’s history and helps in understanding its origin. The painting Saint Leonard of Porto Maurizio is unsigned, as is often found in artworks commissioned by convents and painted by members of the community. As follows from the examination of the whole painting, sample results of which are shown in Figure 4, it was overpainted to a great extent at least twice and revarnished at least three times. This condition must be taken into account when the properties of the inscriptions are examined. In this picture, two texts are visible at present. One, 1797, refers to the possible year of its creation, the other, St. Leonard., to the person depicted. Whereas the date is barely legible, and thus unfortunately not possible to reproduce here, the Saint’s name is very clearly visible. Examination with the aid of OCT reveals a different localization of the two inscriptions within the stratigraphy of the painting. The date (Figure 5a, between arrows) is visible in the tomogram as a bright strip under the varnish and possibly under a semitransparent, thin overpainting covering all of the area under investigation. The exact position of the date within the structure is not entirely clear. It may be painted on top of the primary varnish (some traces of a transparent layer are visible underneath), but there is no doubt that the layer containing the date belongs to a deep stratum of the structure. The second inscription (see letter S from St. Leonard, Figure 5b, between arrows) is painted on top of the thick coating of varnish covering the original black background which absorbs probing light strongly, so that only its upper interface is visible. The text St. Leonard must therefore be considered to have been added later, probably during the last renovation around the middle of the nineteenth century as an alteration commemorating Leonard’s canonization. The analysis of the third, overpainted, inscription proved much more difficult, since it is not legible directly, either by front illumination or in raking light, but photography with very strong back-lighting through the painting (Figure 6a) revealed shadows of the letters. Such imaging would not be possible, of course, in the case of a picture painted on a solid support, like a wooden panel. Because of the almost opaque overpainting layer covering the inscription, OCT examination of this area provided less than conclusive Vol. 43, No. 6

June 2010

826-836


833


FIGURE 7. OCT tomogram from Portrait of an unknown woman taken over the letter d of the signature (located between red arrows). Green arrows indicate the primary, opaque paint layer. To the right, three layers of varnish are visible over an opaque paint layer. Bars indicate real distances in media of refractive index 1.5.

direct results. The deciphering of the letters from 3-D OCT was, nevertheless, possible (Figure 6b) by adding together signals from defined depths beneath the surface and presenting them in a false-color scale, a procedure that had previously been tested and proved27 on model paintings. In the present case, the greatest legibility was obtained by visualization of shadows of the thick letters, that is, by collecting signal from underneath the paint layer (137145 µm). It should be emphasized that, since this technique utilizes scattered light, it is not limited to paintings on transparent supports. Moreover, careful inspection of this OCT tomogram (Figure 6c) leads to the conclusion that the whole area of investigation is, as expected, covered by the overpainting, with a continuous layer of varnishes underneath. It is important to note that, in the tomogram, the layer of varnishes in areas that correspond to cross sections of the letter D is visually similar to that in the surrounding background; no shadows cast by the opaque paint of the inscription are present. Thus, the different transparency of the structure, which made it possible to obtain an image from Figure 6a, is not induced by the top layers of the painting. This means that the inscription must be located underneath the subsequent layer of varnish. However, it is not visible as a separate stratum. This effect may be explained by the strong absorption of light in the thick body of this paint (confirmed by inspection of the image in Figure 6a), which, together with signal loss in the overpainting located on the painting surface, renders it impossible to see much of anything below the varnish layer. Taken together, the detailed conclusions concerning the inscriptions in this picture lead to the supposition that the work may have been painted at some time in the year after Leonard’s beatification in 1796. At that time, the picture carried the inscription B. Leonardus d.[a?] Maurzio, with both the date and the caption covered by a thick multiple layer of varnishing. Subsequently, after canonization, the original inscription was painted over and a new, updated, one added. The date was left visible, possibly even emphasized by overpainting it. 834


826-836

June 2010

Vol. 43, No. 6

In the second case study, Portrait of an unknown woman, the OCT examination shows two layers of varnish (Figure 7): a clear lower layer and a more scattering upper one. This confirms the results of the inspection of the painting’s surface by UV-excited fluorescence (Figure 3), which had revealed the existence of two different kinds of varnish. The primary, probably original, varnish had been partially removed, but its residue is clearly seen in Figure 7. There is no doubt that the signature lies on top of the primary layer of varnish, because the thin opaque paint layer of the signature (between arrows in Figure 7) is located at precisely the same level as the interface between the two varnishes. Moreover, although the paint layer of the signature obstructs deeper penetration of the beam, some traces of the original paint layer (marked with green arrows in Figure 7) are visible underneath at a distance corresponding to the thickness of the primary varnish in the proximity. This analysis suggests that the signature is a forgery: it leads to the strong suspicion that removal of the original varnish, of yellow UV-excited fluorescence, was intentional, and carried out in order to create the impression that the forged signature was a genuine original one placed directly on the paint layer before any varnishing.

Summary The aim of this Account has been to show, using examples of studies of original paintings, the potential of optical coherence tomography for solving real problems arising in art conservation studios. The investigation of inscriptions on oil paintings was specifically discussed. For two examples of such paintings, it has been demonstrated that examination by OCT permits the localization of particular paint layers within the stratigraphy of the thin structure of the paintings. If this structure is transparent enough to the infrared light used for the examination, OCT may constitute the optimal method of choice for cross-sectional examination because of its noninvasiveness, which allows its unrestricted use even in such sensitive areas as those of the artist’s signature. Despite the very specific subject of the studies reported here, it is evident that


the kind of approach described in this Account may easily be adaptable to other appropriately similar tasks in forensic or materials science. In the authors’ opinion, the most promising approach is a synergistic one in which the information gained by means of various different noninvasive methods is combined. Among these methods, OCT shows great potential for further application and development. The authors are grateful to Dr. Robert Dale for critical reading of the manuscript and a number of useful comments. This work was supported by Polish Government Research Grants through the years 2008-2011. M.I., E.A.K., and M.S. gratefully acknowledge additional support from the European Social Fund and the Polish Government within their Integrated Regional Development Operational Program, Action 2.6, under the project “Stypendia dla doktoranto´w 2008/2009 - ZPORR” of the Kuyavian-Pomeranian Voivodship. M.I. additionally acknowledges support from the Foundation for Polish Science within the Ventures Program cofinanced by Operational Programme Innovative Economy within the European Regional Development Fund. BIOGRAPHICAL INFORMATION Piotr Targowski is a Professor of Optics and Informatics in the Institute of Physics of the Nicolaus Copernicus University. Magdalena Iwanicka is a Ph.D. student in the Institute for the Study, Restoration and Conservation of Cultural Heritage of the Nicolaus Copernicus University and a Foundation of Polish Science scholar. Ludmiła Tymin´ska-Widmer is an assistant lecturer at the Institute for the Study, Restoration and Conservation of Cultural Heritage of the Nicolaus Copernicus University. Marcin Sylwestrzak is a Ph.D. student in the Institute of Physics of the Nicolaus Copernicus University. Ewa A. Kwiatkowska is a Ph.D. student in the Institute of Physics of the Nicolaus Copernicus University. REFERENCES 1 de la Rie, E. R. The influence of varnishes on the appearance of paintings. Stud. Conserv. 1987, 32, 1–13. 2 Berns, R. S.; de la Rie, E. R. Exploring the optical properties of picture varnishes using imaging techniques. Stud. Conserv. 2003, 48, 73–82. 3 Plesters, J. Cross-sections and chemical analysis of paint samples. Stud. Conserv. 1956, 2, 110–157. 4 van der Weerd, J. Microspectroscopic analysis of traditional oil paint, Ph.D. Thesis, FOM Institute for Atomic and Molecular Physics, Amsterdam, 2002 http://www.amolf.nl/publications/theses/after-2000/ (accessed 07/03/2009). 5 Rouba, B.; Karaszkiewicz, P.; Tymin´ska-Widmer, L.; Iwanicka, M.; Go´ra, M.; Kwiatkowska, E.; Targowski, P. Optical coherence tomography for non-destructive investigations of structure of objects of art. Presented at the 9th International Conference on Non Destructive Testing of Art, Jerusalem, Israel, May 25-30, 2008, http://www.ndt.net/article/art2008/papers/143Targowski.pdf (accessed 07/ 03/2009).

6 de la Rie, E. R. Fluorescence of paint and varnish layers, Part I, II, III. Stud. Conserv. 1982, 17, 1-7; 65-69; 102-108. 7 Thoury, M.; Elias, M.; Frigerio, J. M.; Barthou, C. Nondestructive varnish identification by ultraviolet fluorescence spectroscopy. Appl. Spectrosc. 2007, 61, 1275–1282, 10.1366/000370207783292064. 8 Dupuis, G.; Elias, M.; Simonot, L. Pigment identification by fiber-optics diffuse reflectance spectroscopy. Appl. Spectrosc. 2002, 56, 1329–1336, 10.1366/000370202760354803. 9 Van Asperen de Boer, J. Infrared reflectography: A method for the examination of paintings. Appl. Opt. 1968, 7, 1711–1714. 10 Gavrilov, D.; Ibarra-Castanedo, C.; Maeva, E.; Grube, O.; Maldague, X.; Maev, R. G. Infrared methods in noninvasive inspection of artwork. Presented at 9th International Conference on NDT of Art, Jerusalem, Israel, May 25-30, 2008, http://www. ndt.net/article/art2008/papers/040Gavrilov.pdf (accessed 07/03/2009). 11 Maev, R. G.; Gavrilov, D.; Maeva, A.; Vodyanoy, I. Modern non-destructive physical methods for paintings testing and evaluation. Presented at 9th International Conference on Non Destructive Testing of Art, Jerusalem, Israel, May 25-30, 2008, http://www.ndt.net/article/art2008/papers/042Maev.pdf (accessed 07/03/ 2009). 12 Tornari, V. Optical and digital holographic interferometry applied in art conservation structural diagnosis. e-Preserv. Sci. 2006, 3, 51–57. 13 Complete list of papers on application of OCT to examination of artwork may be found at http://www.oct4art.eu (accessed 07/03/2009). 14 Targowski, P.; Rouba, B.; Wojtkowski, M.; Kowalczyk, A. The application of optical coherence tomography to non-destructive examination of museum objects. Stud. Conserv. 2004, 49, 107–114. 15 Liang, H.; Cid, M.; Cucu, R.; Dobre, G.; Podoleanu, A.; Pedro, J.; Saunders, D. Enface optical coherence tomography-a novel application of non-invasive imaging to art conservation. Opt. Express 2005, 13, 6133–6144, 10.1364/OPEX.13.006133. 16 Arecchi, T.; Bellini, M.; Corsi, C.; Fontana, R.; Materazzi, M.; Pezzati, L.; Tortora, A. A new tool for painting diagnostics: Optical coherence tomography. Opt. Spectrosc. 2006, 101, 23–26, 10.1134/S0030400X06070058. 17 Latour, G.; Georges, G.; Siozade, L.; Deumie, C.; Echard, J. P. Study of varnish layers with optical coherence tomography in both visible and infrared domains. Proc. SPIE 2009, 7391, 73910J, 10.1117/12.827856. 18 Latour, G.; Moreau, J.; Elias, M.; Frigerio, J.-M. Optical Coherence Tomography: non-destructive imaging and spectral information of pigments. Proc. SPIE 2007, 6618, 661806, 10.1117/12.726084. 19 Yang, M. L.; Lu, C. W.; Hsu, I. J.; Yang, C. C. The use of optical coherence tomography for monitoring the subsurface morphologies of archaic jades. Archaeometry 2004, 46, 171-182, 10.1111/j.1475-4754.2004.00151.x. 20 Yang, M.-L.; Winkler, A. M.; Barton, J. K.; B., V. P. Using optical coherence tomography to examine the subsurface morphology of Chinese glazes. Archaeometry 2008, 51, 808–821. 21 Liang, H.; Peric, B.; Hughes, M.; Podoleanu, A. G.; Spring, M.; Roehrs, S. Optical coherence tomography in archaeological and conservation science - a new emerging field. Proc. SPIE 2008, 7139, 713915, 10.1117/12.819499. 22 Kunicki-Goldfinger, J.; Targowski, P.; Go´ra, M.; Karaszkiewicz, P.; Dzierzanowski, P. Characterization of glass surface morphology by optical coherence tomography. Stud. Conserv. 2009, 54, 117–128. 23 Adler, D. C.; Stenger, J.; Gorczynska, I.; Lie, H.; Hensick, T.; Spronk, R.; Wolohojian, S.; Khandekar, N.; Jiang, J. Y.; Barry, S. Comparison of three-dimensional optical coherence tomography and high resolution photography for art conservation studies. Opt. Express 2007, 15, 15972–15986, 10.1364/OE.15.015972. 24 Liang, H.; Cid, M.; Cucu, R.; Dobre, G.; Kudimov, B.; Pedro, J.; Saunders, D.; Cupitt, J.; Podoleanu, A. Optical coherence tomography: A non-invasive technique applied to painting conservation of paintings. Proc. SPIE 2005, 5857, 261–269. 25 Targowski, P.; Go´ra, M.; Wojtkowski, M. Optical coherence tomography for artwork diagnostics. Laser Chem. 2006, 2006, 1–11, 10.1155/2006/35373. http://www. hindawi.com/journals/lc/2006/035373.abs.html, (accessed 07/03/2009). 26 Go´ra, M.; Targowski, P.; Rycyk, A.; Marczak, J. Varnish ablation control by optical coherence tomography. Laser Chem. 2006, 2006, 1–7, 10.1155/2006/10647. http://www.hindawi.com/journals/lc/2006/010647.abs.html, (accessed 07/03/ 2009). 27 Targowski, P.; Rouba, B.; Go´ra, M.; Tymin´ska-Widmer, L.; Marczak, J.; Kowalczyk, A. Optical coherence tomography in art diagnostic and restoration. Appl. Phys. A: Mater. Sci. Process. 2008, 92, 1–9, 10.1007/s00339-008-4446-x. 28 Go´ra, M.; Targowski, P.; Kowalczyk, A.; Marczak, J.; Rycyk, A. Fast spectral optical coherence tomography for monitoring of varnish ablation process. In Lasers in the Conservation of Artworks, LACONA VII Proceedings, Madrid, Spain, Sept. 17-21, 2007; Castilleo, M., Ed.; Taylor & Francis Group: London, 2008; pp 23-26. 29 Targowski, P.; Go´ra, M.; Bajraszewski, T.; Szkulmowski, M.; Rouba, B.; ŁekawaWysłouch, T.; Tymin´ska, L. Optical coherence tomography for tracking canvas

Vol. 43, No. 6

June 2010

826-836


835


30 31

32

33

deformation. Laser Chem. 2006, 2006, DOI: 10.1155/2006/93658, http://www. hindawi.com/journals/lc/2006/093658.abs.html (accessed 07/03/2009). Optical Coherence Tomography: Technology and Applications; Drexler, W., Fujimoto, J. G., Eds.; Springer-Verlag: Berlin, Heidelberg, New York, 2008. Dubois, A.; Grieve, K.; Moneron, G.; Lecaque, R.; Vabre, L.; Boccara, C. Ultrahighresolution full-field optical coherence tomography. Appl. Opt. 2004, 43, 2874– 2883. Dubois, A.; Moreau, J.; Boccara, C. Spectroscopic ultrahigh-resolution full-field optical coherence microscopy. Opt. Express 2008, 16, 17082–17091, 10.1364/ OE.16.017082. Fercher, A. F. Optical coherence tomography. J. Biomed. Opt. 1996, 1, 157–173, 10.1117/12.231361.

836


826-836

June 2010

Vol. 43, No. 6

34 Leitgeb, R.; Hitzenberger, C. K.; Fercher, A. F. Performance of Fourier domain vs. time domain optical coherence tomography. Opt. Express 2003, 11, 889–894. 35 Sylwestrzak, M.; Kwiatkowska, E. A.; Karaszkiewicz, P.; Iwanicka, M.; Targowski, P. Application of graphically oriented programming to imaging of structure deterioration of historic glass by optical coherence tomography. Proc. SPIE 2009, 7391, 739109, 10.1117/12.827520. 36 Bihl, M. St Leonard of Port Maurice. The Catholic Encyclopedia 1910, Vol. 9, http:// www.newadvent.org/cathen/09178c.htm (accessed 07/03/2009). 37 Stifter, D.; Sanchis Dufau, A. D.; Breuer, E.; Wiesauer, K.; Burgholzer, P.; Ho¨glinger, O.; Go¨tzinger, E.; Pircher, M.; Hitzenberger, C. K. Polarisationsensitive optical coherence tomography for material characterisation and testing. Insight - Non-Destruct. Test. Cond. Monit 2005, 47, 209–212, 10.1784/insi.47.4.209.63154.

Fluorescence Spectroscopy: A Powerful Technique for the Noninvasive Characterization of Artwork ALDO ROMANI,*,†,‡ CATIA CLEMENTI,‡ COSTANZA MILIANI,§ AND GIANNA FAVARO† †

Dipartimento di Chimica, Universita` di Perugia, Via Elce di Sotto, 8, 06123 - Perugia, Italy, ‡Centro SMAArt, c/o Dipartimento di Chimica, Universita` di Perugia, Perugia, Italy, §Istituto CNR di Scienze e Tecnologie Molecolari (ISTM), c/o Dipartimento di Chimica, Universita` di Perugia, Perugia, Italy RECEIVED ON DECEMBER 23, 2009

CON SPECTUS

A

fter electronic excitation by ultraviolet or visible radiation, atoms and molecules can undergo thermal or radiative deactivation processes before relaxing to the ground state. They can emit photons with longer wavelengths than the incoming exciting radiation, that is, they can fluoresce in the UV-vis-near-infrared (NIR) range. The study of fluorescence relaxation processes is one of the experimental bases on which modern theories of atomic and molecular structure are founded. Over the past few decades, technological improvements in both optics and electronics have greatly expanded fluorimetric applications, particularly in analytical fields, because of the high sensitivity and specificity afforded by the methods. Using fluorimetry in the study and conservation of cultural heritage is a recent innovation. In this Account, we briefly summarize the use of fluorescencebased techniques in examining the constituent materials of a work of art in a noninvasive manner. Many chemical components in artwork, especially those of an organic nature, are fluorescent materials, which can be reliably used for both diagnostic and conservative purposes. We begin by examining fluorimetry in the laboratory setting, considering the organic dyes and inorganic pigments that are commonly studied. For a number of reasons, works of art often cannot be moved into laboratories, so we continue with a discussion of portable instruments and a variety of successful “field aplications” of fluorimetry to works of cultural heritage. These examples include studies of mural paintings, canvas paintings, tapestries, and parchments. We conclude by examining recent advances in treating the data that are generated in fluorescence studies. These new perspectives are focused on the spectral shape and lifetime of the emitted radiation. Recent developments have provided the opportunity to use various spectroscopic techniques on an increasing number of objects, as well as the ability to fully characterize very small amounts of sample, either in a laboratory setting or on site. Thus, a new technological highway is open to scientists; it is still difficult to navigate but offers an enormous potential for investigating objects without touching them. Fluorescence spectroscopy is one of the most important of these techniques.

1. Introduction

servation science has been exploited by others. A

In 1982, Rene` de la Rie first investigated the

number of papers concern the understanding of

potential of fluorescence spectroscopy for the non-

the chemical-physical properties of the molecules

invasive characterization of art materials such as

by applying luminescence techniques to standard

1

binders, pigments, and dyes. After his pioneer-

laboratory models and pictorial replicas strictly

ing studies, the fluorescence spectroscopy in con-

reproducing works of art. Other works are more

Published on the Web 04/26/2010 www.pubs.acs.org/acr 10.1021/ar900291y © 2010 American Chemical Society

Vol. 43, No. 6

June 2010

837-846


837

Fluorescence Spectroscopy Romani et al.

specifically focused on testing the potential of the different fluorescence-based techniques to obtain diagnostic information in a noninvasive manner, by developing and projecting new instrumentation or experimental setups. These two main pathways are closely connected since the instrumental output recorded from a diagnostic measurement on an original work of art should take into account the chemical-physical properties of the involved systems. In a recent review, the relevance of fluorescence spectroscopy as a powerful analytical method for the diagnostics of organic dyes in cultural objects has been highlighted.2

2. Laboratory Fluorimetry: Knowledge of Materials and Diagnostics A characterization of the chemical components of artwork should be the first step for the application of luminescencebased techniques in the cultural heritage field. Spectroscopic studies may provide several pieces of information on the molecular properties of materials used in artwork. Steady-state UV-vis absorption and fluorescence spectroscopy supply information about the electronic excitation energy and the nature of the excited states. Time-resolved techniques provide information concerning their relaxation pathways. Preliminary investigations have been carried out in solution, homogeneous medium where quantitative determination of parameters, such as fluorescence quantum yields and lifetimes, can be easily achieved and the effects of changes in environment, pH, or different additives can be controlled. Thereafter, laboratory samples, mimicking multiple execution techniques, have been studied, and the results have been interpreted in light of those obtained in solution. Absorption and luminescence measurements provide complementary information; however, emission is generally more sensitive and selective because detection of each single fluorophore can be achieved using the appropriate excitation wavelength, thus resulting a suitable analytical technique for multicomponent analysis. Moreover, fluorescence excitation spectra may help in recognizing different fluorophores, sometimes present as traces, and unrecognizable by other conventional analytical techniques. Commercial spectrofluorimeters were used for investigations in solutions, whereas, to obtain emission spectra from surfaces out of the fluorimeter sample holder, measurements were carried out using fiber optic systems. To measure fluorescence lifetimes in the laboratory, commercial instruments 838


837-846

June 2010

Vol. 43, No. 6

based on the phase-shift method or on the time-correlated single-photon counting technique have been used. An alternative luminescence-based instrumentation for the diagnostics of artwork is laser-induced fluorescence spectroscopy (LIF),3 consisting of a laser excitation source, optics (lenses/fiber optics), a spectrometer, and a sensitive CCD detector. By using an intensified charge-coupled device (ICCD) as detector or a multichannel analysis (MCA) device,4 time-resolved measurements can be performed. 2.1. Naturally Occurring Organic Dyes. Among the chemical components of artwork, organic dyes are the best candidates to be recognized by fluorimetric methods since they often exhibit fairly intense luminescence. In this context, colorants used in artistic paintings or miniatures or for dyeing precious textiles of the classes of hydroxyanthraquinones,5-9 flavonols,10 phenoxazines,11 carotenoids,12 and indigoids5,13 have been investigated in solution under various experimental conditions by using stationary and time-resolved spectrophotometric and fluorimetric techniques. Yellow colorants, such as flavonols, absorb at λ e 400 nm and fluoresce in the 420-500 nm wavelength range. Yellow-orange carotenoids absorb in the 430-450 nm range and exhibit emissions markedly shifted to the red, 650-700 nm. Typical red colorants, belonging to the family of hydroxyanthraquinones, widely diffused in nature, absorb around 500 nm and emit in a wide spectral region (550-650 nm). Purple color was often obtained using orcein,14 a natural dye that is a mixture of several reddish colorants of the family of phenoxazines. Its absorption ranges from 500 to 570 nm. To the best of our knowledge, orcein is the unique purple color dye that exhibits fluorescence emission that occurs at about 590 nm in solution. Colorants absorbing in the 600 nm region, typically indigoid dyes, are blue; their emission in solution is rather weak and specular to the absorption band. In each class of colorants, substituents but also the environment can markedly affect the emissive properties. Some significant examples of these effects will be described in the following. Flavonoid and anthraquinoid colorants are polyhydroxyl derivatives of molecular colorless structures, flavone and anthraquinone (Scheme 1), which absorb visible light and, therefore, appear yellow and red colored. The number and position of hydroxyls, as well as the surrounding environment, may dramatically affect the fluorescence features. In fact, depending on the above, dual


SCHEME 1

SCHEME 2

FIGURE 2. Emission spectra of the luteolin-Al3+ adducts at increasing concentrations of Al3+ ions: (1) [Al3+] ) 0; (2) [Al3+] ) 5 × 10-6; (3) [Al3+] ) 2.5 × 10-5; (4) [Al3+] ) 4× 10-4 mol dm-3.

red as the pH increases, due to stabilization of anionic forms,6,7,10,18,19 Figure 1. Chelation with metal ions has an effect similar to deprotonation since it induces fairly intense fluorescence emission. For example, luteolin (5,7,3′,4′-tetrahydroxyflavone), the emisfluorescence emissions via intramolecular proton transfer in the excited state (ESPT)15-17 may arise. Moreover, fluorescence shifts and changes in intensity may occur due to either deprotonation, self-aggregation, or interaction with other molecular or ionic species able to form molecular complexes. ESPT can occur from a OH group to a carbonyl oxygen in ortho position, thus originating a tautomeric form. From the double minimum energy surfaces of Scheme 2, it can be seen that after absorption, the molecule can emit from the tautomeric form with a marked bathochromic shift. Occurrence of ESPT for alizarin (1,2-hydroxyanthraquinone) but not purpurin (1,2,4-hydroxyanthraquinone) justifies the more intense emission of the latter.6 Fluorescence generally is weak and often undetectable for the neutral forms but becomes more intense or shifts to the

sion of which is undetectable in the molecular neutral form, fluoresces up to unitary quantum yield attainment upon additions of Al3+ ions,18 Figure 2. The intense fluorescence of chelates explains the light-fastness of the colorant on textiles dyed using a mordant,20 since the absorbed light is mostly given up as emission, reducing photochemical degradation. Metal complexation of hydroxyl dyes is used to obtain lakes, insoluble organic pigments used in painting. Emission spectra of lakes, prepared from weld (Reseda luteola L.) following different recipes,21 allow one to distinguish lakes obtained by different procedures based on the shift of the emission maximum, Figure 3. Contributions to fluorescence of residual chlorophylls from dye extraction are also evident in lakes 1 and 2 at 663 and 675 nm, respectively.21 Although

FIGURE 1. (a) Example of alizarin fluorescence red shift upon deprotonation (normalized spectra): neutral (pH ) 2, s), monoanion (pH ) 9, · · · ), and dianion (pH ) 14, ---) forms. (b) Example of enhancement of the morin fluorescence upon deprotonation: neutral (pH ) 7.5, s), monoanion (pH ) 10, · · · ), and dianion (pH ) 13, ---) forms.

