early stage of weathering of medieval-like potash–lime model glass: evaluation of key factors

RESEARCH ARTICLE

Early stage of weathering of medieval-like potash–limemodel glass: evaluation of key factors

Lucile Gentaz & Tiziana Lombardo & Claudine Loisel &Anne Chabas & Marta Vallotto

Received: 4 February 2010 /Accepted: 23 June 2010 /Published online: 17 July 2010# Springer-Verlag 2010

AbstractPurpose Throughout history, a consequent part of themedieval stained glass windows have been lost, mostlybecause of deliberate or accidental mechanic destructionduring war or revolution, but, in some cases, did notwithstand the test of time simply because of their lowdurability. Indeed, the glasses that remain nowadays are formany in a poor state of conservation and are heavilydeteriorated. Under general exposure conditions, stainedglass windows undergo different kinds of weathering processesthat modify their optical properties, chemistry, and structure:congruent dissolution, leaching, and particle deposition (thecombination of those two leading together to the formation ofneocrystallisations and eventually crusts). Previous researchhas studied the weathering forms and the mechanisms fromwhich they are originated, some others identified the mainenvironmental parameters responsible for the deteriorationand highlighted that both intrinsic (glass composition) andextrinsic (environmental parameters) factors influenceglass degradation. Nevertheless, a clear quantification of

the impact of the different deterioration extrinsic factorshas not been performed.Methods By analysing the results obtained with modelglass (durable and nondurable) exposed in the field, thispaper proposes a simple mathematical computation evalu-ating the contribution of the different weathering factors forthe early stages of exposure of the stained glasses.Results In the case of non durable glass, water runoff wasidentified as the main factor inducing the leaching (83.4±2.6% contribution), followed by gas (6.4±1.5%) andparticle deposition (6.8±2.2%) and adsorbed water (3.4±0.6%). Moreover, it was shown that the extrinsic stimulisuperimposes with the impact of glass composition to theweathering.Conclusions Those results show that the role played by drydeposition, even if less important than that of the wetdeposition, cannot be neglected.

Keywords Stained glass windows . Field exposure .

Model glass . Atmosphere .Weathering . Evaluation

1 Introduction

In usual exposure conditions, the interaction betweenstained glass windows and environmental parametersinduces a deterioration of glass, mainly manifested by lossof colour, transparency and even matter (Bettembourg1976). Medieval glass from the north of Alps is a well-known Si–Ca–K glass, highly vulnerable to the deteriorationphenomena (Newton 1984). It is mainly characterised bysilica contents lower than 60% and by high contents inmodifier cations. Therefore, it is considered a low-durabilityglass included in the type IV of the Hench classification(Hench 1975).

Responsible editor: Philippe Garrigues

L. Gentaz (*) : T. Lombardo :A. ChabasLaboratoire Interuniversitaire des Systèmes Atmosphériques (LISA),Universités Paris-Est Créteil et Paris Diderot CNRS,61 Avenue du Genéral de Gaulle,94010 Créteil, Cedex, Francee-mail: [email protected]

C. LoiselLaboratoire de Recherche des Monuments Historiques (LRMH),Champs-sur-Marne, France

M. VallottoStazione Sperimentale del Vetro (SSV),Via Briati 10, IT-30141 Murano-Venice, Italy

Environ Sci Pollut Res (2011) 18:291–300DOI 10.1007/s11356-010-0370-7

The deterioration of low-durability glass immersed inaqueous solutions was extensively studied and especially forconfined environment (Douglas and El-Shamy 1967). In thoseconditions, two main forms of deterioration occur on glasses:leaching (selective dissolution) and congruent dissolution.Leaching takes place in aqueous solutions with a pH<7 or 9(the value varies with the glass composition: Doremus 1975;Casey and Bunker 1990). The modifier cations present in theglass matrix (alkali and alkaline-earth elements) diffusetowards the glass surface, whereas H+ (or H3O

+) ions diffuseinto the glass. Aweak (and often brittle) hydrated layer formson the surface of the glass. In certain cases, the growth of thislayer is limited by its own thickness, and, after some time, itacts as a barrier against the interdiffusion (Boksay et al.1968). Contrarily, when the glass surface is in contact with apH>7 or 9 (Das 1969; Paul 1977), congruent dissolutionoccurs. Parts of the glass network itself are lost and thestructure of the glass is affected. Pits and craters are commonmorphologies related to this type of deterioration. Althoughwith different rates, this type of deterioration affects any typeof glass, durable or not.

