lable at ScienceDirect

International Biodeterioration & Biodegradation 90 (2014) 152e160

Contents lists avai

International Biodeterioration & Biodegradation

journal homepage: www.elsevier .com/locate/ ibiod

Fungal biodeterioration of stained-glass windows

Alexandra Rodrigues a,b, Sara Gutierrez-Patricio c, Ana Zélia Miller d,Cesareo Saiz-Jimenez c, Robert Wiley b, Daniela Nunes e, Márcia Vilarigues a,b,Maria Filomena Macedo a,b,*

aDepartamento de Conservação e Restauro, Faculdade de Ciências e Tecnologia, Universidade Nova de Lisboa, Campus Caparica, 2829-516 Caparica,PortugalbVICARTE, Research Unit Vidro e Cerâmica para as Artes, Faculdade de Ciências e Tecnologia, Universidade Nova de Lisboa, Campus Caparica,2829-516 Caparica, Portugalc Instituto de Recursos Naturales y Agrobiología, IRNAS-CSIC, Av. Reina Mercedes 10, 41012 Sevilla, SpaindCentro de Petrologia e Geoquímica, Instituto Superior Técnico, Av. Rovisco Pais, 1049-001 Lisboa, PortugaleAssociação Euratom/IST, Instituto de Plasmas e Fusão Nuclear e Laboratório Associado, Instituto Superior Técnico, Av. Rovisco Pais, 1049-001 Lisboa,Portugal

a r t i c l e i n f o

Article history:Received 25 November 2013Received in revised form26 February 2014Accepted 3 March 2014Available online 25 March 2014

Keywords:Stained glass windowsBiodeteriorationFungiHeritage

* Corresponding author. VICARTE, Research Unit ViFaculdade de Ciências e Tecnologia, Universidade Nov2829-516 Caparica, Portugal. Tel./fax: þ351 21 294 83

E-mail address: mfmd@fct.unl.pt (M.F. Macedo).

http://dx.doi.org/10.1016/j.ibiod.2014.03.0070964-8305/� 2014 Elsevier Ltd. All rights reserved.

a b s t r a c t

Biodeterioration of stained-glass windows by fungi was studied using historically accurate glass pro-duction methods. Glass reproductions were made according to the chemical composition determined bymicro energy dispersive X-Ray Fluorescence of two historical glass windows belonging to King FerdinandII’s collection dating from the 15th and 17th centuries. Three distinct glasses compositions with differentcolours were selected and reproduced: i) a mixed-alkali colourless glass: ii) a purple potash-glass withmanganese as chromophore, and iii) a brown potash-glass coloured by iron ions. The reproduced glasssamples, with two initial surface morphologies (corroded and non-corroded), were inoculated with fungipreviously isolated and identified on the original stained-glass windows as species of the genera Peni-cillium and Cladosporium. Physical and chemical glass surface alterations were analysed by means ofoptical microscopy, Raman microscopy, micro Infrared spectroscopy, and scanning electron microscopywith energy dispersive spectroscopy analysis. Results showed that fungi produced clear damage on allglass surfaces, present as spots and stains, fingerprints, biopiting, leaching and deposition of elements,and formation of biogenic crystals. Therefore, the inoculated fungi were able to biodeteriorate glasseswith distinct compositions. Regarding the biodeterioration degree, there were no differences betweenthe initial non-corroded and corroded glass surfaces.

� 2014 Elsevier Ltd. All rights reserved.

1. Introduction

Conservation and restoration of historical stained-glass win-dows is recognized as a complex problem (Drewello andWeissmann, 1997), since not only physical and chemical attackcan occur, but also microbial deterioration, known as biodeterio-ration. Generally, the biological corrosion of historical glass hasbeen underestimated (Garcia-Vallès et al., 2003). There are severalstudies on chemical corrosion of stained-glass windows, located incathedrals and churches all over Europe (e.g. Leissner, 1996;

dro e Cerâmica para as Artes,a de Lisboa, Campus Caparica,22.

Orlando et al., 1996; Garcia-Vallès et al., 1996, 2003; Sterpenichand Libourel, 2001; Melcher and Schreiner, 2004; Farges et al.,2007; Tournie et al., 2008; Gentaz et al., 2011; Vilarigues et al.,2011). However, the biodeterioration of those windows has oftenbeen undervalued, and little is known about the damage producedbymicroorganisms (Rölleke et al., 1999; Schabereiter-Gurtner et al.,2001; Carmona et al., 2006; Piñar et al., 2013). A few works statedthat the biodeteriorative role of microorganisms enhances deteri-oration and accelerates the physicalechemical processes leading toglass decay (Drewello and Weissmann, 1997; Gorbushina andPalinska, 1999; Marvasi et al., 2009). Fungi are among the mostharmful microorganisms associated with biodeterioration oforganic and inorganic materials, including glass (Drewello andWeissmann, 1997). The resistance of fungal spores to desiccation,their adhesion ability to the hydrophobic substrata, as well as the

Table 1Chemical composition used to reproduce the three distinct glasses with differentcolours.

