Systematic testing, verification and validation of laser treatments for unglazed earthenware affected by lichens and fungi biodeterioration

In recent years, there has been a growing interest in exploring environmentally friendly and healthy alternatives to conventional solvent cleaning and biocides in the conservation of stone artworks. Here, we focus on the potential of laser-based photonic methods for treating biodeteriorated earthenware artefacts. The investigation was conducted on Roman dolia (jars) of the International Museum of Ceramics, Faenza, Italy. Three removal methods were tested and compared: (i) brushing using a soft-bristled electric brush and water, referred to as brush cleaning; (ii) a combination of brushing and laser ablation; and (iii) biocide and brushing. Four laser systems with different wavelengths and optimized pulse durations in nanosecond or microsecond regimes were used in the tests. Systematic irradiation tests were conducted to determine the damage thresholds and define safe laser irradiation levels. The characterizations of the surfaces under treatment were carried out pre- and post-laser irradiation using optical microscopy, 3D photogrammetry, and Pulse-Amplitude-Modulated Chlorophyll-Fluorometry. Furthermore, spectroscopic methods based on FTIR, Raman, and LIBS techniques were used to assess the effectiveness of the removal process and the composition of uncovered surfaces. Results have indicated that gentle brushing and water is the most effective approach for safely removing around 60% of the bio-colonization weakly anchored to the substrate over the area under treatment. This comprised viable species, whereas the remaining 40% of the area included endolithic species, mostly thalli of Verrucaria nigrescens and rock-dwelling fungi. The eradication of the latter was the real conservation concern requiring attention. Following the experimentation, the optimal method for safely uncovering the earthenware surface was a combination of water-assisted brushing and 1064 nm laser irradiation as a finishing treatment.