Atmospheric exposures of Si–Ca–K model glasses wereconducted (Römich 2000; Schreiner 1988; Munier et al.2002a; Melcher 2007), and both congruent dissolution andleaching were observed. The deterioration is triggered whenthe glass surface comes in contact with water, due to either thepresence of condensation water (Newton and Bettembourg1976) or to the streaming of rain on the surface, inducing, insome cases, a loss of material and the formation of pits,cracks and scales. A silica gel layer can form, in some cases,as a result of the recombination and precipitation of the glasscomponents (Doremus 1979; Brinker 1988). Nevertheless,the degradation of the glass is not solely caused by a loss ofmaterial. A considerable amount of pollutants (particulate andgas) are transported in the air and are able to react with glassregardless of its durability. First, the addition of atmosphericparticles to the glass surface alters its physicochemicalproperties, leading to a loss of transparency (direct effect),modifying the roughness of the surface and increasing its timeof wetness, as water vapour can condensate around particles(indirect effect). They can exacerbate the leaching impact.This type of deterioration is referred to as the soiling of theglass (Lombardo et al. 2005). Another phenomenonobserved in atmospheric conditions is a crystallisation ofsmall crystals of salts formed at the surface of the glass andcalled neocrystallisations (Munier et al. 2002b; Melcher andSchreiner 2004). They result from the interaction betweenthe atmospheric gases and the cations present on the surface.The cations can be issued by the particle deposited on theglass surface or by the glass itself. On low-durability modelglasses, where the glass can be heavily depleted of itscations, the formation of neocrystallisations is a consequenceof the leaching of the glass. At the beginning of the

exposure, the composition of the salts reflects the composi-tion of the glass matrix and gives an indication of thepreferential extraction of the different cations. The growth ofneocrystallisations can lead to the formation of crusts at thesurface of the glass. On the ancient glasses, the crusts areconstituted of complex mixture of mineral and organicparticles (carbonates, silicates, oxalates, etc.) and very oftenpresent high level of gypsum (CaSO4 · 2H2O;; Gillies andCox 1988; Lefèvre et al. 1998; Perez y Jorba et al. 1978) dueto the relative abundance of SO2 in the ancient atmospherewhen compared with contemporary concentration. Gypsumis poorly soluble and tends to stay on the surface, while othersalts are washed away. The presence of calcium can berelated to the relative abundance of that element in theancient glasses, but considering that black gypsum crustswere found on non-calcium-bearing ancient materials (copperalloys, Ca-free stones, etc.), calcium can also have anatmospheric origin (Lefèvre and Ausset 2002).

As pollution levels vary from one location to another,the intensity of glass deterioration is also variable. Underatmospheric stress, as is the case in an urban atmosphere,it is critical to determine in what proportion the pollutantsand the climate affect the ancient glasses. Quantificationof the influence of environmental factors in the decay ofSi–Ca–K glass was initiated by Melcher (2007) whocalculated the dose–response functions of such environ-mental factors. These functions are effective in evaluatingthe parameters to take into account, and operate adichotomy, between wet and dry deposition impacts. Thepresent paper intends to give a combinated evaluation ofthe two types of deposition. Parallel exposures underatmospheric conditions of a Si–Na–Ca glass (durable)and a Si–K–Ca glass (low durability) have been used inorder to investigate the different effects of compositionand surrounding environment on the decay of glasses. Thedurable modern glass is used as a reference. The survey ofthe environmental parameters during the exposure periodhelped: first, identifying anthropic (pollutants) and meteoro-logical [rain, temperature, relative humidity (RH)] atmo-spheric factors having a role in glass degradation, andsecond, classifying them. It is important to notice that, forthis evaluation, the sole influence of atmospheric parameterswas taken into account. This excludes study of the bioticparameters and omission of microorganisms as specificdegradation factors.

