Sunteți pe pagina 1din 14

The Gate of Paradise: Physical Optimization of the Laser Cleaning Approach Author(s): S. Siano and R.

Salimbeni Source: Studies in Conservation, Vol. 46, No. 4 (2001), pp. 269-281 Published by: International Institute for Conservation of Historic and Artistic Works Stable URL: http://www.jstor.org/stable/1506776 Accessed: 02/08/2010 09:12
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=iich. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

International Institute for Conservation of Historic and Artistic Works is collaborating with JSTOR to digitize, preserve and extend access to Studies in Conservation.

http://www.jstor.org

THE GATE OF PARADISE: PHYSICAL OPTIMIZATION OF THE LASER CLEANING APPROACH S. Siano and R. Salimbeni
Summary-A methodology for the use of lasers in the restoration of the Gate of Paradise, a gilded bronze artwork by L. Ghiberti, was evaluated theoretically and experimentally. Preliminary characterization of the materials to be laser-cleaned made it possible to estimate the thermal regimes involved and to model the ablation process. Calculations aimed at controlling unwanted heating of the gold film were also developed before laser-cleaning tests on large areas were carried out. Finally, the cleaning tests were critically evaluated by means of instrumental inspection. The practicability of the laser approach, either as a preliminary to gentle chemical cleaning involving lower concentrations of chemical agents and shorter exposure times or as a standalone treatment, was demonstrated.

Introduction
The Gate of Paradise by Lorenzo Ghiberti, as well as the other two doors of the Baptistery of Florence, have undergone systematic investigations since 1966, the year of the flood. The studies were aimed at assessing the state of preservation, understanding the deterioration mechanisms, and defining criteria for the future conservation of these gilded bronze artworks [1-5]. The urgent need to make choices also stimulated new historical researches on Ghiberti's works [6, 7]. The large number of multidisciplinary investigations performed on the Gate of Paradise up until the beginning of the 1980s, dealing with the various aspects of the conservation problem, were briefly summarized in a report published by Baldini [8], then Superintendent of the Opificio delle Pietre Dure (OPD), the institution responsible for the restoration work. Basic data about the composition of the copper alloy, gold film, corrosion products and encrustation were obtained through detailed analyses. The main outcome with respect to the conservation treatment was the establishment of a chemical cleaning protocol, which was formulated following several optimization phases at the beginning of the 1980s [9, 10]. After being dismounted from the single casting bronze framework, each sculptural element undergoes a specific sequence of chemical treatments in xylene-acetone (dimethylbenzene-propanone), distilled water, and sodium potassium tartrate solution baths (for details see [8]). This last treatment eliminates dangerous copper corrosion products from the gilded surface and, to some extent, underneath it. The dismounting of the sculptural elements from the framework represents the most critical phase of such a procedure. On the other hand, this step is necessary to avoid the penetration of the chemical Received May 2000 Studies in Conservation46 (2001) 269-281 agents and encrustation products inside the framework cavities behind the sculptures, which would cause serious corrosion problems. As six of the 10 sculptured panels (79.5 x 79.5cm) were displaced due to the banging of the door caused by the violent flow of water during the flood of 1966 and were afterwards fixed with screws, their removal for the conservation treatment was relatively simple. Up until 1985, four of these seriously damaged panels had been dismounted and three of them restored, while the door was still in use at the Baptistery [11]. The last two panels were restored after Ghiberti's artwork was replaced by a copy in 1990, when the original door was brought to the OPD laboratories. This first group of restored sculptures is exhibited today in a controlled environment at the Opera del Duomo museum in Florence. The problem of dismounting the remaining elements is not yet completely solved. So far, two more panels and eight smaller sculptures surrounding the scenes of the Old Testament have been dismounted. Thus, the last two panels and 40 smaller elements would still have to be removed before being subjected to the cleaning procedure described above. The mechanical dismounting takes a long time and appears dangerous for the integrity of the sculptures because of several macroscopic fractures and the very fragile internal structure of the bronze that was revealed in gamma-ray images [8]. This problem, besides the many questions about the possible consequences of an overall chemical cleaning following known treatments as well as previous, undocumented ones, led the OPD to investigate alternative treatment procedures. The main criteria were that the treatment should: 1 2 be non-invasive allow in situ cleaning of the remaining sculptures

269

S. Siano and R. Salimbeni 3 4 achieve a degree of cleaning comparable with that obtained with the present protocol indistinctly seen, with the exception of projecting figures and the floor of the scene, where the encrustation is significantly thicker (Figure 1). The most exposed gilding appears veiled by a dark greenish film, which alters the natural colour of the gold alloy. Furthermore, several brown areas, not showing any trace of gilding, can be clearly observed in the sky. Since the flood did not detach the present panel, these brown zones seem to be the result of previous aggressive cleaning treatments. The shape of these zones is fairly circular or rectangular, sometimes presenting straight lines like scratches. Finally, a diffuse distribution of a large number of reddish-brown spots (not distinguishable in Figure 1) was also observed. The encrustation distribution is similar to that on the previously restored panels and it was reported in the literature together with the specific deterioration dynamics driven by the copper-gold galvanic couple [2, 10, 18]. Chemical analysis Besides the general importance of characterizing the original materials and understanding their deterioration, it should be underlined that knowledge of the composition of the materials that are going to pct 3 pct4 A (pct2) pct 1