Oxide components (w/w) Swiss panel (SP) German glass panel(GP)

Colourless Purple Brown

SiO2 58.5 64.4 62.5CaO 18.0 16.8 17.0K2O 5.0 12.6 11.2MnO 0.7 1.2 0.8Fe2O3 0.8 0.3 0.5Na2O 17.0 4.7 8.0

A. Rodrigues et al. / International Biodeterioration & Biodegradation 90 (2014) 152e160 153

ability to metabolize a wide range of carbon sources and thepreferred acidic to neutral pH values can be important factors to thedevelopment of fungi on glass surfaces. Moreover, fungal metabolicversatility enhances their efficiency to colonize a wide range ofsubstrata, such as stained-glass windows that were exposed to highrelative humidity and temperature fluctuations, water, and otherenvironmental factors such as aerosols and dust that carry outinorganic and organic matter (e.g. microorganism spores). There-upon, the main conditions that promote fungal growth on stainedglass windows are: i) the presence of organic matter of variousorigin, such as dust deposits, dead fungal and bacterial materials,and microbial metabolites; and ii) high temperature and relativehumidity values that allow glass to hold adsorbed water. Theamount of adsorbed water depends on the type of glass and on therelative humidity. According to Walters and Adams (1975), soda-lime glass adsorbs more water than borosilicate glass or lead sili-cate glass. The water activity (aw) on the surface increases in par-allel to the formation of the gel layer and the incorporation of watermolecules. This higher supply of water makes the glass moreattractive for microbes (Drewello et al., 2000).

Fungal biodeterioration studies on historical stained glass areimportant in understanding its processes and to help find a way toavoid and/or control this type of deterioration. Therefore, the mainobjective of this work was to appraise the biodeterioration ofstained-glass reproductions by fungi under laboratory conditions.In order to achieve this goal:

� two historical stained-glass windows, belonging to King Ferdi-nand II’s collection, with known chemical composition wereselected;

� the fungi that colonized the panels under study were identifiedand themore common generawere selected to be used in a glassbiodeterioration experiment;

� glass reproductions with three distinct compositions and withnon-corroded and corroded surfaces (simulating recently pro-duced and weathered glass surfaces) were manufactured andwere then inoculated with the selected fungi;

� different analytical techniques were used to study biodeterio-ration of the inoculated glass surfaces.

2. Materials and methods

2.1. Stained glass windows, storage conditions and its environment

King Ferdinand II (1816e1885) of Portugal gathered a vastcollection of art works, including a collection of stained-glasswindows. This collection was divided between his two principalresidences: the Pena National Palace (Sintra, Portugal) and theNecessidades Palace (Lisbon, Portugal). In the Necessidades Palace,the king had the majority of his collection of stained-glass panelsmounted in five windows and in three transomwindows. He had afurther assemblage installed in the great hall of Pena NationalPalace. In 1910, the stained-glass panels from the NecessidadesPalace were removed from their original position and sent to theAjuda National Palace (Lisbon, Portugal), where they remained instorage for several decades. In 1948, the complete collection arrivedat Pena National Palace with the aim of being installed in thewindows of the Stag Room. This plan was never implemented, andthe windows were kept in a storage room in a poor state of pres-ervation at the Pena National Palace for the next six decades(Rodrigues et al., 2013). Two panels belonging to this collectionwere selected for this study. One of the panels is from the 15thcentury and is thought to be of German origin (GPeGerman Panel).The other is a Swiss panel from the 17th century (SP e Swiss Panel)(Martinho and Vilarigues, 2011).

Pena National Palace, located in Sintra (38_47�N, 9_25�W), isinfluenced by oceanic conditions (it is only 8e9 km from the sea),and surrounded by a vast forested area. Temperature and relativehumidity were measured inside the storage room at the Pena Na-tional Palace, using a humidity/temperature data logger (RotronicHW3), from January to August 2011. Determinations of these twoparameters are important in order to reproduce, under laboratory,the approximate environmental conditions to which the glasswindows were kept in the Palace.

2.2. Manufacture of glass reproductions

Micro energy dispersive X-ray Fluorescence (m-EDXRF) analyseswere performed on the two stained-glass panels (Rodrigues et al.,2013). Glass reproductions were made based on the chemical com-positions determined for the German and Swiss historical stained-glass panels (Rodrigues et al., 2013). From these data, three distinctcompositions, with three different colours, were selected to bereproduced and used in the fungal biodeterioration experiment.Table 1 presents the chemical composition of the selected glasses.From this table it can be noticed that the colourless glass is a mixed-alkali while purple and brown glasses are potash-glass. The amountofMnoxide in the purple glass is higherwhen compared to the othertwo glasses, since this is the colouring element of this glass. Thecolouringelements inbrownglass aremanganese (Mn) and iron (Fe).

For the glass synthesis, pure raw laboratorymaterials were used,and each glass was melted in a refractory ceramic crucible at1300 �C, over 24 h. The glass obtained was blown using traditionaloff-hand glass tools and techniques analogous to the productionperiod of the original samples. The resulting roundels (discs) kept afire polished surface nearly identical to the historic glass production.

The discs were annealed at 500 �C (for 4 h), which was followedby slow cooling (approx. 20 h), and then they were cut into pieceswith no further polishing. The pieces had the approximate di-mensions of (10 � 10 � 2) mm3.

In order to understand glass biodeterioration and the influenceof surface texture, a set of 18 glasses for each composition (9 withnon-corroded surfaces and 9 with corroded surfaces) were pro-duced. The glass corrosion was performed by immersion of thesamples in 40 mL of distilled water for periods of 4e6 months. Thereaction containers were made of inert material (polyethylene) inorder to avoid any reactions between the electrolyte and containerwalls, and sealed to prevent interactions with the atmosphere.

After glass reproductions were made, their chemical composi-tions were analysed using micro energy dispersive X-Ray Fluores-cence (m-EDXRF).

2.3. Biological sampling

Samples for microbial identification were collected in fourdistinct areas in each glass window. The panel surfaces wereswabbed using sterile tubes swabs (one for each sample) that were

A. Rodrigues et al. / International Biodeterioration & Biodegradation 90 (2014) 152e160154

previously wetted in sterilized water in order to allow the removalof all organic and inorganic matter deposited in the panels. Theeight samples collected from both stained glasses were used forfungal isolation and identification.