2 Material and methods

2.1 In situ model glass exposures

Two different glass types were exposed in the field (Table 1).The first type, a low-durability potash–lime model glass, was

292 Environ Sci Pollut Res (2011) 18:291–300

created in order to simulate the composition of transparentmedieval stained glass windows. However, it shows higherconcentrations in modifier cations than the latter, whichallowed quantification of the weathering in the time frame ofthe present research. It is highly concentrated in K and Caand will be hereafter referred to as Si–K–Ca glass. It wasmanufactured at the Stazione Sperimentale del Vetro inMurano. The glass melt was poured in a cylinder crucible,allowed to cool down, and cut in 2 cm × 2 cm × 5 mmsamples. Finally, the side that would face directly theenvironment was polished. The second glass, durable one,referred to as Si–Na–Ca, is a typical modern window glassand was purchased at a hardware store and cut into 2 cm ×2 cm × 2 mm samples.

Both samples were exposed to urban atmosphere,sheltered or unsheltered from the water runoff (ISO 85651992), simulating the exposure conditions that can besimultaneously encountered in churches or even in singlepanels. The exposure site is located in the centre of Paris atca. 44 m high, on the roof of the north tower of St.Eustache church. A parallel air quality and meteorologicalsurvey was carried out. The first one was undertaken by theAirparif agency (the service responsible for the air qualitysurveillance in Paris) at the station “Les Halles” located justbelow the exposure site. Meteorological monitoring wascarried out on site, using a Davies instrument stationimplanted on the church roof. An experimental setup wasdeveloped in order to measure the pH of the raincontinuously.

A first set of samples was withdrawn after 15 days ofexposure and other sets each month during a year and ahalf. Finally, a last set was withdrawn after 2 years ofexposure. The frequent withdrawals and extended exposureperiod allowed documenting the early stage of weatheringat a fine time resolution.

2.2 Analytical techniques

After withdrawal, the samples were placed for at least 24 hin a vacuum desiccator conditioned at 50% RH in order tomaintain the stored glass in similar conditions to those itexperienced on site. Indeed, RH variations could allowwater adsorption/desorption cycles to take place at thesurface of the samples and induce further degradation of theglass after withdrawal. The samples were then analysedusing various destructive and nondestructive techniques in

order to identify the different forms of alteration occurringon their surface and structural modifications.

Gravimetric measurements were conducted in order toquantify the mass variations in both exposure conditions.Measures were obtained with Sartorius Expert LE225Dscales (±0.1 mg accuracy). Samples were weighted beforeand after exposure and, in the case of the sheltered samples,after rinsing the deposited particles from the surface.

The observation of the glass surface involved differentmicroscopic methods. The optic microscopes wereequipped with a Sony 3CCP colour video camera.Primarily, optic microscopy (Leica MZ12 binocularmicroscope, magnification ×80, focus 7 μm) was usedto observe the presence of various deterioration featureson the glasses such as salt crystals, particles, cracks, etc.The images obtained with the microscope were analysedby the software Histolab (Microvision) in order to give aquantification of the percentage of the glass surfacecovered by the deposition over the total analysed surface(%CS). A Nikon Eclipse E200 microscope was used toobserve the optic manifestation of the glass hydration:the irisation of surface. The glass surface irisation wasobserved by reflecting an exterior light source (FortGLI154) on the glass. Further identifications were madeby scanning electron microscope (SEM) including count-ing and characterisation of particles on the surface of theglasses, observation of alteration morphologies anddetermination of chemical compositions of altered andpristine glasses. The upper surface of the glass wascoated with platinum and observed by a Jeol JSM-6301F linked with an energy-dispersive spectroscopy (EDS)detector (Link ISIS 300 for Zelement > ZB), operating at anaccelerating voltage of 20 kV, with a working distance of15 mm and a probe current of 6×10−11 A. The cross-sections were observed after carbon coating by a JeolJSM-5900, linked with an EDS detector Oxford ISIS 300,operating at accelerating voltage of 20 kV, with a workingdistance of 8–10 mm and a probe current of 5×10−10 A.

The identification of the deposited matter was largelyrelated to the specific identification of the soluble depositions.The sample surfaces were hence rinsed in order to solubilisethe deposits. The solutions obtained by that process were thenfiltered (∅filter=0.2 μm) and analysed by both liquid phaseion chromatography and induced coupled plasma atomicemission spectrometry (ICP-AES). Both instruments wereused in a clean room to avoid any contamination. Similarly,