Vol. 43, No. 6

June 2010

837-846


839


FIGURE 3. Normalized emission spectra (λexc ) 366 nm) of powdery weld lakes (lake 1, s; lake 2, · · · ; lake 3, ---) prepared according to three different historical recipes.

it is not probable to find chlorophylls in ancient works of art, due to their instability, this confirms the selectivity of the technique. Studies of purpurin and alizarin lakes, spread on different supports and with various binders, have been also carried out using confocal microfluorescence spectroscopy by Melo et al.22 The results obtained by the authors showed microfluorescence to be a promising analytical tool for the identification of red lake pigments combined with various binding media. Of particular interest was the study of dyed textiles where, in addition to the colorant, a mordant (generally alum) may be present. For example, alizarin dyed wool samples8 do not fluoresce in the absence of mordant. When alum mordant is present (Figure 4, top), the emission matches that detected in solution at pH ≈ 9 (monoanion). Silk and wool samples dyed with orcein showed the fluorescence of both the fiber (∼440 nm) and the colorant (∼630 nm). These samples were also used to reproduce the changes that textiles undergo in time when irradiated with a 175 W xenon lamp for 100 h in order to produce an accelerated aging, which induces decoloration accompanied by decrease of dye emission and increase of fiber emission, Figure 4, middle.8 The aged sample emission was exploited to identify orcein on an original fragment of Renaissance tapestry from Bruxelles (16th century), Figure 4, bottom.11 Another interaction affecting fluorescent properties is selfaggregation, which generally results in a bathochromic shift of the absorption and emission spectra. Detected for indigo in concentrated solution (λmax ) 720 nm) and in powders (λmax ) 750 nm),5 its knowledge was useful to recognize indigo on tapestry (see below). 2.2. Inorganic Pigments. Since the 1990s, fluorescence spectroscopy has been applied to organic materials in the field 840


837-846

June 2010

Vol. 43, No. 6

FIGURE 4. Left column, top, reflectance emission spectrum (s, λexc ) 440 nm) and excitation spectrum of the emission ( · · · , λem ) 620 nm) of a wool sample dyed with alizarin in the presence of mordant; middle, emission of a natural orcein dyed silk sample before (black) and after (gray) 100 h accelerated aging; bottom, comparison between normalized emission spectra from the original tapestry fragment (black) and the orcein-dyed wool standard aged for 100 h (gray). Right column, alizarin dyed wool without (a) and with (b) mordant, orcein dyed silk before (c) and after (d) aging, and (e) original 16th century tapestry fragment.

of cultural heritage; only recently has attention also been devoted to luminescent inorganic pigments. They include ancient synthetic compounds, like some copper based silicates, and more recent semiconductor materials, zinc oxide and cadmium sulfoselenides. While emissions of organic molecules are generally due to π* f π electronic transitions, the color and the luminescence of inorganic compounds may originate from d-d transitions, charge transfer transitions, or transitions between the conduction and the valence band in semiconductors. Egyptian blue is considered to be the first synthetic pigment; it was widely used in the Mediterranean basin from the Fourth Dynasty in Egypt until the end of the Roman period and beyond. Based on X-ray diffraction data, the currently accepted chemical structure for Egyptian Blue corresponds to the naturally occurring mineral cuprorivaite. The absorption and luminescence properties of this pigment have already been assigned by Pozza and co-workers.23 Recently reinvestigated, it revealed the highest quantum efficiency of luminescence (10.5%) ever found for a fluorophore emitting in the


low and red organic pigments whose emissions typically fall in the visible. For the diagnostics of inorganic pigments in works of art, LIF has also been employed to examine a set of CdS and CdSe sulfide-based pigments in some oil painting replicas. The technique was suitable for differentiating among various cadmium pigments and identifying individual components in mixtures.3 FIGURE 5. Diffuse reflectance (a) and normalized luminescence (b) spectra (λexc ) 500 nm) of cadmium-based pigments of different chemical composition.

800-1100 nm range.24 This characteristic has important diagnostic implications in the field of cultural heritage. In fact, Egyptian blue can be easily identified by means of both luminescence imaging25 and spectrofluorimetry (see below, Domus Aurea), because its intense NIR emission does not overlap signals of other luminescent pigments. Among more recent materials, a peculiar emission behavior is shown by zinc oxide. This compound, known and produced since the Roman age, was first used as an antiinflammatory agent. After the mid-19th century, it became a white pigment widely used in modern and contemporary art. Zinc oxide, in the hexagonal wurtzite structure, is a semiconductor with a band gap energy of 3.3 eV at room temperature. Two emissions are reported in the literature, one in the UV and the other in the visible. The UV emission, assigned to the band gap transition, is associated with exciton annihilation, whereas the broad band in the visible, called the green emission, seems to be related to oxygen vacancies.26 When zinc oxide is applied on a painted surface with a binder, the green emission is generally undistinguishable from the binder emission; nevertheless, the band gap can be assumed as a fingerprint of zinc oxide presence. Among semiconductors, cadmium sulfide-based pigments have received particular attention by artists since the early 19th century thanks to the wide variety of shades they offer, from yellow, through orange and red, to maroon. Cadmium sulfide (CdS) is a yellow pigment with a band gap energy of 2.42 eV (513 nm) at room temperature.27 The corresponding transition is due to direct electron-hole recombination, while a second transition, in the 670-800 nm range, known as red emission, is related to a deep-trap resulting from structural defects or impurities like sulfur vacancies.28 The replacement of different ions in the lattice produces a variation of band gap energy that reflects in color (Figure 5a) and emission maximum (Figure 5b) changes. From a diagnostic point of view, it is important that several cadmium-based pigments emit in the NIR and can be thus distinguished from yel-

3. Portable Fluorimetry The investigation of works of art that cannot be moved from their location due to their dimension or for insurance purposes necessitates on-site measurements through the use of portable instrumentation. Under the stimulation of results obtained in diagnostics using laboratory luminescence-based techniques and taking advantage of the technological development in the miniaturization of almost all instrumental parts, a great effort has been addressed to create portable instruments suitable for in situ measurements. 3.1. The Instruments. A portable fluorimeter has been assembled consisting of a xenon lamp as an excitation source, a monochromator for selecting the excitation wavelength, and a fiber-optic cable that directs the exciting light on the surface under study. The emission signals are revealed and transferred to the detector passing through suitable filters, if necessary.11,29 A further improvement in the luminescence-based techniques applied to cultural heritage diagnostics was the realization of a patented prototype of a portable instrument, based on the time-correlated single-photon counting (TCSPC) method, purposely assembled for on-site measurements of luminescence lifetimes on artwork surfaces.30 The usefulness of coupling steady-state and time-resolved luminescence techniques in the diagnostics of artwork emerges when different colorants having similar emission spectra have to be recognized. For example, two structurally similar red dyes belonging to the anthraquinoid family, laccaic acid and carminic acid, have fluorescence maxima at the same wavelength (∼625 nm). They come from far geographical areas and are extracted from different insects (laccaic acids from Kerria lacca Kerr, an Indian insect; carminic acid from Dactylopius coccus Costa, a Mexican insect). Even if not distinguishable from spectra, they can be distinguished on the basis of their different lifetimes (τ1 ) 0.9, τ2 ) 2.6, and τ1 ) 1.5, τ2 ) 5.0, respectively).31,32 Other fluorescence techniques, such as laser imaging detection and ranging (LIDAR), used with application of fluoresVol. 43, No. 6

June 2010

837-846


841


FIGURE 6. Right, decoration details of the Gilded vault room of Domus Aurea. Left top, absorption ( · · · ) and emission (s, λexc ) 540 nm) spectra collected in the blue area. Inset: video-microscopy image (magnification 50×) of the blue area. Left bottom, emission spectra (gray, λexc ) 440 nm; black, λexc ) 540 nm) collected on the purple area. Inset, video-microscopy image (magnification 50×) of the purple area.

cence-multispectral imaging,33,34 and fluorescence lifetime imaging and spectroscopy,4 are amenable to be set up in a portable format. 3.2. The Applications. During the past few years several interventions on original works of art have been carried out using portable instrumentation within the transnational access activity of MOLAB.35 In the following, meaningful applications of portable fluorimetry to objects of artistic and historical importance (painting, tapestry, parchment) are reported. 3.2.1. Mural Paintings: Domus Aurea. The Domus Aurea (Golden House) is the largest imperial palace built in the heart of Ancient Rome by the Roman emperor Nero after the Great fire of Rome in 64 A.D. The huge complex is unique for its architectural composition and richness in fresco paintings and stuccoes that burial hid and preserved for centuries. The composition and execution techniques of the precious decorations were, for the first time, studied in 2007, during a MOLAB access.35 In some blue decorations of the Gilded vault room, the absorption and emission features of Egyptian blue were observed (Figure 6). The colorant could be identified even when it was mixed with other luminescent pigments like hydroxy-anthraquinone red lakes that were found in some purple-violet areas. 3.2.2. Canvas Paintings: Victory Boogie-Woogie (Piet Mondrian, 1942-1944). The Victory Boogie-Woogie is the last unfinished work by Piet Mondrian, which is recognized as one of the most important works of modern art. The paint842


837-846

June 2010

Vol. 43, No. 6

FIGURE 7. Mondrian’s Victory Boogie-Woogie. Left column, fluorescence emission spectra of a (top) white rectangle, (middle) a yellow rectangle, and (bottom) a red painted tape on a blue rectangle. Right column, visible light (a, c, e) and UV-fluorescence (b, d, f) images corresponding to the areas where the spectra were obtained.

ing consists of 574 colored areas, most of which are painted and others made up of composite materials such as paper, commercial tapes, homemade tapes, and cellophane. The study of the pigments employed in this masterpiece is a relevant example of a multitechnique on-site noninvasive approach to modern materials.36 The UV-induced fluorescence image of the surface highlighted different luminescence emissions in localized areas. Fluorescence spectral analysis of some white paints (apparently homogeneous) revealed the presence of two different emissions, Figure 7, top. XRF measurements carried out on white paints recognized the presence of zinc, titanium, and barium. Titanium and barium can be related to the use of titanium dioxide and barium sulfate, the latter also identified by mid-FTIR spectroscopy. The presence of zinc may be due to ZnS and ZnO whose discrimination is by far more complicated. However, UV-vis fluorescence spectroscopy allowed for the identification of zinc oxide through the characteristic band gap emission. In some yellow squares, where cadmium and sulfur were revealed by XRF measurements, red fluorescent spots were detected under UV light (Figure 7, middle). The red fluorescing areas showed an intense emission at about 700 nm,


FIGURE 8. The portable fluorimeter at work on the tapestry Earthquake in Filippi.

which is related to a deep-trap transition resulting from structural defects or impurities in the cadmium sulfide pigment. According to elemental analyses, red squares are rich in cadmium, selenium, and sulfur, suggesting the use of sulfoselenide cadmium red. Fluorescence spectra showed in fact a broad emission centered at about 810 nm, due to structural defects or impurities. A different emission was observed on several transparent cellophane tapes, probably painted in red and glued by Mondrian himself. An emission maximum at 620 nm and the absence of key elements suggest the presence of an organic pigment, Figure 7, bottom. 3.2.3. Tapestries. In-situ fluorimetry was applied to the Renaissance tapestries, designed by Raffaello Sanzio, which are exposed in the Gallery of Tapestries at the Vatican Museum. Most colorants used for dyeing fibers were recognized through their luminescence.37 With a 350 nm excitation, a band around 450 nm was always present, and it was assigned to the wool fiber. Different color regions were analyzed on the observe and reverse sides; the latter was often preferred due to partial bleaching of color on the externally exposed side. Three tapestries (Earthquake in Filippi, Emmaus’ supper, and Listra’s Sacrifice) were analyzed; the portable fluorimeter at work on Earthquake in Filippi is shown in Figure 8. Red-purple regions of the three tapestries exhibited emission maxima in the 624-630 nm range, which were assigned to orcein by comparison with laboratory standards (Figure 9, top). A modest blue shift of the emission from the tapestries with respect to the standard was attributed to aging, based on comparison with aged standards. Emission spectra monitored on blue regions (350 nm excitation), in addition to the band of the fiber, exhibited a batochromic, very weak band (λmax ≈ 750 nm), Figure 9, mid-

FIGURE 9. Top, emission spectra (λexc ) 350 nm) monitored in purple regions on the reverse side of different tapestries, compared with the emission of an aged wool laboratory sample dyed with orcein (orange dotted line). Middle, emission spectra monitored on the reverse of the blue frame decoration of Listra’s Sacrifice tapestry (λexc ) 350 nm). Inset, zoom on the long wavelength region (λexc ) 650 nm); the signal from the warp is completely flat (black line). Bottom: emission spectrum monitored on green regions (λexc ) 350 nm) of the frame decoration of the Earthquake in Filippi tapestry, front side.

dle, which was assigned to the colorant. By comparison with laboratory samples, this emission was assigned to indigo. Two emissions were found in green areas; one centered at 500 nm and the other, much weaker, at 750 nm (Figure 9, bottom). The latter is due to indigo, while the hypsochromic one corresponds to emission from some yellow dye, probably a flavonol, partially overlapped to the emission of the fiber. This indicates that green color was obtained by a mixture of some yellow dyes and the blue indigo. 3.2.4. Parchment: The Book of Kells. The precious Book of Kells is an early medieval manuscript from around the 800, held at the Trinity College Library of Dublin. Investigations were carried out on selected parchment folios of the Book, coupling the portable fluorimeter with the time-correlated single-photon counting equipment.31 Two colorants, blue and purple, were recognized from their emission spectra and lifetimes. The blue colorant showed a very weak emission at about 730 nm, matching that of a laboratory standard of natural indigo in Arabic gum, while the emission spectrum of the Vol. 43, No. 6

June 2010

837-846


843


purple colorant (∼630 nm) was consistent with the fluorescence of orcein.11 These assignments were confirmed by lifetime determinations (τindigo ) 2.4 ns; τorcein ) 2.2 ns).

4. New Perspectives in Data Treatment Generally, experimental data have to be treated and elaborated in order to obtain the widest information concerning the object of interest, especially when analyzing a complex matrix such as a painted layer. Therefore, several works have been focused on searching for the right approach and the best interpretation of the instrumental raw data. Recent advances in this field have been orientated in two directions: the spectral shape and the lifetime of luminescence. 4.1. Corrections of Emission Spectra. In reading emission spectra, the most important improvements have concerned a correct definition of fluorescence maxima to identify either the colorants or the surrounding substances. For identifying on-site some different natural varnishes, the use of both maximum position and full width at half-maximum of the emission spectra as a function of the excitation wavelength have been considered.38 Recognition of binders and varnishes has been also carried out using laser-induced fluorescence coupled with total emission analysis.39 Concerning colorants, self-absorption effect, that is, absorption by the colorant itself of the emitted light, thus erasing a portion of the emission spectrum on the short wavelengths side, is the main source of uncorrected maximum wavelength reading. This effect can be obviated in solution by decreasing the probe concentration, but it is unavoidable for fluorophores in the solid state. The problem of the correction for self-absorption of fluorescence spectra collected on pictorial works has been recently addressed, and a method for the treatment of the fluorescence signals has been developed.40,41 In Figure 10, an application of the self-absorption correction to several carmine lake emission spectra collected on Pala Albergotti by Vasari is shown. 4.2. Reading of Fluorescence Decays. A kinetic approach to the study of luminescence of artistic objects may have a determining role in the diagnostics of colors on surfaces. Compared with previous works,4 the home-assembled TCSPT portable instrument has introduced important improvements by increasing the time resolution by about 1 order of magnitude and applying more advanced methods for the interpretation of the experimental decay profiles. In the approach, proposed in 2004 by Comelli et al. based on MCA, fluorescence decays had been handled by first-order fitting algorithms. This application, although useful to distinguish 844


837-846

June 2010

Vol. 43, No. 6

FIGURE 10. Top, fluorescence spectra (λexc ) 480 nm). The emission maximum moves from 642 nm on the first spectrum to 657 nm on the sixth. Bottom, normalized fluorescence spectra corrected for self-absorption and multiple scattering phenomena. The correction moves the spectra toward shorter wavelengths making them quite similar. Photos show the examined points.

chemically different fluorophores, is not completely exhaustive when describing the complexity of a painted layer, where several different emitting species could coexist. In the TCSPCbased instrument, multiexponential fit of fluorescence decays allows for a better description of such kinetics. To gain more insights into kinetically complex systems, the maximum entropy method (MEM),42 which considers a distributed set of logarithmic functions, is now being applied to luminescence decay analysis. The heterogeneity of the environment surrounding a fluorophore embedded in a painting layer gives rise to complex decays that can originate from different fluorophores but also from the same chemical species in a different microenvironment. In this context, MEM can be used to recover the right distributions of luminescent lifetimes from the experimental data.43

5. Conclusion Over the last 15 years, the scientific community has provided restorers and conservators with a wide range of completely noninvasive techniques able to disclose several secrets concerning the chemical nature of the materials used in creating works of art in the course of the centuries. Luminescencebased spectroscopic techniques were developed, following different approaches, in order to extract the greatest amount of information on several artwork components such as binders, varnishes, and pigments in paintings and organic dyes in tapestries and manuscripts. At first, each luminescence-based technique was generated and tested as a stand-alone inves-


tigation method, able to provide information for diagnostic purposes. Then, scientists coupled this with other complementary spectroscopic methods (for instance vibrational spectroscopy) or complementary applications in the same field (for example, time-resolved luminescence measurements) to collect the widest data sets to design at best the investigated materials. Consequently, today, to the pressing requests by conservators and restorers for noninvasive diagnostic approaches, each piece of information obtainable from any on-site technique must be taken into account: luminescencebased techniques are among the best candidates for this role. The strong engagement into the double pathway of instrumental and technical development, combined with ever increasing knowledge of material properties, have become general rules of the scientific effort in the world of conservation science, and this is what is going to occur for luminescence techniques applied to works of art. This research was supported by Eu-ARTECH (Contract Eu-ARTECH, RII3-CT-2004-506171) and the Bilateral Joint Research and Technology Programme 2005-2007 ItalyGreece. The authors wish also to thank all co-workers and collaborators who are cited in the references of this Account. BIOGRAPHICAL INFORMATION Aldo Romani graduated in Chemistry at the University of Perugia in 1987 and received his Ph.D. in Chemistry in 1992. Currently, he is a Researcher at the Chemistry Department of the University of Perugia. He has authored about 70 scientific papers concerning both basic and applied subjects principally involving characterization of the excited states of organic molecules by means of the parameters that govern their radiative and nonradiative processes using spectroscopic techniques in absorption and emission. These techniques have been also applied, for nondestructive diagnostic purposes, in the field of the cultural heritage. Catia Clementi graduated in Chemistry at the University of Perugia in 2002 and received her Ph.D. in Chemistry in 2006. She currently holds a postdoctoral position at the Department of Chemistry of the University of Perugia. Her research activity concerns the characterization of the excited states of organic dyes and inorganic pigments of artistic and historical relevance by means of spectroscopic techniques in absorption and emission. She has enhanced her expertise thanks to different internships and collaborations with different European laboratories and museums. Costanza Miliani graduated in Chemistry at the University of Perugia in 1995, there receiving the PhD in Chemistry in 1998. Currently, she is a researcher at the CNR-ISTM (Istituto di Scienze e Tecnologie Molecolari) in Perugia. Her scientific interests are mainly focused on the development and application of noninva-

sive spectroscopies for the study of artwork. She has authored over 40 articles concerning structural, electronic, and vibrational properties of materials of interest for cultural heritage. Gianna Favaro graduated in Chemistry from the University of Padua in 1958; she has been lecturer in Physical Chemistry and related subjects at the Universities of Padua, Bologna, and Perugia. Full professor of Physical Chemistry at the University of Perugia, she has authored over 130 articles concerning electronic spectroscopy, photoinduced energy transfer, photosensitization, photokinetic studies on photochromic compounds, and photochemical and photophysical characterization of materials of interest for cultural heritage. Retired since 2008, she is still involved in several scientific engagements. FOOTNOTES * To whom correspondence should be addressed. E-mail: [email protected]. REFERENCES 1 Rene` de la Rie, E. Fluorescence of paint and varnish layers (Part I). Stud. Conserv. 1982, 27, 1-7; Fluorescence of paint and varnish layers (Part II). Stud. Conserv. 1982, 27, 65-69; Fluorescence of paint and varnish layers (Part III). Stud. Conserv. 1982, 27, 102-108. 2 Degano, I.; Ribechini, E.; Modugno, F.; Colombini, M. P. Analytical methods for the characterization of organic dyes in artworks and historical textiles. Appl. Spectrosc. Rev. 2009, 44, 363–410. 3 Anglos, D.; Solomidou, M.; Zergioti, I.; Zafiropulos, V.; Papazoglou, T. G.; Fotakis, C. Laser-induced fluorescence in artwork diagnostics: an application in pigment analysis. Appl. Spectrosc. 1996, 50, 1331–1334. Anglos, D.; Balas, C.; Fotakis, C. Laser spectroscopic and optical imaging techniques in chemical and structural diagnostics of painted artwork. Am. Lab. (Shelton, CT, U.S.) 1999, 31, 60–67. 4 Comelli, D.; D’Andrea, C.; Valentini, G.; Cubeddu, R.; Colombo, C.; Toniolo, L. Fluorescence lifetime imaging and spectroscopy as tools for nondestructive analysis of works of art. Appl. Opt. 2004, 43, 2175–2183. Comelli, D.; Valentini, G.; Cubeddu, R.; Toniolo, L. Fluorescence lifetime imaging and Fourier transform infrared spectroscopy of Michelangelo’s David. Appl. Spectrosc. 2005, 59, 1174– 1181. 5 Miliani, C.; Romani, A.; Favaro, G. A spectrophotometric and fluorimetric study of some anthraquinoid and indigoid colorants used in artistic paintings. Spectrochim. Acta, Part A 1998, 54A, 581–588. 6 Miliani, C.; Romani, A.; Favaro, G. Acidichromic effects in 1,2- and 1,2,4hydroxyanthraquinones. A spectrophotometric and fluorimetric study. J. Phys. Org. Chem. 2000, 13, 141–150. 7 Favaro, G.; Miliani, C.; Romani, A.; Vagnini, M. Role of protolytic interactions in photo-ageing processes of carminic acid and carminic lake in solution and painted layers. Perkin Trans. 2 2002, 192–197. 8 Clementi, C.; Cibin, F. R.; Romani, A.; Favaro, G. A spectrophotometric and fluorimetric study of red madder on dyed wool. Dyes Hist. Archaeol. 23, in press. 9 Clementi, C.; Nowik, W.; Romani, A.; Cibin, F.; Favaro, G. A spectrometric and chromatographic approach to the study of ageing of madder (Rubia tinctorum L.) dyestuff on wool. Anal. Chim. Acta 2007, 59, 646–654. 10 Romani, A.; Zuccaccia, C.; Clementi, C. An NMR and UV-Visible spectroscopic study of the principal colored component of Stil de grain lake. Dyes Pigm. 2006, 71, 224– 229. 11 Clementi, C.; Miliani, C.; Romani, A.; Favaro, G. In situ fluorimetry: A powerful noninvasive diagnostic technique for natural dyes used in artefacts Part I. Spectral characterization of orcein in solution, on silk and wool laboratory-standards and a fragment of Renaissance tapestry. Spectrochim. Acta, Part A 2006, 64, 906–912. 12 Vickackaite, V.; Romani, A.; Pannacci, D.; Favaro, G. Photochemical and thermal degradation of a naturally occurring dye used in artistic painting. A chromatographic, spectrophotometric and fluorimetric study on saffron. Int. J. Photoenergy 2004, 6, 175–183. 13 Seixas de Melo, J.; Moura, A. P.; Melo, M. J. Photophysical and spectroscopic studies of indigo derivatives in their keto and leuco forms. J. Phys. Chem. A 2004, 108, 6975–6981. 14 Cardon, D. Natural Dyes. Sources, Tradition, Technology and Science; Archetype Publications Ltd.: London, 2007; p 485. 15 Sengupta, P. K.; Kasha, M. Excited state proton transfer spectroscopy of 3hydroxyflavone and quercetin. Chem. Phys. Lett. 1979, 68, 382–385.

Vol. 43, No. 6

June 2010

837-846


845


16 Van Benthem, M. H.; Gillispie, G. D. Intramolecular hydrogen bonding. Dual fluorescence and excited-state proton transfer in 1,5-dihydroxyanthraquinone. J. Phys. Chem. 1984, 88, 2954–2960. 17 Strandjord, A. J. G.; Smith, D. E.; Barbara, P. F. Structural effects on the protontransfer kinetics of 3-hydroxyflavones. J. Phys. Chem. 1985, 89, 2362–2366. 18 Favaro, G.; Clementi, C.; Romani, A.; Vickackaite, V. Acidichromism and ionochromism of luteolin and apigenin, the main components of the naturally occurring yellow weld: A spectrophotometric and fluorimetric study. J. Fluoresc. 2007, 17, 707–714. 19 Romani, A.; Favaro, G. Dyes and Pigments: New Research. The beauty of colors: the yellow flavonols in science and art; Nova Science Publisher, Inc.: New York, 2009; pp331-349. 20 Smith, G. J.; Thomsen, S. J.; Markham, K. R.; Andary, C.; Cardon, D. The photostabilities of naturally occurring 5-hydroxyflavones, flavonols, their glycosides and their aluminium complexes. J. Photochem. Photobiol. A 2000, 136, 87–91. 21 Clementi, C.; Doherty, B.; Gentili, P. L.; Miliani, C.; Romani, A.; Brunetti, B. G.; Sgamellotti, A. Vibrational and electronic properties of painting lakes. Appl. Phys. A: Mater. Sci. Process. 2008, 92, 25–33. 22 Claro, A.; Melo, M. J.; Schaefer, S.; Seixas de Melo, J. S.; Pina, F.; Van den Berg, K. J.; Burnstock, A. The use of microspectrofluorimetry for the characterization of lake pigments. Talanta 2008, 74, 922–929. 23 Pozza, G.; Ajo`, D.; Chiari, G.; De Zuane, F.; Favaro, M. Photoluminescence of the inorganic pigments Egyptian blue, Han blue and Han purple. J. Cult. Herit. 2000, 1, 393–398. 24 Accorsi, G.; Verri, G.; Bolognesi, M.; Armaroli, N.; Clementi, C.; Miliani, C.; Romani, A. The exceptional near-infrared luminescence properties of cuprorivaite (Egyptian blue). Chem. Commun. 2009, 3392–3394. 25 Verri, G. The spatially resolved characterisation of Egyptian blue, Han blue and Han purple by photo-induced luminescence digital imaging. Anal. Bioanal Chem. 2009, 394, 1011–1021. 26 Vanheusden, K.; Warren, W. L.; Seager, C. H.; Tallant, D. R.; Voigt, J. A. Mechanism behind green photoluminescence in ZnO phosphor powders. J. Appl. Phys. 1996, 79, 7983–7990. van Dijken, A.; Meulenkamp, E. A.; Vanmaekelbergh, D.; Meijerink, A. The luminescence of nanocrystalline ZnO particles: The mechanism of the ultraviolet and visible emission. J. Lumin. 2000, 87-89, 454–456. 27 Kittel, C. Introduction to Solid State Physics; John Wiley & Sons, Inc.: Hoboken, NJ; 2005. 28 Artem’jeva, O. O.; Vakulenko, O. V.; Dacenko, O. I. Amphoteric center of luminescence in CdS. Semicond. Phys., Quantum Electron. Optoelectron. 2005, 8, 58–60. 29 Clementi, C.; Miliani, C.; Romani, A.; Favaro, G. Spettrofluorimetria UV-VIS in riflettanza: Una tecnica non distruttiva per la diagnostica dei manufatti artistici. Proc., Ravenna ′03 2003, 110–118. 30 Romani, A.; Favaro, G. Italian patent request nr. RM2008A000002, registered January 4th, 2008.

