Analytical and Bioanalytical Chemistry

, Volume 395, Issue 7, pp 2161–2166

Second and third harmonic generation measurements of glues used for lining textile supports of painted artworks


    • Institute of Electronic Structure and Laser, Foundation of Research and Technology—Hellas
  • K. Melessanaki
    • Institute of Electronic Structure and Laser, Foundation of Research and Technology—Hellas
  • C. Fotakis
    • Institute of Electronic Structure and Laser, Foundation of Research and Technology—Hellas
Original Paper

DOI: 10.1007/s00216-009-3060-x

Cite this article as:
Filippidis, G., Melessanaki, K. & Fotakis, C. Anal Bioanal Chem (2009) 395: 2161. doi:10.1007/s00216-009-3060-x


In this study, we present the implementation of non-linear spot measurements for obtaining specific and novel information related to various types of natural and synthetic glues used for lining of painted artworks. Third harmonic generation measurements were employed, in transmission mode, for the accurate and non-destructive thickness detection of lining glues. Furthermore, second harmonic generation signals were collected, in reflection mode, providing complementary information for the discrimination between different types of lining glues.


SHGTHGLining gluesPainted artworks


An essential problem in conservation of painted artworks is the removal of old linings. In this procedure, a piece of fabric, that has been glued for stabilizing reasons on the verso of the original “textile support” of the painting, needs to be replaced due to deterioration effects [1, 2]. The original “textile support” of an artwork is a fragile and complex structure mainly consisted of cellulose (textile, animal glue and filler like gypsum). “Textile support”, due to certain properties of cellulose, aged and deteriorates [1]. Cellulose absorbs atmospheric oxygen and radiation, and is under attack by acid sources (such as CO2) and microorganisms. Those processes initiate chemical reactions and with the catalytic presence of humidity, “textile support” oxidizes and decays. These effects accelerated with the time, forcing “textile support” to lose its elasticity, firmness and become brittle. Tensions and deformations are created to the painted layer, and the “textile support” is no longer able to carry and stabilize the delicate painted surface [2]. This fact raises the need to take actions and protect paintings against damage by enforcing the old “textile support”. To achieve that, an additional textile is glued on the verso of the original one using a very delicate and complex technique called lining, which is traced back to the seventeenth century [1].

It is well known that initially natural adhesives were applied in lining procedures. Natural adhesives are organic materials extracted from animals such as glutoline, bone, casein glues and from plants such as waxes, starch flours, gums and natural resins [2]. These materials are under attack by microorganisms, a procedure that leads to discolor effects and looseness of their mechanical properties. Thus, natural adhesives are not currently in use in conservation.

In the early 1900s, the discovery of synthetic resins, polyvinyl acetates, acrylic solutions and microcrystalline waxes open the road for the replacement of natural adhesives [3]. Aging resistance, controllable setting, hardening procedure, improved adhesion and flexibility are some of the main advantages of synthetic materials [1].

However, the lining glues and the lining textiles also deteriorate and usually suffer from fresh deformations, after a certain period of time, rising up the need to be replaced with new ones, an operation called relining [1]. This conservation stage take place in two steps: a relatively easy operation, the separation of the lining textile from the old “textile support” by employing mechanical methods and a more elaborate one, the removal of the remaining lining glue via chemical and mechanical procedures. In order to assist and guideline, the latter step is crucial to understand and categorize the adhesives used and also be able to measure their thicknesses. In this respect, non-linear (second and third harmonic generation) spot measurements were applied to explore the feasibility of precise thickness detection and composition identification of different natural and synthetic lining glues.

Second and third harmonic generation (SHG–THG) processes are scattering phenomena. Two or three photons of angular frequency ω are destroyed and a photon of angular frequency 2ω (for SHG) or 3ω (for THG) is simultaneously created in a single quantum-mechanical process. SHG and THG modalities present the capability of intrinsic three-dimensionality, high axial resolution, the ability to section deep within the sample and the reduction of “out of focal plane” photobleaching in the specimens. Furthermore, optical higher harmonic generation does not deposit energy to specimens due to its energy-conservation characteristics, providing minimum sample disturbance (e.g. thermal, mechanical side-effects) which is desirable for art conservation studies.