Table 1 Composition of the pristine model glasses (in wt.%), given by SEM-EDS

% SiO2 Al2O3 K2O MgO CaO P2O4 Na2O SO3 Fe2O3

Si–K–Ca 50±0.2 2±0.05 25±0.05 3±0.05 18±0.1 2±0.05 – – –

Si–Na–Ca 71.6±0.2 0.7±0.05 0.4±0.2 4.1±0.1 9.6±0.1 0.1±0.05 13.1±0.2 0.3±0.1 0.1±0.1

Environ Sci Pollut Res (2011) 18:291–300 293

samples were also rinsed in a clean environment and keptrefrigerated in Teflon recipients to avoid any adsorption ofions on the recipient surface. In the specific case of ICP-AESmeasurements, 5 ml of solution were acidified, prior to theanalysis, using 2 μl HNO3 (14.44 M) (Desboeufs et al.2003). A Dionex 4500i ion chromatography was used toanalyse the rinsing water in order to quantify the followinganions SO4

2−, NO3−, Cl−, F−, and HCOO− and cations Ca2+,

Na+, K+, Mg2+, Sr2+ and NH4+. The error (%RSD) was

determined to be lower than 10%, for each cation. ICP-AESanalyses were carried out, using a Perkin-Elmer Optima 3000spectrometer, complementarily to the ion chromatographyanalyses in order to determine whether other cations werepresent in relatively high concentration in the rinsing water.The CO3

2− concentration was calculated from the rinsingsolution pH using the following formula: [HCO3

−] = −log(11.24-pH). The insoluble part of the deposit was assessedusing the difference between the mass of the deposit and themass of the soluble fraction.

The hydration level of glasses was determined in anondestructive manner by using Fourier transform infrared(FTIR) spectrometry. The spectrometer (Perkin-Elmer FTIR2000) is equipped with a 4,000 à 370 cm−1 DTGS detector.The FTIR spectrometer measures the OH absorption band.The comparison of the absorption (ΔE) at 3,300 cm−1

before and after exposure was used to quantify the OHvariation due to the hydration of the glass and to thecrystalline water (Fuchs et al. 1991).

3 Results and discussion

As the main purpose of this article is to quantify theweathering of potash–lime medieval-like glass (Si–K–Ca)in the early stages of exposure, results will be presentedusing Si–Na–Ca glass as a reference of the sole atmosphericdeposition.

3.1 Sheltered samples

On the sheltered samples, hydration of the subsurface of theglass over the first 15 months of exposure was 3 timeshigher on Si–K–Ca than on Si–Na–Ca (Table 2). A similarremark can be made for mass variation as an increase insample mass was observed, on average 24% lower on theSi–Na–Ca glass (Table 2). Hydration of the glass implies thediffusion of hydrated species in the glass matrix and clearlyindicates the process of leaching; the migration of glass-borne cations results from this interaction. Observation of theglass surface showed the presence of particles, and over thefirst 2 years of exposure, the %CS of the Si–K–Ca glasses(Table 2) was found to be on average 2.4 times superior tothe one of Si–Na–Ca glasses. SEM observation confirmed

this tendency as the number of deposited particles washigher on the Si–K–Ca glasses. After a year of exposure, theoverall particle load on the surface of the sheltered Si–K–Caglasses was 3.1 times greater than on the Si–Na–Ca glass(Fig. 1) and greater for every particle fraction studied. It has tobe noted that the number of deposited particles on the glassesalso increased with time, especially in the fine (ϕ<1 μm) andmedium (ϕ<10 μm) fractions. The surplus in mass, thehigher %CS and the number of particles on the Si–K–Caglass is attributed to the presence of soluble salts, confirmingthat leaching is more strongly affecting these glasses thanSi–Na–Ca ones. This mechanism is induced by the presenceof condensation water at the glass surface.

Morpho-chemical analyses of the deposit, performed bySEM+EDS, showed that K and K+Ca-rich species weremore abundant on the Si–K–Ca glass: spindle-shapedcrystal characteristic of syngenite and prismatic crystalsusually associated with gypsum were observed on the glass.Those findings are in agreement with previous studies(Munier et al. 2002a). Furthermore, early micro X-raydifraction (μXRD) results seem to confirm the mineralogy.