846


837-846

June 2010

Vol. 43, No. 6

31 Romani, A.; Clementi, C.; Miliani, C.; Brunetti, B. G.; Sgamellotti, A.; Favaro, G. Portable equipment for luminescence lifetime measurements on surfaces. Appl. Spectrosc. 2008, 62, 1395–1399. 32 Romani, A. Steady-state and time-resolved luminescence for in-situ characterization of polychrome artworks. Luminescence 2008, 23, 262–263. 33 Lognoli, D.; Cecchi, G.; Mochi, I.; Pantani, L.; Raimondi, V.; Chiari, R.; Johansson, T.; Weibring, P.; Edner, H.; Svanberg, S. Fluorescence lidar imaging of the cathedral and baptistery of Parma. Appl. Phys. B: Lasers Opt. 2003, 76, 457–465. 34 Palombi, L.; Lognoli, D.; Raimondi, V.; Cecchi, G.; Hallstrom, J.; Barup, K.; Conti, C.; Gronlund, R.; Johansson, A.; Svanberg, S. Hyperspectral fluorescence lidar imaging at the Colosseum, Rome: Elucidating past conservation interventions. Opt. Express 2008, 16, 6794–6808. 35 http://www.eu-artech.org/. 36 Miliani, C.; Kahrim, K.; Brunetti, B. G.; Sgamellotti, A.; Aldrovandi, A.; van Bommel, M. R.; van den Berg, K. J.; Janssens, H. MOLAB, a mobile facility suitable for noninvasive in-situ investigations of early and contemporary paintings: the case-study of Victory Boogie Woogie (1942-1944) by Piet Mondrian. Proc. 15th Trienn. Conf. ICOM-CC, New Dehli 2008, 2, 244–251. 37 Clementi, C.; Miliani, C.; Romani, A.; Santamaria, U.; Morresi, F.; Mlynarska, K.; Favaro, G. In-situ fluorimetry: A powerful non-invasive diagnostic technique for natural dyes used in artefacts. Part II Identification of orcein and indigo in Renaissance tapestries. Spectrochim. Acta, Part A 2009, 71, 20572062. 38 Thoury, M.; Elias, M.; Frigerio, J. M.; Barthou, C. Nondestructive varnish identification by ultraviolet fluorescence spectroscopy. Appl. Spectrosc. 2007, 61, 1275–1282. 39 Nevin, A.; Anglos, D. Assisted interpretation of laser-induced fluorescence spectra of egg-based binding media using total emission fluorescence spectroscopy. Laser Chem. 2006, 82823. 40 Verri, G.; Clementi, C.; Comelli, D.; Cather, S.; Pique´, F. Correction of ultravioletinduced fluorescence spectra for the examination of polychromy. Appl. Spectrosc. 2008, 62, 1295–1302. 41 Clementi, C.; Miliani, C.; Verri, G.; Sotiropoulou, S.; Romani, A.; Brunetti, B. G.; Sgamellotti, A. Application of the Kubelka-Munk correction for self-absorption of fluorescence emission in carmine lake paint layers. Appl. Spectrosc. 2009, 63, 1323–1330. 42 Siemiarczuk, A.; Ware, W. R. Temperature dependence of fluorescence lifetime distributions in 1,3-di(1-pyrenyl)propane with the maximum entropy method. J. Phys. Chem. 1989, 93, 7609–7618. Brochon, J. C. Maximum entropy method of data analysis in time-resolved spectroscopy. Methods Enzymol. Part B 1994, 240, 262–311. 43 Gentili, P. L.; Clementi, C.; Romani, A. UV-VIS absorption and luminescence properties of quinacridone/BaSO4 solid mixtures. Appl. Spectrosc., submitted for publication.

Scanning Multispectral IR Reflectography SMIRR: An Advanced Tool for Art Diagnostics CLAUDIA DAFFARA,* ENRICO PAMPALONI, LUCA PEZZATI, MARCO BARUCCI, AND RAFFAELLA FONTANA Istituto Nazionale di Ottica (CNR-INO), Largo E. Fermi 6, 50125 Florence, Italy RECEIVED ON OCTOBER 30, 2009

CON SPECTUS

S

pectral imaging technology, widely used in remote sensing applications, such as satellite or radar imaging, has recently gained importance in the field of artwork conservation. In particular, multispectral imaging in the near-infrared region (NIR) has proved useful in analyzing ancient paintings because of the transparency of most pigments and their varied reflectance changes over this spectral region. A variety of systems, with different detectors and filtering or dispersing technologies, have been implemented. Despite the recognized potential of multispectral NIR imaging, which provides information on both spectral and spatial domains (thus extending the capabilities of conventional imaging and spectroscopy), most of the systems currently used in art diagnostics have limitations. The technology is still in its early stages of development in this field. In this Account, we present the scanning multispectral IR reflectography (SMIRR) technique for artwork analysis, together with an integrated device for the acquisition of imaging data. The instrument prototype is a no-contact optical scanner with a single-point measurement of the reflectance, capable of simultaneously collecting a set of 14 spatially registered images at different wavelengths in the NIR range of 800-2300 nm. The data can be analyzed as a spectral cube, both as a stack of wavelength resolved images (multi-NIR reflectography) and as a series of point reflectance spectra, one for each sampled pixel on the surface (NIR spectrometry). We explore the potential of SMIRR in the analysis of ancient paintings and show its advantages over the wide-band conventional method. The multispectral option allows the choice of the most effective NIR bands and improves the ability to detect hidden features. The interband comparison aids in localizing areas of different pictorial materials with particular NIR reflectance. In addition to the analysis of single monochromatic images, enhancement procedures involving the joint processing of multispectral planes, such as subtraction and ratio methods, false color representation, and statistical tools such as principal component analysis, are applied to the registered image dataset for extracting additional information. Maintaining a visual approach in the data analysis allows this tool to be used by museum staff, the actual end-users. We also present some applications of the technique to the study of Italian masterpieces, discussing interesting preliminary results. The spectral sensitivity of the detection system, the quality of focusing and uniformity of the acquired images, and the possibility for selective imaging in NIR bands in a registered dataset make SMIRR an exceptional tool for nondestructive inspection of painting surfaces. The high quality and detail of SMIRR data underscore the potential for further development in this field.

Introduction and Background Near-infrared (NIR) methods have been increasingly used for noninvasive analysis of art objects and a number of different techniques are currently adopted for various tasks in this field. Since its introduction,1 infrared reflectography (IRR) continues to be a well established technique for paintPublished on the Web 03/15/2010 www.pubs.acs.org/acr 10.1021/ar900268t © 2010 American Chemical Society

ing analysis, routinely utilized in any conservation laboratory as part of the diagnostic process before an intervention. Since most of the pigments are transparent to NIR wavelengths, this radiation allows features underlying the paint layer, such as retouching, underdrawings, and pentimenti (changes done by the artist himself during the Vol. 43, No. 6

June 2010

847-856


847

Multispectral IR Reflectography Daffara et al.

painting composition process, dealing with either the drawing or the pictorial layer), to be revealed. Reflectographic imaging is performed by irradiating the artwork with NIR radiation and by detecting the backscattered radiation with suitable devices. According to the interaction properties of the painting materials in the employed spectral range and to the composition and thickness of the specimen, the distribution of absorbed and reflected radiation allows us to extract information about painting layers and artist’s technique. Currently, different kinds of systems are available to acquire a NIR image, ranging from commercial to experimental apparatus, and response depends on the performance of the imaging device, as well as on the spectral band used.2-9 Traditionally, reflectography is being performed in wide-band modality by acquiring the NIR image in a single large band, which corresponds to the spectral range of the device. The early systems using PbO-PbS vidicon cameras have sensitivity up to 2.2 µm, whereas updated Si-based CCD cameras are limited to 1.1 µm. More suitable systems use detectors InGaAs or PtSi with broader sensitivity. Recently, the multispectral approach, widely used in remote sensing, has been proposed in NIR imaging of artworks and has found successful applications in this field.10 Traditional wide-band IRR is improved by the multiband option, which allows the most effective range of wavelengths to be tailored to fit the specific case.11,12 Spectral imaging technique allows the simultaneous collection of both spectral and spatial information thus enlarging the perspectives of IRR to new applications for the study of artists’ materials. In combination with data from analytical techniques,13,14 such as probe-based reflectance spectroscopy, it allows the nondestructive identification of particular pigments and provides information on their spatial distribution across the entire painting.15-17 Despite the recognized potential of this innovative method of investigation and the results achieved, most of the multispectral NIR imaging systems commonly used in painting diagnostics present limitations, and the technology for this specific application is still an open field. Various NIR multispectral devices based on different detectors and on a filtering or dispersing system are being used by a number of research groups.15,18-23 Waveband selection is mostly achieved by means of a scheme of filters fitted with an infrared camera, which records the spatial distribution, and the set of multiband images is stacked as a sequence of acquisitions. Such systems suffer from systematic errors such as blurring of the multispectral channels, which affects the reliability of spectral data and require optical calibration.24 When filters are tuned by mechanical systems, such as rotating wheels, the 848


847-856

June 2010

Vol. 43, No. 6

spectral image set is difficult to register.25 Systems based on sensor array suffer from pixel-to-pixel biases, lens distortion, and nonuniform illumination and require proper correction procedures.15,23 Detector size limits in practice the spatial resolution because of mosaicking.26 Moreover, current multispectral devices used in conservation have limited spectral range, which is related to the available detector technology and cost. To cover a more extended NIR range, vidicon tube or a plurality of detectors are still being used instead of the most advanced solid-state detectors.4,20 As a matter of fact, there is still a need for a specific developed technology in the spectral imaging of artworks covering the extended range of 800-2300 nm. In this Account, we present a novel integrated device for multispectral imaging of paintings in the NIR spectral range up to 2300 nm. After the description of the acquisition system, some applications to the study of Italian masterpieces are reported with the aim of showing the potential of the technique.

Instrument and Method Multispectral imaging in the NIR is carried out by irradiating the painting surface with a broadband source and collecting the backscattered radiation within narrow spectral NIR bands. The imaging method is based on point-by-point scanning of the surface. Reflectance of each sampled pixel is simultaneously acquired at different wavelengths and a set of multispectral images is constructed. Stacking multispectral images provides information in both the spatial and spectral domain that can be easily visualized for the analysis and further manipulation (Figure 1). Each point of the multispectral image cube can be extracted as a reflectance spectrum (NIR spectrometry), while the slices correspond to images at different wavelength bands (multi-NIR reflectography). Multi-NIR Scanning Device. Scanning multispectral IR reflectography (SMIRR) is performed by using the multi-NIR scanner whose schematic diagram is shown in Figure 2. An optical head, moved by a scanning system, both illuminates the painting and collects the backscattered radiation that is carried to the detection unit by means of a fiber bundle. To avoid chromatic aberration, the collecting optics is a catoptric system made of two faced spherical mirrors with f# ) 3.75 to have acceptance cones of about 15° (according to CIE suggestions for the 45°/0° configuration that we follow for the visible module, which is integrated in the device but is not treated in this Account). The EFL is 63 mm; the depth of field is about (1 mm, and we work in a 2f-2f configuration to have a unitary magnification factor. The radiation scattered


FIGURE 1. Sketch of the spectral cube consisting of as many image layers as NIR channels used. It can be analyzed as a set of wavelength resolved images (multi-NIR reflectography) or as a series of spatially resolved reflection spectra (NIR spectrometry).

from the measured point on the painting is imaged on the bundle made of 16 fibers assembled in square array with optical axes distance of 250 µm. Because the fibers have a 200 µm core and the magnification factor is equal to 1, the spot size at the painting is 200 µm for each channel (see Figure 2). The use of optical fibers allows separation of the optical head from the detection unit, thus minimizing the load on scanning stages. The lighting system is composed of two low-voltage halogen lamps, with (5° beam divergence, and is stabilized in current. The irradiated area is about 5 cm2 at the safe working distance of about 12 cm. Single point detection makes both the geometrical aberration negligible and the illumination uniform (the maximum fiber off-axis displacement is about 0.25°). Moreover, paint surface heating is minimized by scanning the lighting system. The detection unit is composed of 3 Si and 12 InGaAs photodiodes, each fitted with an interferential filter. One InGaAs sensor covers the wide-band NIR range 800-1700 nm. The remaining sensors cover the multispectral NIR range 800-2300 nm split in 14 bands with spectral resolution ranging from about 50 to 100 nm. Last fiber is connected to the visible module, which is going to be integrated in the device but is not treated in this paper. The central wavelength of the channels are resumed in Table 1, together with the bandwidth of the corresponding filter. Channels equalization was performed with certified standards with reflectance from 99% to 2.5%. The scanning unit is composed of two motorized XY translation stages orthogonally mounted, with a maximum stroke of 1 m and a precision less than 0.1 mm. The device allows continuous measurement of an area up to 1 m2 with a spa-

tial resolution of 0.5 mm. At an acquisition rate of 2 kHz and a sampling step of 250 µm, it takes about 3 h for maximum scanned area. The effect of motion blur due to electronic bandpass is negligible being less than 50 µm. In the acquisition phase our scanner is probably time-consuming compared with devices based on extended sensors, but this time loss is regained a posteriori because image data are aberration-free and hardware registered, thus saving postprocessing. The instrument has been assembled following the requirements for in situ measurement of paintings, namely, compactness, robustness, and transportability (Figure 3). The device controlling software is a custom application capable of real-time acquisition through a user-friendly graphic interface. Main routines include that for controlling the scanning mechanics and that for acquiring synchronized multichannels data. Additional modules allow post processing of raw 16-bit imaging data for gray levels optimization. During the measurements, the images can be visualized in real-time for each of the spectral bands. Multispectral Data and Processing. NIR images are reconstructed point-by-point by associating to each pixel on the image plane the intensity reflected by the sampled point on the painting surface. At each scanning section a reference image with a reflectance standard, Iref, and a dark image with closed optics, Idark, are acquired. Conversion to reflectance units is then made by using the in-scene reflectance standard with the formula

Rλ ) Fref

Isample - Idark Iref - Idark

where Rλ is the spectral reflection factor, Isample the detected radiation, and Fref the certified value for the reflectance standard. This relation, generally used in the visible range, can be extended to the NIR range because up to 2.5 µm thermal emission is negligible. Reflectance accuracy for a single measurement was obtained using seven SphereOptics standards (2.5%, 5%, 10%, 25%, 50%, 70%, 99% diffuse reflectance) and resulted in the range 5-10% rms of white standard 99%. The spectral image cube was computed by means of algorithms developed in Matlab environment. We worked with the 16-bit data sets sampled by the instrument. The monochromatic images were analyzed separately, as well as jointly. Well-known methods, such as pixel-by-pixel subtraction and the ratio between pairs of single-wavelength images, were carried out to highlight changes in the reflectance of pigments from one wavelength to another. To explore the capability of the multispectral imaging, false color composites were elaborated with either single band or difference/ratio images in Vol. 43, No. 6

June 2010

847-856


849


FIGURE 2. Schematic diagram of the multi-NIR scanning device (left). The instrument is a modular system composed of illumination/collection module, fiber bundle, detection module and scanning module which can be easily assembled (a). Sketch of multichannel acquisition of the painting (b). The single measure is carried out on a square array of sampled points that are simultaneously acquired through the fiber bundle. TABLE 1. Central Wavelength and FWHM for the NIR Channels CH

0

1

2

3

4

5

6

7

8

9

10

11

12

13

14

λ [nm] ∆λ [nm]

wide (900-1700)

800 10

850 67

952 67

1030 55

1112 66

1200 66

1300 90

1400 90

1500 90

1600 90

1700 90

1820 100

1930 112

2265 590

statistical method principal component analysis (PCA). PCA concentrates the significant features into a few representative images and makes somehow automatic the extraction of information enabling a straightforward interpretation of the results.27 A false-color representation was also used to mix the images produced by the PCA transformations to display the painting characteristics in a useful and intuitively appealing way. Interband comparison and operation with different bands are possible thanks to the superimposing property of our images that do not require registration. Despite the drawback of being time and disk-space consuming, this approach has the advantage of being easy to use and of maintaining a visual correspondence with the painting. Moreover, it can be performed also with commercial software for image processing.

Application and Results FIGURE 3. Multi-NIR scanning device during the demonstration measurements at the final meeting of the Eu-Artech project (Instituut Collectie Nederland ICN, Amsterdam, May 13th 2009).

The multi-NIR scanner was evaluated on real artworks to test

both trichromatic RGB and quadri-chromatic CMYK spaces. The false color image, which includes all the information from the constituent monochromatic NIR images, is used to examine the features varying with wavelengths and to visualize information from more than one channel simultaneously. The spectral image cube was also processed using the well-known

hosted at the Opificio delle Pietre Dure (OPD) in Florence, one

850


847-856

June 2010

Vol. 43, No. 6

the system performance on field. First, measurements have been carried out at the INO Optical Metrology Laboratory of the largest restoration facilities in Europe. The device was then moved to Italian and European museums for in situ diagnostics in the framework of the Eu-Artech project (2004-2009).28 The multi-NIR scanner is part of the mobile instrumentation of the Charisma project (2009-2013)29 and


FIGURE 4. Cimabue, Madonna con Bambino, 13th century, Santa Verdiana Museum, Castelfiorentino (Florence, Italy). Detail of the Child’s face at different wavelengths: (a) visible image, (b) wide 900-1700 nm channel, (c) CH2@850 nm, (d) CH6@1200 nm, (e) CH11@1700 nm, (f) CH14@2265 nm. Details, such as the mark of the golden leaf that appears as a white spot above the Child’s right eye, appear more clearly with increasing wavelength and reach the best visibility in the 2265 nm image. On the contrary, the two little spots on the Child’s forehead disappear at higher wavelengths.

can be accessed through its program by the scientific and the conservation community. The preliminary results reported below are intended to give an application overview of the SMIRR technique with emphasis on the advantages over the conventional reflectography.

Multispectral imaging enables the analysis of features that are not detectable in the wide-band reflectography, allowing the choice of the most effective NIR bands according to the different case study. Interband processing is used to enhance the presence of retouches and repaintings or, more generally, the Vol. 43, No. 6

June 2010

847-856


851


ing of multispectral image planes, such as image subtraction and ratio methods, are applied to the registered data set for extracting hidden features. Such processing of image planes is straightforward because, as mentioned above, the multiNIR scanner output is an high-quality data set (spatially registered, uniform illumination, aberration free). Difference/ratio images allows the visualization of all the extracted information in a unique synthesized plane, in addition to enhancing details that are scarcely visible in the monochromatic or in the wide-band images (Figure 5). In the difference image, the reflectivity variation between two bands is accentuated, thus allowing localization of areas of different materials. This is shown in the analysis of a XVI century wooden panel by an anonymous Italian painter (Figure 6). In a similar approach, false color representation with either single channels or difference/ratio images was used for the two paintings La Gravida by Raffaello and Madonna con Bambino, respectively, to differentiate regions which are then visualized in an effective and impactive way (Figures 7 and 8). PCA was profitably applied to a panel painting by Cosme` Tura for extracting the various score images. The first 91.9% score image contains the same information as the wide-band reflectography. The other three score images (5.5%, 1.9%, and 0.3%) combined in a false color representation improve the detection of details as shown in Figure 9.

Conclusion

FIGURE 5. Cimabue, Madonna con Bambino (13th century). (a) Difference and (b) ratio images obtained with CH2@850 nm and CH14@2265 nm. Both elaborations allow the visualization of all the extracted information in a unique synthesized image plane: the golden leaf above the Child’s right eye that appears with increasing wavelength, the two retouches on the Child’s forehead that appear with decreasing wavelength, and the dark dots on the Child’s chest that appear with increasing wavelength.

areas of different pictorial matter. This ability is well-known,11,12,17 but it has not been fully explored with an integrated instrument in such an extended spectral range, 800-2300 nm. Evidence of the potential of the multispectral approach is given in the analysis of the Cimabue’s painting Madonna con Bambino. In Figure 4, multi-NIR images and wide-band reflectography are compared to show how spectral segmentation beyond 1700 nm reveals retouches and details otherwise not visible. Enhancement procedures involving the joint process852


847-856

June 2010

Vol. 43, No. 6

An integrated imaging device operating in the extended NIR range 800-2300 nm has been developed with the specific application of scanning multispectral IR reflectography (SMIRR) of painted surfaces in mind. The multi-NIR scanner overcomes the limitations of most of the systems currently used for multispectral imaging. Point-by-point sampling for each of the NIR channels solves the problems related to the use of extended detectors and filter tuning and provides precise information in the spatial and spectral domain, namely, a metrically and optically corrected set of images that are mutually registered. As the potential of analysis of the multispectral data is enormous and involves many techniques, from the use of easy concepts to advanced mathematical tools, it is very important to start having high-quality data sets. Some applications to the analysis of ancient Italian masterpieces have been shown. The results show that the multispectral option together with the extended spectral range improve the traditional IRR technique. SMIRR data can be analyzed both as multi-NIR reflectography and NIR spectrometry. In this work, emphasis was given to the former approach. The complex aspect of pigment mapping is not treated in this work and will be treated in the


FIGURE 6. Anonymous Italian painter, Madonna con Bambino, XVI century, Gallery of Motti-Bardini Palace, Florence, Italy. Detail of the inscription “Ecce Agnus Dei”: (a) visible image, (b) wide-band reflectogram 900-1700 nm, (c) CH13@1930 nm, (d) CH14@2265 nm, (e) CH14CH13, (f) false color image in the CMYK space (C)CH3/CH2, M)CH7/CH5, Y)CH9/CH8, K)CH14/CH13; the K channel was inverted to enhance the inscription readability). In the difference image the reflectivity variation between two bands is accentuated thus allowing discovery of hidden details, such as the inscription that can be partially seen only in the visible image, but it is nearly invisible in the single IR images.

future. As far as pigment identification is concerned, the optical spectrometric techniques are restricted to particular cases because the reflectance response in a real artwork depends on several factors. Toward this aim and to provide an advanced and integrated diagnostics of paintings in the VIS-NIR spectrum, a version of the instrument including also the multi-VIS module is going to be implemented.

This research has been funded by EU within the 6th Framework Programme, project EU-ARTECH. We are indebted with Dr Cecilia Frosinini of the Opificio delle Pietre Dure in Florence for giving us the possibility to apply the SMIRR technique on real artworks and with the restorer Roberto Bellucci for useful discussions and suggestions. Thanks to Sara Micheli and Dr. Mattia Patti for spending lots of hours looking after the multi-NIR scanner during measurements.

BIOGRAPHICAL INFORMATION Claudia Daffara was born in Rome, Italy, in 1968. She studied theoretical Physics at the University of Padua and obtained a Ph.D. in Physics at the University of Bologna working in computational and applied physics. In 2003, she joined the Art Diagnostic Group at the National Institute of Optics CNR-INO in Florence, where she still works as researcher. Her research interests include radiation transport and imaging techniques for art diagnostics with major focus on data computation and modeling for 2D IR reflectography, thermography, and 3D microprofilometry confocal microscopy. Enrico Pampaloni was born in Florence, Italy, in 1961. He studied Physics at the University of Florence, where he got his degree in 1989 and his Masters degree in Optics in 1992. From 1990, he has been a researcher at the National Institute of Optics CNR-INO in Florence. His principal research interests in the field of applied optics for Cultural Heritage are imaging techniques (IR reflectography, VIS-NIR multispectral Vol. 43, No. 6

June 2010

847-856


853


FIGURE 7. Raffaello, detail of La Gravida, Galleria Palatina, Palazzo Pitti, Florence, Italy. Visible image (a), wide-band reflectogram 900-1700 nm (b, d), and false color composite with CH14@2265 nm, CH5@1112 nm, CH3@952 nm (c, e). In the wide-band reflectogram, no relevant details come out, whereas the false color image enhances different materials both in the face and in the dress.

FIGURE 8. Anonymous Italian painter, detail of Madonna con Bambino, XVI century, Gallery of Motti-Bardini Palace in Florence. (a) Color image, (b) wide-band reflectogram 900-1700 nm, and (c) false color image obtained in the CMYK space using the ratio images CH3/CH2, CH7/CH5, CH9/CH8, CH14/CH13. False color image allows the visualization of all the extracted information in a unique synthesized image plane and differentiation of regions that are profitably visualized in an effective and impactive way. The presence of different pigments at the right shoulder of the Madonna’s mantle are clearly visible.

imaging, UV-fluorescence, thermography), three-dimensional survey (time-of-flight laser scanning, laser-line scanning, and 854


847-856

June 2010

Vol. 43, No. 6

conoscopic microprofilometry), data analysis, and development of diagnostic methodologies.


FIGURE 9. Cosme` Tura, Madonna con Bambino, Galleria dell’Accademia Carrara, Bergamo, Italy. PCA analysis is profitably applied to the multispectral data set for extracting the information: (a) visible image, (b) “standard” false color image, that is, made with wide-band IR, red, green images, and (c) false color image made with the 2nd, 3rd, 4th score images (5.5%, 1.9%, 0.3% information, respectively). Because the first score image contains the same information as the wide-band IRR, by combining the score images other than the first, we extract more features, such as the couple of light spot at the Madonna’s neck side and the inhomogeneity in the background, the Madonna’s right arm mantle, and the chair drapery on the back.

Luca Pezzati Born in Florence, Italy, on March 19th, 1964. He obtained his degree in Physics (1990) from the University of Florence and his specialization in optics (1995). Since 2003, he has been a Senior Researcher at the CNR-INO (National Institute of Optics of the National Research Council), which he joined in 1995 after having worked at the Officine Galileo in Florence. He has been the Coordinator of the Gruppo Beni Culturali (Art Diagnostic Group) of the INO since 1999 and Head of the INO Division in Lecce. He has managed many research projects in the field of optical technologies applied to Cultural Heritage. Among his research interests are optics applied to safeguard of cultural heritage, optical metrology, 3D-measurement of optical techniques, optical systems design, interferometry, phase-analysis techniques, and scientific software design. He is the author of many journal articles and conference proceedings. Marco Barucci was born in Florence, Italy, in 1970. He studied physics at the University of Florence, where he got his Ph.D. on material science in 2004. He was a postdoctoral fellow until the end of 2008. Now, he is a researcher at the National Institute of Optics CNR-INO in Florence. His research interests include material characterizations for low temperature detectors and superconducting devices, optical techniques for material property analysis, and recently, development and realization of optic devices for the diagnostics of artworks. Raffaella Fontana was born in Pavia, Italy, in 1964. She studied physics at the University of Florence, where she got her Ph.D. in 1997 and her Specialization School (Master-like) in Medical Physics in 1999. Subsequently, she was a postdoctoral fellow until 2003. Since then she has been a researcher at the National Institute of Optics CNR-INO in Florence. Her research interests include imaging techniques for the diagnostics of works of art, such as IR reflectography, VIS-NIR multispectral imaging, UV-fluorescence,

confocal microscopy, and optical coherence tomography, thermography, as well as techniques for three-dimensional survey of objects, such as time-of-flight laser scanning, laser-line scanning, and microprofilometry.