produce results aesthetically similar to those obtained for the recently restored panel 5 minimize risks of future deterioration triggered by cleaning agents After introducing the problem and stating the restoration goal [12], the OPD requested us to investigate, from the point of view of physics, the laser cleaning approach. As is well known, the laser technique has undergone significant developments during the last decade in terms of basic studies aimed at understanding the specific physical dynamics involved in the cleaning process, and in terms of technological improvements made to laser systems dedicated to conservation [13-17]. Thus, the situation was considered ideal for an in-depth discussion of the possible application of this new approach to an important artwork such as the Gate of Paradise. In what follows we present a preliminary evaluation of the possibility of meeting the specific requirements by a suitable integration of the laser cleaning technique in the restoration procedure. The experimentation was carried out on the panel entitled 'The Story of Noah' that was recently dismounted from the door. At the present stage, our cleaning tests were aimed at optimizing the laser procedure, then at evaluating the potential of this technique with respect to the specific restoration problem. The work included the following steps: 1 characterization of deterioration 2 3 selection of a suitable laser source preliminary tests to estimate the average reflectance of encrustation and gilded surface, cleaning thresholds, laser cleaning effectiveness, possible side-effects as a function of laser fluence and irradiation conditions physical modelling to estimate in a general way the laser-material interaction effects in order better to control them cleaning tests of different surface morphologies: flat surface, bas-relief, and projecting sculptural elements evaluation of the cleaning results

'The Story of Noah' panel


To the naked eye, the panel appears covered with a blackish-green encrustation which still allows the underlying gilding on most of the surface to be

B(pct 5) C(pct 6) Figure 1 'The Story of Noah' panel. 'pet' indicates the preliminary test sites, while areas A, B and C were laser cleaned after optimization of the irradiation conditions and parameters. Studies in Conservation46 (2001) 269-281

270

The Gate of Paradise: physical optimization of the laser cleaning approach Table I Metal* Cu Sn Zn Pb Analysis of the copper alloy Technique Absorption spectrophotometry by flame in air/acetylene Emission spectrophotometry by flame in nitrogen/acetylene oxide Absorption spectrophotometry by flame in air/acetylene Absorption spectrophotometry in graphite oven Concentration (%) 90.9 6-6 1-2 1.1

*Otherelementsdo not exceed0-2%. be laser cleaned helps in choosing suitable laser parameters and in understanding the specific physical dynamics involved in the removal processes. In particular, it allows realistic assumptions about thermal and mechanical properties and estimation of behaviour under different irradiation conditions. These concerns are very important in situations where it is difficult or impossible to perform direct measurements of the main physical parameters such as temperature and pressure associated with laser-material interaction.
.. fr7r- :r.. . . . ::>:g:. !.' :. m", ..

-,197~?~ ;~a-?. ... <: .:.b: :`.::... .. : ^,... iiiiPiiSiii...... :~~:::~ii;.: ::. ................ii... : . , ;.i;..:. ;...X.~::./:::::..: ......
*i:... : ........ ...... . . '. .. ;* ... . ..... .....
...... .... ... ..........

pP~
?

-.S

x : : a` i[i

? ...

.....

....

..
.:...':

..-. B ;z .. , . ... ...i .. .: .::.: . . . ....

.gE.3,. .

.~

.. ......... ....:. , ~ : .:. ::v

?':.~?iiii}!!?.:i:--'""iZiiiii~

Figure 2 Backscattering SEM images of the external (a) and internal (b) surfaces of the gold film on two different gilding fragments. Studies in Conservation46 (2001) 269-281

The metals As reported,the door was made with a quaternary copper alloy and gilded by the mercury-amalgam method [5, 19]. In orderto verify the alloy composition employedfor the 'Story of Noah' casting, a small bronzefragment(10-lmg) was removedat the bottom edge of the back of the panel, near the right edge. The concentrationof copper, tin, zinc and lead was measuredby atomic spectrophotometry (Table 1). The copper concentrationwas found to be similarto that of the 'Story of Joseph'panel [8], whereas the total amount of alloying metals was slightlyhigher. Severalmicroscopicfragmentsof gold film were collectedfrom the waterused for wettingand washing, to analyse the gilding. These fragmentscame from exfoliation micrositesthat are distributedon the panel surface at relief edges and corrosion microblisters. The smallest fragments, hardly separable from the encrustationproducts,were analysed by scanning electronmicroscopy(SEM) and energydispersive X-ray spectrometry (EDX). Backscattering imagesof fragmentsshowingboth sides of the gold film are shown in Figure 2. The external surface appears smooth and presents straight marks. Conversely,the internal surface is irregularwith protuberancesprobably caused by the amalgam flowing over the texturedsurfaceof the underlying bronze substrate.The observationof verticalfragments and of the broken edge shown in Figure 2a allowed estimationof the averagethicknessof the gold film as about 61im. However, it should be noted that the fragmentsoriginatingfrom microblisters are expected to be thinner with respect to the undeformed film. Microprobeanalysis revealeda residualmercury content (Figure 3), higher on the internal surface than on the externalsurface.The relativeconcentrationsof gold, mercuryand lead were determined by atomic spectrophotometry analysis of the fragments that were separated from the encrustation products:98-3%Au, 1.3%Hg and 0.4% Pb. The copper was not measuredbecause it would have been difficultto determineits origin, since it could 271

S. Siano and R. Salimbeni

2
AuLa

1 VM bma

CuKa

Q.