In order to confirm the nature of the different types ofdeposit and to assess their contribution to sample massvariations, the soluble fraction (salts) of the deposition wasinvestigated by IC and ICP-AES. It represented 79±13% ofthe total deposited mass for the Si–K–Ca glasses against 34±11% for Si–Na–Ca (Fig. 2). On Si–K–Ca glasses, the cationscomposing those salts were confirmed to be K and Ca, themajor modifier cations leached from the underlying glass.On Si–Na–Ca glasses, the calcium is the most abundantcation and, due to the limited leaching of the glass matrix, isvery likely to have an atmospheric origin. Concerning theanion content, analyses showed that on Si–K–Ca glasses,SO4

2− was the major anion (22±8%), followed by HCO3−

(16±7%), NO3− (6±1%) and finally Cl− (2±0.5%). On Si–

Na–Ca, HCO3− is the most abundant anion (9±5%) directly

followed by SO42− and NO3

− (7±3% each).The concentration of the anions on the glass also reflects

the respective reactivities of the atmospheric gases towardsthese glass surfaces. Different studies listed further belowhave shown that atmospheric gases are able to dissolve insmall quantities of water (dew, condensation water oradsorbed water) present on various monument surfaces.The more the material surface attracts water, the higher thereactivity of the gases towards the material (Camuffo andGiorio 2003). Nitrates and sulphates are known to bereactive on building material (Hamilton 1995; Kumar 2005)and the greater reactivity of HNO3 as compared to them ofNOx has been highlighted by Ferm (2005). As glassmaterial has a very high-energy surface, its contact anglewith water is perfect (0°), hence a high wettability. Glassreactivity towards vapour water, and other atmosphericgases, can therefore be very high. Indeed, this reactivity


has been assessed by comparing the anion concentrationsfound on the glass surface to the respective concen-trations in the surrounding air (Fig. 3). Results showedthat on the Si–K–Ca glass, SO4

2− and NO3− displayed

good correlations with SO2 (R²=0.92) and HNO3 (R²=0.87), respectively. For the Si–Na–Ca glasses, the corre-lation coefficients of the sulphate and nitrate and theirrespective atmospheric precursors are still high (0.61 forboth), which confirms the reactivity of both SO2 andHNO3 towards any glass. The correlation coefficients arenevertheless higher for the Si–K–Ca glasses due to theformation of neocrystallisation on the surface, which bindsmore efficiently the gases to the glass. The fluxes of bothgases on the Si–K–Ca glass were hence calculated:0.48 μmolcm−2s−1 for SO2 and 0.20 μmolcm−2s−1 forHNO3. In comparison, the reactivity of the SO2 towardslimestone is lesser with a flux of 0.26 μmolcm−2s−1. Thevalues of deposition velocity calculated for the glasses areconsistent with the rate commonly found for SO2 andHNO3 in the natural environment, when the depositionvelocity is mostly influenced by the surface resistance(Rc), which makes deposition a substrate-limited deposi-tion (Kumar et al. 2008).

3.2 Unsheltered samples

Unsheltered samples presented evidence of leaching as wellas congruent dissolution. Both processes induce a loss ofmaterial. Leaching provokes a subsurface hydration ofglasses. This one is 10 times higher for the Si–K–Ca, thenfor Si–Na–Ca (Table 2; average value for the 15 firstmonths of exposure). On Si–K–Ca glass surface, irisationswere observed, after 4 months of exposure and increasingwith time (Fig. 4b). A three-dimensional network of cracksdeveloped, as the leaching increased, isolating portions ofthe glass (scales) that were periodically lost (Fig. 5a and c).Chemical modifications of the glasses were detected at theuppermost surface, highlighting a cations depletion gradient(Table 3). Loss of scales and cation–proton diffusion in theleached layer resulted in an important mass loss (Table 2).This loss was not compensated by the addition of materialto the glass, such as gas (neocrystallisations) and particledeposition (soiling), since those were washed away byrunning water (open system). No evidence of scaling wasfound on Si–Na–Ca glass, whose mass variation is minute(Table 2).