FOOTNOTES * To whom correspondence should be addressed. E-mail: [email protected]. Phone: +39 055 23081. Fax +39 055 2337755. Web: http://arte.ino.it.

REFERENCES 1 Van Asperen de Boer, J. R. J. Infrared reflectography: A method for the examination of paintings. Appl. Opt. 1968, 7 (9), 1711–1714. 2 Van Asperen de Boer, J. R. J. Reflectography of paintings using an infra-red vidicon television system. Stud. Conserv. 1969, 14, 96–118. 3 Bertani, D.; Cetica, M.; Poggi, P.; Puccioni, G.; Buzzegoli, E.; Kunzelman, D.; Cecchi, S. A scanning device for infrared reflectography. Stud. Conserv. 1990, 35, 113– 117. 4 Walmsley, E.; Fletcher, C.; Delaney, J. Evaluation of system performance of nearinfrared imaging devices. Stud. Conserv. 1992, 37, 120–131. 5 Saunders, D; Billinge, R.; Cupitt, J.; Atkinson, N.; Liang, H. A new camera for highresolution infrared imaging of works of art. Stud. Conserv. 2006, 51, 277–290. 6 Consolandi, L.; Bertani, D. A prototype for high resolution infrared reflectography of paintings. Infrared Phys. Technol. 2007, 49 (3), 239–242. 7 Gargano, M.; Ludwig, N.; Poldi, G. A new methodology for comparing IR reflectographic systems. Infrared Phys. Technol. 2007, 49 (3), 249–253. 8 Daffara, C.; Gambino, M. C.; Pezzati, L. Performance analysis of imaging systems for NIR reflectography. Proceedings of the 2nd International Topical Meeting on Optical Sensing and Artificial Vision (OSAV 2008), St. Petersburg, Russia, May 1215, 2008; International Commission for Optics: Boca Raton, FL, 2008; pp 307314. 9 Falco, C. M. Invited Article: High resolution digital camera for infrared reflectography. Rev. Sci. Instrum. 2009, 80, 071301.1-9. 10 Fischer, C.; Kakoulli, I. Multispectral and hyperspectral imaging technologies in conservation: current research and potential applications. Rev. Conserv. 2006, 7, 3–16. 11 Walmsley, E.; Metzger, C.; Delaney, J. K.; Fletcher, C. Improved visualization of underdrawings with solid-state detectors operating in the infrared. Stud. Conserv. 1994, 39, 217–231.

Vol. 43, No. 6

June 2010

847-856


855


12 Fontana, F.; Bencini, D.; Carcagni, P.; Greco, M.; Mastroianni, M.; Materazzi, M.; Pampaloni, E.; Pezzati, L. Multi-spectral IR reflectography. Proc. SPIE 2007, 6618, 661816.1–12. 13 Mansfield, J. R.; Sowa, M. G.; Majzels, C.; Collins, C.; Cloutis, E.; Mantsch, H. H. Near infrared spectroscopic reflectance imaging: supervised vs. unsupervised analysis using an art conservation application. Vib. Spectrosc. 1999, 19, 33–45. 14 Bacci, M. UV-Vis-NIR, FT-IR, and FORS spectroscopies. In Modern Analytical Methods in Art and Archaeology; Ciliberto, E., Spoto, G., Eds.; Wiley Interscience: New York, 2000; pp 321-362. 15 Baronti, S.; Casini, A.; Lotti, F.; Porcinai, S. Multispectral imaging system for the mapping of pigments in works of art by use of principal-component analysis. Appl. Opt. 1998, 37 (8), 1299–1309. 16 Casini, A.; Lotti, F.; Picollo, M.; Stefani, L.; Buzzegoli, E. Image spectroscopy mapping technique for non-invasive analysis of paintings. Stud. Conserv. 1999, 44, 39–48. 17 Delaney, J. K.; Walmsley, E.; Berrie, B. H.; Fletcher, C. F. Multispectral imaging of paintings in the infrared to detect and map blue pigments. In Scientific Examination of Art: Modern Techniques in Conservation and Analysis; National Academies Press: Washington, DC, 2005; pp 120-136. 18 Mansfield, J. R.; Attas, M.; Majzels, C.; Cloutis, E.; Collins, C.; Mantsch, H. H. Near infrared spectroscopic reflectance imaging: a new tool in art conservation. Vib. Spectrosc. 2002, 28, 59–66. 19 Balas, C.; Papadakis, V.; Papadakis, N.; Papadakis, A.; Vazgiouraki, E.; Themelis, G. A novel hyper-spectral imaging apparatus for the non-destructive analysis of objects of artistic and historic value. J. Cult. Herit. 2003, 4 (1), 330–337. 20 Bacci, M.; Casini, A.; Cucci, C.; Muzzi, A.; Porcinai, S. A study on a set of drawings by Parmigianino: integration of art-historical analysis with imaging spectroscopy. J. Cult. Herit. 2005, 6 (4), 329–336.

856


847-856

June 2010

Vol. 43, No. 6

21 Liang, H.; Saunders, D.; Cupitt, J. A new multispectral imaging system for examining paintings. J. Imaging Sci. Technol. 2005, 49 (6), 551–562. 22 Karagiannis, G.; Salpistis, C.; Sergiadis, G.; Chryssoulakis, I. Non-destructive multispectral reflectoscopy between 800 and 1900 nm: An instrument for the investigation of the stratigraphy in paintings. Rev. Sci. Instrum. 2007, 78 (6), 065112.1-7. 23 Delaney, J. K.; Zeibel, J. G.; Thoury, M.; Littleton, R.; Morales, K. M.; Palmer, M.; de la Rie, E. R. Visible and infrared reflectance imaging spectroscopy of paintings: pigment mapping and improved infrared reflectography. Proc. SPIE 2009, 7391, 739103.1–8. 24 Mansouri, A.; Marzani, F. S.; Hardeberg, J. Y.; Gouton, P. Optical calibration of a multispectral imaging system based on interference filters. Opt. Eng. 2005, 44 (2), 27004.1–12. 25 Gat, N. Imaging spectroscopy using tunable filters: A review. Proc. SPIE 2000, 4056, 50–64. 26 Corsini, M.; Bartolini, F.; Cappellini, V. Mosaicing for high resolution acquisition of paintings. In Proceedings of the 7th International Conference on Virtual Systems and Multimedia (VSM’01), Berkley, CA, October 25-27, 2001; IEEE Computer Society: Washington, DC, 2001; pp 39-48. 27 Bonifazzi, C.; Carcagni, P.; Fontana, F.; Greco, M.; Mastroianni, M.; Materazzi, M.; Pampaloni, E.; Pezzati, L.; Bencini, D. A scanning device for VIS-NIR multispectral imaging of paintings. J. Opt. A: Pure Appl. Opt. 2008, 10, 064011.1-9. 28 EU-ARTECH: Access Research and Technology for the Conservation of the European Cultural Heritage, European Project 2004-2009, Contract no. RII3-CT-2004506171, www.eu-artech.org. 29 CHARISMA: Cultural Heritage Advanced Research Infrastructures: Synergy for a Multidisciplinary Approach to Conservation, European Project 2009-2013, Grant Agreement no. 228330.

Bright Light: Microspectrofluorimetry for the Characterization of Lake Pigments and Dyes in Works of Art MARIA J. MELO* AND ANA CLARO REQUIMTE-CQFB and Department of Conservation and Restoration, Faculty of Sciences and Technology of the New University Lisbon, Campus da Caparica, Portugal RECEIVED ON JULY 7, 2009

CON SPECTUS

C

olor is an important component in the perception of beauty and in an artist’s original intent when creating a work. Better conservation of our cultural heritage requires detailed knowledge of artwork materials and the complex evolution they have endured over time. Organic dyes have been used from ancient times, and their characterization is a challenge that has been successfully addressed over the past few years by the development of advanced techniques, such as microspectrofluorimetry. In this Account, we describe the application of microspectrofluorimetry to the study of medieval illuminations, paint cross sections, millenary textiles, and wall paintings. In our research into color in medieval Portuguese illuminations, we chose to emphasize the importance of the experimental design and the use of microspectrofluorimetry in the context of other analytical techniques, such as microFTIR, microRaman, and micro-X-ray fluorescence (microXRF). Within this framework, we were able to unveil the full complexity of a medieval colorant and to address issues not yet explored, such as the influence of Arab, Jewish, and Christian cultures on the production and underlying technology of Portuguese illuminations. The analysis of individual pigment particles or aggregates (by excitation with an 8 µm diameter spot) in paint cross sections from works by Vincent van Gogh and Lucien Pissarro highlights the technique’s advantage of high spatial resolution. Its high spectral resolution proved to be useful not only for better characterizing the dyes used to color Andean textiles but also for detecting mixtures of relevant chromophores; the emission signals for the reds in Paracas and Nasca textiles were shown to be due to the presence of purpurin and pseudopurpurin. Finally, the complexity of the study of yellow dyes and the importance of accurate historical reproductions is addressed in a study of Asian organic colorants on historic Chinese wall paintings. Microspectrofluorimetry offers high sensitivity, selectivity, fast data acquisition, good spatial resolution, and the possibility of in-depth profiling. It has proved to be an invaluable analytical tool in identifying dyes and lake pigments in works of art. As SaintExupe´ry’s protagonist said in Le Petit Prince, “L’essentiel est invisible pour les yeux,” or “What is essential is invisible to the eye”sbut it may be unveiled with kind love, a prepared mind, and a little help from microspectrofluorimetry.

1. Introduction

works of art. The passionate debate he had

To create a little flower is the labor of ages. William Blake In one of his last public interventions in the 1980s, Brandi challenged scientists to find how to characterize the organic compounds present in

opened in the 1960s with “The cleaning of pic-

Published on the Web 05/06/2010 www.pubs.acs.org/acr 10.1021/ar9001894 © 2010 American Chemical Society

tures in relation to patina, varnish and glazes”1 was certainly on his mind. Without this knowledge, how could one argue that in a certain picture a precious, colored, original glaze was still Vol. 43, No. 6

June 2010

857-866


857

Characterization of Lake Pigments and Dyes Melo and Claro

there and that color was not a consequence of aging but an artist’s choice? How could one avoid the possible irreversible damage in a restoration of a Michelangelo, which includes the removal of an “aged” varnish, without knowing if it is an aged varnish or a colored velatura? Brandi’s challenge was indeed taken up, and in the past two decades the field of conservation science has grown and matured to the extent that it is regarded as a discipline in its own right, in which new advanced methods are developed to unveil some of the last secrets of the works of art. Projects such as the Molart and Molab,2,3 employing very different strategies, played a crucial role in the development of a critical mass and in the shaping of the field of conservation science. The Molart project, with its follower De Mayerne, focused on molecular aging studies, whereas the Molab proposed a holistic approach to the characterization of the work of art: enough data must be collected in situ using noninvasive methodologies in order to be representative of the materials and techniques of a certain work of art. Both approaches are complementary. In the past few years, we have been particularly interested in the development of methodologies that will enable a complete characterization of the organic colorants used in the past as well as their degradation products. Changes in pigments, whether used pure or mixed with other pigments, can alter the appearance of a painting significantly; as a consequence, the identification and state of degradation of colorants is of fundamental interest, since it provides critical information about the artists’ aesthetic perspective, conceptions, and choices, and how the work has changed over time. Therefore, it is desirable to develop methods that can characterize these materials noninvasively or from small samples that may be available from works of art. The potentiality of microspectrofluorimetry in the field of conservation science was tested within this framework: first, in historically accurate reproductions and, finally, in works of art, such as textiles, medieval illuminations, or oil paintings.4,5 Microspectrofluorimetry offers high sensitivity and selectivity, combined with good spatial resolution and fast data acquisition; it can also be used in situ without any contact with the sample or work of art to be analyzed, for movable objects that can be transported in the laboratory. The importance of sensitivity is clear when the following facts are taken into account: some of the dyes used in the past to create bright colors may have faded or may have been applied as very thin coats over, or mixed with, an inorganic pigment or extender, and as a consequence they may be present in very low concentrations. The possibility of in situ analysis of ancient colorants is a considerable advantage, particularly when considering that the techniques currently 858


857-866

June 2010

Vol. 43, No. 6

employed for dye analysis (HPLC-DAD-MS, microFTIR, and microRaman) usually require microsampling. Microspectrofluorimetry also presents some drawbacks, namely, the absence of a molecular fingerprint as disclosed in infrared spectra. This limitation may be overcome by using consistent data-basis buildup with historically accurate reproductions. A brief overview will be presented of the dyes used in the past for dyeing and painting and on the methods currently employed for their analysis. We will address the importance of understanding a dye’s photochemistry and photophysics as well as that of historically accurate reproductions. Finally, case studies will be described to highlight the successful stories where emission fluorescence proved invaluable and the incomplete ones where work is still in progress. 1.1. Natural Dyes in Context: Textiles, Medieval Illuminations, and Paintings. Natural dyes and their metalion complexes have been used for textiles, manuscript illuminations, paintings, and other works of art. For example, anthraquinones and their hydroxy derivatives have been used as red dyes and pigment lakes from prehistoric times, and we can find written accounts of the use of anthraquinone reds and purples as dyes in ancient Egypt;6 anthraquinone lakes (e.g., madder red) were also very popular with Impressionist painters, including Vincent van Gogh. Pure dyes such as indigo were also used as painting materials, for example, in medieval illuminations.7 To be used as a textile dye, the chromophore must be absorbed as much as possible into the fibers; that is, it must be resistant to washing. Dyes can bind to the surface of the fiber or be trapped within. They are often bound to the textile with the aid of metallic ions known as mordants, which can also play an important role in the final color. Alum, as a source of aluminum ion, is an important historical mordant, and it was widely used in the past.8 Dyes, such as indigo, are trapped in the fibers due to an oxidation-reduction reaction, without the aid of a mordant. It is useful to distinguish between dyes and pigments based on their solubility in the media used to apply the color; dyes are generally organic compounds that are soluble in a solvent, whereas pigments, used in painting, are usually inorganic compounds or minerals which are insoluble in the paint medium (oil, water, etc.) and are dispersed in the matrix. Lake pigments can be prepared by precipitating the dye extract with aluminum salts, such as alum.4,7 1.2. Chromophores: The Classic Palette. By the time of the founding of the Mediterranean civilizations, what we would consider the classic palette for natural dyes had already


FIGURE 1. Anthraquinone reds: (i) from plants, alizarin (1,2 dihydroxy anthraquinone) and purpurin (1,2, 4 trihydroxy anthraquinone), first row; (ii) from insects, kermesic acid, carminic acid, and laccaic acid A, second row. The blues and purple chromophores are depicted in the third row: indigo (indigotin), purple (6,6′-dibromoindigotin), and indirubin. For the yellows, the flavone luteolin, the carotenoid crocetin, and the chalcone marein (okanin 4′-O-glucoside), fourth row.

been established, and most valued colors were indigo for the blues, anthraquinone-based chromophores for the reds, and 6,6′-dibromoindigo for purple7 (Figure 1). The natural sources for yellows were much more diverse, so yellows could generally be obtained locally (Figure 1). With the exception of some browns, all other colors, including green and orange, could be obtained with these blue, red, purple, and yellow dyes. This classic palette was preserved over centuries, if not millennia. The first adjustment resulted from the loss of Tyrian purple following the fall of Constantinople and the subsequent collapse of the Roman social and commercial web. This was followed by a new entry, cochineal red, brought by the Spanish from the New World.7 However, even with the introduction of cochineal, the chemical nature of the classic palette was maintained, as carminic acid is still a substituded 1,2dihydroxy anthraquinone. This classic palette was only challenged by the audacity of chemists, who created new molecules, and colors never seen before, from the mid-19th century on.9 1.3. Fading, Lifetimes, Photochemistry, and Photophysics. The processes taking place in the excited state, after absorption of light by a molecule, are usually complex and in heterogeneous systems such as the ones offered by works of

art they are still a challenge. This may explain why so few systematic studies have been carried out and published.10 Both photochemistry and photophysical studies will be necessary to fully exploit the strength and test the limitations of UV-VISNIR emission spectroscopy as an analytical technique for the identification of dyes in works of artistic and historic importance. Also, this knowledge will enable more information on the dye’s environment and its conservation state to be extracted. 1.4. Current Methods and the Development of New Methods of Analysis. Identifying the ancient dyes and dye sources has only been possible with the development, in the past two decades or so, of sensitive new microanalytical techniques.11 It is possible to group the methods as in situ (UV-VIS reflectance and florescence) or requiring microsampling (Raman, IR, HPLC-DAD-MS). Currently, the most efficient method for unequivocal dye characterization is HPLC-DADMS, but it requires sampling (Table 1). Chromophores are first extracted and then separated chromatographically and characterized by UV-VIS spectrophotometry and mass spectrometry; whenever possible, a comparison is made with authentic references. Currently, the use of HPLC-DAD enables dyestuff characterization from as little as 0.1 mg of thread. Recently, Vol. 43, No. 6

June 2010

857-866


859


the results obtained with the historical reproductions and the testing of the analytical methodologies. The information necessary to produce the historical reproductions is to be found both on documentary sources covering technical information contemporary with the period being investigated as well as on analytical data obtained by the works themselves.14,15

TABLE 1a

2. Microspectrofluorimetry

a * Emission and excitation spectra may not allow for an unequivocal identification due to the low S/N. Together with lifetimes, identification is possible. ** It is necessary to extract the dye. *** In textiles, madder lake was identified by shifted Raman spectroscopy. [15d] (d) Depends on the concentration of the dye in the alumina matrix; usually unequivocal identification is not possible.

developed mild extraction methods allow more detailed chemical information to be obtained on the historical natural dyes, and as a consequence it is sometimes possible to identify the natural sources, even for the more complex analyses of yellow dyes. Raman microscopy has also been actively explored for dye analysis and new methods, that allow for higher S/N, developed, such as surface-enhanced Raman scattering (SERS).12 Most promising are the analytical techniques based on fluorescence detection that take advantage of the opportunity for high sensitivity and selectivity, combined with in situ analysis.4,5,13 Microspectrofluorimetry as well as transportable fiber-optic spectrofluorimeters have been systematically tested in the past few years. 1.5. The Importance of Accurate Historical Reproductions and a Database. When studying the materials used to create a work of art, complexity may be addressed not only by using complementary techniques but also by knowing the materials and techniques employed in the original process of art creation. Recreating old recipes with as much historical accuracy as possible provides representative samples that may be used as standards.14 Having the Historical Recipe Reconstructions assembled in a database is essential when operating with in situ techniques. Model samples prepared with pure materials will in turn enable both a better understanding of 860


857-866

June 2010

Vol. 43, No. 6

2.1. The Technique. The measurements were obtained with a microSPEX instrument using a setup described elsewhere,16 where the Spex Fluorog apparatus 3-2.2 is connected to an Olympus BX51 M confocal microscope, with spatial resolution controlled with a multiple-pinhole turret, corresponding to a minimum 2 µm and maximum 60 µm spot with a 50× lens. For steady-state fluorescence spectra, a continuous 450 W xenon lamp, providing an intense broad spectrum from the UV to near-IR, is directed into a double-grating monochromator. The incident excitation beam is directed onto the sample, and its fluorescence is directed back up into the microscope. To view the sample’s fluorescence directly, a binocular eyepiece and a digital camera are used. Beam-splitting is obtained with standard dichroic filters used at 45°; they are located in a two place filter holder. For a dichroic filter of 570 nm, excitation may be carried out until about 560 nm and emission collected after about 580 nm. For the study of red and yellow dyes, two filter holders with two sets of dichroic filters are employed, 500 and 570 nm in one set and 430 and 470 nm in the other set. This enables both the emission and excitation spectra to be collected with the same filter holder for each group of dyes. Spectra are collected after focusing on the sample (eye view) followed by signal intensity optimization (detector reading). The pinhole aperture that controls the area of analysis is selected based on the signal-to-noise ratio. For weak to medium emitters, it is set to 8 µm, which is appropriate for the analysis of individual pigment particles or aggregates in a paint film, and more generally it offers an optimized signalto-noise ratio and good spatial resolution for artists’ paints. 2.2. Comparison with Fiber Optic. With regard to the comparison of the method with the currently available transportable fiber-optic spectrofluorimeters, the more significant advantages of microspectrofluorimetry are as follows: (i) the possibility to acquire both emission and excitation spectra, (ii) high spatial resolution, (iii) high signal-to-noise ratio, enabling high spectral resolution and to obtain very well resolved spectra, and finally the possibility of (iv) in-depth profiling. It is important to stress the fact that an excitation spectrum repro-


FIGURE 2. Acid base equilibria for alizarin. The 1.10 keto tautomer, formed in the excited state (alizarin is a 9.10 quinone) is also depicted. For more details please see ref 4.

duces the chromophore absorption spectrum; a relevant parameter for the characterization of a dye or a pigment. The simultaneous acquisition of emission and excitation (absorption) spectra together with high spectral resolution facilitates a more accurate identification of dye molecules and lakes. In common with what is obtained using a fiber-optic setup, microemission fluorescence displays high sensitivity and very good reproducibility together with the possibility of performing semiquantitative analysis. When working with the fiberoptic and higher areas of analysis, a higher light throughput is achieved and this may be an advantage when studying very weak emitters. On the other hand, in the absence of confocal excitation, emission from the support or underlayers may turn signal interpretation difficult. 2.3. Photophysics and Photochemistry as a Tool to Explore with Higher Efficacy the Information Obtained. Anthraquinone derivatives and yellow dyes, flavonoid or carotenoid based, are complex systems displaying photophysical properties that may be strongly influenced by pH and metal ion complexation.4,5,17,18 As regards indigo, a full understanding of its photophysics as well as photochemistry has emerged from the comprehensive studies on the photophysics of the keto and leuco forms of indigo and its derivatives carried out during the past 5 years.10a,b Photophysical and photochemistry studies provided important information for the understanding of the photostability of indigo blues as well as of 1,2-dihydroxy anthraquinone reds. For both systems, efficient radiationless processes, possibly involving excited state proton transfer (ESPT) and excited state intra- and intermolecular proton transfer (ESIPT), play a determinant role in the overall stability of the molecules. ESPT and ESIPT with excited state tautomer formation (Figure 2) can be considered to induce a photoprotective mechanism for the molecule, enabling a very fast and effective dissipation of the excess energy of the excited state. Less is known about the excited state properties of pigment lakes in the solid state,4,5 and only recently were the first systematic studies published on the solid state photophysics and photochemistry of these ancient colorants, reds,4 blues,10 and yellows.18 Nevertheless, this positive trend will enable the disclosure of the full potential of fluorescence spectroscopy applied to the study of ancient dyes.

From these first photophysical studies on alizarin and purpurin lakes, it emerged that more is needed to be known from the structure of the aluminum and other metal ion lakes in order to have a full rationalization of the observed excited state properties. From the studies we are presently conducting, by solid state NMR, a complex pattern emerged, where the final structure depends on the methods used for the lake formation.19

3. Case Study: Medieval Portuguese Illuminations Portuguese codices date from the formation of Portugal as a kingdom and are testimonies to medieval ideas, religion, and politics. Color use and production was a consequence of the technology available as well as of cultural and artistic options; defining the specificities of color will contribute to fingerprint the influences of the three different cultures that coexisted in Portugal at that time, Arab, Jewish, and Christian. This subject is approached within an interdisciplinary framework, from an art history and molecular sciences point of view, aiming to explore issues related with the symbolic and social meaning of color in medieval Portuguese illuminations, produced during the twelfth and first quarter of the thirteenth century in Portuguese monasteries. In the past 5 years, we have undertaken a systematic study of the materials and methods of Portuguese medieval illuminations,20 where microspectrofluorimetry has been used in context with other techniques. 3.1. Microspectrofluorimetry in Context: Experimental Design and Modus Operandi. MicroRaman, microFTIR, microXRF, and microXRD are powerful complementary techniques for the characterization of colorants in medieval manuscripts. More recently, analytical techniques based on emission fluorescence have also been experimented with for the identification of dyes and lake pigments with rewarding results, and were used to study the color of Portuguese medieval illuminations from three important Portuguese monasteries.21 The global approach and experimental design for a comprehensive study on color production for medieval Portuguese illuminations will be briefly described (Figure 3). Missions, where the equipment is transported to the institutions where Vol. 43, No. 6

June 2010

857-866


861


FIGURE 3. Main steps in a mission for the study of the medieval manuscripts from Alcobaça monastery (BNP, May 2009). MicroXRF and microRaman are techniques complementary to microspectrofluorimetry, that are used in situ. Microsamples will be afterward analyzed in laboratory, by microFTIR, microspectrofluorimetry, and microXRF. Photos by Duarte Belo.

the manuscripts are preserved, are prepared after a careful selection by the art history experts. From these manuscripts, a relevant number of folia will be analyzed in order to ensure statistical relevance. During the mission, the first screening is carried out by microXRF, which indicates the possible colorants and extenders present and allows a first quantification of these elements; moreover, its 70 µm enables us to obtain data that are representative for the distribution, in the manuscript, of a certain paint color. MicroRaman, which allows for high spatial resolution (1-5 µm spot) and where the diverse paint components may be excited separately, as well as emission fluorescence techniques will be used to address specific points, such as the molecular characterization of an inorganic pigment or a dye, respectively. Together with the spectroscopic investigation, paints are also analyzed by optical microscopy which allows us to understand how the final color is built up (by layers or by mixture), to detect possible degradation phenomena, and to sample the color paints that will be subjected to a more detailed characterization in the laboratory, as regards the colorants, binders, and additives. Designed microsampling presents several advantages, namely it keeps to a minimum the handling of the manuscript and allows for more detailed studies in the laboratory, without the time constraint that in situ mission implies. Typically, in the laboratory, a sample will be first analyzed by microFTIR, allowing for binder characterization and to gain an insight into the full paint formulation. If a dye is present, microspectrofluorimetry may be carried out first, as it requires no contact with the sample and employs a low-intensity radiation. MicroRaman and microXRD may be used to address specific points; with both methods, sample destruction may occur. 3.2. Microspectrofluorimetry in Context: The Medieval Portuguese Palette. From the study of all three monasteries’ manuscripts, carmine and bottle-green colors emerged as pos862


857-866

June 2010

Vol. 43, No. 6

sibly characteristic of the Portuguese production, and the first medieval Portuguese palette was proposed (Figure 4). In the Lorvaõ manuscripts, namely, in The book of birds (1183-1184), the carmine color is based on an organic dye, and infrared analyses indicate that it was possibly obtained from lac dye. The paint is of a beautiful deep red color, displaying a glassy appearance when observed under the microscope. Paint historical reconstructions were produced based both on the infrared results and the recipes found in the medieval treatise “The book on how to make colors”. The emission and excitation spectra obtained from three original samples matched well the signals obtained with a paint reconstruction where lac was extracted from the raw material at pH ) 11 and the paint solution was applied with a pH ) 6. Even if the infrared fingerprint was also similar, it was not possible to reproduce the glassy appearance of the original paint. Moreover, taking into account the quantum yields of fluorescence, measured in solution, for laccaic acid A22 (φf ≈ 10-3) and its aluminum complex (φf ≈ 10-2), it is possible to suggest that the lac dye was not applied as a pigment lake. In this case, microspectrofluorimetry enabled the dye identification as well as to gather information on the characterization of the paint recipe used. Historical reconstructions based on medieval Arabic sources are currently in progress, and eventually will enable to obtain both a match for the molecular fingerprint and the original color appearance. The study of color in the Santa Cruz monastery collection was carried out in the framework of a Molab mission, and,23 with the aid of a portable fluorimeter, fluorescence emission spectra and lifetimes were collected in situ when there was evidence for the presence of a dye or pigment lake. Indigo is easily identified both by fluorescence emission (φf,DMF ≈ 10-3) as well as by Raman spectroscopy, and it was possible to conclude that the dark blues were obtained mixing lapis lazuli with indigo. With regard to the carmine or dark red color, the presence of emission was detected


FIGURE 4. The Lorvaõ palette. Macrophotos for the most relevant colors and respective details, highlighting the way color is constructed in the Lorvaõ illuminations.

but it was low and no match was found in the Molab database. However, it was possible to conclude that the dark red was obtained again by mixing an inorganic pigment, vermilion, with a dye, not yet identified. The detection of these mixtures for the dark blue and red colors is in itself important evidence for the characterization of the Santa Cruz palette, as it was not detected in the manuscript collections of the two other monasteries, Lorvaõ and Alcobaça.