2.4. Fungal isolation and identification by molecular techniques

The samples collected from both stained-glass windows wereplated on malt extract agar (MEA) using a sterile loop, and incu-bated at 22 �C in the dark for 4e5 days. After incubation, colonieswere removed from the agar surface using a sterile loop andtransferred to fresh MEA medium. The pure cultures were thenused for molecular identification.

Molecular identification of fungal strains was based on theanalysis of internal transcribed spacer regions (ITS) sequences fromrDNA. Genomic DNA from isolated fungi was extracted by scrapingthe mycelium from the plates and transferring it to a 1.5 mlEppendorf tube containing 350 ml of lysing buffer and glass beads.The mixture was shaken in a cell disrupter Fast Prep-24 (MP Bio-medicals, Solon, OH, USA) through 3 cycles at 4.5 m/s for 30 s. TheDNA was purified by phenol/chloroform extraction and ethanolprecipitation.

Amplifications of fungal internal transcribed spacer (ITS) re-gions were attempted using forward primer ITS1 (50-TCC GTA GGTGAA CCT GCG G) and reverse primer ITS4 (50-TCC TCC GCT TAT TGATAT GC-30) as described previously by White et al. (1990). PCR re-actions were performed in 50 mL volumes, containing 2 mL of theextracted DNA used as template DNA, 5 mL of 10� PCR buffer Biotaq(Bioline, Randolph, Massachusetts, USA), 1.5 mL of 50 mM MgCl2(Bioline), 1 mL of 10 mM deoxyribonucleoside triphosphate mixture(dNTPs) (Invitrogen, Carlsbad, California, USA), 0.5 mL of 50 mM ofeach primer and 0.25 mL of Taq DNA polymerase enzyme (Bioline),made up to 50 mL with nuclease-free water (Sigma-Aldrich, USA).PCR amplifications were performed with a thermal cycler iCyclerBioRad (BioRad, California, USA) using the following thermocyclingprogram: 2 min denaturing step at 95 �C, followed by 35 cycles ofdenaturing (95 �C for 15 s), annealing (55 �C for 15 s) and elonga-tion (72 �C for 2 min). A final elongation step of 10 min at 72 �C wasadded at the end. All amplification products were electrophoresedon 1.5% (w/v) agarose gels, stained with SYBR Green I (Roche Di-agnostics, Mannheim, Germany) and visualized under UV light. Theamplified PCR products were purified using themi-PCR PurificationKit (Metabion, Germany), according to the manufacturer’s protocol,diluted in 20 mL nuclease-free water (Sigma-Aldrich, USA) andstored at �20 �C. Finally, the purified products were sequenced bySecugen Sequencing Services (CSIC, Madrid, Spain).

Fig. 1. Experimental design: (a) Petri dish with distilled water at the bottom and a net to septhat were inoculated (example of the purple glass). (c) Top view of the Petri dish with the creader is referred to the web version of this article.)

The ITS sequences were edited using the biological sequenceeditor (Bioedit) v7.0.4 software (Technelysium, Tewantin, Australia),and similarity search was performed using the NCBI (Basic LocalAlignment Search Tool) BLASTn algorithm (Altschul et al., 1990) foridentification (National Centre for Biotechnology Information,http://www.ncbi.nlm.nih.gov/).

2.5. Glass biodeterioration experiment

All the glass samples were placed in glass Petri dishes, withdistilled water at the bottom. Between the water and the glasssamples there was a net to avoid the direct contact with the water(Fig. 1a). This net was divided in distinct areas to avoid samplesmixture with each other. In this study a total of 4 glass Petri disheswere used. Three of these Petri dishes contained a different glasscomposition (corresponding to the different colours) with corrodedand non-corroded surfaces in a total of 18 samples (see example forpurple samples in Fig. 1b). In one of the Petri dishes all the controlsamples, with the three different colours, were placed (Fig. 1c) in atotal of 6 samples for each distinct glass composition.

Before inoculation, all Petri dishes were autoclaved at 120 �C for20 min. After cooling, two fungal strains, previously isolated fromthe historical glass windows and grown on potato dextrose agar(PDA), were inoculated on the glass surfaces together with some ofPDA medium. This carbon source was used to simulate the organicmatter that was deposited on the stained-glass windows over theyears. Control samples were treated the same way as the othersamples and placed under the same conditions but were notinoculated with fungi. A set of 9 samples (3 non-corroded þ 3corroded þ 3 control) of each type of glass was removed from thesterilized environment and analysed after 2.5 months (t2.5, where tdesignates time in months) of incubation, after 4.5 months (t4.5)and, finally, at the end of the experiment (t6).

All the glass samples (control samples and inoculated samples)were kept under the same conditions (22e23 �C and 75e95%relative humidity (RH)) in order to simulate the environmentaltemperature and humidity measured inside the Pena NationalPalace.

2.6. Post-experiment analyses of glass surface alterations

A set of techniques was used to characterize the morphologicaland chemical alteration of the glass surfaces during the biodeteri-oration experiment.

The analyses were performed on the glass samples after acleaning procedure. Cleaning was separated in two steps. Some of

arate the water from the glass samples. (b) Top view of the Petri dish with glass samplesontrol samples. (For interpretation of the references to colour in this figure legend, the

A. Rodrigues et al. / International Biodeterioration & Biodegradation 90 (2014) 152e160 155

the analyses were performed right after the first cleaning step;others were only performed after the two cleaning steps werecarried out. Cleaning Step 1 consisted in themechanically removingof the upper part of the fungal biofilm colonizing the glass surfaceusing a scalpel parallel to the surface, leaving a thin biomass filmover the glass.

