Applied Surface Science 211 (2003) 128–135

Spectroscopic studies of laser ablation plumes of artwork materials M. Oujja, E. Rebollar, M. Castillejo* Instituto de Quı´mica Fı´sica Rocasolano, CSIC, Serrano 119, 28006 Madrid, Spain Received 15 October 2002; accepted 4 February 2003

Abstract Studies on the plasma plume created during KrF laser (248 nm) ablation of dosimeter tempera samples in vacuum have been carried out to investigate the basic interactions of the laser with paint materials. Time resolved optical emission spectroscopy (OES) was used to measure the translational velocity of electronically excited transients in the plasma plume. Laser-induced fluorescence (LIF) studies using a probe dye laser, allowed to determine the velocities of non-emitting species. The propagation velocities of C2 in the a3pu and d3pg electronic states and of excited atomic species are indicative of a high translational temperature. Differences between the velocities of organic and inorganic species and between emissions from the tempera systems and from the pigments as pellets allow to discuss the participation of photochemical mechanisms in the laser irradiation of the paint systems. # 2003 Elsevier Science B.V. All rights reserved. PACS: 42.62.-b laser applications; 42.62.Fi laser spectroscopy; 61.80.Ba ultraviolet, visible, and infrared radiation effects; 79.20.Ds laserbeam impact phenomena Keywords: Ultraviolet laser ablation; Cleaning of paintings; Pigments; Spectroscopy of ablation plume

1. Introduction Pulsed-laser ablation is a widespread method of analysis and surface processing. Two important applications include laser cleaning, by removal of external contaminant layers, and laser-induced deposition of materials ejected on the ablation plume in thin films [1,2]. To optimise the conditions of operation, it is crucial to characterise and understand the underlying mechanisms of interaction between the laser and the substrate. Several methods are used for the diagnosis *

Corresponding author. Tel.: þ34-91-561-9400; fax: þ34-91-564-2431. E-mail address: [email protected] (M. Castillejo).

of surface ablation processes, including the study of physical and chemical effects induced on the surface and the analysis of the plasma plume by means of optical spectroscopic techniques [3–5]. Pulsed-laser ablation is investigated as a tool for the restoration of pictorial artworks, because amongst other advantages, it allows the controlled elimination of thin extended contamination surface layers with high spatial resolution both lateral and in depth [6]. The use of excimer UV lasers focuses in the removal of degraded superficial varnish layers [7]. More recently the fundamental radiation of the Nd:YAG laser at 1064 nm, and its harmonics at 532, 355 and 266 nm, covering the spectral range from the IR to the UV, are being explored to remove other contaminants,

0169-4332/03/$ – see front matter # 2003 Elsevier Science B.V. All rights reserved. doi:10.1016/S0169-4332(03)00245-9

M. Oujja et al. / Applied Surface Science 211 (2003) 128–135

as soot or polymerised dirt, from pictorial artworks [8,9]. In the worst case scenario of removal of varnish using KrF excimer laser radiation, the laser beam directly interacts with the paint layer. If the laser fluence overcomes the threshold for ablation, ejection of material from the surface takes place. Direct laser irradiation induces various effects over the paint constituents, pigments and binding medium. Discoloration, alterations in the molecular composition of pigments or degradation of the binding medium may take place on the surface layer of the sample [10–14]. The analysis and assessment of the relative importance of thermal versus photochemical mechanisms on ablation is crucial to understand the mentioned effects and imperative in order to validate the cleaning of paintings with UV laser radiation. In particular, it is important to determine the product distribution and the kinetic energy of the species ejected in the plume in order to understand the ablation process [15,16]. Aiming at the elucidation of the above-mentioned processes, this work reports on optical spectroscopic studies of the plume accompanying pulsed UV excimer laser ablation of dosimeter tempera systems. These systems consist in a mixture of pigments, used in the traditional painting practice, and an egg-based binding medium. Measurements were also carried out in pigments as pellets to investigate how the binding medium modifies the interaction between the laser and the pigments in the paint. Two techniques have been used to study the plume with spectral, spatial, and temporal resolution. Time resolved optical emission spectroscopy (OES) served to determine the velocity of flight of emitting fragments. Laser-induced fluorescence (LIF), in which a second dye laser, conveniently delayed with respect to the ablation laser, excites the fluorescence of selected species, allowed the analysis of ejected components that are formed in nonemitting fundamental or low energy states. The corresponding distribution of translational energies of species ejected in the plume was thus determined.