On Si–K–Ca, pits were found underneath the scales(Fig. 5b). Congruent dissolution, being a precursor of the

Fig. 1 Morphology of the deposits observed on the glass Si–Na–Ca(a) and Si–K–Ca (b) (scales=50 μm), and particle number (c) after12-month exposure assessed by SEM

Table 2 Values of the hydration (ΔE) given by infrared (IR) spectrometry, the mass variation and the percentage of covered surface (%CS) of Si–K–Ca and Si–Na–Ca glasses, in sheltered and unsheltered conditions, averaged over the first 15 months of exposure

IR (ΔE) Mass variation (μgcm−²) %CS

Si–Na–Ca Si–K–Ca Si–Na–Ca Si–K–Ca Si–Na–Ca Si–K–Ca

Sheltered 0.008±0.006 0.024±0.014 110±22 145±20 8.7±2.5 21.2±3.3

Unsheltered 0.006±0.008 0.059±0.010 −9±30 −1134±29 3.2±1.0 4.7±1.4


pit formation, the occurrence of alkaline pH, was investi-gated. Rainwater did not seem to have a major role as onlyeight rain episodes (over 167 events in the first year ofexposure) had a rain pH superior to 7. Therefore, theformation of pits can be explained as follows: waterinfiltrates in cracks, and when it reaches the base of thescales, it is submitted to specific conditions (isolatedsystem); the leaching of the surrounding glass depletes thewater in protons; the water pH rises and congruentdissolution can take place. In such condition, congruentdissolution is a direct consequence of leaching. Since thetwo mechanisms are closely related here, their specificcontributions could not be accounted for independently.

3.3 Exposure comparison

3.3.1 Influence of glass composition on weathering

Comparing the two exposure conditions, the impact of wetdeposition is highlighted by the higher hydration values inthe unsheltered mode (Table 2). The observation of theirisated surfaces on unsheltered samples and their absenceon sheltered ones (Fig. 4) gives a good appreciation of theextent of the surface hydration. The intensity of irisationalso increases with time from 4 to 12 months on theunsheltered samples, although the degree of hydrationseems to be restricted by the scaling. This induces the

uncovering of a fresh and less hydrated surface of the glass(Fig. 5). Results show that the glass composition influencesthe behaviour of the two model glasses as a higherdegradation is always found on the Si–K–Ca glass whencompared to the Si–Na–Ca glass. Quantification of theeffect of the chemical composition on the deterioration ofglass has highlighted that in sheltered conditions, weather-ing of Si–K–Ca glass is 3 times higher than the decay ofSi–Na–Ca glass. Even more so in unsheltered conditions,deterioration of Si–K–Ca glass is up to 126 times higherthan the weathering of Si–Na–Ca glass (Table 2).Although chemical composition plays an important roleon the deterioration of glass (especially Si–K–Ca glass),discrepancy between the two exposure modes in Table 2shows that the environmental conditions are crucial ondefining glass decay. For example, weathering of Si–K–Caglass in unsheltered conditions is 3 times higher than insheltered ones.

3.3.2 Influence of abiotic extrinsic factors on weathering

Evaluation of the contribution of the different atmosphericfactors inducing glass weathering is done, considering themass parameter, by comparison of the different types ofglass and exposure. Indeed, mass is commonly affected byany deterioration phenomena and is an easily measurablevalue. A mass loss is induced by depletion of cations due to

Fig. 2 Average composition of the soluble fraction of the deposits over the 15 first months of exposure for Si–K–Ca (a) and Si–Na–Ca (b)

Fig. 3 Concentration ofSO4

2− (a) and NO3− (b) on the

glass over the atmosphericconcentrations of the relativeatmospheric precursor SO2, andHNO3, cumulated over theexposure time


leaching, congruent dissolution and scaling, whereas a massgain results from particle deposition and neocrystallisationgrowth (addition of atmospheric gases). Within the experimenttime frame, the evolution of the glass hydration and thedeposition of particle were found to follow a linear progression.Thus, the weathering can be considered here a steady-stateprocess; an evaluation of relative influence of specificparameter could be performed.

A simple mathematical computation has been thus set upto quantify the contribution of environmental parameters:wet deposition (rainwater) and dry deposition (condensa-tion water, particles and gases). First, it must be remindedthat, even though leaching is affecting the Si–Na–Ca glass(Table 2), it is acting in such a minute propensity(extremely low mass variation and hydration) that, whencompared to the Si–K–Ca glass, it is possible to assume itreflects the direct impact of particles on the deterioration(soiling). Thus, the particle contribution to degradation (Cp)

was approximated by the mass added to the Si–Na–Caglasses in sheltered condition:

Cp ¼ mSi�Na�Ca sð Þa � mSi�Na�Ca sð Þ

b ð1Þwhere Cp = the contribution of the particles, mb = the massbefore exposure, ma = the mass after exposure, and (s)denotes sheltered samples.