Introduction
A relevant aspect in the conservation of cultural heritage concerns biodeterioration and treatments to eliminate the biological growths damaging stone heritage exposed to outdoor conditions [1][2][3].Like stones, the identified biodiversity in earthenware materials is extensive, ranging from microorganisms to plants.It encompasses bacteria, cyanobacteria, algae, lichens, mosses and fungi, having their peculiar biochemical and physical properties [4,5].Moreover, the composition and properties, such as surface texture, granulometry, porosity, and water sorption, are crucial factors in determining the bioreceptivity of ceramics.Particular attention has been paid in the literature to construction and decorative materials, such as bricks and roofing tiles, which have exhibited the greatest biodeterioration diversity [1,5,6].It has been found that the type of colonization in earthenware objects depends more on the chemical composition rather than the state of sintering of the ceramic substrate [7].For example, calcium-rich unglazed earthenware roof tiles have shown higher bioweathering than alumina-rich ones.In the biodeterioration process, lichens and rock-dwelling fungi present a significant issue, as hyphal penetration can develop up to a depth of a millimetre and more within the substrate, causing aesthetic, physical, and chemical damage to the ceramic material.This can be a major concern for the preservation of pottery and other ceramic artefacts.Various techniques have been developed and experimented to mitigate such bio-induced damages and maintain the structural and aesthetic integrity of artefacts for as long as possible [8,9].The techniques available include chemical methods, namely traditional biocides, solvent gels and nanoparticles [9,10]; physical methods, such as mechanical removal, UV irradiation [11], LED illumination systems with various emissions [12], gamma radiation, laser cleaning, heat shocking, microwaves [13,14], and dry ice treatment; and biological methods, such as natural molecules (e.g.essential oils) with biocidal activity [15], enzymes, and microorganisms.With such a wide choice of solutions, traditional biocidal and chemical products, paired with mechanical cleaning, are still the most effective in the long run.Phenomena of recolonization by green algae and cyanobacteria were observed in outdoor exposed stonework after a few years from treatment (around two years), whereas lichens require almost 4-5 years [16][17][18].Biocide products can also interact with the substrates and cause, following repeated treatments over time, the growth of microorganisms resistant to the biocides themselves [19].Some biocides can have adverse effects that cause physical damage, such as changes in colour, structure and permeability, or even chemical damage, such as mineral solubilization and pH alteration [20].The use of these products also raises concerns about their toxicity to the environment and public health.
Among promising options, laser methods that employ either ablative or non-ablative mechanisms are undeniably advantageous as they do not involve adding chemicals to the substrate, which is known to impact significantly weak and fragile deteriorated surfaces.Laser removal is a widely used method for removing various build-ups from the surfaces of artistic and historical stone monuments [21][22][23][24].Laser methods are highly controllable, selective, contactless, and environmentally friendly.First laser treatments on earthenware objects have been applied to terracotta sculptures and architectural decorations for removing dark soil deposits and encrustations [22,[25][26][27].The IR laser radiation of Q-Switched Nd:YAG laser emitting at 1064 nm proved more effective and selective than the harmonic wavelengths of 532 or 266 nm.More recently, the Nd:YAG laser radiation at 1064 nm was used to remove dirt deposits on a very compromised earthenware statue not allowing conventional mechanical and solvent cleaning that might have caused detachments and losses of material [28].Laser irradiation allowed dirt removal, avoiding the pre-consolidation of the degraded and deteriorated surface.Polić et al have studied laser-material interaction mechanisms on Iranian medieval ceramic samples using TEA CO 2 (10.6 µm) and Nd:YAG (1064 nm) lasers with fluences above the damage thresholds.The results have demonstrated that laser-induced surface modifications depend on the laser applied and whether the surface is glazed or not.The CO 2 laser has proven suitable for repairing structural failures, such as cracks, by welding, whereas the Nd:YAG laser revealed effective in removing contaminants such as oxides, oils, and encrustations [29].
In contrast, eradicating lichens and fungi from earthenware objects represents a more heterogeneous and complex scenario regarding laser-material interactions compared to soil and dirt.The use of different lasers and their combination with biocide, microwaves, solvents and mechanical methods for eliminating biological growths has been extensively investigated on various lithotypes, such as carbonate rocks [23,[30][31][32], granite [33,34], sandstone [35], and dolostone [36].Comparatively, less attention has been paid to the biodeterioration of earthenware artefacts.This can be attributed to the lower frequency of ceramic works exposed to outdoor conditions compared to stone.However, the usefulness of combined treatments using dual IR-UV sequential irradiation at 1064 and 266 nm and biocides was successfully demonstrated to remove lichens thalli from roofing tiles [7].
In this respect, the present investigation aimed to explore the potential of laser treatments for earthenware artefacts affected by biodeterioration.The study was carried out on a Roman earthenware dolia (large jars) from the garden of the International Museum of Ceramics (MIC) in Faenza (IT).The presence of significant biological growths on the dolia represented an opportunity to explore such conservation issues from a broader perspective.
The experimentation concerned the application and comparison of diversified removal methods which included (i) mechanical cleaning (i.e.brush cleaning) (ii) combination of lasers and brush cleaning (iii) biocide followed by brush cleaning.Laser that differed in wavelength and pulse duration were tested.The characterization of the dolium surface after the biological removal was carried out through optical microscopy, 3D photogrammetry, and Pulse-Amplitude-Modulated Chlorophyll Fluorometry (CF-PAM).Moreover, FTIR, Raman, and LIBS spectroscopic techniques were employed to evaluate the removal effectiveness and to characterize the effects of alteration that should be prevented.

The Roman dolium of the MIC
The MIC houses five Roman ceramic dolia in its inner garden.The dolium (h 57 × Ø 61 cm) selected for this study (inventory n. 19076), as well as the other four, had its surface extensively colonized by biological growths mainly including a diffuse greenish and black-greyish layer of crustose lichens and biofilms of varying thickness and areal coverage.
A preliminary diagnostic investigation was conducted to characterize the biodeterioration patterns.Then, the minero-petrographic characterization and identification of the biological species growing on the surface of the dolia stored at MIC was deeply carried out, and the results were recently published [37,38].It has been reported that the composition of the five dolia was characterized by similar percentages of quartz, anorthite, pyroxene, gehlenite, illite/mica and traces of hematite.The surface in the upper half of the dolium body and on the rim close to the opening was extensively covered mainly by a black-grey layer of Verrucaria nigrescens and Acarospora gallica.On top of this layer, large thalli of Circinaria hoffmanniana were present.Differently, in the lower half, the surface was colonized by a thin green biofilm layer mostly composed of algae.Some of these species are shown in figure 1.As emerged from the previous study by Pinna et al, among all the species grown on the dolium surface, Verrucaria Nigrescens was the most well-anchored to the substrate and, therefore, the most difficult to remove with traditional treatment.