2. Experimental The samples studied in this work are well-defined egg tempera dosimeter system samples. Details on the preparation procedure of similar samples have been

129

given elsewhere [12,13,17] and a short description is provided here. The pigmented tempera paint, in which the pigments or colours are mixed with an emulsion of egg yolks (removed from their sacs) and mastic, a resin of vegetal origin, was deposited on a sheet of Melinex using a Byk Gardner (Geretsried, Germany) film applicator at 200 mm wet layer thickness. After preparation, the samples were stored in the dark for a period of 3 weeks in a dry atmosphere at ambient temperature. Results reported here correspond to tempera samples containing inorganic and organic pigments used in traditional painting practice. These are azurite (basic copper carbonate, 2CuCO3Cu(OH)2), lead white (basic lead carbonate, 2PbCO3Pb(OH)2), lead chromate (PbCrO4), zinc white (ZnO) and curcumin (organic C21H20O). Samples of unpigmented (egg and mastic only) tempera were also measured. Measurements were also performed in some of the above pigments in powder after being pressed in the form of pellets in a special holder. Measurements of the ablation plume of the samples were carried out in vacuum. The samples were placed in a cross-shaped glass cell evacuated to 9 Pa as measured directly in the cell with a Pirani gauge. At this pressure, the visible plume extended around 20 mm above the surface, allowing measurements of the emissions with spatial resolution. Ablation of the samples was carried out with a pulsed KrF excimer laser operating at 248 nm. The pulses have a duration of 20 ns (FWHM) and energy up to 40 mJ. The laser beam is focussed on the surface of the sample with a 10 cm focal length quartz lens. The fluence of irradiation was determined through measurements of the energy per pulse using a joulemeter (Gentec ED-200) and of the size of the irradiated area. Typical fluences used in this work are in the range of 0.8–2 J/cm2. These fluences are well above the ablation threshold measured for the samples that range between 0.11 and 0.40 J/cm2 [12]. The spontaneous emission from a selected region of the emitting ablation plume was collected at right angles with respect to the laser beam by a f ¼ 4 cm quartz cell and imaged onto the entrance slit of a high intensity monochromator coupled to a photomultiplier (Hamamatsu R928) with a rise time of 2.2 ns. The monochromator slits were set to provide measurements with resolution of 2.5 nm. A cut-off filter at 300 nm was used in front of the entrance window of

130

M. Oujja et al. / Applied Surface Science 211 (2003) 128–135

the monochromator to reduce the scattered light resulting when the UV laser illuminates the surface of the sample. To select the emissions from regions of the plume at different distances from the surface, the assembly formed by the vacuum glass cell containing the sample and the lens that focuses the ablation laser on the surface, were placed in a translational stage that could move perpendicularly to the fixed axis of detection. The signal from the photomultiplier was fed into a 100 MHz digital oscilloscope for measurements of the time resolved emissions. The oscilloscope was interfaced with a PC for data storage and analysis. For LIF measurements, the plume was probed by a dye laser beam at a given distance from the surface of the sample, around 8 mm, large enough to avoid interference from the spontaneous emission that accompanies plume formation. The probe dye laser explored the plume parallel to the surface of the sample and perpendicularly to the ablating laser. The delay between the ablation UV laser pulse and the probe dye laser was controlled with a BNC555 delay generator in the range of 0–10 ms, allowing the measurement of products in the volume of the plume that was being probed at various times after the start of the ablating UV pulse. Time-resolved laser-induced fluorescence, following the excitation of the selected transition, was observed through a lateral window of the glass cell in a direction perpendicular to both the ablating and the probe beams and detected with the photomultiplier behind a combination of a cut off and a band pass interference filters. A narrow bandwidth dye laser pumped by the third harmonic of a Q-switched Nd:YAG (Quantel Brilliant B) laser was used. Excitation of the (0,0) band of the C2(d3 pg a3 pu ) transition was performed with Coumarin 522 dye at 516.5 nm. The dye laser pulse energy in the interaction region was 0.5 mJ. The fluorescence corresponding to the (0,1) band of the same electronic transition, was observed at 560 nm, through an interference filter with a bandwidth of 10 nm (Specac 560FS10-50). In this arrangement the cut-off filter at 300 nm was also used to reduce the scattered light from the UV laser.