4. Case Study: Cross Sections Paint cross sections are used to gain more insight into the technique used to create a painting, its stratigraphic structure, from the preparation layer covering the support, into the color construction and varnish protection. Microsamples are removed from the object and mounted as cross sections, which involves embedding in a synthetic resin followed by careful polishing. Paint cross sections are commonly examined by microscopy, and therefore, the strengths of microspectrofluorimetry were tested for their analysis, first in model samples, in which a chalk (CaCO3) preparation and a white lead (PbCO3 · Pb(OH)2) imprimatura were used for a fluorescent vinyl commercial paint, GEO.4 Remarkably good excitation and emission spectra were acquired with a high spatial resolution of 8 µm, which is appropriate for the analysis of individual pigment particles or aggregates in a paint film. This was followed by analysis in cross sections from paintings by Vincent van Gogh (Van Gogh Museum) and Lucien Pissarro (Courtauld Institute of Art) (Figure 5). Red lake pigments and dyes were characterized by microspectrofluorimetry; emission and excitation spectra were obtained with high spatial resolution (8-30 µm). The fluorophores, purpurin, and

FIGURE 5. Painting cross section from Old Mark’s field by Lucien Pissarro (Courtauld Institute), detail from a Paracas mantle (Museum of Fine Arts-Boston), and coupon reproducing the materials and techniques used in a wall painting of the Mogao Grottoes (Getty Conservation Institute).

Vol. 43, No. 6

June 2010

857-866


863


eosin lakes were identified by comparing their spectra with those from historical reconstructions assembled in a database. Some of these samples were also analyzed by other techniques requiring microsampling such as HPLC-DAD and SEM/ EDS, and the results were consistent with those found by microspectrofluorimetry.

5. Case Study: Textiles Pre-Columbian textiles are unique as a cultural and historical record, representing the longest continuous textile record in world history24 (Figure 5). Fortunately, in extremely arid archeological sites, the cultural heritage of different Andean cultures such as Paracas, Nasca, Chancay, and Lambayeque has been preserved. The reds used to dye in Peru before the Inca

FIGURE 6. Emission and excitation spectra, acquired in a 30 µm area, by exciting directly on a single fiber for the Paracas skirt (MFA 21.2581), 200B.C. to 200 A.D.; λexc ) 500 nm and λem ) 590 nm. In the inset, a detail showing the 30 µm excitation spot during data acquisition.

Empire (14th-15th century) were based on purpurin chro-

6. Case Study: Wall Paintings

mophore obtained from Relbunium sp.8,25 Microspectrofluorimetry was used to analyze the 76 micro-

Located near the old Silk Road, in the oasis city of Dunhuang

samples taken from different Andean textiles, dated from 200

(northwestern China), a group of caves covered with wall

B.C. to A.D. 1476, in the collection of the Museum of Fine Arts-

paintings, the Mogao Grottoes, were studied by the Getty Con-

Boston. The majority of the samples present a red color, but

servation Institute (GCI). In order to develop strategies to iden-

fibers with pink and purple color were also analyzed. SEM-

tify the possible Asian organic colorants applied, a collection

EDX screening enabled them to confirm the use of aluminum

of reference samples was built, in the GCI laboratory, with sev-

ion, Al3+, as a mordant and also to conclude that all the red

eral Asian biological sources (Figure 6). Those could be used

samples studied were made of camelid fibers.5

in different sorts of samples, among them: pigments, dried

Emission and excitation spectra were obtained in a 30 µm spot. The fluorophores were identified by comparing their

extracts, and wall painting mock-ups of different paint combinations.26

spectra with those from historical reconstructions assembled

HPLC-DAD-MS and FTIR were used, and good results were

in a database. In the Paracas and Nasca textiles, dated from

obtained for most of the reference samples and paint sam-

200 B.C. to A.D.1476, purpurin and pseudopurpurin were the

ples collected in Cave 85,26 although some of them are still

25

red dyes used.

Carminic acid was detected in textiles dated

to be discovered. Without the requirement of microsampling,

close to the Inca Empire, A.D. 1000-1476. The results

microspectrofluorimetry proved to be a good analytical tech-

obtained with this new technique were confirmed and are

nique to characterize even small quantities of colorant

consistent with those obtained with conventional methods,

material not detectable with the traditional techniques, such as

requiring microsampling, such as HPLC-DAD-MS and SEM-

Gardenia augusta.

EDX. It is worth stressing that it was the excellent spectral res-

In order to be able to identify the botanical species applied

olution obtained for the reds dyed with Relbunium that

in the mock-ups, two kinds of lakes were prepared with the

enabled to assess both purpurin and pseudopurpurin in the

main chromophores of each species: the first ones in homo-

emission as well as in the excitation spectra (Figure 6). This

geneous media (solution) and the other ones in solid state

was confirmed by HPLC-DAD-MS and by comparison with the

(powder and painted with aqueous media).

spectra obtained from a reconstructed pseudopurpurin lake

Emission and excitation spectra were obtained in an 8 or

(obtained after extraction followed by HPLC separation and

30 µm spot for both reconstructions and the mock-up. It was

finally complexation with Al3+). In this study, emission and

possible to obtain the two spectra for all the paint combina-

excitation spectra were obtained directly from fiber set in the

tions containing a Gardenia augusta layer and realize that the

microscope stage, but analysis could have been carried out on

underlayers (clay and ground) do not interfere with the emis-

the entire textiles.

sion signal from Gardenia augusta. However, some small shifts

864


857-866

June 2010

Vol. 43, No. 6


in the emission spectra were observed when Gardenia augusta glaze was applied over the inorganic pigment paints. The painted mock-ups provided the creation of a database which in the future can be useful to determine the presence of Gardenia augusta glaze over complex stratigraphies, including alum layer, seven different pigments (vermillion, atacamite, red lead, carbon black, red earth orpiment, and azurite), ground, and clay.

7. Future Perspectives Work is currently in progress to maximize the extraction of the information present in fluorescence emission and excitation spectra acquired by microfluorimetry. The possibility of obtaining the chromophores’ lifetimes together, in the same microspot, will also be fully explored. Research on the photophysics and photochemistry of historic dyes is necessary to exploit better this powerful tool and is being actively pursued. Although the molecular structures of lakes are difficult to determine by fluorescence emission alone, combined with data from other analytical techniques information about the environment of the colorant and its state of degradation can be obtained. This is one of the most promising issues to be explored, because it will enable us to characterize the structure of the metal-dye complex, which in turn may be used to fingerprint particular color recipes and to gain an insight into the paint technology of past times, a technology that did not leave any records other than the objects themselves. It will also enable us to understand better the degradation mechanisms responsible for color fading. Together this information and knowledge will promote the development of better and sustainable conservation strategies: it is only possible to keep safe what we know well. The importance of historically accurate reconstructions assembled in a database was evidenced during this account. This pool of standards will be enriched by the entering of “original” data that will have been fully processed and rationalized. In a symbiotic process, the original data will enable advances in art technological source research that, in turn, will enable a deeper understanding of the work of art and its conservation state. When sufficient data have been collected, and the consistency of the database fully tested, search algorithms will be developed making this new advanced analytical tool accessible to the conservation community, and not just to the photophysics experts. The works of art we have patiently studied are preserved in museums, archives, and other institutions; the research we have carried out has been possible by the generous support of the

directors and staff from DGARG-TT (Torre do Tombo), BPMP (Biblioteca Pu´blica Municipal do Porto), and BNP (Biblioteca Nacional de Portugal); the friendly environment they offer us, their trust in our work and in what science may contribute for a better conservation has been a continuous stimulus. It is a pleasure to thank Meredith Montague, Richard Newman, Aviva Burnstock, Klaas Jan van den Berg, Cecily Grzywacz, Jan Wouters, and the Molab team for all the colorful discussions and Professor Heather Lechtman for her kind interest. To them, Blake’s little flower. BIOGRAPHICAL INFORMATION Maria Joaõ Melo obtained her Ph.D. in Physical Chemistry, in 1995, from New University of Lisbon. In 1999, after a Post Doc at ICVBC-CNR in Florence, she joined the Conservation Unit at the New University of Lisbon, where she is responsible for the C&R scientific laboratory. Since 1999, she has also been a researcher at Requimte. Her research interests include the conservation of medieval illuminations and of Modern Art, namely, the study of the mechanisms of photodegradation in polymer systems and color paints. Other areas of interest are Color in Art and Nature and Semiochemistry. Ana Claro was born in 1978, in Portugal. She obtained her degree (2004) and her Ph.D. (2009) in Conservation and Restoration from Universidade Nova de Lisboa. She became a trainee assistant at this University (2005-2009) and joined the Associate Laboratory REQUIMTE-CQFB (FCT-UNL) in 2004 and the Medieval Studies Institute (FCSH-UNL) in 2005. Currently, she is at the Getty Conservation Institute as a visiting Post Doc, doing research on Asian organic colorants. Her research also concerns the study of materials applied mainly in illuminated manuscripts. FOOTNOTES * To whom correspondence should be addressed. E-mail: [email protected]. REFERENCES 1 Brandi, C. The cleaning of pictures in relation to patina, varnish and glazes. The Burlington Magazine 1949, 91, 183–188. 2 Reporting Highlights of the De Mayerne Programme; Boon, J. J., Ferreira, E. S. B., Eds.; NWO: The Hague, 2006. 3 http://www.eu-artech.org/. 4 Claro, A.; Melo, M. J.; Scha¨fer, S.; Seixas de Melo, J. S.; Pina, F.; van den Berg, K. J.; Burnstock, A. The use of microspectrofluorimetry for the characterization of lake pigments. Talanta 2008, 74, 922–929. 5 Claro, A.; Melo, M. J.; Seixas de Melo, J. S.; van den Berg, K. J.; Burnstock, A.; Montague, M.; Newman, R. Identification of Red Colorants in Cultural Heritage by Microspectrofluorimetry. J. Cult. Heritage 2010, 11, 27–34. 6 Halleux, R. Les Alchimistes Grecs: Papyrus de Leyde, Papyrus de Stockholm, Recettes; Les Belles Lettres: Paris, 2002 (1st ed., 1981). 7 Melo, M. J. History of natural dyes in the ancient mediterranean world. In Handbook of Natural Colorants; Bechtold, T., Mussak, R., Eds.; John Wiley & Sons: Chichester, 2009; Chapter 1. 8 Cardon, D. Natural dyes: sources, tradition, technology and science; Archetype Books: London, 2007. 9 (a) Seixas de Melo, J. S.; Takato, S.; Sousa, M.; Melo, M. J.; Parola, A. J. Revisiting Perkin’s Dye(s): The Spectroscopy and Photophysics of Two New Mauveine Compounds (B2 and C). Chem. Commun. 2007, 2624–2626. (b) Sousa, M.; Melo,

Vol. 43, No. 6

June 2010

857-866


865


10

11

12

13

14

15

M. J.; Parola, A. J.; Morris, P. J. T.; Rzepa, H. S.; Seixas de Melo, J. S. A study in mauve. Chem. Eur. J. 2008, 14, 8507–8513. (c) Ball, P. Bright Earth: Art and the Invention of Color; Farrar, Straus and Giroux: New York, 2002. (a) Seixas de Melo, J. S.; Moura, A. P.; Melo, M. J. Photophysical and Spectroscopic Studies of Indigo Derivatives in Their Keto and Leuco Forms. J. Phys. Chem. A 2004, 108, 6975–6981. (b) Sousa, M. M.; Miguel, C.; Rodrigues, I.; Parola, A. J.; Pina, F.; Seixas de Melo, J. S.; Melo, M. J. A photochemical study on the blue dye indigo: from solution to ancient Andean textiles. Photochem. Photobiol. Sci. 2008, 7, 1353–1359. (c) Clementi, C.; Nowik, W.; Romani, A.; Cibin, F.; Favaro, G. A spectrometric and chromatographic approach to the study of ageing of madder (Rubia tinctorum L.) dyestuff on wool. Anal. Chim. Acta 2007, 596, 46–54. Zhang, X.; Laursen, R. A. Development of mild extraction methods for the analysis of natural dyes in textiles of historical interest using LC-diode array detector-MS. Anal. Chem. 2005, 77, 2022–2025. (a) Jurasekova, Z.; Domingo, C.; Garcia-Ramos, J. V.; Sanches-Cortes, S. In situ detection of flavonoids in weld-dyed woll and silk textiles by surface-enhanced Raman scattering. J. Raman Spectrosc. 2008, 39, 1309–1312. (b) Canãmares, M. V.; Leona, M. Surface-enhanced Raman scattering study of the red dye laccaic acid. J. Raman Spectrosc. 2007, 38, 1259–1266. (c) Whitney, A. V.; Van Duyne, R. P.; Casadio, R. Identification and characterization of artists’ red dyes and their mixtures by surface-enhanced Raman spectroscopy. J. Appl. Spectrosc. 2007, 61, 994–1000. (d) Rosi, F.; Paolantoni, M.; Clementi, C.; Doherty, B.; Miliani, C.; Brunetti, B. G.; Sgamellotti, A. Subtracted shifted Raman spectroscopy of organic dye and lakes. J. Raman Spectrosc. 2009, 41, 452–458. Romani, A.; Clementi, C.; Miliani, C.; Brunetti, B. G.; Sgamellotti, A.; Favaro, G. Portable Equipment for Luminescence Lifetime Measurements on Surfaces. Appl. Spectrosc. 2008, 62, 1395–1399. Carlyle, L. Historically Accurate Reconstructions of Oil Painters’ Materials: An overview of the Hart Project 2002-2005. In Reporting Highlights of the De Mayerne Programme; Boon, J. J., Ferreira, E. S. B., Eds.; NWO: The Hague, 2006. Marke, C. The Art of All Colours: Mediaeval Recipe Books for Painters and Illuminators; Archetype Publications: Dorchester, 2001.

866


857-866

June 2010

Vol. 43, No. 6

16 http://www.biosciencetechnology.com/Archive/2003/07/Fluorescence-Mapping/. 17 Please see refs 7-10 from ref 4. 18 Favaro, G.; Clementi, C.; Romani, A.; Vickackaite, V. Acidichromism and ionochromism of luteolin and apigenin, the main components of the naturally occurring yellow weld: A spectrophotometric and fluorimetric study. J. Fluoresc. 2007, 17, 707–714. 19 Silva, L. C.; Claro, A.; Melo, M. J.; Cabrita, E. J.; Mafra, L. Using Solid and Liquid State NMR Techniques to Unveil the Secrets of Alizarin Lakes, 27th Annual Meeting on Dyes in History and Archaeology, 8-11 October 2008, Istanbul, Turkey. 20 (a) Moura, L.; Melo, M. J.; Casanova, C.; Claro, A. A Study on Portuguese Manuscript Illumination: The Charter of Vila Flor (Flower Town), 1512. J. Cult. Heritage 2007, 8, 299–306. (b) Miguel, C.; Claro, A.; Gonçalves, A.; Muralha, V. S. F.; Melo, M. J. A study on red lead degradation in the medieval manuscript, Lorvaõ Apocalypse (1189). J. Raman Spectrosc. 2009, 40, 19661973. 21 St. Cruz, St. Mamede of Lorvaõ, and St. Maria of Alcobaça are important in the context of the political strategy of the Reconquest, which experienced the creation of monasteries for the maintenance, under the Christian domination, of territories recently won back from the Muslims. 22 Laccaic acid A is the main chromophore present in lac dye, and as lake pigment it may be considered a weak emitter, but still detectable by microspectrofluorimetry. 23 http://www.eu-artech.org/files/MEDMAN-UserReport.pdf. 24 Paul, A. Paracas Ritual Atire: Symbols of Authority in Ancient Peru; University of Oklahoma Press: Norman, OK, 1990. 25 In all the samples analyzed by HPLC-DAD, the major chromophore was purpurin, but pseudopurpurin at 30% to 45% was always detected. 26 Grzywacz, C.; Bomin, S.; Yuquan, F.; Wouters, J. Development of identification stategies for Asian organic colorants on historic Chinese wall paintings. In 15th Triennial Meeting ICOM-CC; Bridgland, J., Ed.; Allied Publishers: New Delhi, 2008, 528-535.

Immunodetection of Proteins in Ancient Paint Media LAURA CARTECHINI,*,† MANUELA VAGNINI,‡ MELISSA PALMIERI,‡ LUCIA PITZURRA,§ TOMMASO MELLO,| JOY MAZUREK,⊥ AND GIACOMO CHIARI⊥ †

Istituto di Scienze e Tecnologie Molecolari - CNR, c/o Dipartimento di Chimica, Universita` di Perugia, via Elce di Sotto 8, 06123 Perugia, Italy, ‡Dipartimento di Chimica, Universita` di Perugia, Via Elce di Sotto 8, 06123 Perugia, Italy, §Dipartimento di Medicina Sperimentale e Scienze Biochimiche, Universita` di Perugia, Via del Giochetto, 06126 Perugia, Italy, |Dipartimento di Fisiopatologia Clinica, Universita` di Firenze, Viale Morgagni, 85, 50134 Firenze, Italy, ⊥Getty Conservation Institute, 1200 Getty Center Drive, Suite 700, Los Angeles, California 90049 RECEIVED ON NOVEMBER 27, 2009

CON SPECTUS

D

iagnostic immunology is a powerful tool, widely used in clinical and biochemical laboratories for detecting molecules. In recent years, the technique has been adaptated to materials sciences as a result of the extensive advances achieved in immunology. Today, many companies supply custom antibodies as well as new high-performance bioprobes for virtually any use. The idea of using immunodetection in the field of conservation science is not new. This analytical methodology is, in fact, particularly attractive for investigating biopolymers in painting materials; it is highly sensitive and selective with respect to the biological source of the target molecules. Among biopolymers, proteins have been widely used in the past as painting binders, adhesives, and additives in coating layers. An accurate assessment of these materials is necessary to obtain deeper insights into an artist’s technique as well as to design proper restoration and conservation methods. In spite of the diagnostic potential offered by immunodetection-based techniques, some analytical drawbacks had, until recently, limited their use in routine applications in conservation science. In this Account, we highlight the most important results achieved in our research on the development of analytical methodologies based on the use of enzyme-linked immunosorbent assay (ELISA) and immuno-fluorescence microscopy (IFM) techniques for the highly sensitive and specific identification of proteins in artistic and archeological materials. ELISA and IFM offer two alternative analytical routes to this final goal: ELISA provides a fast, cost-effective, quantitative analysis of microsamples put in solution, whereas IFM combines the immunodetection of the targeted molecules with the characterization of their spatial distribution. The latter approach is of great value in the stratigraphic investigation of paintings. We discuss the limits and strengths of these methodologies in the context of the complex matrixes usually found in the investigated materials and the prolonged aging that they have undergone. Immunology is a relatively new technique in conservation science, providing a rich new field for innovation. We see two areas that are particularly ripe for future contributions. The commercial manufacture of antibodies specifically tailored for use in cultural heritage studies holds enormous potential. Moreover, the need for further refinement of detection systems in immuno-fluorescence techniques, especially the suppression of the autofluorescence background in painting materials, offers an abundance of opportunities for researchers. Immunology is a relatively new technique in conservation science, providing a rich new field for innovation.

Published on the Web 05/03/2010 www.pubs.acs.org/acr 10.1021/ar900279d © 2010 American Chemical Society

Vol. 43, No. 6

June 2010

867-876


867

Immunodetection of Proteins in Ancient Paint Media Cartechini et al.

Introduction Until recently, scientific studies in the field of cultural heritage focusing on paintings have primarily concerned themselves with inorganic compounds.1,2 The investigation of organic components has been subjected to increased interest only in the past decade thanks to the development of new nondestructive and microdestructive techniques.3,4 Even so, a detailed characterization of natural organic materials in ancient paintings is still a challenging issue because of their intrinsic chemical complexity and their tendency to easily undergo degradation over long periods of time.5 Among naturally occurring organic substances, proteins have been widely used not only as binders but also as adhesives or as additives in coating layers. In particular, animal glue, egg (both yolk and albumen), and milk (or its byproduct casein) are mostly encountered in paintings. They contain collagen, ovalbumin (egg white), and casein, respectively, as the main distinctive proteins. Until now chromatography analysis seems to be the best established approach to provide detailed information on the wide class of natural organic compounds found in paintings (proteins but also oils, resins, waxes, and plant gums).6 In regard to proteins, procedures for both high-performance liquid chromatography (HPLC) and gas chromatography (GC) techniques, mainly combined with mass spectrometric (MS) detection, have been developed to distinguish among egg, glue, and milk/casein on the basis of the quantitative determination of their amino acidic profiles.7,8 This approach can be affected by the alteration of amino acid relative amounts induced by sample contamination. Furthermore, protein mixtures are difficult to be analytically resolved, and different biological sources of the same protein cannot be established. Recently, several new methods based on the proteomic approach have been developed, but they still need further experimentation for routine applications.9,10 Therefore, an alternative analytical approach that is simple, cost-effective, has minimal sample manipulation, and that can possibly resolve a complex mixture while being selective with respect to the biological source is highly desirable for protein recognition in artworks. Immunological techniques have the potential to become a powerful diagnostic tool in cultural heritage for highly specific and sensitive identification of proteins in microsamples of art and archeological materials.11-17 To that end, enzymelinked immunosorbent assay (ELISA),18,19 immuno-fluorescence microscopy (IFM),20 and immuno-chemiluminescence microscopy21 have all begun to be systematically used to 868


867-876

June 2010

Vol. 43, No. 6

develop analytical methodologies for applications in conservation science. This Account provides an overview of our research on the application of ELISA and IFM to the investigation of proteins in painting materials. ELISA is an accurate and sensitive immunochemical assay that can be used for the fast, easy, and costeffective measurement of targeted antigens in a painting sample brought to solution. ELISA typically uses antibodies conjugated to enzymes as detection reagents. These enzymes act on chromogenic or fluorogenic substrates that produce an amplified detectable signal. In the case of IFM, immuno-fluorescent probes are used to obtain a highly sensitive and specific detection of the targeted molecule and, at the same time, to image its spatial distribution. This is a particularly valuable approach for the stratigraphic investigation of paintings, although at the expense of sample preparation time. Our research work based on both methodologies is presented here in two separate sections where the main experimental aspects are described, and the most relevant results are reported and discussed for each approach.

ELISA Direct ELISA is the simplest among various methods,22 as the protein (antigen) is attached to a plastic solid phase (normally a 96-well microtiter plate) and an enzyme-labeled detection antibody is added. The detection antibody (named “primary antibody”) binds specifically to antigens in a recognized molecular sequence: the epitope of the antigen. In indirect ELISA, the primary antibody is not conjugated with enzymes but instead is targeted by enzyme-conjugated secondary antibodies specific to the primary antibody class (IgG, IgM) of the primary antibody. The use of secondary antibodies increases the specificity of ELISA and reduces the unspecific background due to the linkage of antibodies on the plastic surface of the well plate. The ELISA technique is used at the Getty Conservation Institute (GCI) to detect egg, mammal glue, casein, and plant gum in paint samples from works of art. It is a modified procedure based on literature17,23 and uses the indirect method; the procedure19 is summarized in Figure 1. A paint sample (between 100 and 300 µg, while wall painting samples should be larger, up to 1 mg, as they are likely to have less binding media) is first dissolved in a elution buffer and then diluted and added to wells of an ELISA well plate. The binder containing the antigen to be detected attaches to the plastic of the well and is processed for immunologic recognition.


FIGURE 1. Sketch of the ELISA methodology adopted for protein detection in painting samples: (1) a paint sample is first dissolved in an elution buffer and then diluted and added to wells of an ELISA well plate; (2) a primary antibody targeted to the particular bound antigen is added to the well (2 h, room temperature); (3) an enzyme-conjugated secondary antibody is added (2 h room temperature). Both steps 2 and 3 are followed by sample rinsing with water to remove any unbound antibody; (4) p-NPP is added to the wells, and a reaction stop buffer (0.75 M NaOH) is added. Absorbance is read at 405 nm using a multiwell spectrophotometric plate reader.