Morphological evaluations
The three-dimensional model of the entire object was reconstructed using the Structure from Motion (SfM) technique through Close Range Photogrammetry [39].According to the well-established image processing, approximately 165 photos were taken using a Nikon D3300 digital reflex camera with a 45 mm lens for the dolia model reconstruction.
In figure 2, the horizontal section and vertical section with sizes data are displayed, which evidences that it was crafted on the turning wheel.
Ultra-Close Range Photogrammetry was employed to evaluate the morphological characteristics of the surface after removal treatments, using different laser sources and operative modes [40,41].The technique is based on the same principles as classical photogrammetry but is applied on a different scale.The system comprises a Canon 7D digital camera with a 60 mm macro lens mounted on a 260 mm motorized dolly.
After selecting the shot parameters, a set of adjacent images is collected (distance from the surface: 320 mm; step: 15.5 mm).For each treatment area, the corresponding textured 3D model was reconstructed.The tool's software allows exporting the model in xyz format to evaluate the surface characteristics such as geometry, morphology, and roughness.The latter was determined using a suitable MATLAB algorithm.
The roughness of the surface was evaluated using the statistical parameters R a , R z and R Max according to the standard ISO 4287:1997 (ISO 4287:1997-Geometrical Product Specifications (GPS): • R a is the arithmetical mean deviation of the surface pattern (arithmetic average roughness): where z i is the measured quote • R z is the average among peaks (R p ) to valley heights (R v ) of the surface, determined on the samples outside of the 2σ interval (σ, standard deviation): • R Max is the difference between the maximum peak to valley height.

Laser testing and procedure for the biocolonization treatment
Laser systems with different wavelengths and pulse durations were used in the experimentation.These are listed in table 1.
Two LQS laser regimes were employed in single-peak pulse mode (LQS 1 ) and triple-peak pulse mode (LQS 3 ).Notably, the latter burst mode emits three power peaks with the same energy (150 mJ per pulse), duration of 120 ns each and temporally spaced of 40 µs.The QS Nd:YAG (532 nm) laser was used with an optimized homemade fibre-coupled setup, enabling a maximum energy per pulse of 50 mJ.The FR Er:YAG (2940 nm) laser system was coupled to an articulated arm.
The exact measurement of the laser-spot size was obtained under the optical microscope by measuring the imprinted mark on photosensitive ZAP-IT ® paper, which is suited for recording pulsed laser source beam characteristics in the UV-IR wavelength range.
The determination of safe laser irradiation parameters for biodeteriogen removal was carried out using a well-established and standardized laser testing methodology, which has been thoroughly described elsewhere [42].The methodology involves accurately measuring the laser damage threshold, F th (in J cm −2 ), i.e. the minimum fluence causing detectable damage or modification under the microscope.The laser damage threshold, F th , was achieved in dry conditions for both a single laser pulse Fth (1) and a series of five pulses Fth (5) by gradually increasing the fluence (at 1 Hz pulse repetition frequency, prf) until a surface alteration was detected.The area close to the internal edge of the dolium's rim (below the cover) was selected to determine Fth, as deemed the most well-preserved and least affected by biological damage.An Olympus OCM SZX7 stereo microscope equipped with a magnification range of 8×-56× and a field of view of 27.5-3.9mm was utilized to monitor laser-induced changes.Image processing was performed using Fiji software.
Once determined fluence thresholds Fth for each laser system, colonized areas of approximately 2 × 2 cm 2 were selected for bioremoval treatments.The procedure consisted of two steps: (1) removal of superficial and 'less-anchored' biomasses by water-assisted mechanical cleaning (MT) using a soft-bristled brush; (2) alternating laser irradiation scans below Fth (1) and MT to remove the remaining 'more-anchored' species.An electric brush (600 rpm) with soft bristles was used for MT.
For comparison with the photonic treatments, the action of a soft brush (mechanical cleaning) and the biocide Biotin T (3%) applied with cellulose pulp for 48 h were also independently evaluated.The biocide treatment was included in the experimentation to improve the speed of bideteriogen removal, concerning both the initial step using MT as well as the removal of the residual biological growth using the laser.