down spectroscopy (LIBS) [12,13], contains spontaneous emissions due to atomic species, present in the pigment composition, atomic calcium and molecular organic fragments as C2(d3pg), CN(B2S) and CH(A2D). We report here velocity of flight measurements of atomic and molecular fragments observed through their emission in the plume of ablation of tempera paints and pellets. We determined the velocity of flight of the more prominent neutral atomic species emitting in the ablation plume of inorganic pigment paints: CuI at 521.82 nm in azurite, PbI at 405.78 nm in lead white, CrI and PbI at 520.99 and 405.78 nm, respectively in lead chromate and ZnI at 491.30 nm in zinc white. For the latter, the ZnO emission, observed in the plume as a broad intense band at 383 nm, was also measured. Additionally, we measured the velocity of flight of C2(d3pg), and in some systems of CN(B2S), transient emissions at 516.5 and 386 nm, respectively. Measurements on C2 were motivated by the fact that this molecule is considered to be a building stone for carbonisation or charring, an effect that has been observed on the surface of samples under direct KrF laser irradiation [12,13]. It can also give insight into the processes of interaction of the laser with the organic binding medium. The weakness of the C2 emission in the ablation plume of azurite, lead chromate and zinc white prevented measurements of this species in the mentioned systems. Representative time resolved transient emissions are shown in Fig. 1. They correspond to C2(d3pg) produced in the ablation plume of unpigmented tempera. The

3. Results The KrF ablation plume of tempera samples, recorded using the technique of laser-induced break-

Fig. 1. C2(d3pg) transient emissions at 516.5 nm from 248 nm laser ablation of unpigmented tempera at various distances from the surface of the sample.

M. Oujja et al. / Applied Surface Science 211 (2003) 128–135

different transients were obtained at various distances from the surface of the sample at a laser fluence of 2 J/ cm2. Each transient was obtained by focussing the laser in a new spot of the sample and averaging over four consecutive laser pulses. At distances close to the surface, the scattering of the laser and the background continuous emission produced in the plume contribute, giving rise to an intense peak during the first 200 ns. Additionally, a flying transient emission is observed, its maximum shifts to latter times as the distance of observation from the surface increases. From the correlation of the distance to the surface with the temporal position of the maximum of the emission, it is possible to determine the velocity component, perpendicular to

131

the target surface, of the species flying in the plasma plume. Fig. 2 shows the corresponding measurements for CuI and C2(d3pg) emissions from azurite and curcumin tempera samples, respectively, obtained at a laser fluence of 2 J/cm2. For CuI, a linear correlation was found up to 20 mm from the surface. The distanceversus time dependence could be fitted by a straight line in all atomic and molecular emissions measured in this work, except in the case of the pellet of curcumin. For this system, as observed in Fig. 2, the behaviour was nonlinear and the slope decreases with distance. A linear law is indicative of a free expansion of species in the plume without collisions with background molecules and the slope of the fitted straight line gives directly the value of

Fig. 2. Distance of observation vs. time of maximum intensity for: (a) C2(d3pg) from curcumin and (b) CuI from azurite. (*) Linear fit as dotted lines, and (&) linear fit as continuous lines, represent emissions from the samples as pellets and tempera paints, respectively. The slopes of the fits give the velocities of flight of the species (see text).

132

M. Oujja et al. / Applied Surface Science 211 (2003) 128–135

Table 1 Velocity of flight of emitting and of C2(a3pu) species (in units of 103 m/s) formed in the 248 nm laser ablation plume of pigmented tempera samples and pigments pellets obtained at a laser fluence of 2 J/cm2 and pressure of 9 Pa System

C2(d3pg)

CN(B2S)

Unpigmented tempera Azurite tempera Azurite pellet Lead white tempera Lead white pellet Curcumin tempera Curcumin pellet

7.7

8.8 8.0

Lead chromate tempera Lead chromate pellet Zinc white tempera Zinc white pellet

a

a b

a a

7.6 a

Excited atomic species 10.4 10.8 9.1 8.2

(CuI) (CuI) (PbI) (PbI)

8.0 4.8 (t < 1.2 ms), 3.0 (t > 1.2 ms) a a a

C2(a3pu) 3.8 3.4 1.7 3.4 b

3.4 4.8 11.7 9.3 10.5 7.4

(PbI), 11.1 (CrI) (PbI), 8.3 (CrI) (ZnI), 11.5 (ZnO) (ZnI)

b b b b

Very low intensity. Not measured.