The antibody conjugated enzyme alkaline phosphatase is used for colorimetric detection by enzymatic conversion of the substrate p-nitrophenyl phosphate (p-NPP) to p-nitrophenol, which is a yellow dye at pH 8.5. The reaction color development depends on the amount of antigen present in the well and the amount of time the enzyme is left to be active. The addition of a buffer solution to stop the reaction allows for comparison of quantitative results from different ELISA assays. Specific research strategies at the GCI have focused on finding suitable antibodies to detect proteins in paint. Experimental efforts have focused primarily on (i) testing antibody specificity to recognize denatured aged proteins in paints; (ii) evaluating detection limits of the technique to estimate the appropriate sample quantity to be collected; (iii) decreasing contamination events (false positives) and investigating cases of false negatives; and (iv) selecting a suitable blocking solution able to reduce background due to antibody unspecific binding, or to its cross-reactivity with different proteins having similar epitopes. Finding optimal ELISA conditions requires extensive testing for each type of binding media to identify in one assay all those present in a sample, with the lowest background levels and the highest detection sensitivity. “Sea Block” (Pierce Chemical) in phosphate buffer solution (PBS) was chosen as the blocking agent giving the lowest background levels. The primary antibodies, corresponding secondary antibodies, dilutions, and detection limits are shown in Table

1. Sample absorbance is expressed as optical density at 405 nm (OD405nm). For each antibody pair, background levels were evaluated in 20 blank wells; they gave OD405nm averaged values ranging between 0.11 and 0.16. The threshold values for binder detection in Table 1 were estimated as the blank average OD405nm plus three times the blank OD405 nm standard deviation. Each antibody chosen for use in ELISA was tested on several animal sources of glue, egg, and casein.19 Eggs from several species of bird were tested for ovalbumin in whole egg, egg yolk, and egg white; all were positive except for egg yolk, as it lacks significant amounts of ovalbumin. Milk was tested from human, cow, goat, and buffalo, and all reacted positive for casein except human milk. The limit of detection is dependent on the conservation degree of the binder as shown in Table 1 by the increase of the detection limit in the ELISA assays performed on 47 day artificially aged binders. Aging and pigments adversely affect the sensitivity of ELISA due to loss of antigenic sites, crosslinking, and nonsolubility.8,25 The GCI’s research initially focused on nonpigmented samples exposed to different conditions of UV light and humidity: exposure to UV light with high relative humidity (69-75% RH) consistently had the most dramatic effect on all binders, while exposure to UV light at low humidity (20-25% RH) or to fluctuating humidity only (12-85% RH) had a much less effect. More recently, ELISA was used to evaluate aging degradation effects on binders in the presence of pigments. Paint samples were produced by mixing three common proteinaceous binders with ten pigments usually found in wall paintings. They were prepared using traditional recipes,26 applied to glass slides, and artificially aged for 47 days. The decrease of the detectable amount of binding media after artificial aging is shown in Figure 2. The measurement is based on known wt % values of binder in paint before aging and measured values after aging. Quantitation of the detectable amount of binder was obtained by calibration curves from 47 day artificially aged unpigmented standards. As evidenced by Figure 2, pigments interacted differently with binding media. Results in charcoal/egg, charcoal/glue, red ochre/casein, red ochre/glue, and vermilion/egg exhibited no measurable change after aging. Even though a few of the paints showed up to a 100-fold decrease in the expected values, ELISA was still able to identify casein and egg in all aged paints. Only the animal glue antibody turned out to be very sensitive to binder aging, so that half of the samples gave a complete loss of signal. The specific role played by different pigments in the degradation of binders is still not clear.5,8,25 Vol. 43, No. 6

June 2010

867-876


869


TABLE 1. Primary Antibodies, Corresponding Secondary Antibodies, Dilutions, and Detection Limits of Binders (Fresh and after 47 Days of Artificial Aging) Tested in ELISA primary antibody catalogue # (company) # AB1225 (Chemicon) #RCAS-10° (Immunology Consultants) #T40113R (Gentaur) #JIM13 (Carbo Research Univ. of Georgia)

threshold values (mean OD405nm+ 3 SD)b

detection limit (ngc)

binder tested

species reaction

dilutiona

secondary antibody (dilutiona)

0 day

47 days

ovalbumin, whole egg casein, milk curd

most birds (not yolk) buffalo, cow, and goat (not human)

1:800 1:800

rabbit IgG (1:500) rabbit IgG (1:500)

0.21 0.15

0.6 0.3

1 4

collagen, animal glue plant gum, gum arabic

most mammals (not rabbit) most Acacia spp. and fruit tree (not Tragacanth)

1:400 1:50

rabbit IgG (1:500) rat IgG (1:500)

0.19 0.17

3 7

12 12

a Primary and secondary antibodies were diluted in a solution of Sea Block in PBS (1:10 in volume). b Values are the blank mean OD405nm + 3 SD of 20 independent determinations. c ELISA was performed on solutions of pure binder in an elution buffer (see Figure 1). The detection limit was calculated by dividing the weight of the binder by the highest dilution factor that still yielded a positive result.

FIGURE 2. Changes in detectable amounts of binding media when exposed to pigments after 47 days of artificial aging. A Xenon-arc Ci400 weatherometer, with filters to stop far UV-light, irradiance of 0.5 w/m2, RH of 60-80%, 40 °C chamber temperature, was used. X-axis shows the pigments and their binding media: black ) casein, light gray ) egg, and dark gray ) animal glue. The Y-axis shows the ratio between the measured concentration of binder (expressed as wt % binder in the paint) to the known binder wt % concentration before artificial aging. No change ) no significant decrease in % binder; 10×, 100×, and 1000× decrease ) 10, 100, and 1000 times decrease in % binder, respectively; complete loss of signal ) no binder detected.

Recently, we also tested an antibody (see Table 1) for the detection of gums. Their recognition is a challenging task even for chromatographic techniques.6 ELISA results from several plant sources showed positive for gum Arabic and fruit tree gums, while gum Tragacanth was negative. Pigment and aging degradation effects were also evaluated after 47 days of artificial aging of pigmented plant gums (same conditions of Figure 2), obtaining positive results for half of the samples. These highlighted the need to test all new antibodies both for aged gums and species reactivity. During the course of research, various works of art were tested by ELISA;19 the results are summarized in Table 2. GC-MS24,25 was used to support ELISA results, and it proved that the assay is valid for ancient materials. Yet, in some cases, 870


867-876

June 2010

Vol. 43, No. 6

ELISA provided discrimination of different proteinaceous components, not distinguished by GC-MS (Table 2). The oldest successfully analyzed paint sample to date is from the Tomb of Nefertari, nearly 3000 years old; it tested positive for plant gums. Other Egyptian objects dating from 200 B.C. to 50 A.D. tested positive for mammal glue and egg as the binder. Overall, the results were very encouraging, but false negatives were obtained for few samples. More recently, the focus of our research has been to eliminate false negatives when testing artificially aged pigmented samples. For example, robust extraction procedures such as sonication and heat were tried in an attempt to increase the solubility of cross-linked binders. Most likely, new antibodies need to be made specifically to identify aged binders.27 Of course, those


TABLE 2. Results of Various Artworks Tested by ELISA GC/MSa

ELISA artwork

origin

pigment

identified binder

OD405

identified binder

wt %

Tempera portraits on wood, Getty museum #74.AP.20

Egypt, c. 200 A.D.

black red

glue glue egg

0.6 0.4 0.5

animal glue animal glue

Jacob Lawrence, Paper Boats

American, 1949

black

egg casein

0.6 0.8

NTc

Potala Palace wall paintings

Tibet 17th C. to present

copper green yellow lead red

glue glue NRb

0.9 0.6 –

animal glue animal glue animal glue

2 1.5 0.8

model of the plaza (maquetas) with inlayed and painted wood

Tomb of Huaca de la Luna, Peru c.14th C.

red gold white

plant gum plant gum plant gum

0.6 0.4 0.2

polysaccharide polysaccharide polysaccharide

1 1 1

Maria, das Kind anbetend by Bevilacqua Gema¨ldegalerie Alte Meister in Dresden

1480 A.D.

brown

0.6 0.6 0.4 0.6 0.7

protein

7

dark green

glue egg egg plant gum egg

protein

gold leaf on blue

NRb

Wall painting, Church of the Mission of St. Frances by Andrea Pozzo

Italy c.17th C.

copper green

egg

0.8

egg

Nefertari tomb

Egypt c. 1000 B.C.

red iron oxide Egyptian green charcoal black Egyptian blue yellow ochre

plant gum plant gum plant gum plant gum NRb

0.7 1 1 0.5 –

Acacia gum NTc NTc NTc Acacia gum

Cartonnage, #79374 and #79385 Petrie Museum of Egyptian Archaeology

Egypt c. 200 B.C.

Green Earth

egg glue

2.0 2.0

NTc

painted wood head HUCSM A0939 Skirball Center

Egypt c. 200 A.D.

red

glue

2.0

Animal glue

2 2 –

– 2 – 4 – – – 2 – 9

a

Paint samples analyzed by GC-MS are reported as wt % of binder in paint. GC-MS sample compositions were compared to those of standard reference binding media using the method of correlation coefficients.25 b No result: the analysis failed to yield data or no binder was detected. c Not tested: analysis not done.

antibodies that are readily commercially available and applicable to binding media are highly desirable, but the need for tailored antibodies for cultural heritage is compelling, yet expensive.

IFM We focused on the development of an IFM protocol to identify and localize proteins in painting stratigraphy based on the indirect method. Similarly to indirect ELISA, in this approach, the primary antibody forming the immuno-complex with the targeted protein is detected by a secondary antibody which is conjugated with a fluorophore (Figure 3). The analytical protocol has been optimized for recognition of egg white, bovine casein,20,28 and, recently, animal glue by comparing different conditions for antibody dilution, incubation time, and temperature. From one microsample (100 µm and needle-shaped, cubic, rectangular, hexagonal, or shapeless, respectively. Glasses of composition M1 were attacked at a much higher degree due to their higher concentration of network modifier ions and exhibited deep cracks and detached parts of hydrated glass material in some cases.39,40 Similar results are also reported by Munier et al.41 for potash-lime-silica glass, as well as mixed alkali soda-potash-silica glass, which were exposed for times up to 12 months in the polluted urban atmosphere of Paris. Simultaneous measurements of the concentrations of pollutants should enable a correlation with the chemistry of the weathering products. Generally, the formation of sulfates was favored by high contents of SO2 in the ambient atmosphere and higher humidity. The important role of SO2 in glass weathering was also demonstrated by investigations of historical glasses from the Cathedral of Leoń, Spain, which were removed during restoration works in the 19th century and stored in the meantime. Interestingly, the weathering crusts consisted mainly of carbonates,42 which are assumed to be an intermediate product in the degradation mechanism. However, Weigel43 demonstrated that these carbonates are converted to sulfates as soon as SO2 is present in the weathering atmosphere. As mentioned before, also the deposition and reactions of airborne particulate matter onto glass surfaces has to be considered in glass deterioration assessment. In a study on the influence of particulate matter in combination with SO2 and humidity on the weathering process of potash-lime-silica glass of composition M1 by means of SEM/EDX44 in the framework of the MULTI-ASSESS project (Model for multipollutant impact and assessment of threshold levels for cultural heritage, project number EVK4-CT-2001-0004445) a significantly higher degree of glass deterioration could be observed on samples treated with atmospheric dust in comparison to reference specimens without particulates (Figure 8). Again, the dominant weathering products were syngenite and gypsum. Of special importance was the observation that the weathering rate of dust-treated glass samples was approximately the same when exposed at 50% relative humidity compared with 70% or 100%, whereas for exposures without particulates significantly fewer and smaller weathering products have formed at the lower RH-level. This effect is probably due to the presence of deliquescent constituents in the dust and may be fur-

FIGURE 8. SE-images of weathered glass surfaces after 250 h of exposure (glass composition M1, 100% RH, no SO2) without (a) and with (b) atmospheric dust. A significantly higher degree of weathering and larger weathering products can be observed in the latter case.

ther evidence for the risk potential of dusty urban atmospheres for historical glasses. Investigations concerning the initial stages of the glass weathering process were carried out using AFM, SIMS, and SEM/EDX33,46,47 for potash-lime-silica glasses of composition M1. While the glass surface remained unaltered under an inert atmosphere of N2 for exposure times of more than 1 h, the application of a humid N2 atmosphere (50% RH) lead to the formation of hydrated and swollen glass material sized approximately 100-200 nm within minutes. Additions of SO2, NO2 or both to the humid gas stream showed secondary crystalline phases as additional weathering products, which completely cover the glass surface after typically some 10 h (Figure 9). Figure 10 depicts the depth distributions of K, Ca, and H in polished model glass M3 (60 wt % SiO2, 25 wt % CaO, and 15 wt % K2O). The leached layer, where the mono- and bivalent modifier ions are depleted and hydrogen is incorporated into the silicate structure, can be clearly seen. In comparison to Figure 10, Figure 11 shows the results of leaching experiVol. 43, No. 6

June 2010

916-926


923

Degradation of Glass Artifacts Melcher et al.

FIGURE 10. Time-of-flight (TOF)-SIMS depth profile of glass M3 (composition 60 wt % SiO2, 25 wt % CaO, and 15 wt % K2O) after leaching for 30 min in 10-3 N HNO3. Decreased intensities of the K and Ca signals and increased H intensities (secondary axis) can be observed in the outmost glass layers. The thickness of the hydrated layer is approximately 25 nm.

FIGURE 11. TOF-SIMS depth profiles of glass M3 for K: untreated and after 30, 60, and 90 min of leaching in 10-3 N HNO3. An increase of the thickness of the leached layer with leaching time can be observed.

FIGURE 9. AFM images depicting a sketch (10 × 10 µm2) of a cleaved glass surface (glass composition M1) under dry nitrogen (image a) and after exposure to humid (RH ) 50%) N2 stream containing 1 ppm SO2 (images b and c after t ) 14 and 401 min of exposure, respectively). Within minutes, two kinds of features can be observed: swollen glass material sized in the submicrometer range, as well as crystalline-like features covering large parts of the surface after approximately 6 h of exposure.

ments of glass M3 in 10-3 N HNO3 for 30, 60, and 90 min, which leads to an increase of the thicknesses of the leached layers. 4.2. Quantification of Glass Weathering. A final important question concerns the quantitative estimation of glass 924


916-926

June 2010

Vol. 43, No. 6

deterioration by atmospheric corrosion. As a numerical measure for the degree of weathering of a glass specimen, the leaching depths of the network modifier ions K+ and Ca2+ (determined by linescan measurements of the cross-sectioned specimens using SEM/EDX) proved to be reliable for potash-lime-silica glasses.48-50 As also the climatic and environmental parameters such as temperature, relative humidity, pH, and amount of precipitation and the concentrations of the acidifying gases SO2, NO2, and O3 were measured at the exposure sites, dose-response relationships could be set up using multiple linear regression (MLR). It is based on the assumption of a relationship between a dependent or response variable, y, and p predictor variables, xi (p g 1) of the form


y ) b0 + b1x1 + b2x2 + ... + bpxp + ε

(6)

with a random error component ε and the regression coefficients bi. Its solution is the so-called least-squares estimator of β () βˆ ), which is given in matrix notation by

βˆ ) (XTX)-1XTy

(7)

In the present case, the leaching depths of K and Ca, d(K) and d(Ca), served as response variable y, whereas products of the exposure time and the climatic and environmental parameters were used as predictor variables (“dose”). Results of these calculations for the glass M1 exposed for 6 and 12 months are shown in the eqs 8 and 9. The concentrations of NO2 and SO2 (in µg/m3) as well as the climatic parameters temperature (T, in °C) and relative humidity (RH, %) enter the equation and can be considered as significantly influencing the leaching depths of K and Ca (measured in micrometers) and hence the rate of deterioration of glasses of this composition. Typically, slightly higher values for the leaching depth of K were measured (and are predicted by the eqs 8 and 9) compared with those of Ca. This can be explained by the stronger bonding of the bivalent ion Ca2+ to the glass network.

d(K) ) -0.64 + (0.03RH + 0.04c(SO2))√t -

(

0.05T + 2.03

d(Ca) ) -0.79 + (0.03RH + 0.03c(SO2))√t -

(

0.04T + 1.91

)

1 t (8) c(NO2)

)

1 t (9) c(NO2)

5. Outlook and Conclusion The preceding chapters made obvious that estimations of the durability of glasses under ambient conditions are difficult to obtain and are only valid under certain assumptions concerning the glass composition and the exposure conditions. Dose-response relations presented in the previous chapter might therefore be a useful tool for mapping areas of increased risk for glass materials of specific composition. Similar approximations exist for soda-lime-silica glass.26,51 Hence a first action to be undertaken for an enduring preservation of objects of our cultural heritage must be a substantial reduction of pollutants and particle emissions, as it was the case for SO2 in the last decades. It shall be mentioned that today several other concepts for a reduction of deterioration of outdoor glass artifacts caused by weathering are in use. For instance, protective glazings have proved to be an efficient way to reduce the corrosive attack to window panes at Sainte Chapelle in the city center

of Paris.52 A further approach for future protection of endangered glass artifacts are superficial coatings. Promising results were obtained for sol-gel silica coatings, which are transparent, colorless, and low-reflective and which show an effective adhesion to the glass.53,54 The authors want to thank Prof. Dr. Gernot Friedbacher and Prof. Dr. Herbert Hutter (Institute of Chemical Technologies and Analytics, Vienna University of Technology) for enabling the SEM/EDX, AFM, and SIMS measurements and Prof. Dr. Johann Wernisch (Institute of Solid State Physics, Vienna University of Technology) for the ESEM investigations. Furthermore, the authors want to thank all partners in the UNECE/ICP-Materials and MULTI-ASSESS projects and the European Union for the financial support. BIOGRAPHICAL INFORMATION Michael Melcher received his M.Sc. (2002) and Ph.D. (2005) in chemistry from the Vienna University of Technology for works on investigations on archaeological silver findings and the weathering of low-durability potash-lime-silica glasses, respectively. His research interests also include multivariate statistics of chemical and archaeometric data. Currently he is a co-worker at the Institute of Science and Technology in Art at the Academy of Fine Arts, Vienna, Austria. Rita Wiesinger studied biotechnology at the University of Applied Sciences Krems, Austria, where she received her diploma in 2006. Since October 2006, she has been a Ph.D. student of technical chemistry at the Vienna University of Technology and research assistant at the Academy of Fine Arts, Vienna. Her research interests include the development and application of in situ methods to study the mechanisms occurring at the metal/ atmosphere interface during atmospheric corrosion. Manfred Schreiner studied chemistry at the Vienna University of Technology, where he received an engineering degree. After his Ph.D. in materials science, he was as a postdoctoral fellow at UCSD (University of California, San Diego, CA). In 2000, he became full professor and head of the Institute of Science and Technology in Art of the Academy of Fine Arts in Vienna, Austria. His main research interests are degradation (corrosion) of materials in historic and contemporary art, nondestructive (noninvasive) material analysis, and documentation of objects of art and archaeology using visible, infrared, UV, and X-ray radiations. FOOTNOTES * Tel +43-(0)1-58816-8619, e-mail [email protected] (M. Melcher); tel +43-(0)158816-8616, e-mail [email protected] (R. Wiesinger); tel +43-(0)1-58816-8600, e-mail [email protected] (M. Schreiner); fax +43-(0)1-58816-8699. REFERENCES 1 Newton, R. G. The durability of glass - a review. Glass Technol. 1985, 26, 21–38. 2 Frank, S. Glass and Archaeology; Academic Press: London, 1982. 3 Zachariasen, W. H. The atomic arrangement in glass. J. Am. Chem. Soc. 1932, 54, 3841–3851.

Vol. 43, No. 6

June 2010

916-926


925


4 Scholze, H. Glas - Natur, Struktur und Eigenschaften; Springer: Berlin, 1977. 5 Douglas, R. W.; Isard, J. O. The action of water and of sulphur dioxide on glass surfaces. J. Soc. Glass Technol. 1949, 33, 289–335. 6 Clark, D. E.; Pantano, C. G.; Hench, L. L. Corrosion of glass; Books for industry: New York, 1979. 7 Rana, M. A.; Douglas, R. W. The reaction between glass and water. Part 1. Experimental methods and observations. Phys. Chem. Glasses 1961, 2, 179–195. 8 Rana, M. A.; Douglas, R. W. The reaction between glass and water. Part 2. Discussion of the results. Phys. Chem. Glasses 1961, 2, 196–205. 9 Boksay, Z.; Bouquet, G.; Dobos, S. The kinetics of the formation of leached layers on glass surfaces. Phys. Chem. Glasses 1968, 9, 69–71. 10 Doremus, R. H. Interdiffusion of hydrogen and alkali ions in a glass surface. J. NonCryst. Solids 1975, 19, 137–144. 11 Doremus, R. H. Glass Science, 2nd ed.; John Wiley & Sons: New York, 1994. 12 Douglas, R. W.; El-Shamy, T. M. M. Reactions of glasses with aqueous solutions. J. Am. Ceram. Soc. 1967, 50, 1–9. 13 Lanford, W. A.; Davis, K.; Lamarche, P.; Laursen, T.; Groleau, R.; Doremus, R. H. Hydration of soda-lime glass. J. Non-Cryst. Solids 1979, 33, 249–266. 14 March, P.; Rauch, F. Hydration of soda-lime glasses studied by ion-induced nuclear reactions. Nucl. Instrum. Methods Phys. Res. B 1986, 15, 516–519. 15 El-Shamy, T. M. The chemical durability of K2O-CaO-MgO-SiO2 glasses. Phys. Chem. Glasses 1973, 14, 1–5. 16 Greiner-Wronowa, E.; Stoch, L. Influence of environment on surface of the ancient glasses. J. Non-Cryst. Solids 1996, 196, 118–127. 17 El-Shamy, T. M.; Lewins, J.; Douglas, R. W. The dependence on the pH of the decomposition of glasses by aqueous solutions. Glass Technol. 1972, 13, 81–87. 18 Schreiner, M. Secondary ion mass spectrometer analysis of potash-lime-silica glasses leached in hydrochloric and sulfuric acids. J. Am. Ceram. Soc. 1989, 72, 1713–1715. 19 Ruska, E. The Development of the Electron Microscope and of Electron Microscopy. Nobel Lecture, December 8, 1986. 20 Castaing, R. Application des sondes electronique a une methode d’ analyse ponctuelle chimique et cristallographique. Ph.D. Thesis, University Paris, 1951. 21 Ogilvie, R. E. In Proceedings of the Seminar (September 7-16); Young, W. J., Ed.; Museum of Fine Arts: Boston, MA, 1965, pp 223-229. 22 Ogilvie, R. E. In Proceedings of the Seminar (June 15-19); Young, W. J., Ed.; Museum of Fine Arts: Boston, MA, 1970, pp 84-87. 23 Bethge, H., Heydenreich, J., Eds.; Electron Microscopy in Solid State Physics; Elsevier: Amsterdam, 1987. 24 Reed, S. J. B. Electron Microprobe Analysis; University Press: Cambridge, U.K., 1987. 25 Fearn, S.; McPhail, D. S.; Oakley, V. Room temperature corrosion of museum glass: an investigation using low-energy SIMS. Appl. Surf. Sci. 2004, 231-232, 510– 514. 26 Fearn, S.; McPhail, D. S.; Hagenhoff, B.; Tallarek, E. TOF-SIMS analysis of corroding museum glass. Appl. Surf. Sci. 2006, 252, 7136–7139. 27 Darque-Ceretti, E.; Aucouturier, M. In Non-destructive Microanalysis of Cultural Heritage Materials; Janssens, K., Van Grieken, R., Eds.; Wilson & Wilson’s: Amsterdam, 2004; pp 397-461. 28 Dowsett, M.; Adriaens, A. The role of SIMS in cultural heritage studies. Nucl. Instrum. Methods Phys. Res. B 2004, 226, 38–52. 29 Niehuis, E.; Grehl, T. In TOF-SIMS; Vickermann, J. C., Briggs, D., Eds.; IM Publications: Chichester, U.K., 2001; pp 753-778. 30 Grehl, T.; Moellers, R.; Niehuis, E. Low energy dual beam depth profiling. Appl. Surf. Sci. 2003, 203, 277–280. 31 Binnig, G.; Quate, C. F.; Gerber, Ch. Atomic force microscope. Phys. Rev. Lett. 1986, 56, 930–933. 32 Arribart, H.; Abriou, D. Ten years of atomic force microscopy in glass research. Ceram.-Silik. 2000, 44, 121.

926


916-926

June 2010

Vol. 43, No. 6

33 Schmitz, I.; Schreiner, M.; Friedbacher, G.; Grasserbauer, M. Tapping-mode AFM in comparison to contact-mode AFM as a tool for in situ investigations of surface reactions with reference to glass corrosion. Anal. Chem. 1997, 69, 1012–1018. 34 Schmitz, I.; Schreiner, M.; Friedbacher, G.; Grasserbauer, M. Phase imaging as an extension to tapping mode AFM for the identification of material properties on humidity-sensitive surfaces. Appl. Surf. Sci. 1997, 115, 190–198. 35 Geilmann, W.; Berthold, H. J.; To¨lg, G. Beitra¨ge zur Kenntnis alter Gla¨ser V. Die Verwitterungsprodukte auf Fensterscheiben. Glastechn. Ber. 1960, 33, 213–219. 36 Bettembourg, J.-M. In Conservation within historic buildings: preprints of the contributions to the Vienna Congress, 7-13 September 1980; Bromelle, N. S., Thompson, G., Smith, P., Eds.; London, 1980; pp 93-95. 37 Newton, R. G. The weathering of medieval window glass. J. Glass Stud. 1975, 17, 161–168. 38 Swedish Corrosion Institute, ICP on effects on materials including historic and cultural monuments, Report No. 1: Technical manual, Stockholm, 1988. 39 Melcher, M.; Schreiner, M. Statistical evaluation of potash-lime-silica glass weathering. Anal. Bioanal. Chem. 2004, 379, 628–639. 40 Melcher, M.; Schreiner, M. In Proceedings of the MULTI-ASSESS Project Workshop “Cultural Heritage in the City of Tomorrow”. Kucera, V., Tidblad, J. Yates, T. Watt, J., Eds.; Swedish Corrosion Institute: Stockholm, 2004; pp 103-133. 41 Munier, I.; Lefe`vre, R.; Losno, R. Atmospheric factors influencing the formation of neocrystallisations on low durability glass exposed to urban atmosphere. Glass Technol. 2002, 43C, 114–124. 42 Carmona, N.; Villegas, M. A.; Fernandez Navarro, J. M. Characterization of an intermediate decay phenomenon of historical glasses. J. Mater. Sci. 2006, 41, 2339–2346. 43 Weigel, H.-J. Korrosionsuntersuchungen an Modellgla¨sern fu¨r mittelalterliche Zusammensetzung. Ph.D. Thesis, University Erlangen, 1980. 44 Melcher, M.; Schreiner, M.; Kreislova, K. Artificial weathering of model glasses with medieval compositions - an empirical study on the influence of particulates. Phys. Chem. Glasses: Eur. J. Glass Sci. Technol. 2007, 49, 346–356. 45 Please visit the project’s web-page for more information: http://www.corr-institute.se/MULTI-ASSESS/web/page.aspx. 46 Schreiner, M.; Woisetschla¨ger, G.; Schmitz, I.; Wadsak, M. Characterization of surface layers formed under natural environmental conditions on medieval stained glass and ancient copper alloys using SEM, SIMS and atomic force microscopy. J. Anal. Atom. Spectrom. 1999, 14, 395–403. 47 Schreiner, M.; Schmitz, I. Surface analytical investigations on naturally weathered medieval stained glass. Riv. Staz. Sper. Vetro 2000, 6, 15–22. 48 Melcher, M.; Schreiner, M. Evaluation procedure for leaching studies on naturally weathered potash-lime-silica glasses with medieval composition by scanning electron microscopy. J. Non-Cryst. Solids 2005, 351, 1210–1225. 49 Melcher, M.; Schreiner, M. Leaching studies on naturally weathered potash-limesilica glasses. J. Non-Cryst. Solids 2006, 352, 368–379. 50 Melcher, M.; Schreiner, M. Quantification of the influence of atmospheric pollution on the weathering of low-durability potash-lime-silica glasses. Pollut. Atmos. 2007, 49, 13–22. 51 Rogers, P.; McPhail, D.; Ryan, J. A quantitative study of decay processes of Venetian glass in a museum environment. Glass Technol. 1993, 34, 67–68. 52 Godoi, R. H. M.; Kontozova, V.; Van Grieken, R. The shielding effect of the protective glazing of historical stained glass windows from an atmospheric chemistry perspective: Case study Saint Chapelle, Paris. Atmos. Environ. 2006, 40, 1255– 1265. 53 Bianco, B. D.; Bertoncello, R.; Bouquillon, A.; Dran, J.-C.; Milanese, L.; Roehrs, S.; Sada, C.; Salomon, J.; Voltolina, S. Investigation of sol-gel silica coatings for the protection of ancient glass: interaction with glass surface and protection efficiency. J. Non-Cryst. Solids 2008, 354, 2983–2992. 54 Carmona, N.; Villegas, M. A.; Fernandez Navarro, J. M. Protective silica thin coatings for historical glasses. Thin Solid Films 2004, 458, 121–128.