Analytical assessments
The efficacy of cleaning treatments was assessed by evaluating (1) their ability to eradicate remnants of thalli and other biological debris from the ceramic surface and (2) their effects on the surface texture and composition.Various techniques were utilised to examine the treated samples.The characterization of the biospecies colonizing the earthenware surface, as well as information on their phototrophic activity, were achieved through stereomicroscope examinations and CF-PAM measurements, respectively.CF-PAM imaging was performed using a Handy FluorCam FC 1000-H, a portable device by Photon Systems Instruments provided with four light panels containing 25 × 4 pulsed LEDs (k max = 620 nm), a longpass optical filter, and a 512 × 512 pixels CCD sensor.CF-PAM is a well-recognized method to gain information on the spatial distribution of photosynthetic activity and to study the stress effect on phototrophic organisms [31].One of the parameters used is the maximum quantum yield of the photosystem PSII, defined as QY max = (F max -F 0 )/F max , where F 0 and F max represent the minimum and maximum fluorescence of previously dark-adapted samples, respectively.For most plants, the optimal value of QY max on a scale of 0-1 corresponds to approximately 0.8, whereas for lichens and microalgae the values reported in the literature are significantly lower (0.2-0.6).
Reflectance FTIR spectra were collected with a portable Bruker Optics ALPHA FT-IR Spectrometer equipped with a SiC Globar source and a DTGS detector.All spectra were acquired in total reflection mode, collecting 128 scans, with a resolution of 4 cm −1 in the 7000-375 cm −1 range and a measuring spot of 6 mm in diameter.The spectra were processed using OPUS 7.2 software.
Raman measurements were carried out using the portable system i-Raman (B&W Tek, CA, USA), which is equipped with a fibre-coupled diode laser @785 nm and a spectrometer which covers the range 175-3200 cm −1 with a spectral resolution of 8 cm −1 .
LIBS measurements were performed using a homemade portable LIBS instrument previously described and employed in several archaeometric studies (see, for example [43],).In this study, it was used to characterize the near-surface compositional alterations possibly due to weathering, biofilm activity or a burial phase.

Laser treatments and morphological evaluations
In table 2, damage threshold fluences Fth for all laser types are presented.
As shown, the F th values of different laser types may vary significantly depending on the laser wavelength and pulse duration.The SFR laser exhibited the highest single-pulse Fth (1) value, whereas the 532 nm QS laser had the lowest.An incubation effect was observed with threshold fluence decreasing with pulse number, i.e.Fth (5) ≪ Fth (1).This phenomenon is caused by local discoloration of relatively absorbing centers.Successive laser pulses further increase the optical absorption of the latter, thus decreasing Fth.Instead, the decrease of the Fth (5)/Fth (1) ratio (see table 2) with pulse duration indicates a thermal ablation mechanism with microseconds pulses.
Figure 3, lines I-VII, shows the areas of the dolium selected for performing and comparing differentiated removal treatments.The areas were situated in the upper half of the dolium above the major curvature, where most of the biological growth was present.
CF-PAM measurements conducted before the removal treatments underlined the presence of photosynthetic active species.The range of calculated QY max images in figure 3(B I-VII) was between 0.67 and 0.71.
As described in section 2.2, the procedure for biocolonization removal consisted of two phases.Preliminary MT tests revealed that the surface texture of the original dolium substrate had been affected by the colonising biomasses, resulting in decreased cohesion and hardness.The uncovered surface showed well-anchored residual colonies, as black roundish blobs, consisting of thalli from Verrucaria nigrescens that had deeply penetrated the ceramic substrate, along with black fungi inhabiting rocky environments.CF-PAM detected no chlorophyll fluorescence emission upon MT.Some areas featured the phenomenon of biopitting, which is widely found in stone artefacts [44].
As depicted in figure 3(C I), the MT resulted in an incomplete yet considerable removal, about 58%-60% of the area covered by biomasses, including biofilms and lichens, was easily uncovered without applying any pressure.The treated surface displayed a spotty, uneven distribution of lichens and fungi ranging between 250 and 500 µm in average diameter.The estimated removal rate required to reach this level was approximately 1 cm 2 min −1 .
Figure 3 in areas C(II-VI) shows the results of the two-stepped cleaning procedure.All the areas received an initial water-assisted MT to remove biomasses grown on top, followed by the laser treatment to facilitate the eradication of fungal bodies anchored to the ceramic substrate.
As can be seen, all the laser techniques tested achieved satisfactory material removal while maintaining the integrity of the surface topography.No CF-PAM signals could be detected after the removal of the residual species.Stereomicroscopy (figure 3(E)) inspections confirmed that the MT-laser procedure led to an increased removal degree compared to the MT alone.Most importantly, protruding fungal bodies could be almost completely eradicated without significantly altering the topography of the uncovered surface.In other cases where hyphal rooting was more difficult to remove, the laser treatment facilitated the separation of the fungal body from the substrate-embedded hyphae.The remnants were identifiable as dark spots filling micro-pits (60-80 µm diameter).Regarding laser testing, area C(II) was achieved using the QS laser at 532 nm with alternating passes of water-assisted MT.Initial tests with fluences equal to or lower than F th (1), i.e. between 0.4 and 0.65 J cm −2 (2 Hz prf), were found poorly effective and time-consuming.The removal was effective just above F th (1), at F of 0.8 J cm −2 and 2 Hz prf.It is worth noting that, under the cleaning condition F ⩾ F th (1), it is essential to consistently spray an appropriate amount of water to avoid the risk of whitening and mechanical harm to the ceramic substrate.
Area C(III) was treated with the LQS 1 laser at a fluence of 1.8 J cm −2 .The prf was increased from 2 to10 Hz to thoroughly clean the tested area while using water and brushing.Compared to the other tests, the surface could appear slightly more abraded under raking light (figure (3DIII)), likely due to the use of a stiff-bristled brush.LQS nanosecond pulses did not effectively remove the residual fungal thalli, which remained anchored to the ceramic substrate, although they underwent laser-induced structural Table 2. Lasers, corresponding pulse durations and spot diameters used in the present tests, along with corresponding damage threshold fluences (expressed in J cm −2 ) for single F th (1) and five F th (5) repetitive pulses at the pulse repetition frequency (prf) of 1 Hz.Uncertainty in the size of the spot is within 2.5%.