the velocity. A curved behaviour is interpreted as the result of the propagation of a shock wave through the background gas [18,19]. The C2(d3pg) emission obtained by ablation of pellets of curcumin was observed to be twice as intense as the corresponding emission in the paint system. If the higher intensity is due to a higher number of emitters, the increasing amount of gas collisions that slow down the flight velocities could explain the observed difference. Table 1 lists the estimated velocities of flight of the different atomic and molecular transient emissions. Two values are derived for the C2(d3pg) transient emitted by the pellet of curcumin, corresponding to the two slopes represented in Fig. 2 for this system. The obtained propagation velocities, in the range of (3–12)  103 m/s, indicate that highly energetic particles are ejected in the plasma plume. The velocity of flight of organic molecular transients, C2 and CN, are somewhat lower (around 20%) than the corresponding values for atomic species present in the pigment composition. LIF studies of C2 species were performed upon excitation of the fluorescence of a3pu which is a non-emitting state, only 716.24 cm1 above the fundamental ground state of the C2 molecule. Fig. 3 shows a typical time resolved C2 transient recorded at 8 mm from the surface. The first two peaks, similar to those shown in Fig. 1, correspond to the scattering and background continuous emission (first peak) and to the spontaneous emission produced by the ablating

pulse (second peak). The third peak corresponds to the laser-induced fluorescence by the probe dye laser, that, in the case of Fig. 3, was delayed 3.24 ms from the ablating pulse. The intensity of the LIF signal was monitored as a function of the delay that was converted to velocity of flight, since the ejected material was sampled at a known distance from the surface. A distribution of velocities of flight of the C2(a3pu) species was thus obtained in each case. In Fig. 4, the corresponding distributions of curcumin as pellet and lead white tempera are given as representative examples of the obtained results. The velocity dis-

Fig. 3. Lead white tempera spontaneous and LIF C2(a3pu) signals at 8 mm from the sample surface. Delay between ablating and probe lasers is 3.24 ms. Laser ablation fluence is 1 J/cm2.

M. Oujja et al. / Applied Surface Science 211 (2003) 128–135

133

Fig. 4. LIF signal of C2(a3pu) as a function of flight velocity from: (a) pellet of curcumin, (b) lead white tempera. The continuous lines represents fits to Maxwell–Boltzmann distributions with the indicated most probable velocity. Laser ablation fluence is 2 J/cm2.

tributions of the ablated C2(a3pu) particles can be described [20] in terms of Maxwell–Boltzmann functions. Fits to such distributions gave values of the most probable velocities of flight as summarised in Table 1. The most probable velocity of flight of C2(a3pu) particles ejected in the plume of tempera paints reaches similar values for all systems and ranges between (3.0 and 3.8)  103 m/s. For pellets of azurite and curcumin, the most probable velocity of flight of C2(a3pu) is outside this range with values of 1:7  103 m/s and 4:7  103 m/s, respectively. These differences might reveal how the production of C2 particles, through the interaction of the laser with the organic binding medium, is affected by the presence of the pigment in the paint. On the other hand, the most probable velocities of flight of the C2(a3pu) particles are, for all cases, smaller than the C2(d3pg) values determined through measurements of spontaneous emission.

4. Discussion OES and LIF studies of the 248 nm laser ablation plasma plume of model tempera samples have allowed

measurements of the velocities of ejected species. These velocities account for the character of the processes that take place in the irradiated material leading to ablation, as well as for the interaction processes that occur post ablation as the plume expands after the ejection events [21]. In UV ablation of organic substrates, it is generally accepted that the direct photodissociation of chemical bonds of the substrate compounds is coupled with thermal processes in which laser energy is converted into heat in the ablation process [1,22]. The velocities of flight of energetically excited fragments measured in this work using OES, at a fluence above the threshold for ablation of the substrate, correspond to high kinetic energy values. These velocities are about 20% lower for the organic species (C2 and CN) than for the metallic elements present in the pigment composition. For organic paints, as curcumin tempera, the laser will interact similarly with both the binding medium (unpigmented tempera) and the pigment, mainly through photochemical interactions. Indeed, previous mass spectrometric measurements [12,13] indicate that laser ablation originates a low degree of chemical change in the material of the unpigmented tempera paint, providing a proof for the dominance of photo-