Step 2 consisted of a repeated cleaning of the glass surface witha cotton swab embedded in a 1:1 water ethanol solution, forremoving any biofilm trace (this is a regular conservation proce-dure used as a soft wet cleaning of stained-glass panels).

2.6.1. Optical microscopyThe microscopic documentation was carried out using a light

microscope (Axioplan 2, Zeiss) with digital camera (Nikon DMX) onsamples with biofilm, right after Step 1 of the cleaning methodol-ogy described above, and again after the Step 2.

2.6.2. Scanning electron microscopySamples were also analysed by scanning electron microscope

(analytical SEM Hitachi S2400 with Rontec standard EDS detector)at MicroLab, Instituto Superior Técnico (Lisbon, Portugal). Glasssamples were observed after the two cleaning steps have beenmade to access the corrosion patterns on the glass surface and theleached elements.

2.6.3. Fourier Transformed Infrared spectroscopyThe glass surfaces were analysed by Fourier Transform Infrared

(FTIR) spectroscopy after both cleaning procedures have beenmade, since this technique is sensitive to both long-range ordercrystalline materials and short-range order amorphous materials(Efimov, 1996; Efimov and Pogareva, 2006). These analyses wereperformed using a Nicolet Nexus spectrometer equipped with aContinuum microscope, in either one of two modes: in reflectanceusing an Attenuated Total Reflectance (ATR) slide-on accessorywith Si crystal, which can only give information from the nearsurface region of the samples.

The spectra were obtained for the spectral region between4000 cm�1 and 650 cm�1 in mid-infrared (Efimov, 1996; Vilariguesand Da Silva, 2006). FTIR spectra were normalized and a lorentziancurve fitting was carried out in order to ‘isolate’ the broad vibra-tional band at 800e1200 cm�1 in two main peaks from the struc-ture of the glassy network: the SieOeSi vibrational mode (about1000 cm�1 for quartz, and for vitreous silica ca. 920 cm�1) and theSieO� vibrational mode (ca. 900e950 cm�1) (Vilarigues and DaSilva, 2006).

Table 2Phylogenetic affiliations of the ITS sequences obtained from the fungal strains isolated fr

German panel Swiss panel

S1 S2 S3 S4 S1 S2 S3 S4

xx

xx

xx x

xx x

xx

xx x x

xx

aClosest relative obtained by comparison with the NCBI database and corresponding accession number.

2.6.4. Raman microspectroscopyA Raman microspectroscopy analysis was performed on the

glass surfaces, allowing for the identification of the corrosion andbiocorrosion products, since the following information can bepresent in spectra: (i) nature and phases: identification can be veryeasily obtained from databases; (ii) structure and composition ofglassy phases (Colomban, 2004). Raman spectra were obtainedusing a Raman Horiba Jobin Yvon, Labram 300, equipped with a500 mW HeeNe laser operating at 532 nm. The laser beam wasfocused with a 50� Olympus objective lens. All analysis was carriedout without filter and in 15 cycles of 30 s, between 200 and1300 cm�1.

In curve fitting Raman spectra, the baseline was first subtractedusing Labspec (DILOR) software, which was assumed to be poly-nomial. The number of attach points was minimized in order to‘isolate’ at the best the two strong peaks and to attach the baselineto each spectrum at about the same wavenumber position. AGaussian shape was assumed for the disordered state of glass(Colomban, 2003) and a lorentzian approach was assumed forcrystalline materials also found as overlapping peaks. The samespectral windows were used for the extraction of the componentsusing QtiPlot software peak-fitting module.

3. Results and discussion

3.1. Pena National Palace environmental conditions

The temperature and relative humidity measured inside PenaNational Palace from January to August 2011 presented thefollowing values: temperature average was 15 �C ranging from 9 �Cto 28 �C. The RH had an average value of 78% and ranged from 42%to 96%. The high values of relative humidity provide optimal con-ditions for fungi growth.

3.2. Identification of fungal strains by molecular techniques

Several fungi were cultured and isolated, and further identifiedby DNA-based molecular analysis. The identification of the isolatedfungal strains is presented in Table 2. The comparative sequenceanalyses revealed very high similarities between 99% and 100% todifferent members of the Ascomycota phylum and one member ofBasidiomycota. The identified fungi grouped into five generawithinAscomycotae Alternaria, Chaetomium, Cladosporium, Didymella andPenicillium, and the genus Sistotrema within Basidiomycota.Regarding the fungal strains isolated from both glass panels, five

om stained glass windows. S is the sample.

Fungal strains

Closest relativea Accession number Similarity (%)

Alternaria tenuissima JN624884 100Chaetomium coarctatum HM365260 100Cladosporium langeronii DQ780380 99Cladosporium sphaerospermum HQ248189 99Cladosporium sphaerospermum AB572904 99Cladosporium sphaerospermum AB572911 99/100Didymella phacae EU167570 99Penicillium sp. JQ775564 99Penicillium brevicompactum HM210834 99Penicillium brevicompactum HM469408 99Penicillium citreonigrum EF198646 99Penicillium citreonigrum EF198647 99/100Penicillium roseopurpureum JN246025 99Sistotrema sp. GU062211 99

A. Rodrigues et al. / International Biodeterioration & Biodegradation 90 (2014) 152e160156

different genera were obtained from the German panel and threefrom the Swiss panel. Members of the genus Penicillium dominatedon both panels, followed by species of the genus Cladosporium.However, the dominance of Cladosporium spp. on the Swiss panelwas higher than in the German panel. As a summary, Penicilliumand Cladosporium were the two predominant fungal genera iso-lated from both glass panels. These fungi are commonly isolatedfrom air samples either in indoor or outdoor environments, as wellas from historical window panels (Drewello and Weissmann, 1997;Piñar et al., 2013). They are known to play an important role in glasscorrosion (Krumbein et al., 1991; Drewello and Weissmann, 1997;Piñar et al., 2013) and therefore Penicillium sp. and Cladosporiumsp. were chosen for the glass biodeterioration experiment.