Laser type
Pulse duration Spot diam.(mm) F th (1) (J cm −2 ) F th (5) (J cm −2 ) F th (5)/F th ( modification.This weakening effect, combined with a few MT passes, favored the removal of the remaining biomasses, reducing the loss of ceramic fragments from the biodeteriorated surface.Area C(IV) of figure 3 shows the result of the LQS 3 treatment.The burst mode of LQS 3 led to slightly attenuated mechanical effects while maintaining good removal efficiency.Using a fluence of about 3.5 J cm −2 , 10 Hz prf the complete removal of residual biospecies was achieved.
Despite lower ablation efficiency as compared to short pulses, long-pulse laser treatments such as SFR and Er:YAG systems have produced satisfactory results.Area C(V) and C(VI) show the results of SFR (F = 4.2 J cm −2 , 10 Hz prf) and Er:YAG (F = 4 J cm −2 , 10 Hz prf) laser treatments, respectively.CF-PAM measurements were carried out in all six areas after MT and laser cleaning, revealing any phototrophic activity after the treatments.
Table 3 presents the results of surface roughness analysis using ultra close-range photogrammetry.The statistical parameters R a , R z and R Max indicate that surface roughness increased for shorter laser pulse durations, while it improved for long-pulsed lasers (i.e.area V is comparable to VI, followed by areas IV, III, and II).The results can be explained by considering that treatments with similar R a can be further classified by the average among maximum peaks (R p ) to valley heights (R v ) of the surface, i.e. by R z .For instance, area-treated III and VI exhibit similar R a values, but the R z and R max values of area-treated III are twice that of area-treated VI.Thus, the greater the differences between peaks and valleys, the more the material is excavated by the action of the laser.
Regarding using Biotin T (3%) for biocide treatment, the results are displayed on line VII of figure 3.After 48 h from the biocide application, the color of the treated area shifted from green to black, and no photosynthetic activity was observed by CF-PAM.Mechanical removal of remaining endolithic species in a portion of the treated area C(VII) did not improve the removal areal ratio, which was again around 60%. Due to their thick, melanized cell walls, fungi also resist chemical attacks by biocides or other anti-microbial treatments [14].Applying the biocide before the laser treatment did not enhance the effectiveness of eliminating the remaining species.
Finally, the lower half of the dolium, facing downwards towards the ground and predominantly colonized by green algal biofilms did not show any issues in terms of cleaning.The algal film could be removed easily using a brush and water.

Reflectance FTIR spectroscopy
Biofilm removal with the different cleaning methods was investigated through its main bands, typical of proteins and esters (aliphatic C-H stretching at 2931 cm −1 , C=O stretching at 1710 cm −1 and amide II at 1558 cm −1 ).The band at 1668 cm −1 , which is also well visible in uncolonized earthenware, was attributed to the overlap of amide I vibration in proteins and to the -OH bending in Si-OH groups [31,45,46]; therefore, it was not considered for monitoring biofilm removal.
The depletion of the above-mentioned diagnostic bands after the laser treatment was quite evident, with SFR (figure 4, spectrum V) showing its better performance compared to the other laser modes.Significant residues were detected after mechanical cleaning (spectrum I).Instead, MT cleaning significantly improved the removal of biofilm after biocide application (spectrum VII), compared to the use of biocide only (spectrum not shown).
Similar observations were made in the NIR spectral range, where CH combination bands at 4335 and 4263 cm −1 of the biological component were no longer detectable after the removal treatments applied [47].The bands at 5085 and 5235 cm −1 due to OH groups in clay minerals, as well as at 4525 cm −1 (ν OH + δ Al-OH ) were well visible and not affected by the laser irradiation treatments.