IgLa

I
298 8.00 CNT 9.00 9.92KeV 10.00 11.00 12.00 10 eV/ch B EDAX

20

40

60

80

100

Figure 3 Evidence of residual mercury content by EDX analysis. have been related to copper compounds attached to the gold film. From the previous data one could expect thermal properties of the gilding film close to those of pure gold, in particular if the heat pulse is released on the external surface as happens during laser cleaning. Encrustation products Encrustation samples were removed with a small stick and analysed by ionic chromatography and Fourier transform infrared (FTIR) spectroscopy. The average soluble salt concentrations are reported in Table 2, revealing a predominance of sulphates and a lower quantity of chlorides. The insoluble fraction contains mostly carbon compounds and silicates. FTIR spectra were also dominated by sulphate absorption bands (with main peaks at 3408, 1621 and 1118cm-'), with traces of nitrates. This analysis provided the reference data for the ones performed on the ablated material in order to assess any chemical changes caused by laser heating, in particular on the copper compounds. Table 2 Soluble salt concentrations (%?) by ionic * chromatography Sample
Eo
Eb Ed

t [ns]
Figure 4 Laser intensity IL as a function of time, as measured by means of a fast photodiode. The curve was fitted with an arbitrary exponentialfunction f(t) (completely superimposed) that was employedfor the thermal estimations. Choice of a suitable laser source In the near infrared spectral region the gold reflectivity is quite high (up to 99% for polished surfaces [20]), whereas encrustation products such as carbon black particles and copper compounds exhibit high optical absorption. Thus, in principle, Nd:YAG lasers emitting at 1064nm could represent a good choice in order to realize the auto-termination of the removal process at the gold film surface, and to limit heating of the film. However, it was important to measure the reflectance of the encrustation and gilded surfaces in order to achieve realistic estimates of the heating effects. The pulse duration of a suitable Nd:YAG laser should be short enough to allow a fast encrustation removal, thus preventing conduction heating of the underlying gold film. On the other hand, the transient heating of the gilding by direct irradiation is expected to be higher for short pulses compared to long ones. In the present case, we chose a commercial Q-switched Nd:YAG laser with a pulse duration TL = 28ns (full width at half maximum, Figure 4). As will be shown in the next section, it represents a good compromise and allows the optimization of the operating regime. The output beam was coupled to an optical fibre (core diameter 15mm) terminated by a home-made optical manipulator designed for easy handling and changing the size of the irradiation spot during cleaning. Reflectance measurements Figure 5 shows the experimental set-up employed to Studies in Conservation46 (2001) 269-281

Sulphates 25-05 22-36


25-56

Chlorides 6-85
4-05

Nitrates

5.89
3-85

Ef

25-12

3-01 5-7

stick. Eb, Ed and Ef are laser-ablated products removed at the beginning, during, and in the last phase of the cleaning process, respectively.

an *Eorepresents encrustation sampletaken with a small

272

The Gate of Paradise: physical optimization of the laser cleaning approach

Preliminary cleaning tests IS


9

_-

'

a-

Six small areas, representing the deterioration features described above, were selected for preliminary cleaning tests. They were selected as follows: 1 2 along the top edge of the panel, including some reddish spots in the sky, to try cleaning the typical encrustation distribution and brown spots at the bottom of the scene, to test the removal of thick encrustation.

scone
From the laser To the scope

The locations are indicated in Figure 1 as pct 1-6. Figure 5 Experimental set-up for the mweasurement During the first tests, it was found that the use of of the optical reflectance. OF: optical fib re, L: lens, water during laser irradiation was of crucial imporF: neutral optical filter, IS. integrating sphere, P: tance, whereas cleaning in dry conditions was considered dangerous for the gilding. Indeed, in dry photodiode. conditions the laser irradiation produced in some measure the optical reflectance, Rop,of thLeencrusta- cases a slight reddening effect of the underlying tion and gilding film. Low energy pulses are gold film that could be reasonably associated with a focused onto the surface through a lens (L) and an sputtering of copper caused by the overheated integrating sphere (IS). The reflected beaim from the encrustation. Conversely, when the surface was ies irradiated area is reflected several tim on the sprayed with water, the previous side-effect disapinternal wall of the sphere, achieving a homoge- peared and the cleaning result was surprisingly neous energy distribution inside the sphere. To good. The minimal laser fluence FL (pulse energy per obtain the absolute reflectance of the surface being tested, the signal amplitude measured by a photodi- unit area, usually measured in J.cm-2) allowing ode (P) is compared with the one obtalined for a material removal was found to depend on the thickness of the water layer produced by wetting the surstandard reflectance sample (see for exam iple [21]). We measured the variation of R for the differ- face. The numerical values of FL reported in what ent encrustation typologies describeA abc>ve and for follows, which were obtained by direct measurethe gold film. In this last case the reflectance was ment of the pulse energy and of the irradiation spot measured on very localized cleaning spots. The diameter, are average values with an uncertainty of ach surface about +10%, essentially due to the pulse to pulse average values reported in Table 3 for e; typology correspond to independent measurement laser fluctuations. The most efficient ablation occurred when the le on different sites. The error on the sing] measurement (<5%), not reported in Table 3, was mini- encrustation was covered with no more than several mized by pulse to pulse monitoring o f the laser tens of gm of water. Conversely, thicker water layalue to the ers inhibited material removal, while ineffective energy, then normalizing each energy v; incident energy. As can be seen, in each case there microfragmentation of the encrustation material was a significant difference in reflectance between was observed. In the first case, the initial ablation the encrustation and the gilded surfi ace, which threshold resulted at about 300mJ.cm-2. This fluence allowed an overall encrustation removal, with makes the laser ablation process selective the exception of small spots of copper corrosion products with typical sizes of 100-500[tm, mostly Table 3 Typical optical reflectances (at 1064nm) of localized on gilding microblisters. Thus, a finishing irradiation phase at about 500mJ.cm-2 was required different encrustation typologies and gildin I ~g to complete the cleaning. This fluence value will be referred to as 'cleaning threshold'. R (0o ) some sites) Surface typology (for O~P Damage to the gold film, which resulted in surface texturing and formation of microsize melting 7 Brown spots 14, 10, spheres inside the irradiated area, occurred at about 15, 20 Slight encrustation 2-3J.cm-2. We would like to underline that this inva25, 21 Reddish spots sive test, that did not remove the gilding film, was Thick encrustation 10, 9 performed on two spots of diameter of about 72, 80 69, Gilding 25mm at the top edge (pct 1). The spots were afterStudies in Conservation46 (2001) 269-281 273