3.3. Biodeterioration of the glass reproductions

3.3.1. Glass surface morphologyBefore inoculation, the glass surfaces were analysed by optical

microscopy and SEM-EDS analysis in order to evaluate their surfacemorphology. The non-corroded surfaces showed smooth surface asnew, while the corroded ones, which were immersed in water inorder to accelerate the corrosion process, showed development of amicro-crack network, a corroded surface layer and some corrosionproducts, mostly calcium (Ca) enriched crystals. These representedthe initial state of the corroded samples used for the biodeterio-ration experiment.

3.3.2. Microscopic observationsOptical microscopy results of initially non-corroded glass sur-

face are presented in Fig. 2. This figure shows the initial (Fig. 2(a)and (b)) and final (Fig. 2(c) and (d)) glass alterations observed. Afteronly 2.5 months of incubation (t2.5) it was possible to verify greatalterations over all the inoculated glass surfaces, Fig. 2(a) shows themycelia growth associated with crystal formation that occurred

Fig. 2. Optical microscopy results of initially non-corroded colourless glass surface at t2.5: (a)the surface and the white arrows show the mycelia growth; (b) after the second cleaning (microscopy results of initially non-corroded glass surfaces at t6, after the second cleaning (glass with staining. (For interpretation of the references to colour in this figure legend, the

after the mechanical cleaning (Step 1) After cleaning Step 2 wasmade a similar pattern can also be seen in Fig. 2(b), where the areasbeneath the mycelia show more crystalline substances which wereformed. Microorganisms under investigation seemed to have agood affinity for adhesion and growth on the glass surface. Theywere able to create a well-developed mycelia film on the glasssurfaces (Fig. 2(a)), in a relatively short time (2.5 months). The highdegree of fungal adhesion was observed over all the glass samples.

The SEM-EDS analysis allowed us to observe physical andchemical alterations caused by the inoculated fungal strains on theglass surfaces, complementing the optical microscopy observa-tions. After only 2.5 months of incubation (t2.5) the formation oflarge quantities of crystalline substances was observed in all threetypes of inoculated glasses, in concordance with the optical mi-croscopy results. Also at t2.5, the initially non-corroded surfacesshowed some prints whichwere attributed to the presence of fungi.

The observed surface alteration that occurred after 2.5 months(t2.5) increased slowly until 6 months of incubation (t6) (Fig. 2(c)and (d)). In the case of the colourless glass, the formation of crys-talline substances was the major alteration observed over thesesurfaces (Table 3, see Raman results below). This is associated withan increasing of fungal mycelia over time, which did not happen forthe brown and purple glass types. In fact, fungal growth seemed tobe almost the same at t2.5, t4.5 and t6 for these two types of glass. Onthe other hand, these two coloured glasses showed an increasingstaining and iridescence (Fig. 2(c) and (d)). This can be related to achemical alteration, possibly caused by the mobilization of someions due to the interaction between the metabolic products pro-duced by the fungi and the glass substratum. Some crystal forma-tion was also noticeable over the glass surfaces.

It was already shown that some fungi are able to acquire theelements needed for growth from the deteriorating glass(Krumbein et al., 1991), and that chemical changes in the glass(accumulation of oxidized manganese and iron salts in the surface

After mechanical cleaning (Step 1), the black arrows indicate some crystal formation onStep 2): crystalline structures are clearly visible. Figures (c) and (d) present the opticalStep 2): (c) brown glass (the black arrow indicates some crystal formation); (d) purplereader is referred to the web version of this article.)

Table 3Crystalline substances found over the glass samples, mostly identified by m-Raman(n.i. ¼ non identified crystal).

Glass samples Colourless Purple Brown

Inoculated Inoculated Inoculated

t0 Non-corroded e e e

Corroded e e e

t2.5 Non-corroded CaCO3 CaCO3 e

Corroded CaCO3 CaCO3 e

t4.5 Non-corroded NaSO4$H2O CaCO3 e

Corroded NaAlSi3O8 e e

t6 Non-corroded NaSO4$H2O CaCO3 e

Corroded CaSiO3 þ n.i. e SiO2 þ n.i.

A. Rodrigues et al. / International Biodeterioration & Biodegradation 90 (2014) 152e160 157

layer) occur rapidly upon a short period of inoculation with fungi(Gorbushina and Palinska, 1999). Drewello and Weissmann (1997)also found that purple medieval glass could be less durablebecause of high quantity of the transition metal (Mn). Our resultsare consistent with the possibility that the microscopically alter-ation on brown and purple sample surfaces was due to thesephenomena, since they were rich in Fe and Mn.

Fig. 3. Results of the SE and BSE SEM images of the glass surfaces at t6: (a) initially non-c(b) initially non-corroded brown glass surface: biopiting (white arrow) and micro-cracksnon-corroded surface which shows fingerprints of hyphae; (d) initially non-corroded purplepresence of the fungi are clearly visible.

After 6 months, etching produced by hyphae was visible on alltypes of glasses by SEM (Fig. 3). This phenomenon is associatedwith micro-cracking and enrichment of some elements, namely Ca,(Fig. 3a), biopiting (Fig. 3b), hyphae fingerprints (Fig. 3c), crystalsformation (Fig. 3d) and some elements depletion or redeposition(Fig. 4).