134

M. Oujja et al. / Applied Surface Science 211 (2003) 128–135

chemical ablation mechanism by the KrF laser. Additionally, absorption of 248 nm photons by the curcumin molecule should be followed by excitation and energy relaxation mechanisms involving singlet and triplet molecular states [23]. The photochemical activity of the paint is most probably enhanced by the presence of curcumin, resulting in a more efficient coupling of the pigmented paint with laser photons. The differences in velocity of ejection of the C2(d3pg) and C2(a3pu) transients originated in the pellet of curcumin, as compared with those determined in the corresponding tempera paint, and the non-linear behaviour observed in the spontaneous emission from the pellet, may reflect the different interaction of the laser with the pigment in absence of binding medium. For inorganic paints, the interaction of laser photons with the pigments is likely to be qualitatively different from the mostly photochemical interaction with the organic binding medium in the paint. The compounds that constitute the pigments azurite, lead white and lead chromate are inorganic salts and ZnO is a semiconductor. Interaction of the laser with the inorganic salts will result in vaporisation and melting. The fact that the metallic elements taking part in the pigment composition are ejected with faster velocities than the organic diatomic counterparts (C2 or CN), indicate a more efficient coupling of the laser photons with the pigment itself. Interaction of inorganic pigments with 248 nm photons, and the subsequent discoloration effect, is thought to take place by reduction of the original salts and decomposition in constituent elements [10,12,13]. The low oxygen content of the atmosphere in which the ablation plume expands in the present experiments will favour these processes. On the other hand, direct absorption of 248 nm laser photons by ZnO crystals in the paint mixture or in the pellet is possible, as the room temperature band gap of this compound semiconductor (3.3 eV) is below the energy of a laser photon (4.99 eV). The velocity of flight determined in this work for ZnI and ZnO species ejected in the ablation plume are in agreement with values reported in the ArF excimer laser ablation of ZnO in vacuum [24]. This indicates that laser pigment interaction governs the ablation process of the tempera paint. Differences between the velocities of ejection of excited elements from the different pigments, Cu from azurite, Pb from lead white, Pb and Cr from lead chromate and Zn from ZnO are negligible. The excess

of laser energy seem to govern the ablative behaviour, the pigment properties being of minor importance at high fluences above the ablation threshold. On the other hand, the velocity distributions of C2(a3pu) particles in the ablation plumes of the studied systems, as determined through LIF measurements, approach Maxwell–Boltzmann distributions with average energies over 1 eV. As already noticed above, the most probable velocities of C2(a3pu) particles are smaller than the C2(d3pg) values determined by OES, most probably due to the fact that collisions between particles ejected in the primary ablation result in electronic excitation and consequent transfer of kinetic energy from the plume expansion. Kinetic energy values similar to the ones reported here for C2(a3pg) have been obtained in the UVablative decomposition of polymers [5]. The high translational temperatures indicate that direct bond excision reactions by the 248 nm photons are responsible for the dynamics of the small fragments which occur in the plume above the ablation site. At the laser intensities used in this work, highly efficient multiphoton mechanism should be considered in operation. The energy of a 248 nm photon (4.99 eV) is above the binding energy of C–C (3.6 eV), C–N (3.2 eV) and C–O (3.7 eV) [25]. This gives further evidence that photochemical breaking of these bonds is driving the interaction mechanisms between the 248 nm laser and the tempera samples [12,13,26]. The present results support this idea.

5. Conclusions In this work, we have determined the velocity of flight of C2 in the d3pg and a3pu electronic states and of excited atomic species produced in the 248 nm excimer laser ablation of model tempera paints in vacuum. The obtained high translational temperature values are not in the range of a thermal mechanism and are more compatible with a photochemical decomposition mechanism of the tempera paint. Differences between the velocities of organic and inorganic species and between emissions from the tempera systems and from the pigments as pellets have allowed to discuss the contribution of this mechanism in the laser irradiation of the paint systems. Experiments at laser fluences approaching the ablation threshold, at longer laser wavelengths and high

M. Oujja et al. / Applied Surface Science 211 (2003) 128–135

speed photography of the plume using a gated, intensified coupled charged device (ICCD) camera are in progress to get further insight into the dynamics of the ablation process.

Acknowledgements Thanks are given to CORESAL, Spain, for sample preparation, to Drs. M. Martı´n, M. Santos and L. Dı´azSol for useful discussions and to D. Silva for assistance with experiments. Support from Project BQU2000-1163-C02-01 is gratefully acknowledged. MO and ER thank Comunidad de Madrid and the Thematic Network on Cultural Heritage of CSIC, respectively for fellowships.