The gaseous pollutants contribution (Cg) is calculatedconsidering the mass of the anions present in theneocrystallisations as issued from the atmospheric gasdeposit:

Cgx sð Þ ¼

XXi½ �Vð Þx � Xi½ �Vð ÞSi�Na�Ca

n oð2Þ

where Cg = the contribution of the gases, x = the glasstype (Si–Na–Ca or Si–K–Ca), [Xi] = concentration in μg/L,of the anion i in the rinsing solution, and V = the volume ofrinsing solution.

Fig. 5 SEM photographs of the unsheltered Si–K–Ca glass surface (a) (scale=50 μm), detail of a gap created by a scale loss (b) at 6-monthexposure (scale=5 μm), cross-section detail of the leached layer and cracks (c) at 12-month exposure (scale=10 μm)

Fig. 4 Optical microscopy observation, in reflection mode, of the irisated surface of Si–K–Ca glasses in a sheltered and b unsheltered conditions


As in sheltered conditions, leaching is conditioned bythe dry deposition (adsorbed condensation water), the massof the cations from the neocrystallisations can be used toassess the contribution of this water (Caw):

Cawx sð Þ ¼

XXj

� �V

� �x � Xj

� �V

� �Si�Na�Can o

ð3Þ

where Caw = the contribution of the adsorbed condensationwater, [Xj] = concentration, in μg/L, of the cation j in therinsing solution.

Durable glass scarcely reacts with adsorbed water orgases (as testified by a very low hydration level); hence, theeffects of those two were not taken into account. Conse-quently, this calculation will lead to a null value for Caw

and Cg on this particular glass.As leaching of glasses exposed in unsheltered condition

is strongly related to the presence of rainwater at theirsurface, the contribution of rainwater (Crw) can beestimated through the mass lost during the exposure, thatis to say the difference between the mass before and afterexposure minus the mass of particles remaining on thesurface [mrp; Eq. 4]. This last mass is calculated assumingthat the effect of the natural rinsing of the surface (rainrunoff) is the same as the effect of rinsing performed at thelaboratory. As the latter is being possibly more effectivethan the former, the values obtained with Eq. 5 mightunderestimate the impact of the particles, but in a lesserextent, than if they were completely neglected:

Crwx uð Þ ¼ mb

x uð Þ � max uð Þ � m rp ð4Þ

where Crw, = the contribution of the rainwater, and (u)denotes unsheltered samples;

mrp ¼ mrx sð Þ � mb

x sð Þh i

þX

Xj

� �V

� �x ð5Þ

where mr = the sample mass after rinsing of its surface.For both model glasses, the contribution of wet and dry

deposition factors to mass variation (value after 15 months ofexposure) is presented in Fig. 6a and b. As previouslyobserved, the deterioration of the Si–K–Ca glass is driven bythe presence of running water on its surface, with an 83.4±2.6% contribution to the mass variation of the model glassover the first 15 months. Nevertheless, the contribution ofthe dry deposition of gases and adsorbed water (6.4±1.5%

and 3.4±0.6%, respectively, of the mass variation) must notbe neglected, showing that leaching is also active in thesheltered samples. Particle deposition, that is to say soiling,is accounting for 6.8±2.2% of the deterioration. On the Si–Na–Ca glass, the trend reverses and soiling is the leadingdeterioration phenomena. Nevertheless, rainwater still playsa strong role to play in the deterioration with almost a thirdof the contribution (32.0±13.4%).

4 Conclusion

The study of medieval-like model glasses, exposed toatmospheric conditions, led to the evaluation of the role ofboth the composition and the surrounding environment onthe potash–lime glass weathering in the early stages ofexposure.

As expected, independently of exposure conditions, thelow-durability glass (Si–K–Ca) always presents a greateralteration than the durable one (Si–Na–Ca). In shelteredconditions, for each parameter used and assuming that Si–

Fig. 6 Contribution of the four atmospheric deterioration factors (Cp forparticles, Cg for gas, Caw for adsorbed water and Crw for rainwatercontributions), to the mass variation, after 15 first months of exposure(a) and their relative percentage contribution for each model glass (b).n.a. Data nonavailable through the calculation

Table 3 Composition of the Si–K–Ca glass after 6 months exposure at different depths, in the leached layer and of the pristine glass (in oxide %),given by SEM-EDS

SiO2 Al2O3 K2O MgO CaO P2O4

Upper leached layer (scale) 73±3 5.7±0.3 10.8±0.5 0.7±0.05 9.5±0.5 n.a.