Drewello and Weissmann (1997) reported that the enrichmentof some elements in specific areas can be produced by the mobi-lization or redeposition of elements by microorganisms or inducedby biomineralization. In this study, several crystals rich in Ca werefound by Raman microspectroscopy and we suggest they could berelated with similar phenomena.

The observed depletion of Na, K and Ca from glass surfaces insome pittings (Fig. 4) can be attributed to microbial corrosion(Krumbein et al., 1991). These elements can be leached by organ-isms or their exuded acids. On the other hand, the elements Na, Kand Ca were also found in higher concentrations, mostly at thesurface layer, compared to the adjacent biopitted surface. This canindicate that the fungi removed the surface layer and part of theglass matrix, exposing the bulk glass. These glass alterations havebeen already referred by several authors: alteration of optical

orroded brown glass surface: the black arrow indicates an area with Ca enrichment;are visible, as well as fingerprinting of the spores of the fungi; (c) colourless initiallyglass: the white arrow shows a crystalline substance, and the areas with prints of the

Fig. 4. SEM-EDS analysis spectra of the brown glass surface at t6. Surface chemicalalteration e depletion of Na, K and Ca e versus biopiting e bulk glass is exposed.

A. Rodrigues et al. / International Biodeterioration & Biodegradation 90 (2014) 152e160158

properties (Drewello and Weissmann, 1997; Garcia-Vallès et al.,2003), etching, pitting, leaching (Drewello and Weissmann, 1997;Gorbushina and Palinska, 1999; Carmona et al., 2006; Ehrlich,2010) and biocracking (Müller et al., 2001). Some of these alter-ations were considered similar and sometimes hard to distinguishfrom glass corrosion without the influence of microorganisms,while others were considered typical patterns: the crater formationby coalescence of dense biopiting, fingerprinting and internalcorrosion followed by breaking of the surface into fresh craters(Gorbushina and Palinska, 1999).

3.3.3. Raman microspectroscopyThe m-Raman analysis allowed for the identification of small

crystalline substances formed over the glass surfaces (Table 3). Theclean control samples showed no crystal formation, while in theinoculated ones the formation of calcium carbonate, a commonwater corrosion product, was observed (Greiner-Wronowa andStoch, 1996; Vilarigues and Da Silva, 2006). Despite the fact thatthis is themost common origin of this crystal, themajor densitywasobserved on inoculated surfaces and this leads us to presume thatfungiwere also involved in this crystal formation. On the otherhand,

Fig. 5. FTIR spectra obtained from colourless glass samples at the region of 400e4000 cm�1

1020 cm�1) and the bond SieO� (ca. 900 cm�1). (For interpretation of the references to co

in some areas we have been able to identify some sulphates (hy-drated sodium sulphate), and also some silicates (CaSiO3, NaAlSi3O8,SiO2) using m-Raman. According to Krumbein et al. (1991), thepresence of these crystals can be due to fungal acids excreted duringthe complex mechanisms involved in biocorrosion.

3.3.4. Fourier Transformed Infrared spectroscopyFTIR analyses were carried out only in colourless and brown

samples since purple samples did not provide a signal with thistechnique. Both colourless and brown samples showed clear sur-face alterations, which can be attributed to the presence of fungi,since the control samples show a very different and less intensealteration (see Fig. 5).

The two main bands considered in this discussion are related tovibrational mode n of the bond SieOeSi (ca. 1020 cm�1) and thebond SieO� (ca. 900 cm�1) (Abo-Naf et al., 2002; De Ferri et al.,2012). The band shifts indicates a structural modification which isassociated with the deterioration of the surface and formation of asilica-enriched layer (Vilarigues and Da Silva, 2006). The intensityof Si-O-Si band increases due to the decrease of alkali ions (De Ferriet al., 2012). On the other hand, the shift of the second peak to alower wavenumber indicates that the number of non-bridgingoxygen and of alkali ions decreases (Vilarigues and Da Silva,2006). These phenomena can sometimes be cyclic when thesilica-gel layer is totally leached and another glass layer is on top.Although the corrosion is complex and incongruent when thecatalytic action of fungi and other microorganisms is taken underconsideration (Krumbein et al., 1991), the complexity of corrosionof the samples can be related with the non-linear variation of thebands along the incubation time, and the role of fungi is evidentwhen looking at the FTIR spectra obtained (Fig. 5).

Comparisons of the inoculated and non-inoculated samplesshowed clear differences, with the inoculated samples havinghigher surface alteration that can be mainly attributed to fungalbiodeterioration. Moreover, the short period of time in whichevident biodeterioration was noted (2.5 months) is a previouslyunseen result for this type of glass. Since the glass samples were notanalysed before this time, it is possible that biodeteriorationoccurred before 2.5 months.

. The two main bands were associated with vibrational mode n of the bond SieOSi (ca.lour in this figure legend, the reader is referred to the web version of this article.)

A. Rodrigues et al. / International Biodeterioration & Biodegradation 90 (2014) 152e160 159

Regarding fungal biodeterioration, no significant differenceshave been observed between the corroded and non-corrodedinoculated samples. This departed from the expected results as itwas previously thought that the corroded surface would naturallylend itself to higher fungal growth; since it is known that a highersupply of water promotes, theoretically, a more attractive glasssurface formicrobes. But, the results do not seem to indicate that theinitial silica-gel layer promoted the formation of a denser biofilm ormore physical adhesion to the surface, as indicated by Krumbeinet al. (1991). They rather seemed to show that the inoculated non-corroded surfaces often had more crystalline structures and stain-ing in the early stages. After 6 months of incubation both corrodedand non-corroded surfaces seemed to achieve the same degree ofsurface biodeterioration.