S. Siano and R. Salimbeni wards 'erased' by smoothing the surface with a small stick. In the best operating conditions the ablated material, including a significant fraction of atomized water, was projected 10-20cm away from the irradiated surface. In order to obtain information about the ablation dynamics we performed chemical and structural analysis of the ablated material. The samples were taken by irradiating the surface through a glass slide, then collecting the ablated material on it. Three samples were taken: at the beginning (Eb), during (Ed) and at the end of the surface cleaning (E). The ionic chromatography data are reported in Table 2 together with those of the encrustation samples taken mechanically (EO). The sulphate and chloride concentrations were similar, whereas in samples Ed and Ef nitrates were also detected that are therefore localized in close proximity to the gold film. These data, along with optical and SEM microscopy of the ablated material, which did not reveal structural changes, melting or vaporization effects, demonstrated the absence of thermal and chemical transformations of the copper compounds and atmospheric particulates generated by the laser heating driving the ablation process. The analysis also suggests that the photomechanical contribution plays a significant role in the material removal process. Cleaning tests on red spots revealed interesting features. Most of them looked like soft wax and were easily atomized or vaporized by laser irradiation. However, in a number of cases, they were hard and the laser irradiation caused flaking, after a few pulses, of large pieces, up to about 0.5cm2 and some hundreds of tm thick. This material, its distribution and stratigraphy are interesting from an historical point of view and are now under investigation. Finally, the cleaning tests of the brown spots revealed even more interesting features. In fact, while, as expected, the small spot (pct 3 in Figure 1) did not show gold underneath, the larger one (pct 4) covered an area of gilding that was not suspected. In spite of a significant gold content, confirmed by absorption spectrophotometry, this area behaved differently when submitted to the laser cleaning operation. Besides a lower cleaning threshold (about 300mJ.cm-2), we also observed the occurrence of gilding atomization at a fluence as low as 600mJ.cm-2. After cleaning, the reflectance at the laser wavelength was observed to be about 55%, i.e., significantly lower than the typical values measured for the gilding (Table 3). All these peculiarities suggested that the large brown areas correspond to old integration treatments (for example with gold leaf) to repair damages, in agreement with the hypothesis formulated by Bearzi [2].

Laser-material interaction modelling On the basis of the materials characterization, and preliminary reflectancemeasurements cleaning tests, it was possible to describe,from the point of view of physics, the cleaning process of the encrusted gildedsurface. observaAccordingto the previousexperimental tions, at the beginning of the surface irradiation most of the laserenergyis dissipatedto heat and to whereasduringthe finishremovethe encrustation, there is a non negligibledirectheatingof ing phase area is already the gold film. Most of the irradiated cleanedduringthis last phase. Thus, it is useful to discuss separately two interaction regimes: laser ablationand directlaserheatingof the gold film. Ablation mechanism In the presentcase, as well as in most conservation applications,laser ablation of the encrustationhas a thermal nature, i.e., it arises by conversion of photon energy to heat inside the materialsunder irradiation. Dependingon the laser fluencesand on the distribution of optical absorbers within the encrustation, different material removal regimes At could be involvedduringthe cleaningtreatment. the cleaning threshold derived above, the most to importantcontributions the materialremovalare

Laser beam

IA-

w
E
GF CO

B
Figure 6 Schematization of the panel surface and encrustation and of the laser-inducedphotomechanical action in wet conditions. B: bronze substrate, CO: copper oxide layer, GF: gilding film, E: encrustation layer, W: water layer. The arrows indicate the propagation of the pressure wave. Studies in Conservation46 (2001) 269-281

274

The Gate of Paradise: physical optimization of the laser cleaning approach from localized vaporization of water and from photoacoustic generation. To describe the dynamics, let us consider the stratigraphy shown in Figure 6 where the encrustation products are completely immersed in a slightly thicker water layer. From an optics point of view, the water does not absorb at the Nd:YAG laser wavelength, while the encrustation is an inhomogeneous absorber with a distribution of absorption centres containing various 'chromophores'. Among these, carbon black particles are very effective, providing a reflectance of 15% and an absorption coefficient of 8-3 x 104cm- [22]. This corresponds to an optical penetration depth of 120nm (i.e., the penetration length inside the material providing an attenuation of the incident fluence of l/e, with e = 2.73). Laser irradiation produces a temperature rise of the surface absorbing centres, driving their rapid expansion (Figure 6) on a nanosecond time-scale. During the same time the generated heat spreads in water along a characteristic distance given by the thermal diffusion length [23]:
l,w =2VDw

'

rL

(1)

reflected at the gold surface (Rac 0.95) without phase change. The absence of visible vapour jets, the observation of material projected far away from the irradiated surface, besides the analysis results of the ablated material and the estimate provided by equation (1), suggest that cold spallation plays a significant role in the present irradiation conditions. When the water thickness exceeds a certain limit, the first phase of the interaction process is similar to that in the previous case, but the increased distance between the water-air interface and the encrustation can result in an independent spallation of water or a strong amplitude attenuation of the pressure wave by diffraction [24]. In such confined conditions, the encrustation fragmentation can also occur by free cavitation bubble collapses [25], whereas fast material removal is inhibited by mechanical impedance of the water layer. This is in agreement with the experimental observations. During the following cleaning tests we used just enough water to achieve a cold spallation mediated regime; thick water layers were found both ineffective and dangerous, because of possible temperature and cavitation damages. Heating of the gold film In the ablation regime occurring at the beginning of the laser cleaning, the encrustation shields the gold film to some extent, whereas during the finishing phase several pulses can reach it directly. It is therefore necessary to evaluate the temperature distributions due to the irradiation of the film. Since the thermal conductivity of gold (K 312W.m-lK [28]) is much higher with respect to air or water (K = 0O03W.m-'K and K = O-6W.m-'K [29]), heat may propagate inside the film during the laser pulse. The thermal diffusion length, calculated by replacing Dw with the gold thermal diffusivity Dg in equation (1), is lthg = 3-7m. Thus, at least for areas showing a continuous gilding (i.e., without exposed substrate), heat propagation is confined within the gold film during laser irradiation. That allows the use of the semi-infinitive solids approximation where an ideal thermal contact between gold and water is assumed [23]. The heat supplied per unit area and unit time at the interface can be written as:
f(t) = AOP IL (t)