The Coordinated Use of Synchrotron Spectroelectrochemistry for Corrosion Studies on Heritage Metals ANNEMIE ADRIAENS*,† AND MARK DOWSETT‡ †

Department of Analytical Chemistry, Ghent University, Krijgslaan 281-S12, B-9000 Ghent, Belgium, and ‡Department of Physics, University of Warwick, Coventry CV4 7AL, United Kingdom RECEIVED ON OCTOBER 30, 2009

CON SPECTUS

C

orrosion is a major source of degradation in heritage metal objects, and any remedial measures are subject to a strong (Western) ethic that favors conservation as opposed to restoration. Accordingly, major scientific challenges exist for developing appropriate treatment methods to stabilize and protect artifacts after they are recovered from an archaeological site, both before and during their display or storage in a museum. Because inappropriate treatments can cause irreversible damage to irreplaceable objects, it is crucial that the chemical processes involved are fully understood and characterized before any preservation work is undertaken. In this regard, large infrastructural facilities such as synchrotrons, neutron sources, and particle accelerators provide a wealth of analytical possibilities, unavailable in smaller scale laboratories. In general, the intensity of the radiation available allows measurements on a short time scale or with high spatial resolution (or both), so heterogeneous changes induced by a chemical process can be recorded while they occur. The penetrative nature of the radiation (e.g., X-rays, protons, or neutrons) also allows a sample to be studied in air. If necessary, complete artifacts (such as paintings or statuettes) can be examined. In situ analysis in a controlled environment, such as a liquid or corrosive atmosphere, also becomes an exciting possibility. Finally, there are many complementary techniques (local atomic structure or crystal structure determination, macroscopic 3-D imaging (tomographies), imaging chemical analysis, and so on) so the many distinct details of a problem can be thoroughly explored. In this Account, we discuss the application of this general philosophy to studies of corrosion and its prevention in cultural heritage metals, focusing on our recent work on copper alloys. More specifically, we use synchrotron-based techniques to evaluate the use of corrosion potential measurements as a possible monitoring method for copper-based objects recovered from marine environments. The extraction of chlorides from such artifacts is a process that must take place before the artifacts are put on display or stored, because air exposure of untreated metal will result in severe damage or loss in as little as a few weeks. Chloride is removed by soaking the artifact for up to two years in tap water or dilute sodium sesquicarbonate, with regular solution changes. Our research supports the effectiveness of this treatment for thin nantokite (copper(I) chloride) layers, but it raises questions for copper hydroxychlorides (atacamite and paratacamite), especially when these minerals are trapped in fissures. Electrochemical parameters such as the corrosion potential are shown to be insensitive to the physical presence of large hydroxychloride coverages if they overlie a cuprite (Cu2O) layer. X-ray absorption spectroscopy proves to be a good monitor for the chloride in solution over the working electrode, whereas X-ray diffraction offers the potential for real-time measurement of the surface chloride composition. In principle, the two techniques together offer the possibility of monitoring surface and fluid levels simultaneously.


Vol. 43, No. 6

June 2010

927-935


927

Synchrotron-Based Corrosion Studies Adriaens and Dowsett

Introduction The physical part of our cultural heritage is deteriorating faster than it can be conserved, restored, or studied. Assets are being lost, or are at risk, through natural processes of decay (sometimes accelerated by poor environmental control); environmental disasters (sometimes exacerbated by human activity); direct effects of increased public access (without commensurate conservation measures); conservation/preservation procedures whose long-term effects were damaging; and simple negligence, looting, and war. While there is a general agreement that action is required to halt or mitigate decline, few people actually realize that high level research and technology play an essential role in protecting our cultural heritage. In fact, advanced analytical methods are an essential prerequisite in this field, as they provide the means to study the structural chemistry of the processes involved. They can, additionally, contribute to the development of simple diagnostic techniques necessary for practical applied conservation, help establish authenticity, and, by identifying materials and processes, allow us to reach back through time and develop a deeper understanding of the craftsmanship and the technology used. Infrastructural facilities such as synchrotrons, neutron sources, and particle accelerators provide a wealth of analytical possibilities which are unavailable in smaller scale laboratories. In general, the intensity of the radiation available allows measurements on a short time scale or with high spatial resolution, or both, so that heterogeneous changes induced by a chemical process can be recorded while they are happening. Again, the penetrative nature of the radiation (e.g., X-rays, protons, neutrons) allows a sample to be studied in air, and, because there is no need for it to fit within a confined space in an analytical chamber, whole artifacts may sometimes be studied, provided the logistical problems of transporting them to the facility can be solved. The probe intensity also opens up the exciting possibility of analysis in situ in a controlled environment, for example, during immersion in a liquid or while exposed to a corrosive agent. Additionally, there are many complementary techniques available, giving local atomic structure, crystal structure, macroscopic 3-D imaging (tomographies), imaging chemical analysis, and so on, so that the details of a problem can be thoroughly explored. The use of ion beam analysis (IBA) techniques in the field of cultural heritage goes back for more than 20 years, with particle induced X-ray emission (PIXE) being the most popular fingerprinting technique for trace elements. One of the first proceedings of a workshop on the use of small accelerators in 928


927-935

June 2010

Vol. 43, No. 6

archaeometry was published in 1986 in the journal Nuclear Instruments and Methods in Physics Research, Section B.1 The workshop coincided with the installation of a 2 MV tandem accelerator AGLAE (Accelerateur Grand Louvre pour l’Analyze Ele´mentaire) in the basement of the Louvre museum in Paris which has since been dedicated exclusively to cultural heritage research.2 Since 1986, several other proceedings, for example, refs 3 and 4, and individual papers have been published in peer-reviewed journals, all demonstrating the vast interest in using IBA techniques in the area of heritage science. In 1995, COST (one of the EU instruments for funding research networks) dedicated a specific action to the topic. It ran for five years (1995-2000) under the title “COST G1: Application of Ion Beam Analyses to Art and Archaeological Objects” and involved the participation of 12 European countries. The output of this network resulted in ca. 30 joint papers and two monographs.5,6 Within the sixth and seventh Framework Program of the EU, the EU-ARTECH project and its successor CHARISMA, a consortium among 13 internationally distinguished European infrastructures dedicated to artwork conservation, offer a coherent set of transnational access programs including one to AGLAE.7 At the same workshop in 1986, a paper was presented on new possibilities in cultural heritage research using synchrotron facilities,8 and by 1996 the number of synchrotron-related papers in the field of cultural heritage was increasing rapidly. Today, more than 50 papers per year9 are published in this area. The main techniques used are X-ray fluorescence (SR-XRF), absorption (XAS), and diffraction (SR-XRD) and infrared spectroscopies. In 2004, an interface dedicated to archeology and cultural heritage was launched at Synchrotron SOLEIL, which aims at facilitating the access of researchers to the synchrotron.10 Other synchrotron infrastructures also welcome cultural heritage related projects. The use of neutron imaging techniques is becoming increasingly popular. Paintings (neutron autoradiography) and metallic artifacts such as statuettes (radiography and tomography) are suitable objects of study. Especially with regard to metal or large objects, the advantage of neutron imaging compared to X-ray imaging is the fact that neutrons usually have a much higher range in the material, and in cases where intense X-ray fluxes cause damage, neutrons may not. It is well-known that the attenuation of X-rays increases strongly with higher mass numbers of the investigated material. If heavy metals are to be investigated, a few millimeters are sufficient to shield the X-ray beam completely. At this point, neutrons become valuable due to higher penetration for most of the relevant metals.11


In this Account, we will discuss our own experience at synchrotron infrastructures for one specific project, namely, the in situ monitoring of the surface and electrochemical behavior of archeological copper alloys during their storage and stabilization. Artifacts recovered from marine environments or from the ground are typically saturated with chlorides whose attack on the metal will be greatly accelerated upon air exposure. Serious degradation or even loss of the artifact can result. To alleviate this problem, objects are stored in tap water or sodium sesquicarbonate solution for periods of up to several years.12–14 This is believed both to remove potentially damaging soluble chlorides from the microstructure of the metal and to change various copper chlorides into more benign compounds such as cuprite. A method which has been proposed for monitoring the progress of the treatment and determining its end-point is the measurement of the corrosion potential Ecorr, the open circuit potential of the electrochemical cell in which, effectively, the artifact is immersed. The storage process in our work was imitated by immersing artificially corroded copper coupons in a 1% (wt/v) sodium sesquicarbonate solution. The possible use of corrosion potential measurements as a monitoring method was evaluated by recording the corrosion potential of the immersed coupons as a function of time while analyzing the surface of the coupon simultaneously using either SR-XRD or XAS. In order to be able to perform these experiments, we developed an electrochemical cell (eCell) which is engineered so as to permit X-rays from a synchrotron beamline to be scattered or absorbed by the surface of a sample while in solution and electrochemical reactions are taking place. X-rays emitted from the surface then carry time-resolved information on the specific reactions as they occur.

Monitoring the Corrosion of Cupreous Artifacts As with many other metals, copper corrodes once it comes into contact with an aggressive environment, for example, the sea or the atmosphere. In the field of art, copper-based objects are often preferred in the corroded state, not only because of the aesthetically pleasing colors but also because the presence of corrosion products provides evidence of time past and time passing, thereby adding extra value to the object.12,16 However, corrosion may also become a problem, especially when specific corrosion products (such as cuprous chlorides) are in contact with the metal core. Under certain conditions, the deterioration of the underlying metal will continue and will lead to the destruction of the object. Corroded archeological copper and copper alloy artifacts recovered from wet saline environments are particularly sus-

FIGURE 1. Schema of the electrochemical cell used in this work.

ceptible to further corrosion. Studies have shown that, despite storage in tap water and stabilization in sodium sesquicarbonate, corrosion layers can subsequently provoke side effects such as the modification of the natural patina and the development of new active corrosion,16,17 hence the need for continuous monitoring of the objects under care. During treatment, the chloride concentration in solution can be measured. When this exceeds a predetermined value, the solution is replaced. This procedure is repeated until the chloride concentration remains low enough. The disadvantage of this method is the fact that it is indirect: the conservator has no idea what is happening with the metal surface. Hence, a different monitoring method is needed. In our work, we have evaluated whether Ecorr measurements may have a role to play. Ecorr is the open circuit potential of the metal object against a stable reference electrode. The potential obtained depends on the solution (electrolyte) in which the object is immersed (which is known as it was chosen by the conservator) and the composition of the clean or corroded metal surface. The hypothesis is that stable Ecorr data imply a stable surface chemistry. The method is easy to use, is of relatively low cost, and may therefore be suitable for museum applications.

The Use of eCell Central to this work is a new environmental cell (eCell), capable of maintaining a controlled liquid, gas, or vapor environment over a sample (Figure 1). The eCell is a second generation of a published design.18 It can be used as a three electrode electrochemical cell when filled with liquid electrolyte, allowing simultaneous real time analysis of the metal surVol. 43, No. 6

June 2010

927-935


929


face using a spectroscopic technique (in our case XRD and XAS).18,19 To ensure that X-rays can reach the sample surface, and that electrochemical and solution processes can take place undistorted by limited electrolyte volumes, the working electrode can be raised and lowered between “X-ray” and “electrochemistry” positions under remote control. In the former position, about 125 µm of electrolyte remains between the sample surface and an 8 mm thick Kapton X-ray window. In the latter, the electrolyte thickness increases to ∼5 mm. Experiments in this study were carried out on simulated materials; that is, copper coupons were artificially corroded. Four corrosion products commonly found on real copper artifacts were considered:16 cuprite, nantokite, atacamite, and paratacamite. Cuprite (Cu2O) is regularly found on copper artifacts and is a stable corrosion product.20 Among the copper chlorides, nantokite (CuCl), atacamite (Cu2(OH)3Cl), and a mixture of atacamite and paratacamite (both isomers of Cu2(OH)3Cl) were selected. Nantokite (CuCl) is considered as the main catalytic agent for active corrosion. The presence of nantokite on the metal surface can jeopardize the long-term stability of an object. In fact, bronze disease or pitting corrosion is usually attributed to this corrosion product.16,20 Atacamite and paratacamite are two other important chlorides in bronze corrosion. They are often considered as end products and are formed on top of the active corrosion areas.16 Atacamite is the most common of the Cu2(OH)3Cl isomers, but it often alters into paratacamite.20 Conversion between copper hydroxychlorides and their relative stabilities has also been discussed by Pollard et al.21 The corrosion layers were made on pure copper coupons (ADVENT, purity 99.9%) according to protocols listed in the literature.19,20 The corroded coupons were immersed in a sodium sesquicarbonate solution in order to imitate the stabilization treatment. Various concentrations of sodium sesquicarbonate solutions have been used by conservators to stabilize bronze artifacts with lower concentrations currently favored, to limit rinsing steps, and to avoid the potential formation of copper carbonate. For this study, a 1 wt % (wt/v) sodium sesquicarbonate solution was prepared by dissolving 11.89 g/L Na2CO3 · NaHCO3 · 2H2O (Sigma) in deionized water (pH ) 10). In addition to the artificial corrosion of the copper coupons, powder samples of Cu2O (for comparison with cuprite, Fluka, > 99%) and Cu(I)Cl (for comparison with nantokite, Fluka, > 97%) were purchased to serve as reference material. Atacamite powder is not commercially available and was prepared according to an existing protocol.20 A first set of simultaneous synchrotron X-ray diffraction (SRXRD) and electrochemical experiments was carried out at sta930


927-935

June 2010

Vol. 43, No. 6

tion 2.3 of the Synchrotron Radiation Source, Daresbury Laboratory (U.K.).22 The amount of data collected as a function of time was limited here by the scan time of the diffractometer (ca. 30 min). Acquisition times of the order of 1 s would, however, would be preferable, so that data for different products can be recorded contemporaneously, fast reactions can be studied, and short-lived byproduct can be observed. Times in this range can be achieved using one- or two-dimensional area detectors which collect a large range of angles in parallel. Parallel collection of spectra also minimizes the X-ray dose; the beam can be off for most of an experiment lasting many hours. Moreover, it is vital to be able to collect the electrochemical data in a mode where the reaction takes place in the bulk of the electrolyte (i.e., allowing the optimization of transport processes in the electrochemical cell), and to raise the working electrode to the cell window for a short time for X-ray analysis. This is not possible with a scanning diffractometer, as spectra must be recorded continuously throughout the experiment. Further sets of SR-XRD experiments were therefore performed at SRS station MPW6.2 and beamline 28 (XMaS) at the ESRF.23 The advantage of the former beamline lies in the fact that data were acquired using a RAPID II detector system which is one-dimensional and collects all of the in-plane diffraction simultaneously. It allows the acquisition of spectra with a time resolution of under a second.24 The latter beamline makes use of a 2D Mar CCD 165 detector (Mar USA Inc., Evanston, IL) for the acquisition of surface diffraction patterns. The advantage over a 1-D in-plane detector is that preferential crystallographic orientation shows up immediately as do time dependent changes in crystal orientation. Spectra can be extracted which average over or highlight these effects as one wishes. XAS measurements were performed at station BM26A (DUBBLE) at the ESRF.25,26 Cu K-edge (8.979 keV) XAS spectra were recorded as a function of energy. The scan time was 20 min, and measurements were made in fluorescence mode.

Can Corrosion Potential Measurements Be Incorporated in a Trustworthy Sensor? Figure 2 shows a typical SR-XRD spectrum (gray) of a copper sample initially covered with synthetic nantokite and some cuprite (which is also formed by the protocol). The offset spectrum (black) shows the sample after 10.3 h in the sesquicarbonate. The copper signal increases by a factor 2.4, nantokite disappears, and cuprite diminishes slightly. Figure 3 shows the variation of the SR-XRD peak heights of nantokite and cuprite as a function of time, together with the simultaneously measured Ecorr.18 Again, nantokite disappears. This time, however,


FIGURE 2. Typical spectra from copper coated with nantokite before and after treatment in 1% w/v sodium sesquicarbonate for 10.3 h. In the gray spectrum taken prior to treatment, cuprite and nantokite are evident. After treatment, only cuprite remains. In this experiment, the copper intensity increased by a factor of 2.4.

For the open circuit potential (OCP) data, the right-hand y-axis gives the corrosion potential. The hypothesis that a stable Ecorr means a stable surface chemistry is not supported. Although there is a rough correlation between the rise in Ecorr and the rapid loss of nantokite during the first 30 min, Ecorr then becomes more or less stable while the surface composition continues to change. The ongoing processes were studied in further detail using XAS.25,26 The latter provides an independent means of surface characterization, as the technique is not only sensitive to the presence and evolution of amorphous surface compounds FIGURE 3. Variation of the SR-XRD peak heights of nantokite and cuprite as a function of time, together with the simultaneously measured Ecorr. The reference electrode was 3 M Ag/AgCl.

the cuprite signal grows by more than a factor of 2.5. Many repeated measurements show behavior between these two extremes, probably dependent on the thickness and morphology of the nantokite, and other factors. The mechanism of cuprite formation remains unclear at this stage, as several routes are possible in the presence of chloride ions. For example, Oddy and Hughes12 believe that nantokite can react with water to form cuprite through the following reaction:

2CuCl + H2O f Cu2O + 2Cl- + 2H+ However, cuprite can also be formed through a precipitation reaction.27 2CuCl2 + 2OH f Cu2O + H2O + 4Cl

The increase of the copper signal in Figure 3 may either be due to the fact that the final cuprite layer is thinner than the original nantokite, or, alternatively, give evidence for the precipitation of copper as a superficial layer over the cuprite.28

but will also give information on the presence of potential complex ions in the solution. As with the SR-XRD experiments, eCell was filled with a 1% (wt/v) sodium sesquicarbonate solution and XAS data were taken as a function of time. Overall, the spectra show two significant features: First, for the thin corrosion layers, such as nantokite and cuprite, the XAS signal is strongly influenced by the presence of the underlying copper, which is less the case for the corrosion layers atacamite and paratacamite. The latter is clearly seen in Figures 4 and 5.25 Figure 4 shows spectra for the four powder references and a bare copper sample, with an inset to show the details of the edge region. Figure 5, on the other hand, shows the data obtained in fluorescence mode from the corrosion products on the copper surface. The cuprite and nantokite spectra are strongly influenced by the presence of the underlying copper, simply because the layers are thin. Nevertheless, the shoulder in the nantokite at 8985 eV (A) is clearly a relic of the post edge peak in the CuCl, and the spectrum from the cuprite coated sample is modified from its dominant copper shape at (B). The thicker atacamite spectrum Vol. 43, No. 6

June 2010

927-935


931


FIGURE 4. XAS spectra for copper, cuprite, CuCl (the analogue for nantokite), and synthesized atacamite. The inset shows the edge region in detail. Reproduced with permission from ref 25. Copyright 2009 The Royal Society of Chemistry (http://dx.doi.org/10.1039/ b814181a).

FIGURE 5. XAS spectra for a clean copper electrode and copper electrodes covered with cuprite, nantokite, and atacamite corrosion layers. Reproduced with permission from ref 25. Copyright 2009 The Royal Society of Chemistry (http://dx.doi.org/10.1039/b814181a).

remains distinctive, although it too is clearly influenced somewhat by the underlying copper. Second, our attention was immediately drawn to the fact that XAS spectra of the corroded copper samples, collected during their immersion in the sesquicarbonate solution, all show increasing signals as a function of time. The increase is observed both at the absorption Cu K edge and in a proportionate increase in the X-ray absorption near edge structure (XANES) and the extended X-ray absorption fine structure (EXAFS) modulation, but the shapes of the spectra do not appear to evolve strongly, even though the chemical composition of the surface is undoubtedly changing. The effect is largest for the atacamite layer, as described in what follows. Figure 6 shows the successive XAS spectra recorded for this 932


927-935

June 2010

Vol. 43, No. 6

FIGURE 6. Sequential XAS spectra of the copper/atacamite sample in a sodium sesquicarbonate solution. The scan numbers are indicated on the right. The figure in brackets is the elapsed time in minutes at the end of the scan. Reproduced with permission from ref 26. Copyright 2008 Maney Publishing (www.maney.co.uk/journals/sur and www.ingentaconnect.com/content/maney/se).

FIGURE 7. Ecorr data taken simultaneously with each immersion stage of the experiment in Figure 8. The reference electrode was 3 M Ag/AgCl. Reproduced with permission from ref 25. Copyright 2009 The Royal Society of Chemistry (http://dx.doi.org/10.1039/ b814181a).

sample.26 The copper edge height of the first measurement is set to be equal to 1. All subsequent scans are normalized to the edge height of the first scan. Scan number 8 already shows an increase of 4 times with respect to the copper edge of the first measurement. From the ninth spectrum on, the increase between the different spectra becomes rapidly larger. The obvious change in the XANES or EXAFS intensity of the immersed samples could not be observed for the original dry corrosion layers, monitored over similar times in air (i.e., there is no increase in signal over time for dry samples). Figure 7 shows the corrosion potential data as a function of time with complete changes of the electrolyte in the cell at 15 and 35 h (imitating typical conservation treatment).26 Again, these data were collected simultaneously with the XAS


seen when copper(II) chloride is dissolved in water, although the diffraction pattern of the original corrosion layer has no match with copper(II) chloride. Moreover, although solid copper(II) chloride has a totally different XAS spectrum from that of atacamite, the spectrum of dilute CuCl2 solution is similar to that of solid atacamite and the solution observed here (i.e., the copper becomes octahedrally coordinated in solution). The dissolution of the soluble fraction leaves the atacamite only physically connected to the surface.

Conclusions and Future Perspectives FIGURE 8. Webcam images of the eCell during the sequence in Figures 6 and 7. Left hand plate shows the blue haze flowing off the sample after 20 min immersion. The right-hand plate shows the situation at the end of the first immersion (14 h). Reproduced with permission from ref 25. Copyright 2009 The Royal Society of Chemistry (http://dx.doi.org/10.1039/b814181a).

data mentioned above. The results demonstrate that over a time period of 240 min (4 h) the corrosion potential keeps on changing, implying a continuous change of the surface chemistry. The signal stabilizes after ca. 8 h at which moment the XAS signal is still increasing (not shown). Again, the corrosion potential remains relatively stable after the electrolyte changes, but the XAS data (not shown) indicate further dissolution from the layer. While the spectra are being recorded and the electrochemistry is in progress, eCell makes a continuous visual record of the process using a webcam. Figure 8 shows two such webcam images recorded after the sample had been immersed for 20 min (left) and 14 h (right).25 The detachment of material from the atacamite layer is evident as a blue haze which drifts downward as the sample surface is in a vertical plane on DUBBLE. In fact, two sorts of detaching material are clearly visible: relatively large crystalline fragments up to hundreds of micrometers in size and the haze reminiscent of dissolving CuCl2. The sudden increase in signal as of scan number 9 can in this context be explained by the fact that at this stage more of the substance becomes detached. It should be noted, however, that the spectrum remains characteristic of atacamite (or an octahedrally coordinated copper complex ion) throughout this process. Our very recent work has shown that this behavior is due to a major flaw in the corrosion protocol itself. It produces atacamite only as a minority constituent, and a highly soluble, but as yet unidentified, copper salt in the majority. The soluble component is removed by the soaking process, leaving nearly pure atacamite behind. The solution process as observed in the webcam images is nearly identical to that

We have developed an electrochemical cell (eCell) which is engineered so as to permit X-rays from a synchrotron beamline to be scattered or absorbed by the surface of a sample, or by a liquid in which it is immersed, while electrochemical reactions are taking place. X-rays emitted from the system then carry time-resolved information on the specific reactions as they occur. During the analysis, the surface can remain immersed in electrolyte or exposed to air so as to study the process in the most relevant way, and electrochemical data can be measured coincidentally. Although similar in situ cells for the study of idealized (e.g., atomically flat, single crystal) surfaces have been described, this is, so far as we are aware, the first time such experiments have been done on rough (on the 1-100 µm scale), polycrystalline, impure metals typical of real artifacts. In this work, we have used eCell to evaluate a potential monitoring method for copper-based objects stored in solution. Results have shown that SR-XRD and XAS analyses of a surface with a thin layer of electrolyte above give the possibility of probing both the surface structure and the complex ions or colloidal material in the liquid. The addition of Ecorr measurements shows that changes in chemistry of the bulk corrosion product are not necessarily linked to this parameter which appears to be dominated by the chemistry of a (semi)conducting interfacial layer of cuprite. Therefore, Ecorr is not a good indicator of chloride removal; that is, we have not yet found a reliable and straightforward monitoring method (a sensor device) to replace the laborious testing of chloride concentration in solution. Nevertheless, we will continue our X-ray investigation of the soaking treatment with a view to looking at optical absorption or a more refined electrochemical measurement as the basis for a possible sensor, and more generally to study the details of reactions leading to the formation and removal of corrosion layers, and passivation of copper alloy surfaces. In addition, we will explore the use of X-ray excited optical luminescence (XEOL) to obtain XAS spectra for monitoring corroVol. 43, No. 6

June 2010

927-935


933


sion, passivation, and coating in situ. XEOL has a greatly improved surface specificity compared to X-ray diffraction and X-ray fluorescence absorption measurements using beams with energies > 8 keV where the penetration and escape depths exceed the thickness of the layers of interest. It is wellknown that the escape depths of optical photons and electrons produced by Auger de-excitation and other processes occurring as a consequence of the X-ray photoionization event are in the range 100s nm to 3.0% sticky rice solution in the lime mortar, the development of the strength of lime mortar will be restrained. The elasticity modulus of mortars was tested according to the Standard for Test Method of Performance on Building Mortar (JGJ/70-90), Part 8.47 In the presence of sticky rice soup, the elasticity modulus of lime mortar is also improved (Table 3). When a 3.0% sticky rice soup solution is added, the elasticity modulus of lime mortar reaches its peak value of 2400 MPa. However, it is still far below the highest value of the elasticity modulus (1-18 GPa) of ancient brick,48 which suggests that the sticky rice-lime mortar is characterized by an elastic behavior compatible with traditional masonry buildings. 4.2.6. Water Vapor Permeability of Hardened Mortar. The water vapor permeability of the hardened mortars was determined by the methodology described in EN 1015-19.49 Results listed in Table 4 indicate that the water vapor permeability of lime mortar can be reduced by the addition of sticky rice. This is due to the compact structure of sticky rice-lime mortar. A reduced permeability is often a negative factor50 for mortar, since it affects the elimination of water vapor that exists within buildings. However, even when a 3.0% sticky rice soup solution is introduced, the mortar can still have a high water vapor permeability of 2.7 × 10-11 kg m-2 s-1 Pa-1. On

lime mortar

sticky rice-lime mortar

750a and 810b 0.25a and 0.23b

700a and 690b 0.49a and 0.49b

Before climatic cycles. b After climatic cycles.

the other hand, this may indicate other positive characteristics such as lower water absorption due to capillary action or even a lower permeability. 4.2.7. Water Absorption of Hardened Mortar. Water absorption of hardened mortar was measured according to JGJ/T70-2009, Part 14.37 The results are listed in Table 4. It implies that the water adsorption of lime mortar can clearly be decreased via the addition of sticky rice soup. This is partly due to the compact structure of mortar, which is formed in the presence of sticky rice soup. In addition, the interaction between the sticky rice polysaccharide and the calcium carbonate may be another important factor. During the solidification of sticky rice-lime mortar, the hydrophilic hydroxyl group of amylopectin is covalently bonded by calcium cation51 and the alkyl group may provide additional hydrophobicity. Low water absorption can protect the buildings against the erosion of water and soluble salt.52 4.2.8. Compatibility with Substrate. On the basis of EN 1015-21,53 the compatibility of mortars was evaluated. The substrates were clay bricks, the most frequently used building materials in ancient Chinese construction. Before the test, the specimens were cured at 20 °C under 60 ( 5% relative humidity for 2 months. The effects of climatic cycles on the mortars were evaluated via water permeability and adhesive strength. The test results are listed in Table 5. The permeability of sticky rice-lime mortar to liquid water is lower than that of lime mortar. For masonry mortar, a reduced permeability to liquid water is a positive factor, since it limits the penetration of water into the mortar. Besides, compared with lime mortar, sticky rice-lime mortar is less susceptible to climatic cycles. After artificial climatic cycles, the lime mortar deteriorates slightly in permeability (from 750 to 810 mL) and adhesive strength (from 0.25 to 0.23 MPa). The sticky rice-lime mortar, however, seems not to be affected. These results show that sticky rice-lime mortar is more compatible with brick substrates. As a result, the performance of lime mortar can be significantly improved in the presence of an appropriate amount of sticky rice soup. Compared with lime mortar, sticky rice-lime mortar has more stable physical properties, has greater mechanical strength, and is more compatible with substrates, Vol. 43, No. 6

June 2010

936-944


941

Sticky Rice-Lime Mortar Technology Yang et al.

which make it a suitable restoration mortar for ancient masonry buildings.