SHG and THG microscopy measurements are mainly used as tools for the in vivo imaging and mapping of sub-cellular biological structures and processes. SHG modality provides information related to stacked membranes and arranged proteins with organized structures, such as collagen [46]. In addition, SHG is a useful technique for probing membrane-potential-induced alignment of dipolar molecules [7]. THG is proven to be generated from regions with optical inhomogeneity [8] and is used for probing structural and anatomical changes of biological samples at cellular or sub-cellular level [9, 10].

More recently, non-linear imaging techniques were used as diagnostic tools for the precise detection of multilayer structures of painted artworks [11]. Specifically, via the detection of THG and multi-photon fluorescence (MPEF) signals from model painted artworks, the thickness determination of varnish-protective layers and the identification of the chemical composition of painting materials were achieved [12]. Detection took place in both transmission and reflection mode, demonstrating the ability of this non-destructive technique to be applied on the evaluation of original artworks.

In this study, higher harmonic generation (SHG–THG) spot measurements are employed for the extraction of novel information related to thickness determination and composition discrimination of various types of lining glues used in painted artworks.

Experimental apparatus

The experimental set-up is outlined in Fig. 1. A femtosecond t pulse laser (Amplitude Systems) has been used as excitation source at 1,028 nm. The pulse duration was 200 fs and the repetition rate 50 MHz. The beam was directed to a modified optical microscope (Eclipse ME600D-Nikon) using suitable dichroic mirrors, and was focused tightly onto the sample by an objective lens with high numerical aperture (×50 NA 0.8-Nikon). To ensure that the back aperture of the objective was fulfilled, a telescope system was used.
Fig. 1

Schematic of the experimental set-up. L1 and L2 are lenses, DM is dichroic mirror, O is objective lens, C is condenser lens, F1 is colored glass filter, F2 is interference filter and PMT is photomultiplier tube

The samples were placed to a round glass slide of 35 mm diameter and ~45 μm thickness (Marienfeld) that fit into a motorized xyz translation stage (8MT167-100-Standa). The minimum step of the stage in each direction was 1 μm. The focal distance between the objective lens and the sample was controlled by the motorized stage employing a specially designed software (National Instruments, Labview 7.1). The recording time was 10 ms in every step. A CCD camera (PLA662-PixeLINK) was used for the observation of the samples. The average laser power on the specimen was 30 mW (0.6 nJ per pulse). No damage in the sample was observed at this power.

THG signals were collected and collimated in transmission mode by a condenser lens (Plan-Apochromat ×100 NA 1.4 Carl Zeiss). After passing through a 340-nm colored glass filter (U 340-Hoya), the signals were sent to a photomultiplier tube (Hamamatsu H9305-04) connected to a lock-in amplifier (SR810-Stanford Research Systems).

SHG signals were collected in the backward direction using another photomultiplier tube (PMT Hamamatsu R4220) connected to a lock-in amplifier. The photomultiplier tube was attached at the position of the microscope eye-piece. A 514-nm interference filter (F03-514.5-CVI) was placed at the photomultiplier input.


Eight different samples representing natural and synthetic lining glues, casted on thin cover-slips, were analyzed. All the samples were fresh.

Three samples of natural materials from plant and animal origin, were examined.
  • Sample 1 is flour paste, the most common used adhesive. Flour consists of starch and gluten, and when combined with boiling water forms a thick and viscous paste. It is stable and remains water-soluble after drying [3].

  • Sample 2 is rabbit skin glue. Rabbit skin is impure gelatine with some lower molecular weight residues of collagen, keratin and elastin. It forms top-quality and high viscous glue when diluted in warm water, and remains water-soluble after drying [2].

  • Sample 3 is starch paste. It is an adhesive prepared by mixing one part of flour paste, one part of rabbit skin glue and some drops of linseed extracts and garlic juice [1]. Those materials, once added to boiling water forms a thick, viscous paste. It forms a strong and water-insoluble paste, which can be removed only with the aid of enzymes, after drying.

Additionally, four samples from modern synthetic materials and a sample comprises a mixture of synthetic and natural materials (sample 7) were examined.
  • Sample 4 is Beva 371 a synthetic wax. It is a thermoplastic elastomeric polymer mixture, containing ethylene vinyl acetate (EVA) resin with a variety of waxes and ketone resins [13].