References [1] D. Ba¨ uerle, Laser Processing and Chemistry, Springer, Berlin, 1996, Appl. Surf. Sci. 186 (2002) 1. [2] P.R. Willmot, Rev. Mod. Phys. 72 (2000) 315. [3] R. Mitzner, A. Rosenfeld, R. Ko¨ nig, Appl. Surf. Sci. 69 (1993) 180. [4] F. Cleyssens, R.J. Lade, K.N. Rosser, M.N.R. Ashfold, J. Appl. Phys. 89 (2001) 697. [5] R. Srinivasan, B. Braren, Chem. Rev. 89 (1989) 1303. [6] V. Zafiropulos, Laser ablation in cleaning of artworks, in: B. Luk‘yanchuk (Ed.), Laser Cleaning, World Scientific, Singapore, 2002. [7] S. Georgiu, V. Zafiropulos, D. Anglos, C. Balas, V. Tornari, C. Fotakis, Appl. Surf. Sci. 738 (1998) 127–129. [8] S. Klein, T. Stratoudaki, Y. Marakis, V. Zafiropulos, K. Dickmann, Appl. Surf. Sci. 157 (2000) 1.

135

[9] M. Castillejo, M. Martı´n, M. Oujja, E. Rebollar, C. Domingo, J.V. Garcı´a-Ramos, S. Sa´ nchez-Corte´ s, J. Cult. Herit. (2003), in press. [10] P. Pouli, D.C. Emmony, C.E. Madden, I. Sutherland, Appl. Surf. Sci. 173 (2001) 252. [11] V. Zafiropulos, T. Stratoudaki, A. Manousaki, K. Melesanaki, G. Orial, Surf. Eng. 17 (2001) 249. [12] M. Castillejo, M. Martı´n, M. Oujja, D. Silva, R. Torres, V. Zafiropulos, O.F. van den Brink, R.M.A. Heeren, R. Teule, A. Silva, Anal. Chem. 74 (2002) 4662. [13] M. Castillejo, M. Martı´n, M. Oujja, J. Santamarı´a, D. Silva, R. Torres, A. Manousaki, V. Zafiropulos, O.F. van den Brink, R.M.A. Heeren, R. Teule, A. Silva, J. Cult. Herit. 4 (2003) 257s. [14] M.I. Cooper, P.S. Fowles, C.C. Tang, Appl. Surf. Sci. 201 (2002) 75. [15] M. Hauer, T. Dickinson, S. Langford, T. Lippert, A. Wokaum, Appl. Surf. Sci. 197–198 (2002) 791. [16] R. Srinivasan, Appl. Phys. A 56 (1993) 417. [17] O.F van den Brink, Molecular Changes in Egg Tempera Paint Dosimeters as Tools to Monitor the Museum Environment, Ph.D. Thesis, University of Amsterdam, 2001. [18] B. Angleraud, J. Aubreton, A. Catherinot, Eur. Phys. J. AP 5 (1999) 303. [19] P. Serra, J.M. Ferna´ ndez-Prados, J. Navarro, J.L. Morenza, Appl. Phys. A 69 (1999) S183. [20] R. Srinivasan, J. Appl. Phys. 61 (1987) 376. [21] S. Georgiou, C. Kosmidis, A. Koubenakis, W.J. Jia, K.W.D. Ledingham, Phys. Chem. Chem. Phys. 1 (1999) 5339. [22] Y.G. Yinglin, L.V. Zhigilei, B.J. Garrison, J. Photochem. Photobiol. A: Chem. 145 (2001) 173. [23] S.M. Khopde, I. Priyadarsini, D.K. Palit, T. Mukherjee, Photochem. Photobiol. 72 (2000) 625. [24] Y. Kawaguchi, A. Narazaki, T. Sato, H. Niino, A. Yabe, Appl. Surf. Sci. 197/198 (2002) 268. [25] J. Wei, N. Hoogen, T. Lippert, O. Nuyken, A. Wokaum, J. Phys. Chem. B 105 (2001) 1267. [26] T. Lippert, J.T. Dickinson, S.C. Langford, H. Furutami, H. Fukumura, H. Masuhara, T. Kunz, A. Wokaum, Appl. Surf. Sci. 127/129 (1998) 117.