Raman spectroscopy
Figure 5(B) shows Raman measurements carried out in the six areas treated by different lasers to assess the effectiveness of the laser cleaning qualitatively.For the sake of comparison, Raman spectra were also acquired on the edge of the dolium in points without biological growth and taken as a reference (figure 5(A)).Mineral phases identified in the earthenware are indicated with asterisks and consisted mainly of quartz (462 cm −1 ), hematite (220, 290 cm −1 ), amorphous phases of silica (bands at 345 and 1060 cm −1 ).The broad and intense band from 1000-2000 cm −1 could be ascribable to the luminescence emitted by the non-crystalline phase of silicon oxide [48,49].Slight differences among the reference spectra in figure 5(A) are due to the inhomogeneity of the crystallinity of the mineral phases constituting the earthenware.For clarity, an offset has been applied to each spectrum.
The spectra acquired where MT was performed did not show any band (neither Raman nor luminescence) because of the emission of an intense fluorescence background indicating the presence of biodeteriogens on the surface (figure 5(B), area I).The red spectrum was acquired in a spot of the MT-area that was particularly abraded: in this case, the luminescence emission band of amorphous silica begins to be observed in the spectrum.The Raman spectra acquired in the MT + SFR area (figure 5(B), area V) showed a very dynamic situation where the main spectral features due to earthenware were observed.Raman spectra obtained in this area are similar to those acquired in the reference areas.The area treated with QS (532) (area II), LQS (areas III-IV), and Er:YAG (area VI) lasers showed spectral features that were not as clear as in the SFR area.Finally, in the LQS III area (area IV), a flattening of the spectral bands of the mineral phases was noted.

LIBS
LIBS depth profiles were acquired in each laser-treated area (I, II, III, IV, V and VI) as well as in near uncleaned areas (0) for comparison.Several hundred LIBS spectra were acquired during each depth profile analysis of the test areas.Each LIBS spectrum has been normalized to the total intensity to reduce the effect of shot-to-shot variations and differences in laser-target energy coupling with increasing shot numbers, i.e. the crater depth.LIBS atomic lines of the main constituents of the earthenware and of biological growth were selected in the spectra.In particular, emission lines of C(I) (247.9 nm), Si(I) (250.7 nm, 288.2 nm), For statistical reasons, four Raman spectra were recorded in each area.Mineral phases identified in the earthenware (i.e.reference spectrum) are indicated with asterisks and consist mainly of quartz (462 cm −1 ), hematite (220, 290 cm −1 ), amorphous phases of silica (bands at 345 and 1060 cm −1 ).The broad and intense band from 1000-2000 cm −1 could be ascribable to the luminescence emitted by the non-crystalline phase of silicon oxide.Fe(I) (438.3 nm,), Al(I) (308.2 nm, 394.4 nm), Ca(II) (315.9 nm), Ca(I) (422.7 nm, 443.5 nm), Mg(I) (285.3 nm, 518.4 nm), Na(I) (589.2 nm), K(I) (404.4 nm), Ti(II) (334.9 nm) and CN molecular violet bands (386-389 nm) were considered for the analysis.The intensity of the carbon line was used as a marker for evaluating the efficiency of treatments because it is directly related to biological growth.The relative intensity of Ca(II) (315.9 nm) and Si(I) (250.8 nm) lines was used as a markers for assessing calcium carbonate precipitation and infiltration [50,51].Figure 6 shows the comparison of the spectra of the encrusted surface (0) and of the innermost part of the earthenware where the carbon signal has disappeared and the signals from other elements of the encrustation or earthenware alteration (Ca, Fe, Si, Ti, Mg) reduced to the bulk values.
Figure 7 shows the peak intensity changes of the C element in the biodeteriorated area (0) and in the treated area (I, II, III, IV, V, VI) as a function of laser shot number, i.e. crater depth.As expected, the cleaned areas show a quick decrease in the C signal, reflecting the effectiveness of the cleaning processes with residual organic traces up to about ten pulses corresponding roughly to 30 µm in-depth (figure 7, inset graph).
Further analysis of the acquired depth profiles, performed in different areas of the dolium, revealed aspects related to the possible burial of the object originally.Figure 8 dispalys the grouping of the depth  distributions of the calcium to silicon ratio, which showed two distinct behaviors mainly related to the spatial location of the measurement site along the vertical direction.The distinct behavior of the two sets of measurements suggests the possibility of a long-term burial condition with a burial line above the main curvature of the dolium.