where Dw is the thermal diffusivity of water. Therefore, the water temperature increases up to the vaporization limit in a very small depth (lthw 127nm). Rapid expansion of the absorbing components and water vaporization both contribute to generating a compression wave (arrows in Figure 6) that propagates in a multiphase medium (liquid, solid and gas), inducing structural weakening and local microfragmentation of the encrustation. When the pressure transient reaches the water-air interface it is reflected back following the amplitude law [24]: R PaCa-PwCw
Pac a + Pwcw

(2)

where Rac is the acoustic reflection coefficient, Pa and pware the densities of air and water, Ca and c are the acoustic speeds in these two media. Since tion at the surface with a phase inversion. Thus, the compression wave becomes a rarefaction wave which returns back inside the encrustation, contributing to the cavitation expansion induced by water vaporization [25, 26] but also generating more fragmentation. This contribution to the ablation dynamics is usually called 'cold spallation' (see for example [27]). The gilding layer acts as a rigid boundary [24] by increasing the photoacoustic pressure then the spallation phenomenon. In fact, the acoustic impedance of gold is much higher than that of water so the pressure wave is mostly Studies in Conservation46 (2001) 269-281
>> Paca, Rac PwCw

-1, which means a total reflec-

(3)

where IL is the laser intensity (laser power per unit area, usually measured in W.cm-2) and A = 1 R is the optical absorbance which, basel on the numbers in Table 3, ranges between 0-2 and 0.3, while f(t) follows the temporal shape of the laser 275

S. Siano and R. Salimbeni

250
0

? 200 U)

150

300 mJcm2

100

200

300

Time ns ]
Figure 7 Thermal transient induced by laser pulse (dashed line) at the gold-water interface. The curves correspond to different laser fluences and gilding reflectances. From the bottom to the top. FL = 0 7, FL = 0-8 and R 300mJ. cm-2, R p and Rop = 7. 500mJ.cm-2, Rop 08 pulse (Figure 4). The theory provides exact solutions for the temperature rise distributions in gold film and water [23, 30]. The interface temperature transients at the operating fluences are reported in Figure 7. Maximum temperature increases between 143 and 213?C are expected at the cleaning threshold, which are well below the critical value of the gilding film. The experimental damage threshold (2.3J.cm-2) corresponds to peak temperatures of 656-983?C (according to the different values of Ao). The simultaneous heating distributions inside the water layer are plotted in Figure 8. These have to be considered just descriptive because after about 30ns (in the time-

scale given in Figures 4 and 7) the surface temperature exceeded 100?C, generating nanoscale vaporization fluid dynamics. About the thermal wave propagation inside the gold film, the theory provides, in the worst condition (Aop 0-3), a quick peak temperature decrease. By assuming a film thickness of 6itm, one can expect a maximum temperature at the internal film surface not higher than several tens of degrees Celsius (Figure 9). This last evaluation holds when taking into account also the finite size of the film and the decrease in thermal conductivity of the copper oxide substrate (see Figure 6). In general, besides the single pulse effect, the cumulative heating should be considered whenever the time lapse between consecutive pulses is shorter than that required for the complete thermal relaxation of the irradiated surface to the ambient temperature. In the present case, the problem of cumulative heating does not rise by optical absorption of the gold layer because of its high thermal conductivity, whereas it is expected to be significant for exposed copper oxide areas and residual hard encrustation. To give an idea, a pulse frequency rate f = 20Hz on a low thermal conductivity material could generate overheating of some hundreds of degrees [31], that is likely to occur in cases where the low porosity of the irradiated material does not allow an effective cooling by water.

Cleaning of large areas


The physics-based analysis discussed above allowed definition of a cleaning procedure based on a suitable water-assisted laser ablation. It was accomplished through two steps: rough cleaning by a laser fluence FL = 300mJ.cm-2 at a pulse frequency rate

0 o

-, 250
50 ns

, 200

"k l&,

- 250 0 o

- 200
IL.

150 |- 100
Q- 50 E

'
I.

150

30 ns\

*, 100
a)
1.

20 ns

E
0.1 0.2 0.3

E 50 0 2 4 6 8 10

o
0

Distancefromthe interface [ pm]


Figure 8 Thermal distributions inside the water layer during laser irradiation. 276

the Distance from interface pm] [


Figure 9 Propagation of the thermal wave inside the gold film. Studies in Conservation46 (2001) 269-281

The Gate of Paradise: physical optimization of the laser cleaning approach


f' . ---~~_rrux . . u_ i..-'-

f = 5Hz; finishing cleaning by FL = 500mJ.cm-2 and

' ".. ' ? "':'

'. : ~.: "?ni??,. ."

,~

I*~~=~ d~'/*~::~* ? ~, '~'~YI

' ;....~ ;, CII. ~ i[~,'~ : _* ~ %~Ar/%*.,v, ~'4~*'~-'"~


?+~~.,,.~ .,: .. ,~, ~, ,** ~, .~ ..... , .... d",':.,, " '~ -, , ,.a;,, ',," .,*;,~

...."'i',,-~

...~?..,'?'.,* .

r ,,-,..,,,,,?. ... .:' .