  • Sample 5 is Lascaux 498 HV acrylic adhesive. Lascaux 498 HV is a water-based emulsion of thermoplastic acrylic resin containing butyl acrylate thickened with methacrylic acid [14].

  • Sample 6 is Mowilith, a synthetic adhesive. Mowilith is a polyvinyl acetate (PVAC) homopolymer, soluble in certain solvents, such as ketones and aromatic hydrocarbons [1, 3].

  • Sample 7 (Fluor-Lascaux) is a combination of synthetic and natural materials and is prepared according to a modern recipe of flour paste [1]. A relatively small amount of Lascaux 498 HV is added to the flour paste (15 grams to 500 cm3 of paste) forming a water-soluble and a very stable adhesive.

  • Finally, sample 8 is Vinavil, a synthetic adhesive. Vinavil is polyvinyl acetate water dispersion. It is a water-soluble material (in contrast to sample 6) and upon drying remains soluble in ethanol/water mixture [3].


Figure 2 depicts second- and third-harmonic-generation spot measurements from the flour paste of sample 1. The two non-linear signals (SHG and THG) are generated simultaneously from the focal volume at the sample plane. Through THG measurements, three different peaks can be distinguished indicating the interface between the different media. The first one corresponds to the air/glass interface; the second faint one to the glass/material and the third peak represents the material/air interface, respectively. The second peak of THG in Fig. 2 is faint due to the fact that the efficiency of the THG process is lower for glass/material interface than that for air/glass and material/air interfaces. A possible explanation is that the refractive index mismatch is higher for air/glass and material/air interfaces than for the glass/material interface. The measured thickness of the flour paste layer is 35 μm while the thickness of the glass is always around 45 μm. The axial resolution of our system is in the order of a few microns (the axial beam-waist, when the Nikon (×50, NA 0.8) objective lens is used, is about 2 μm). Consequently, the detected THG signals provide the proper information for the precise thickness determination of the lining glue layer. The depth penetration into the samples is limited from the small working distance of the high numerical aperture lenses that are used for the generation of the non-linear phenomena. In our case, the maximum depth that can be reached into the specimens is 300 μm.
Fig. 2

Second- and third-harmonic-generation spot measurements from a flour paste (sample 1)

Complementary information related to the composition of the glues is acquired via the realization of SHG measurements. High SHG signals are collected from the flour paste of sample 1 (Fig. 2). The main ingredient of the flour paste is starch. Starch has been shown to have a naturally high χ2 coefficient [15, 16]. The starch granule, a highly birefringent structure, consists of crystalline amylopectin lamellae organized into effectively spherical blocklets and large concentric growth rings [17]. It is feasible that these optically active structural features are responsible for the strong SHG signals arising from the starch-based samples (like flour paste). Similarly, high SHG signals are obtained from starch glues (data not shown).

Consequently, THG represents a new qualitative way for the precise detection of the thicknesses of various types of natural and synthetic glues. Furthermore, the non-destructive modality of SHG allows the discrimination, rather than the determination, of the composition between different lining glues.

In the current study, THG signals were detected in transmission mode while SHG signals were collected in reflection mode. However, since THG measurements can be realized in reflection mode [12], the non-destructive technique of optical higher harmonic generation (SHG–THG) can be integrated in a compact, novel instrument with the capabilities of diagnosis and laser cleaning for art restoration purposes. The great degree of axial resolution, the increased penetration depth and the minimal sample disturbance that second and third harmonic generation diagnostic modalities present, are essential for the accurate online control of any cleaning procedure and the elimination of the damage effects on the exposed painted surfaces.