Considerations on the laser treatment efficiency
Laboratory tests carried out at 532, 1064 and 2940 nm and related analytical assessments showed that laser irradiation in combination with gentle brushing was an effective way to remove remaining lichens and fungi dwelling on the substrate.However, the different lasers were not found equally efficient.The ns pulsed laser QS Nd:YAG at 532 nm was the most effective, thanks to the high ablation efficiency and the high absorption of lichen and fungi at this wavelength [31].Unfortunately, the ceramic substrate also exhibits high absorption at 532 nm, making it impractical to remove biological species without causing mechanical damage to the substrate, primarily selective ablation of highly absorbing centers (i.e.hematite).532 nm QS laser is only effective when the cleaning condition F ⩾ F th (1) is met.In contrast, at 1064 nm, the removal efficiency is significantly reduced due to the lower absorption compared to 532 nm.Diffuse reflectance Vis-NIR spectrophotometry measurements were conducted to assess the wavelength dependence at 532 and 1064 nm between the substrate and the dark biodeteriogens, mostly Verrucaria and black fungi.The recorded spectra of figure 9 showed that the optical reflection for the biological contaminants was around 15%-20% at 532 nm and 40%-55% at 1064 nm, whereas for the dolium substrate it was 23%-26% at 532 nm and 75%-85% at 1064 nm.
Therefore, the optical contrast in the NIR region and the longer pulse durations enable higher selectivity and a wider safe fluence range for cleaning treatments, i.e.F < F th (1).The LQS 1 and LQS 3 performed quite satisfactorily, with faster removal rates if compared to long-pulsed SFR Nd:YAG (1064 nm).The latter showed a very wide operational fluence range and, although less efficient than LQS was found more efficient than the FR Er:YAG (2940 nm) laser.For a spot size of 2 mm, SFR achieved a removal rate of 0.25 cm 2 min −1 , which is approximately twice as fast as the FR Er:YAG removal rate of 0.1-0.12cm 2 min −1 to achieve an acceptable surface cleaning.These removal rates are to be considered relative to this experimentation, as in real applications, the spot size can be increased up to 1 cm, significantly reducing the time required for the procedure.In addition to shorter application times, the 1064 nm wavelength, in both SFR and LQS modalities, was retained a good candidate due to its wider operating fluence margin compared to the FR Er:YAG.Moreover, longer pulse durations enabled the elimination of photomechanical effects that are typically caused by short ns pulse durations, thereby reducing the increase in surface roughness due to the loss of bioweathered fragments.

Validation and conservation treatment
The results of laboratory testing, morphological evaluations, and the multy-analytical assessment constituted the basis for a decision on the restoration procedure of ceramic dolium (inv.N. 19076).The procedure comprised two phases.First, the thickness of the biological growth layers was reduced by water-assisted mechanical cleaning using a soft-bristled brush and water.The result of this treatment applied to the whole biodeteriorated dolium surface is depicted in figure 10(B).
This was intended as an initial pre-treatment to decrease the thickness of the biological growth while limiting the loss of ceramic material already compromised by endolithic microorganisms.Where the Verrucaria and fungi layer was excessively thick (over 1 mm), mechanical scalpel cleaning was used before the laser finishing treatment.The second step consisted in the laser treatment of the surfaces using a 1064 nm dual pulse duration approach, alternating SFR and LQS pulses (figure 10(B)).The operating parameters were those previously found in the experimental phase and were: F = 4.2 J cm −2 , 5 mm spot size and 10 Hz prf for SFR and F = 1-1.7 J cm −2 , 5 mm spot size and 10 Hz prf for LQS.These parameters were associated with constant water assist to prevent local overheating caused by the absorption of iron oxide load [52].This represented the best compromise between application times and safety of irradiation parameters.Specifically, the long-pulsed laser treatment with SFR was used to induce either thermal decomposition or mechanical damage to the remaining endolithic biostructures.This is important for conservation purposes to prevent or delay biodeterioration in long term [36].Furthermore, maintain the surface roughness of the bioweathered surface is advisable; therefore, complete eradication of any remaining endolithic biostructures was not achieved.The use of LQS laser treatment at moderate fluences allowed for the removal of weakened by-products and debris that had already been treated with SFR.The dual pulse duration approach, using SFR + LQS, reduced the number of brush passes required and, therefore, minimized the loss of ceramic material.The outcome of this procedure is illustrated in figure 10(C).The same procedure was also used for the recovery of two other dolia, inv.N. 19075 and inv.N. 19077, from the courtyard of the MIC.