5. Application of Sticky Rice-Lime Mortar in Historic Preservation Sticky rice-lime mortar has been used in the restoration of some ancient masonry constructions. The Shouchang bridge is a single-arch stone bridge, built during the Song Dynasty, and is currently one of the national cultural heritage protection units. The restoration work on the Shouchang bridge began in 2006. Because of bridge foundation settlement and plant growth, cracks between building blocks appeared and continued to widen in recent years. Necessary conservation, including the consolidation of the bridge foundation and bridge wall, removal of plants, and antiweathering treatment, was conducted. Sticky rice-lime mortar (made of slaked lime, sticky rice soup, and stone powder) was used as joint mortar. On the whole, sticky rice-lime mortar is the right choice. There are not any more cracks or desquamation after almost 5 years of exposure in the open air, indicating the good compatibility of sticky rice-lime mortar. Besides, there are no more plants in the mortar, and this may be attributed to the high alkalinity of slaked lime.

6. Conclusions Sticky rice-lime mortar technology appeared at least 1500 years ago and was extensively adopted in ancient China. Analytical study shows that the ancient masonry mortar is a kind of special organic-inorganic composite material. The inorganic component is calcium carbonate, and the organic component is amylopectin, which should come from the sticky rice soup added to the mortar. Moreover, we found that amylopectin in the mortar acted as an inhibitor: the growth of the calcium carbonate crystal was controlled, and a compact microstructure was produced, which should be the cause of the good performance of this kind of organic-inorganic mortar. The test results of the modeling mortars show that sticky rice-lime mortar has more stable physical properties, has greater mechanical strength, and is more compatible, which make it a suitable restoration mortar for ancient masonry buildings.

This work was supported by the grants from the National Natural Science Foundation of China (20671080) and the National Technology Support Program of China (2006BAK30B02). 942


936-944

June 2010

Vol. 43, No. 6

BIOGRAPHICAL INFORMATION Fuwei Yang is a Ph.D. candidate in physical chemistry. His work is mainly focused on Chinese traditional building materials and the restoration and preservation of soil and stone heritage. Bingjian Zhang is a professor of physical chemistry, the head of the Institute of Physical Chemistry at Zhejiang University, and the vice director of the Heritage Conservation and Identification Center of Zhejiang University. His research interests are centered on materials for the preservation of soil and stone heritage, building stone, and biomineralization. Qinglin Ma is a professor of cultural relic preservation and vice president of the Chinese Academy of Cultural Heritage. His research interests include the reinforcing materials of wall painting, the preservation of metal antiquity, and paint in ancient wall painting and pottery.

APPENDIX A: EXPERIMENTAL DETAILS 7.1. Materials. KI, I2, alcohol, deionized water, NaOH, acetic acid, KBr, CaCO3, Ca(OH)2, and NH4HCO3 (A.R. grade, g99.8% assay) were used in this study. Sticky rice, which is a type of rice grown in Southeast and East Asia mainly composed of amylopectin, is commercially available. 7.2. Sampling. The samples of historical mortars were obtained from the Nanjing ancient city wall from the Ming Dynasty just before its repair. The sampling positions are ∼5-10 cm under the surface of the original mortar to ensure the samples are undisturbed. 7.3. Instrumentation and Operating Conditions. A NICOLET 560 Fourier transform infrared spectrometer was employed to identify the main molecular groups present in mortars, and the mortar samples were analyzed in KBr pellets. The spectra were traced in the range of 2500-500 cm-1, and the band intensities were expressed in transmittance (percent). For the preparation of samples, 1-2 mg of the fraction of mortar was mixed homogeneously with 100 mg of anhydrous KBr in an agate mortar. A pressure of 10 ton was applied to this mixture to yield a transparent pellet. Identification is based on comparison of the bands of the recorded FTIR spectra with those of reference materials or from the literature. For identification of the crystallography of the mortars, XRD analyses were performed with an AXS D8 ADVANCE X-ray diffractometer using Cu KR radiation (λ ) 1.54 Å), a voltage of 40 kV, and a current of 40 mA. The thermal analyses were performed on simultaneous DSC-TG equipment (TA Instruments, model NETZSCH STA 409 PC/PG). The experimental conditions were as follows: (a) continuous heating from room temperature to 1000 °C at a heating rate of 20 °C/min, (b) air-gas dynamic atmosphere (45 cm3/min), (c) alumina, top-opened crucible, and (d) a sample mass of ∼15 mg. The surface morphology of the mortar samples was inspected with an optical microscope and scanning electron microscopy (SEM, FEI SIRION100). In addition, the consistency of fresh mortar was determined with an SC145 mortar consistency tester. The shrinkage of mortar was measured in


the longitudinal dimension with an SP-175 mortar contraction detector. The strength test of modeling mortar was conducted with a Yinchi (Shanghai, China) YC-125B tension meter. 7.4. Methods. Approximately 100 g of the historical mortar sample was ground in an agate mortar, and the fine powder obtained was analyzed by FTIR, XRD, and thermal analysis (TG-DSC). Next, the starch-iodine test was employed to confirm the amylopectin component of the mortar. For the starch-iodine test, 5.0 g of the ground mortar sample was added to 100 mL of deionized water and the mixture heated at 80-90 °C for 10 min with stirring. The formed suspension was then cooled to 25 °C and its pH adjusted to 6.0-7.0 with acetic acid. We created the iodine-KI reagent by dissolving 0.2 g of iodine in 100 mL of deionized water in the presence of 2.0 g of KI. The iodine-KI reagent was dropped into the suspension, and a blue or red brown color (blue for amylose and red brown for amylopectin) was observed if starch existed. We made a sticky rice solution by cooking a mixture of sticky rice powder and distilled water at 100 °C and 1.0 bar for 4 h. For the hydration of quick lime, calcium oxide powder was added to water (quick lime:water mass ratio of 1:3) with agitation and the slaking temperature was held at 80 ( 10 °C. The lime putty obtained was stored for 6 months in an airtight container before being used. To take a deeper look at the historical mortars, several modeling samples with 0, 1.0, 2.0, 3.0, and 5.0% sticky rice were fabricated and dried at room temperature under 60 ( 5% relative humidity. Six months later, the modeling samples were analyzed by FTIR, XRD, and SEM under the same conditions used for the historical samples. For the specimens for the strength test, slaked lime, sticky rice soup, and deionized water were evenly blended and added to the cube-shaped mold (70.7 mm × 70.7 mm × 70.7 mm). The specimens obtained were naturally carbonized at 20 °C under 60 ( 5% relative humidity for 6 months before being tested. FOOTNOTES * To whom correspondence should be addressed. Telephone: +86-0571-87997523. E-mail: [email protected]. REFERENCES 1 Davey, N. A History of Building Materials; Phoenix House: London, 1961; pp 9798. 2 Woolley, L. Excavations at Ur; Ernest Ben: London, 1955; pp 22. 3 Moropoulou, A.; Bakolas, A.; Anagnostopoulou, S. Composite Materials in Ancient Structures. Cem. Concr. Compos. 2005, 27, 295–300. 4 Cowan, H. J. The Master Builders: A History of Structural and Environmental Design from Ancient Egypt to the Nineteenth Century; John Willey & Sons: New York, 1977. 5 Riccardia, M. P.; Duminucob, P.; Tomasic, C. Thermal, Microscopic and X-ray Diffraction Studies on Some Ancient Mortars. Thermochim. Acta 1998, 321, 207– 214. 6 Bo¨ke, H.; Akkurt, S.; |$$I˙pekogˇlu, B.; Ugˇurlu, E. Characteristics of Brick Used as Aggregate in Historic Brick-Lime Mortars and Plasters. Cem. Concr. Res. 2006, 36, 1115–1122. 7 Gu¨leç, A.; Tulun, T. Physico-Chemical and Petrographical Studies of Old Motars and Plasters of Anatolia. Cem. Concr. Res. 1997, 27 (2), 227–234. 8 Qiu, S. H. When Did the Lime Production Begin? The 14C Isotopic Geochemical Method Can Answer the Question. Archeology of Cultural Relics 1980, 5, 126–130.

9 Wang, K. The Ancient History of Road Transportation in Shanxi, China; Communication Press: Beijing, 1989; pp 10. 10 Tie, F. D. Conservation and Restoration of Wall Paintings of Early Western Han Dynasty. Sci. Conserv. Archaeol. 2004, 16 (1), 47–51. 11 Shen, K. The Dream Rivulet Diary; Times Wenyi Press: Changchun, China, 2001; pp 110. 12 Song, Y. Y. Tian Gong Kai Wu; Commercial Press: Shanghai, 1958; pp 197. 13 Li, X. H.; De, Q. S. The Artistic Appreciation of Stone Tower with Sutra Inscriptions. Architect Gard 2005, 3, 32–34. 14 Archeology Crew of Henan Province. A Brick Tomb with Colorized Portrait in Deng country, Henan provice; Cultural Relic Press: Beijing, 1958; pp 20. 15 Wu, Q. Z. The Experience and Implication for the Flooding Control in the Ancient Towns in China. City Planning Review 2002, 26 (5), 76–84. 16 Yang, G. L. Viewing the Remains: Antirot Technique in the Ancient Grave Custom. J. Changsha Soc. Work Coll. 2004, 4, 11–15. 17 Nanjing Museum. The Tomb of Xupu and His Wife. Cultural Relics 1982, 2, 28–33. 18 Lin, J. S.; Lin, Z. J.; Chen, J. F. The Ancient Earthquake and Earthquake-Resistance of the Ancient Buildings in Quanzhou City. World Earthquake Eng. 2005, 21 (2), 159–166. 19 Zhang, Y. H. The History of Building Technology in Ancient China; Science Press: Beijing, 1985; pp 173. 20 Wang, Y. M. The Developing History of Cement in China; China Architecture & Building Press: Beijing, 2005; pp 4-5. 21 Yang, T.; Zhou, J.; Zhou, C. Analysis and design of timeber-pile composite foundation on soft soils. Chin. J. Rock Mech. Eng. 2003, 22 (9), 1560–1565. 22 Li, X. H. Research on the Protection and Repair of Main Structure for Mingyuan Bridge. Build. Sci. 2009, 25 (4), 61–66. 23 Song, J. A Cultural Survey of the Old Bridges in Wudang Mountains Built in the Ming Dynasty. J. Hubei Univ. 2006, 33 (5), 587–590. 24 Zhang, F. H.; Qiang, Y.; Du, F. D. Watchtower-like dwelling in Kaiping County. Archit. J. 2004, 7, 82–84. 25 Cai, J. S. Environment Friendly Castle-Like Dwellings. Archit. J. 1995, 5, 42–44. 26 Zhang, N. A. The Report about the Measurement Results of the Nanjing City Wall. Nanjing culture and culture relics Bureau, 2006. 27 Pavlovic, S.; Brandao, P. R. G. Adsorption of Starch, Amylose, Amylopectin and Alucose Monomer and Their Effect on the Flotation of Hematite and Quartz. Miner. Eng. 2003, 16, 1117–1122. 28 Rundl, R. E.; Foster, E. J. F.; Baldwin, R. R. On the Nature of the Starch-Iodine Complex. J. Am. Chem. Soc. 1944, 66, 2116–2120. 29 Genestar, C.; Pons, C.; Ma´s, A. Analytical Characterisation of Ancient Mortars from the Archaeological Roman city of Pollentia. Anal. Chim. Acta 2006, 557, 373–379. 30 Liu, X. X.; Yu, L.; Liu, S. H.; Chen, L.; Li, L. Thermal Decomposition of Corn Starch with Different Amylose/Amylopectin Ratios in Open and Sealed Systems. Cereal Chem. 2009, 86 (4), 383–385. 31 Chen, J. L. The Researching About the Relation of Materials Prescription of Ancient Mortar and Strength; National Cheng Kung University: Tainan, Taiwan, 2004, pp 14. 32 Zeng, Y. Y.; Zhang, B. J.; Liang, X. L. A Case Study and Mechanism Investigation of Typical Mortars used on Ancient Architecture in China. Thermochim. Acta 2008, 473 (1-2), 1–6. 33 Yang, F. W.; Zhang, B. J.; Pan, C. C.; Zeng, Y. Y. Traditional Mortar Represented by Sticky Rice-Lime Mortar. Sci. China, Ser. E: Technol. Sci. 2009, 52 (6), 1641–1647. 34 Yang, L.; Zhang, X. Y.; Liao, Z. J.; Guo, Y. M.; Hu, Z. G.; Cao, Y. Interfacial Molecular Recognition between Polysaccharides and Calcium Carbonate during Crystallization. J. Inorg. Biochem. 2003, 97, 377–383. 35 Steven, R. D.; McGrath, K. M. Aqueous Precipitation of Calcium Carbonate Modified by Hydroxyl-Containing Compounds. Cryst. Growth Des. 2004, 4 (6), 1141–1148. 36 Vishwas, V.; Hardikar, I.; Egon, M. Influence of Ionic and Nonionic Dextrans on the Formation of Calcium Hydroxide and Calcium Carbonate Particles. Colloids Surf., A 2001, 186, 23–31. 37 Standard for Test Method of Performance on Building Mortar (JGJ/T70-2009). 38 Pavıá, S.; Fitzgerald, B.; Howard, R. Evaluation of Properties of Magnesian Lime Mortar; WIT Press: Southampton, U.K., 2005; pp 375-384. 39 Specification for Mix Proportion Design of Masonry Mortar (JGJ/98-2000). 40 Leemann, A.; Winnefeld, F. The Effect of Viscosity Modifying Agents on Mortar and Concrete. Cem. Concr. Compos. 2007, 29, 341–349. 41 Sjo¨qvist, M.; Gatenholm, P. The Effect of Starch Composition on Structure of Foams Prepared by Microwave Treatment. J. Polym. Environ. 2005, 13 (1), 29–37. 42 Van Balen, K.; Van Gemert, D. Modelling Lime Mortar Carbonation. Mater. Struct. (Bagneux, Fr.) 1994, 27, 393–398.

Vol. 43, No. 6

June 2010

936-944


943


43 Michoinova´, D.; Rovnanı´kova´, D. High-Calcium Lime Mortar: The Effects of Traditional Preparation and Curing. APT Bull. 2008, 39 (4), 23–29. 44 Chandra, S.; Eclund, F.; Villarreal, R. R. Use of Cactus in Mortars and Concrete Research. Cem. Concr. 1998, 28, 41–45. 45 DeSilva, P.; Bucea, L.; Moorehead, D. R.; Sirivivatnanon, V. Carbonate Binders: Reaction Kinetics, Strength and Microstructure. Cem. Concr. Compos. 2006, 28, 613–620. 46 Franco, M.; Rinaldi, G. Poly-Amino-Phenolic Additives Accelerating the Carbonation of Hydrated Lime in Mortar. Environ. Eng. Sci. 2002, 19 (4), 271–276. 47 Bostenaru-Dan, M. Materials, Technologies and Practice in Historic Heritage Structures; Springer: Dordrecht, The Netherlands, 2010; pp 37. 48 Standard for Test Method of Performance on Building Mortar (JGJ/70-1990)

944


936-944

June 2010

Vol. 43, No. 6

49 Methods of test for mortar for masonrysPart 19: Determination of Water Vapor Permeability of Hardened Rendering and Plastering Mortars (BS EN 1015-19:1999). 50 Ziad, A. S.; Mohamed, A. H. Laboratory Evaluation of Various Types of Mortars for the Conservation of Qasr al-Bint monument, Petra, Jordan. Eng. Struct. 2001, 23, 926–933. 51 Fernańdez-Munõz, J. L.; Zelaya-Angel, O.; Cruz-Orea, A.; Sańchez-Sinencio, F. Phase Transitions in Amylose and Amylopectin Under the Influence of Ca(OH)2 in Aqueous Solution. Anal. Sci. 2001, 17, 338–341. 52 Watta, D.; Colston, B. Investigating the Effects of Humidity and Salt Crystallization on Medieval Masonry. Build. Environ. 2000, 35, 737–749. 53 Methods of test for mortar for masonrysPart 21: Determination of the Compatibility of One-Coat Rendering Mortars with Substrates (BS EN1015-21:2002).

ADDITIONS AND CORRECTIONS Corrections to Special Issue on Art Conservation Volume 43, Number 6, 2010

Volume 43, Number 6, 2010

SALVATORE SIANO AND RENZO SALIMBENI

PARASKEVI POULI, ALEXANDROS SELIMIS, SAVAS GEORGIOU, AND COSTAS FOTAKIS

Advances in Laser Cleaning of Artwork and Objects of Historical Interest: The Optimized Pulse Duration Approach

Recent Studies of Laser Science in Paintings Conservation and Research

Page 739. The conspectus graphic and the captions of Figures 4, 5, 7, 8, 10, and 12-14 require the addition of the following credit: Copyright 2010 Salvatore Siano.

Volume 43, Number 6, 2010 EMILIANO CARRETTI, MASSIMO BONINI, LUIGI DEI, BARBARA H. BERRIE, LORA V. ANGELOVA, PIERO BAGLIONI, AND RICHARD G. WEISS

New Frontiers in Materials Science for Art Conservation: Responsive Gels and Beyond Page 751. The conspectus graphic requires the addition of the following credit: Copyright 2010. Photograph courtesy of Richard Weiss.

238

’

Page 771. The conspectus graphic and the caption of Figure 2 require the addition of the following credit: Copyright 2010. Photographs courtesy of Michael Doulgeridis. The caption of Figure 1 requires the addition of the following credit: Copyright 2010 Paraskevi Pouli.

Volume 43, Number 6, 2010 FRANCESCA CASADIO, MARCO LEONA, JOHN R. LOMBARDI, AND RICHARD VAN DUYNE

Identification of Organic Colorants in Fibers, Paints, and Glazes by Surface Enhanced Raman Spectroscopy Page 782. The captions of Figures 2 and 4-8 require the addition of the following credit: Copyright 2010 Metropolitan Museum of Art.



S. PRATI, E. JOSEPH, G. SCIUTTO, AND R. MAZZEO

BERNHARD BLÜMICH, FEDERICO CASANOVA, JUAN PERLO, FEDERICA PRESCIUTTI, CHIARA ANSELMI, AND BRENDA DOHERTY

Noninvasive Testing of Art and Cultural Heritage by Mobile NMR

New Advances in the Application of FTIR Microscopy and Spectroscopy for the Characterization of Artistic Materials

Page 761. The conspectus graphic requires the


addition of the following credit: Photograph courtesy

addition of the following credit: Copyright 2010

of Bernhard Blümich.

R. Mazzeo.


’

238–239

’

2011

’

Vol. 44, No. 3

Published on the Web 02/16/2011 www.pubs.acs.org/accounts 10.1021/ar100124r & 2011 American Chemical Society

Additions and Corrections



SIMONA FANTACCI, ANNA AMAT, AND ANTONIO SGAMELLOTTI

CLAUDIA DAFFARA, ENRICO PAMPALONI, LUCA PEZZATI, MARCO BARUCCI, AND RAFFAELLA FONTANA

Computational Chemistry Meets Cultural Heritage: Challenges and Perspectives Page 802. The conspectus graphic and and the caption of Figure 2 require the addition of the following credit: Copyright 2010 Antonio Sgamelloti.

Volume 43, Number 6, 2010 ALDO ROMANI, CATIA CLEMENTI, COSTANZA MILIANI, AND GIANNA FAVARO

Fluorescence Spectroscopy: A Powerful Technique for the Noninvasive Characterization of Artwork Page 837. The conspectus graphic and the captions of Figures 4, 6-8, and 10 require the addition of the following credit: Photographs courtesy of Aldo Romani.


Scanning Multispectral IR Reflectography SMIRR: An Advanced Tool for Art Diagnostics Page 847. The conspectus graphic and the captions of Figures 1 and 3-9 require the addition of the following credit: Copyright 2010 Claudia Daffara.

Volume 43, Number 6, 2010 YOSHIHIRO KUSANO, MINORU FUKUHARA, JUN TAKADA, AKIRA DOI, YASUNORI IKEDA, AND MIKIO TAKANO

Science in the Art of the Master Bizen Potter Page 906. The conspectus graphic and the caption of Figure 13 require the addition of the following credit: Copyright 2010 Y. Kusano.

Volume 43, Number 6, 2010 FUWEI YANG, BINGJIAN ZHANG, AND QINGLIN MA

Material Aspects of Icons. A Review on Physicochemical Studies of Greek Icons

Study of Sticky Rice-Lime Mortar Technology for the Restoration of Historical Masonry Construction

Page 877. The conspectus graphic and the captions


of Figures 1-6 require the addition of the following

addition of the following credit: Copyright 2010

credit: Copyright 2010 Ormylia Foundation Art Diag-

Bingjian Zhang.

SOPHIA SOTIROPOULOU AND SISTER DANIILIA

nosis Centre.

DOI: 10.1021/ar100124r

Vol. 44, No. 3

’

2011

’

238–239

’

ACCOUNTS OF CHEMICAL RESEARCH ’ 239

Advanced Techniques in Art Conservation

Analytical Techniques in Materials Conservation

Advanced Techniques in Biophysics

Advanced Techniques in Diagnostic Microbiology

Advanced Techniques in Dermatologic Surgery

Conservation education and outreach techniques

Remote Sensing for Ecology and Conservation: A Handbook of Techniques (Techniques in Ecology and Conservation)

Habitat Management for Conservation: A Handbook of Techniques (Techniques in Ecology and Conservation)

Bird Ecology and Conservation: A Handbook of Techniques (Techniques in Ecology & Conservation)

Advanced 3D Photorealism Techniques

LabView: Advanced Programming Techniques

Advanced Data Mining Techniques

Advanced Commodity Trading Techniques

Forest Ecology and Conservation: A Handbook of Techniques (Techniques in Ecology & Conservation)

Conservation and Sustainable Use: A Handbook of Techniques (Techniques in Ecology and Conservation)

Forest Ecology and Conservation: A Handbook of Techniques (Techniques in Ecology & Conservation)

ShaderX5: Advanced Rendering Techniques

Advanced Data Mining Techniques

Advanced 3D Photorealism Techniques

Advanced Commodity Trading Techniques

Advanced Network Analysis Techniques

Advanced Data Mining Techniques

Advanced Data Mining Techniques

Advanced Data Mining Techniques

Advanced Painter Techniques

The Art of Prolog, Second Edition: Advanced Programming Techniques

The Art of Prolog, Second Edition: Advanced Programming Techniques

Techniques of fantasy art

SalonOvations' Day Spa Techniques (S Advanced Techniques)

Advanced Automation Techniques in Adaptive Material Processing

Advanced Techniques in Art Conservation

Analytical Techniques in Materials Conservation

Advanced Techniques in Biophysics

Advanced Techniques in Diagnostic Microbiology

Advanced Techniques in Dermatologic Surgery

Conservation education and outreach techniques

Remote Sensing for Ecology and Conservation: A Handbook of Techniques (Techniques in Ecology and Conservation)

Habitat Management for Conservation: A Handbook of Techniques (Techniques in Ecology and Conservation)

Bird Ecology and Conservation: A Handbook of Techniques (Techniques in Ecology & Conservation)

Advanced 3D Photorealism Techniques

LabView: Advanced Programming Techniques

Advanced Data Mining Techniques

Advanced Commodity Trading Techniques

Forest Ecology and Conservation: A Handbook of Techniques (Techniques in Ecology & Conservation)

Conservation and Sustainable Use: A Handbook of Techniques (Techniques in Ecology and Conservation)

Forest Ecology and Conservation: A Handbook of Techniques (Techniques in Ecology & Conservation)

ShaderX5: Advanced Rendering Techniques

Advanced Data Mining Techniques

Advanced 3D Photorealism Techniques

Advanced Commodity Trading Techniques

Advanced Network Analysis Techniques

Advanced Data Mining Techniques

Advanced Data Mining Techniques

Advanced Data Mining Techniques

Advanced Painter Techniques

The Art of Prolog, Second Edition: Advanced Programming Techniques

The Art of Prolog, Second Edition: Advanced Programming Techniques

Techniques of fantasy art

SalonOvations' Day Spa Techniques (S Advanced Techniques)

Advanced Automation Techniques in Adaptive Material Processing

Recommend Documents