This study also shows that, after six months, the different glasscompositions studied (colourless mixed-alkali and brown andpurple potash-glass samples) presented a similar degree of biode-terioration. However, this biodeterioration exhibited distinct formsin the distinct glasses. The mixedealkali colourless glass presenteda higher precipitation of crystals and more developed fungalmycelia; whereas potash brown and purple glasses showed morechanges in optical properties (spots and fingerprints due to thebiological action) than the mixed-alkali glass. However, purpleglass presented much more pitting with element depletions (K, Caand Na) than the other glasses, giving an indication that maybe thisglass could be more susceptible to fungal biodeterioration. Thesedata corroborate previous studies in which it was observed thatglasses made with potash flux were more easily decayed (Garcia-Vallès and Vendrell, 2002; Garcia-Vallès et al., 2003). Pittingoccurring only on potash glass was also reported by Koestler et al.(1986).

4. Conclusions

All the glass samples inoculated with fungi, presented surfacebiodeterioration after a short period of time (2.5 months). In theglasses used in this study, biodeterioration by fungi occurred inboth corroded and non-corroded samples in a similar form anddegree. This unexpected result should be further studied.

At the end of our experiment (6 months), the inoculated sam-ples showed evidence of strong surface biodeterioration with het-erogeneous spots, abundant pitting, stains, leaching and formationof crystals as well as several fingerprints of fungal hyphae andconidia. Regarding the chemical results, Na, K and Cawere themostleached elements from glass surfaces near pitting areas. Leaching ofFe and Mn was not detected, but the optical alterations seem toindicate an ionic alteration of these elements.

Concerning the conservation of the stained-glass panels, it wasnoticed that the conventional procedure of stained glass cleaningby using a solution of water: ethanol (1:1) does not completelyremove the fungi. Therefore, further research should be done inorder to find a better and safer way to clean these glasses.

Finally, in order to have a deeper understanding of biochemicalprocess involved in glass biodeterioration by fungi, more studiesshould be done towards the identification of metabolic productsproduced by these microorganisms and their chemical interactionwith glass surfaces.

Acknowledgements

This work was supported by The Portuguese Science Foundation(projects PEst-OE/EAT/UI0729/2011 and PTDC /EPH-PAT/3579/2012), and was partially financed through a postdoctoral grant toAZM (SFRH/BPD/63836/2009) and a doctoral grant to AJR (SFRH/BD/84675/2012).

The authors are grateful to Pena National Palace, Parques deSintra, Monte da Lua, for the collaboration.

References

Abo-Naf, S.M., El Batal, F.H., Azooz, M.A., 2002. Characterization of some glasses inthe system SiO2, (Na2ORO)-R- by infrared spectroscopy. Mater. Chem. Phys. 77,846e852.

Altschul, S.F., Gish, W., Miller, W., Myers, E.W., Lipman, D.J., 1990. Basic localalignment search tool. J. Mol. Biol. 215, 403e410.

Carmona, N., Laiz, L., Gonzalez, J.M., Garcia-Heras, M., Villegas, M.A., Saiz-Jimenez, C., 2006. Biodeterioration of historic stained glasses from the Cartujade Miraflores (Spain). Int. Biodeterior. Biodegrad. 58, 155e161.

Colomban, P., 2003. Polymerization degree and Raman identification of ancientglasses used for jewelery, ceramic enamels and mosaics. J. Non-Cryst. Solids323, 180e187.

Colomban, P., 2004. Raman spectrometry, a unique tool to analyze and classifyancient ceramics and glasses. Appl. Phys. A 79, 167e170.

Drewello, R., Weissmann, R., 1997. Microbially influenced corrosion of glass. Appl.Miciobiol. Biotechnol. 47, 337e346.

Drewello, U., Weissmann, R., Rölleke, S., Müller, E., Wuertz, S., Fekrsanati, F.,Trolle, C., Drewello, R., 2000. Biogenic surface layer on historical window glassand the effect of excimer laser cleaning. J. Cult. Herit. 1, S161eS171.

Efimov, A.M., 1996. Quantitative IR spectroscopy: applications to studying glassstructure and properties. J. Non-Crystalline Solids 203, 1e11.

Efimov, A.M., Pogareva, V.G., 2006. IR absorption spectra of vitreous silica and sil-icate glasses: the nature of bands in the 1300 to 5000 cm�1 region. Chem. Geol.229, 198e217.

Ehrlich, H., 2010. Modern views on desilicification: biosilica and abiotic silicadissolution in natural and artificial environments. Chem. Rev. 110, 4656e4689.

Farges, F., Etcheverry, M., Haddi, A., Trocellier, P., Curti, E., Brown Jr., G.E., 2007.Durability of silicate glasses: an historical approach. X-Ray absorption Finestructure e XAFS13: 13th International Conference. AIP Conf. Proc. 882, 44e50.

De Ferri, L., Bersani, D., Lorenzi, A., Lottici, P.P., Vezzalini, G., Simon, G., 2012.Structural and vibrational characterization of medieval like glass samples.J. Non-Cryst. Solids 358, 814e819.

Garcia-Vallès, M., Blazquez, F., Molera, J., Vendrell-Saz, M., 1996. Studies of patinasand decay mechanisms leading to the restoration of Santa Maria de Montblanc(Catalonia, Spain). Stud. Conserv. 41, 1e8.