-.': '"*....t.

.. ....... ~, . : g-: ~...,,T. ~ ~ ~

-.

!It"

.e' ..?

??

f = 2-5Hz. A final washing with deionized water for about 30 minutes was also performed to remove possible encrustation microfragments and soluble compounds trapped in the surface texture, exfoliation microsites and exposed substrate. A flat surface (A), a bas-relief figure (B) and a figure including the previous features and a projecting element (C) were selected to evaluate the quality and feasibility of the optimized laser treatment. These areas are indicated in Figure 1 by solid line rectangles. The flat surface (6 x 6cm) was cleaned in about 30 minutes (Figure 10). Most of the time was spent performing the finishing cleaning, this phase being slow and delicate because it was difficult to distinguish with the naked eye between exposed micro. w . ,+.g,hi~~~~~~~~~~~~~~~~~~~~~~~~. ^. "^. S ?
~ ..:....... :: '7,;i:: w

.... ?? ;:? ? .',.

. .:.' ':i"-:

'.4

hI .K

iI

b Figure 11 Images extracted from the magnetic tape on which the videomicroscopyobservation was stored: (a) microscale image (area A), showing the high quality of the laser cleaning without any side-effects; (b) detail including a microblister exhibiting radial failures and gold loss at the apex, generated by natural deterioration.

Figure 10 Laser cleaning of the flat site A (6 X 6cm): (a) before, (b) after, (c) detail (1-5 X 1cm). Studies in Conservation46 (2001) 269-281

277

S. Siano and R. Salimbeni

_ |*

.. **..^
:i i;

i sd~ 9........... - ''