Conclusions
This study has shown that extremely precise and controlled cleaning of unglazed terracotta surfaces can be achieved using different laser wavelengths and pulse duration.Chlorophyll fluorescence imaging performed after laser treatments confirms the efficiency of the removal of phototroph organisms.The decrease of diagnostic FTIR and Raman bands of biofilm after the laser cleaning was quite evident, showing superior performance as compared to the biocide and mechanical removal.LIBS measurements in laser-treated areas showed no presence of carbon at the surface and through the depth up to 30 µm.Moreover, the ratio Ca/Si highlighted compositional differences among the areas treated due to burial conditions.
Following the laboratory experimentation, a two-step procedure combining mechanical pre-treatment followed by a 1064 nm laser procedure with variable pulse duration ensured an increased removal degree compared to the traditional mechanical cleaning alone, as well as the safeguarding of surface structural properties.Best compromise between application times and safety irradiation parameters was achieved alternating SFR and LQS laser pulsed durations.Finally, this work reports the development and the first large-scale validation study on laser treatment of biodeteriorared Roman unglazed earthenware dolia.The outcomes indicate that an integrated approach using photonic technologies offers a viable and environmentally friendly alternative to traditional methods based on chemical and mechanical processes.Further research will be conducted to assess the long-term efficacy of the treatment.At present, no decolonization occurred one year after the treatment.

Figure 2 .
Figure 2. Horizontal (left) and vertical (right) cross-sectional views of the dolium (inventory n. 19 076).Violet lines indicate internal diameter and wall thickness.Distance is expressed in meters.

Figure 5 .
Figure 5. Raman spectra acquired in areas of the dolium without biological growth (A), taken as reference, and in areas treated with MT and MT + lasers (B): MT (I), MT + QS @532 (II), MT + LQS1 (III), MT + LQS3 (IV); MT + SFR (V), MT + Er:YAG (VI).For statistical reasons, four Raman spectra were recorded in each area.Mineral phases identified in the earthenware (i.e.reference spectrum) are indicated with asterisks and consist mainly of quartz (462 cm −1 ), hematite (220, 290 cm −1 ), amorphous phases of silica (bands at 345 and 1060 cm −1 ).The broad and intense band from 1000-2000 cm −1 could be ascribable to the luminescence emitted by the non-crystalline phase of silicon oxide.

Figure 6 .
Figure 6.Portions of LIBS spectra with identified elements.Spectra are normalized to the total integral.

Figure 7 .
Figure 7. Changes in the intensity of the integral of C(I) (247.9 nm) peak as a function of laser shot number/crater depth: encrusted areas with biological growth (0), MT (I), MT + QS @532 (II), MT + LQS1 (III), MT + LQS3 (IV); MT + SFR (V), MT + Er:YAG (VI).The inset plot shows the decrease of the C(I) line in the cleaned areas.The estimated uncertainty on each plotted point is below 2%.

Figure 8 .
Figure 8. Mean (solid line) and standard deviation (shaded area) of the Ca/Si ratio versus the number of laser pulses/crater depth in areas of the dolium above (black line) and below (red line) the hypothetical burial region (i.e.above the major curvature of the dolium).Each Ca/Si depth profile is an average of six measurements.The ratio of calcium to silicon was calculated by analyzing the intensities of the atomic LIBS emission lines of Ca(II) at 315.9 nm and Si(I) at 250.7 nm.

Figure 9 .
Figure 9. Mean (solid line) and standard deviation (shaded area) of diffuse reflectance Vis-NIR spectra showing the wavelength dependence at 532 and 1064 nm between the dolium substrate and the biological contaminants.

Figure 10 .
Figure 10.Photographs of the Dolium (inv.N. 19 076) taken in the east (E), north (N), west (W) and south (S) exposure directions before (A), after mechanical cleaning using a soft-bristled brush and water (B), after the dual laser irradiation treatment using SFR and LQS Nd:YAG (1064 nm) temporal regimes and MT (C).

Table 1 .
Main technical specifications of laser systems used.

Table 3 .
Comparison of surface roughness analysis for areas treated with only MT and MT + laser.SD is the standard deviation of the surface pattern referred to the Ra.Area treated Treatment type Ra (SD) (µm) Rz (µm) R Max (µm)