Garcia-Vallès, M., Vendrell, M., 2002. The glasses of the transept’s rosette of thecathedral of Tarragona: characterization, classification and decay. Boletín la Soc.Española Cerámica Vidr. 41, 217e224.

Garcia-Vallès, M., Gimeno-Torrente, D., Martínez-Manent, S., Fernández-Turiel, J.L.,2003. Medieval stained glass in a Mediterranean climate: typology, weatheringand associated biomineralization processes and products. Am. Mineral. 88,1996e2006.

Gentaz, L., Lombardo, T., Loisel, C., Chabas, A., Vallotto, M., 2011. Early stage ofweathering of medieval-like potash-lime model glass: evaluation of key factors.Environ. Sci. Pollut. Res. 18, 291e300.

Gorbushina, A.A., Palinska, K.A., 1999. Biodeteriorative processes on glass: experi-mental proof of the role of fungi and cyanobacteria. Aerobiologia 15, 183e191.

Greiner-Wronowa, E., Stoch, L., 1996. Influence of environment on surface of theancient glasses. J. Non-Cryst. Solids 196, 118e127.

Koestler, R.J., Santoro, E.D., Ransick, L., Brill, R.H., Lynn, M., 1986. Preliminary SEMstudy of microbiologically induced deterioration of high alkali-low lime glass.In: O’Rear, C.E., Llewellyn, G.C. (Eds.), Biodeterioration Research 1. Plenum, NewYork, pp. 295e307.

Krumbein, W.E., Urzì, C.E., Gehrmann, C., 1991. Biocorrosion and biodeterioration ofantique and medieval glass. Geomicrobiol. J. 9, 139e160.

Leissner, J., 1996. The effect of corrosion on stained glass windows. Mater. Con-strucción 46, 27e38.

Martinho, B., Vilarigues, M., 2011. Vitrais e Vidros, um gosto de D. Fernando II/Stained Glass and Glass Objects, Ferdinand II’s passion. Parques de Sintra,Monte da Lua, S.A.

Marvasi, M., Vedovato, E., Balsamo, C., Macherelli, A., Dei, L., Mastromei, G.,Perito, B., 2009. Bacterial community analysis on the Medieval stained glasswindow “Nativitá” in the Florence Caterdral. J. Cult. Herit. 10, 124e133.

Melcher, M., Schreiner, M., 2004. Statistical evaluation of potash-lime-silica glassweathering. Anal. Bioanal. Chem. 379, 628e639.

Müller, E., Drewello, U., Drewello, R., Weiszmann, R., Wuertz, S., 2001. In situanalysis of biofilms on historic window glass using confocal laser scanningmicroscopy. J. Cult. Herit. 2, 31e42.

Orlando, A., Olmi, F., Vaggelli, G., Bacci, M., 1996. Medieval stained glasses of PisaCathedral (Italy): their compositionandalterationproducts.Analyst 121, 553e558.

Piñar, G., Garcia-Valles, M., Gimeno-Torrente, D., Fernandez-Turiel, J.L., Ettenauer, J.,Sterflinger, K., 2013. Microscopic, chemical, and molecular-biological investi-gation of the decayed medieval stained window glasses of two Catalonianchurches. Int. Biodeterior. Biodegrad. 84, 388e400.

Rodrigues, A., Martinho, M., Gomes, C., Macedo, M.F., Vilarigues, M., 2013. Thestained-glass collection of King Ferdinand II of Portugal: concept, conservation,and chemical analysis of two panels. In: Roemich, H., van LookerenCampagne, K. (Eds.), Recent Advances in Glass, Stained Glass and CeramicsConservation 2013. Spa Pub., Amsterdam, pp. 259e267.

A. Rodrigues et al. / International Biodeterioration & Biodegradation 90 (2014) 152e160160

Rölleke, S., Gurtner, C., Drewello, U., Drewello, R., Lubitza, W., Weissmann, R., 1999.Analysis of bacterial communities on historical glass by denaturing gradient gelelectrophoresis of PCR-amplified gene fragments coding for 16S rRNA.J. Microbiol. Methods 36, 107e114.

Schabereiter-Gurtner, C., Piñar, G., Lubitz, W., Rölleke, S., 2001. Analysis of fungalcommunities on historical church window glass by denaturing gradient gelelectrophoresis and phylogenetic 18S rDNA sequence analysis. J. Microbiol.Methods 47, 345e354.

Sterpenich, J., Libourel, G., 2001. Using stained glass windows to understand thedurability of toxic waste matrices. Chem. Geol. 174, 181e193.

Tournie, A., Ricciardi, P., Colomban, P., 2008. Glass corrosion mechanisms: a mul-tiscale analysis. Solid State Ionics 179, 2142e2154 (Elsevier B.V).

Vilarigues, M., Da Silva, R.C., 2006. Characterization of potash-glass corrosion inaqueous solution by ion beam and IR spectroscopy. J. Non-Cryst. Solids 352,5368e5375.

Vilarigues, M., Redol, P., Machado, A., Rodrigues, P.A., Alves, L.C., Da Silva, R.C., 2011.Corrosion of 15th and early 16th century stained glass from the monastery ofBatalha studied with external ion beam. Mater. Charact. 62, 211e217.

Walters, H.V., Adams, P.B., 1975. Effects of humidity on the weathering of glass.J. Non-Cryst. Solids 19, 183e199.

White, T.J., Bruns, T., Lee, S., Taylor, J., 1990. Amplification and direct sequencing offungi ribosomal RNA genes for phylogenetics. In: Innis, M.A., Gelfand, D.H.,Sninsky, J.J., White, T.J. (Eds.), PCR protocols. A Guide to Methods and Appli-cations. Academic Press, San Diego, pp. 315e322.