Figure 3 depicts non-linear (SHG–THG) spot measurements from rabbit skin glue (sample 2). The thickness of this natural glue is 21 μm (based on THG signals). Additionally, detectable SHG signals arise from this collagen-based glue. Collagen, which has a highly crystalline not centro-symmetric triple-helix structure, produces SHG extremely effectively [18]. Aging phenomena were not investigated in the framework of the current study. However, sample 2 was tested again 7 months after its preparation. It is feasible to obtain high SHG signals from the aging sample, and the measured thickness is 19 μm (data not shown). Therefore, it is expected that aging impact to the generation of the non-linear phenomena from the samples is restricted.
Fig. 3

Second and third harmonic generation spot measurements from rabbit skin glue (sample 2)

Figure 4 shows SHG and THG measurements from Beva 371 a synthetic wax (sample 4). The precise thickness identification of the glue is possible via THG measurements (121 μm). Second-order non-linear signals were not collected from this synthetic glue due to the absence of efficient SHG sources.
Fig. 4

SHG and THG measurements from a synthetic adhesive (Beva 371 synthetic wax, sample 4)

Figure 5 presents high-order harmonic (SHG–THG) measurements from sample 5 (Lascaux 498 HV acrylic resin glue). It is feasible to calculate the thickness of the glue, which is 38 μm, based on THG spot measurements. In this case, no SHG signals are collected from this acrylic lining glue due to the absence of highly ordered and birefringent molecules (such as collagen and starch granules). Similarly, non-linear measurements were performed on Mowilith, a polyvinyl acetate homopolymer (sample 6). From Fig. 6, it can be seen that the thickness of this sample is 31 μm, while no SHG signals are detectable from this synthetic glue.
Fig. 5

Higher-harmonic-generation spot measurements from an acrylic resin glue (Lascaux 498 HV, sample 5)
Fig. 6

Non-linear measurements from Mowilith, a synthetic glue (sample 6)

Figure 7 depicts non-linear signals collected from Fluor-Lascaux glue (sample 7). The thickness of this glue is 31 μm. The presence of starch granules that are very efficient SHG sources in the composition of this glue allows the collection of high second-order signals.
Fig. 7

SHG and THG measurements from a Fluor-Lascaux glue (sample 7)

Figure 8a shows SHG and THG spot measurements from Vinavil, a polyvinyl acetate water dispersion (sample 8). The thickness of the glue is 190 μm. After initial measurements, a small area of the sample was mechanical-treated in order to remove a part of it. The use of mechanical methods is the most common in such cases and it is chosen as the most appropriate one. We have to note that in certain cases, additional chemical treatment is required. However, the selection of proper solvents is critical, since they can easily penetrate inside the materials and can be extremely harmful for the painted layers causing changes to their properties and chemical composition. In our case, the conservation practice of mechanical treatment was chosen for the realization of the removal of a thin layer of the lining glue. Figure 8b presents non-linear measurements to the same glue (sample 8) after smooth treatment by mechanical etching. The measured thickness is 170 μm. The reduction of glue thickness is 20 μm due to treatment. Therefore, this non-linear technique has the unique capability for accurate on line monitoring of the thickness reduction of the mechanical-treated lining glues, ensuring the usefulness of this non-destructive modality for real-case studies. No SHG signals are detectable from the synthetic glue (before and after treatment). The integration of these non-linear diagnostic techniques in a developed, user friendly, compact workstation, for the realization of measurements from original lining glues of painted artworks, comprises our main future target.
Fig. 8

a Non-linear spot measurements from a synthetic glue (Vinavil, sample 8). b The same measurements after mechanical treatment of the sample


Non-linear modalities (such as SHG–THG) comprise powerful, non-destructive, diagnostic tools that provide detailed information regarding lining glues thicknesses and composition. Eight different types of glues have been investigated. THG modality is an appropriate technique for the precise determination of the thicknesses of the adhesives. SHG measurements attribute complementary information related to the composition of glues (existence or absence of collagen and starch granules). Consequently, these diagnostic non-linear modalities provide essential, unique information for the assessment of the appropriate conservation method that has to be followed on the lining “textile support” of a painted artwork.

Furthermore, we have to note that the compact size of the employed excitation source and the reduced time of data acquisition (each measurement lasts less than a minute) make this innovative technique ideal for in situ laser diagnosis of painted artworks.


This work was supported by the UV Laser Facility operating at IESL-FORTH under the European Commission “Improving Human Research Potential” program (RII3-CT-2003-506350) and by the Marie Curie Transfer of Knowledge project “NOLIMBA” (MTKD-CT-2005-029194).

Copyright information

© Springer-Verlag 2009