?

~~~*,.

'i w_ w 9B9 s
*'s," .r'"

*8rr ^s 1|B
' ,

~ ~ ~ Hi~r~iR
.M...I..,.p"

.~

"v

......'.'*"

:.... .. ....

Figure 13 Site C after laser cleaning. the area A was performed by high magnification optical fibre video microscopy (Figure 11a), to analyse possible residual encrustation and any disturbance of the surface texture. This technique was chosen because it allows easy inspection of sculptural elements by flexible optical fibre bundle and storage of the surface scan on a magnetic tape that is useful for further observation, comparison, monitoring, etc. An optimum degree of cleaning was confirmed, as well as the absence of melting or texturing effects attributable to the laser action. This conclusion is also supported by the SEM analysis of microfragments discussed above. As expected, the time spent to clean site B (plane projection about 2 x 35cm) was relatively longer than required for site A: about 90 minutes. The finishing cleaning of the chiselled lines, such as for example those of the tresses and of the details of the face of the figure, required a careful action because of the larger holding surface of the hard encrustation residues at the bottom of the groves. The detail of Figure 12 shows the good quality of the cleaning Studies in Conservation46 (2001) 269-281

'

. ..':i: .

.*:; ,

._

, _

.... '

Figure 12 Detail of site B: before (top) and after (bottom) the laser cleaning test. sites and encrustation residues. For an overall laser treatment, coupling of the laser beam with a stereomicroscope to perform the finishing cleaning would represent an improvement. The observation of the cleaned area through the stereomicroscope showed a finely textured surface with scratches and microblisters (Figure lOc), hence the high quality and selectivity of the laser removal. In some cases, thinning at the apex of the microblisters associated with radial failures and/or partial gilding loss by exfoliation were observed. These last phenomena, already observed on chemically cleaned panels, are attributable to natural deterioration and not to cleaning, as was also underlined in [10]. Before moving to laser cleaning of the relief figures in the main scene (B, C), further inspection of 278

The Gate of Paradise: physical optimization of the laser cleaning approach

Conclusions
We investigated a safe laser cleaning methodology for the restoration of a number of sculptural elements of the Gate of Paradise. The optimization of the treatment, allowing avoidance of any mechanical or thermal damage, was achieved through preliminary irradiation tests and a detailed physical interpretation of the experimental observations aided by chemical analysis. The cleaning tests performed on large areas demonstrated the applicability of the developed methodology on the different surfaces and encrustation features of the sculptures to be restored. The laser treatment also increased our understanding of the artwork under restoration because of the specific interaction of the laser with the materials of which it is composed. However, we cannot rule out other peculiarities of the encrustation distribution and composition, calling for additional research to improve the technique and better characterize the state of preservation and the historical issues related to the art object. The main advantages of an optimized laser methodology are that it makes it possible to clean the sculptures without the dangerous mechanical dismounting from the framework, and that it avoids or at least limits the use of reactive chemical agents. In collaboration with the team of experts from the OPD we will perform further studies aimed at comparing the results of the laser methodology we have presented with those obtained using an integrated approach combining laser ablation and the use of chemical poultices with lower concentrations of active agents. Figure 14 Details of different surface textures of site C after laser cleaning. treatment, confirmed by surface analysis performed as described for the previous test site (Figure 1lb). The problem of cleaning the inside of the projecting elements, for example the outstretched arm of the figure in site C (Figure 13), was solved by designing a suitable optical manipulator. This was realized by arranging a small dielectric mirror (2.5mm in diameter) on a plastic support allowing the orientation of the output laser beam from the optical fibre. During the cleaning operation the irradiated area was observed through a reflecting foil placed on the panel plane. This allowed cleaning of the inside of the arm and could be employed for other projecting sculptural elements. The cleaning of site C (plane projection about 10 x 35cm) was accomplished in about eight hours. The results were also very good in proximity to the panel fractures, on exposed substrate and on the various textural features of the gilding (Figure 14). Studies in Conservation46 (2001) 269-281

Acknowledgements
This study was supported by the 'Cultural Heritage' Special Project of the Italian National Research Council. The authors also wish to thank Dr Anna Maria Giusti and Dr Mauro Matteini of the OPD for having entrusted them with the present study, and the group of conservators working on the Gate of Paradise: F. Burrini, S. Agnoletti, A. Brini, L. Nicolai, for their kind and passionate participation. Finally, thanks are extended to M. Miccio of the Soprintendenza Archeologica della Toscana, for videomicroscopy analysis and helpful advice.

References
1 LEONI,M., and PANSERI, C., 'Influenza della condizioni ambientali sulla corrosione delle opere d'arte metalliche e problemi relativi alla loro conservazione' in La Metallurgia Italiana,

279

S. Siano and R. Salimbeni Atti e Notizie, Associazione Italiana di Metallurgia, Milan (1968) 3-79. DASSU, G., PIAZZESI, G., and ALESSANDRINI, G., 'Problemi di conservazione delle Porte del Battistero di Firenze' in Conservazione dei 15 Monumenti. Atti del XXIX Congresso Nazionale dell'Associazione Termotecnica Italiana, Italian National Research Council, Milan (1976) 155-165. 16 CESAREO, R., and MARABELLI, M., 'Analisi XRF di antiche porte Italiane in leghe di rame' in Atti del Convegno dei Lincei, Accademia dei Lincei, Rome (1976) 409-420. 17 ALESSANDRINI,G., DASSU, G., PEDEFERRI,P., and RE, G., 'On the conservation of the Baptistery doors in Florence', Studies in Conservation24 (1979) 108-124. 18 LEONI,M., 'Studio metallografico della Porta del Paradiso di Lorenzo Ghiberti del Battistero di Firenze', La Fonderia Italiana 4 (1981) 99-101. Lorenzo Ghiberti,materia e ragionamenti,exhibition catalogue, Centro Di, Florence (1978). Lorenzo Ghiberti nel suo tempo, Atti del Convegno Internazionaledi studi, L.S. Olschki 19 Editore, Florence (1980). 'Lorenzo Ghiberti, Storie di Giuseppe e di Beniamino, Storie di Adamo ed Eva: bassorilievi in bronzo della Porta del Paradiso' in 20 Metodo e scienza operativitd e ricerca nel restauro, ed. U. BALDINI,Sansoni Editore, 21 Florence (1983) 168-206. MATTEINI, and MOLES, 'Kinetic control 22 M., A., of the reactivity of some formulation utilised for the cleaning of bronze works of art' in ICOM Committee for Conservation 5th TriennalMeeting, Ottawa (1981) 23/4.
FIORENTINO, MARABELLI, P., M., MATTEINI,M., 23

M., 'Laser ablation in artworks restoration: benefits and problems' in Proceedings of the Conference on Laser '95, STS Press, McLean VA (1996) 441-444. SIANO,S., and PINI, R., 'Analysis of the blast wave induced by QS Nd:YAG laser photodisruption of absorbing targets', Optics Communications135 (1997) 279-284.
SIANO, S., MARGHERI,F., MAZZINGHI,P., PINI, R., and SALIMBENI, R., 'Cleaning processes of

encrusted marbles by Nd:YAG lasers operating in free-running and Q-switching regimes', Applied Optics 36 (1997) 7073-7079. Proceedings of Lasers in the Conservation of Artworks, LACONA III, Journal of Cultural Heritage 1 (Supplement 1) (2000).
MAZZA, B., PEDEFERRI,P., RE, G., SINIGAGLIA, D., and CIGADA, A., 'Nuovo metodo di

6 7 8

10

valutazione istantanea della velocita di corrosione atmosferica dei bronzi dorati' in Conservazione dei Monumenti: Atti del XXIX Congresso Nazionale dell'Associazione TermotecnicaItaliana, Italian National Research Council (1976) 155-165. E., GARBASSI, and MELLO, 'Surface spectroF., metric studies on patinas of ancient metal objects', Studies in Conservation 29 (1984) 172-180. Handbook of Optics, ed. G. DRISCOLL, McGraw-Hill, New York (1978). BUDDE, W., Optical Radiation Measurements, Vol. 4, Academic Press, New York (1983). ESENALIEV, R.O., KARABUTOV, A.A., N.B., and LETOKHOV, PODYMOVA, V.S., 'Laser ablation of aqueous solutions with spatially homogeneous and heterogeneous absorption', Applied Physics B 59 (1994) 73-81.
CARSLAW, H.S., and JAEGER, J.C., Conduction

and MOLES, 'The condition of the Door of A., Paradise by L. Ghiberti. Tests and proposals for cleanings', Studies in Conservation 27 (1982) 145-153. 11 L'oro del Ghiberti. restauri alla Porta del Paradiso, booklet of the exhibition, ed. G. Edizioni Panini, Modena (1985). LINARES, 12 GIUSTI, and MATTEINI, private commuM., A., nication (2 May 1999).
13 SALIMBENI, R., MAZZINGHI,P., PINI, R., SIANO,

of Heat in Solids, Clarendon Press, Oxford (1978).


24 R.O., OREAEVSKY, ESENALIEV, A.A., LETOKHOV, V.S., KARABUTOV, A.A., and MALINSKY,

T.V., 'Studies of acoustical and shock waves in the pulsed laser ablation of biotissue', Laser in Surgery and Medicine 13 (1993) 470-484.
25 SIANO, S., PINI, R., SALIMBENI, S., and VANNINI, M., 'Imaging and analysis of photo-

mechanical effects induced in water by high S., VANNINI, M., MATTEINI, M., and ALDOVRANDI, 'Laser restoration of paintA., power laser-target interaction', Applied Physics B 62 (1996) 503-510. ings: issue and perspective' in Proceedings of the Congress on Science and Technology for 26 YAVAS, O., SCHILLING, A., BISCHOF, J., the Safeguard of Cultural Heritage in the BONEBERG, J., and LEIDERER, P., 'Bubble nucleation and pressure generation during MediterraneanBasin, Vol. 1, Italian National laser cleaning of surfaces', Applied Physics A Research Council, Palermo (1998) 811-815. 14 SIANO, MARGHERI, MAZZINGHI, PINI, 64 (1997) 331-339. P., S., F., R., SALIMBENI, TocI, G., and VANNINI, 27 PALTAUF, G., and SCHMIDT-KLOIBER, H., R.,
280

Studies in Conservation46 (2001) 269-281

The Gate of Paradise: physical optimization of the laser cleaning approach 'Model study to investigate the contribution of spallation to pulsed laser ablation of tissue', Lasers in Surgery and Medicine 16 (1995) 277-287.
28 INCROPERA, F.P., and DEWITT, D.P.,

Authors
SALVATORE SIANO,born 1962, graduated in physics from the University of Florence. As a researcher of the Quantum Electronics Institute of Florence (CNR) since 1994, he was involved in the field of the diagnostics and modelling of laser ablation. In particular, he participated in the scientific activities of various Italian national research projects devoted to the development of laser technologies and methodologies for artwork restoration. Address: Istituto di Elettronica Quantistica-CNR, Via Panciatichi 56/30, 50127 Firenze, Italy. born 1948, graduated in physics from the University of Florence. He has been the director of the Quantum Electronics Institute of Florence (CNR) since 1991. His scientific activity has been dedicated to laser physics and technology and their application in various fields. In particular, he has been in charge of several research projects devoted to the safeguard of cultural heritage. Address: as for Siano.
RENZO SALIMBENI,

Fundamentalsof Heat and Mass Transfer, 3rd edn, John Wiley and Sons, New York (1991). 29 Handbook of Chemistry and Physics, 60th edition, ed. R.C. WEAST, CRC Press, Boca Raton (1980).
30 M.M., EL-ADAWI, HASSAN, A.F., EL-NICKLAWY, M.K., NASR, E.M., HEMIDA, A.A., and ABD EL-GHAFFAR, O.A., 'Heating effects induced

by a pulsed laser in the semi-infinite target in view of the theory of linear systems', Optics & Laser Technology28 (1996) 337-343. 31 SIANO, FABIANI, PINI, R., SALIMBENI, R., F., S.,
GIAMELLO, M., and SABATINI, G.,

'Determination of damage thresholds to prevent side effects in laser cleaning of pliocene sandstone of Siena' in Proceedings of Lasers in the Conservation of Artworks, LACONA III, Journal of Cultural Heritage 1 (Supplement 1) (2000) S47-S53.

Resume-On a evalue, de facon theorique et pratique, une methodologie de nettoyage au laser pour la restauration de la Porte du Paradis, un bronze dore de L. Ghiberti. La caracterisationpreliminaire des materiaux a nettoyer a permis d'estimer les regimes thermiquesmis en atuvreet de modeliser le processus d'ablation. Des calculs ayant pour but de controler le chauffage indesirable de la feuille d'or ont egalement ete faits avant les tests de nettoyage menes sur des surfaces importantes. Finalement, les tests de nettoyage ont ete evalues de facon critique par inspection instrumentale. On demontre la faisabilite du nettoyage au laser, soit comme etape preliminaire avant un nettoyage leger t base de produits chimiques peu concentres et de courte duree d'action, soit comme traitementprincipal. Zusammenfassung-Eine Methode zur Verwendungvon Lasern bei der Restaurierung der Paradiestiir, einem vergoldeten Bronzeobjekt von L. Ghiberti, wurde theoretisch und praktisch beurteilt. Eine Charakterisierung des mit dem Laser zu reinigenden Materials erlaubte es, die vorherrschendeTemperaturabzuschdtzen und ein Modell far die Ableitung der Wdrme zu entwickeln. Berechnungen, die darauf abzielten, ein ungewolltes Erhitzen der Goldauflage zu vermeiden, wurden ebenfalls entwickelt, bevor Tests der Laserreinigung auf grofien Bereichen durchgefiihrt wurden. Abschliefiend wurden die Reinigungsversuche durch instrumentelle Untersuchungenbewertet. Die Durchfiihrbarkeitder Laserreinigung konnte nachgewiesen werden und zwar sowohl als alleinige Ma,iname, als auch als erster Schritt bei einer Reinigung mit Chemikalien, um so die Konzentrationender Behandlungsmittelzu verringernund die Behandlungsdauerzu verkirzen. de Resumen-Una metodologia para el uso del laser en la restauraciodn la Puerta del Paradiso, una obra maestra de bronce dorado realizada por Lorenzo Ghiberti,fue evaluada teoretica y experimentalmente. La caracterizaci6npreliminar de los materiales que debian ser limpiados por medio del klser hizo posible estimar los niveles termales que debian registrarse, asi como configurar el proceso de separaci6n de los dep6sitos. Los cdlculos se llevaron a cabo antes de las pruebas de limpieza con ldser con el fin de controlar los niveles de calor no deseados para la pelicula de oro. Finalmente, las pruebas de limpieza se evaluaron criticamente por medio de la inspecci6n instrumental. La utilidad de la teoria de la limpieza con laser se demostr6, tanto como un medio preliminar a la suave limpieza quimicapor medio de bajas concentraciones de agentes quimicos y en cortas exposiciones, o como tratamiento individualaislado.

Studies in Conservation46 (2001) 269-281

281

S-ar putea să vă placă și