Lower leached layer (gap) 59±3 2.4±0.1 20.6±1.0 2.1±0.1 16±1 n.a.

Pristine Glass 53±3 1.8±0.1 24.5±1.2 2.7±0.1 17±1 n.a.

n.a. P peak overlapping with coating peak


Na–Ca is almost inert, the impact of the glass compositioncan be estimated to be responsible for two thirds of thedegradation. No general trend is found in unshelteredconditions, the impact of the chemical composition rangesfrom 2 times (for the %CS) to 126 times (for the massvariation). The difference between those deterioration ratiosin the two exposure conditions shows that the environmentalconditions, and especially wet deposition, enhance the role ofthe chemical composition on the glass weathering.

A mathematical computation led to the calculation of thecontribution of both wet and dry deposition. The presenceof running water drives the Si–K–Ca glass deterioration asit is responsible for ~83% of the weathering. The impact ofleaching in dry exposure conditions is ~10% and imputableto the action of gaseous deposition (~6%) and thecondensation water (~4%). The direct effect of particles(soiling) on Si–K–Ca glass degradation is ~7%. On the Si–Na–Ca glass, particle deposition is directly responsible for~68% of the deterioration. The only other contribution tothe weathering of Si–Na–Ca is due to rainwater, whichinduces ~32% of the glass mass variation, whether byleaching or chipping of glass. Therefore, although to amuch lesser extent than with Si–K–Ca glasses, leaching canaffect also Si–Na–Ca glass. Note that this computation wasperformed assuming that after 15-month exposure, thedeterioration processes (soiling and hydration) present asteady-state evolution.

Even when SO2 atmospheric concentration severelydropped during the last decades, it is the first contributorto “dry leaching” of low-durability glasses with the highestdeposition rate. This phenomenon results from the highreactivity of this gas towards surfaces. Its presence in theatmosphere is thus still critical to the deteriorationprocesses happening on the stained glass windows and itsmonitoring has to be carried on.

Whether glasses are ancient and highly sensitive ormodern and durable, the impact of each pollutant isdeterminant for the appreciation of deterioration and mustnot be minimised or neglected. Those results raise thequestion of the indirect effect of particle deposition on thedeterioration of the glass. Further studies will focus ondetermining their impact on glass wettability and theiractive role in leaching and congruent dissolution.

The most commonly found protection system for thestained glass windows is the double-glazing windows. Theyprotect the glasses from the impact of running water. It will bethus important to adapt the conservation strategies in order toprotect glass from both wet and dry degradation. This paperhighlights that the impact of the “dry” leaching is not to beneglected. In that perspective, other complementary measuresof conservation must be taken. Focusing on this idea,preventive conservation (double glazing) is necessary but willneed the backing of regular specific maintenance of stained

glass windows in all the historical monuments in order tocounteract the effect of dry deposition.

Acknowledgements The authors would like to thank the FrenchMinistry of Culture for funding this research via the “national researchprogram for the comprehension and conservation of cultural heritagematerials” (PNRCC). The authors would like to thank M. Verità of theUniversity IUAV of Venice, Laboratorio Analisi Materiali Antichi.

References

Bettembourg JM (1976) Composition et altération des verres devitraux anciens. Verres Réfract 30(1):36–42

Boksay Z, Bouquet G, Dobos S (1968) The kinetics of the formation ofleached layers on glass surface. Phys Chem Glasses 9(2):69–71

Brinker CJ (1988) Hydrolysis and condensation of silicates: effects onstructure. J Non-Cryst Solids 100:31–50

Camuffo D, Giorio R (2003) Quantitative evaluation of water depositedby dew on monuments. Boundary-Layer Meteorol 107:655–672

Casey WH, Bunker B (1990) Leaching of mineral and glass surfacesduring dissolution. In: Hochella MF Jr, White AF (eds) Mineral–water interface geochemistry. Mineral Soc Am 23:397–426

Das CR (1969) Theoretical aspects of the corrosion of glass. The glassindustry. September:422–427

Desboeufs KV, Losno R, Colin JL (2003) Figures of merit of pneumaticand ultrasonic sample introduction systems in inductively coupledplasma-multichannel-based emission spectrometry in an ultra-cleanenvironment. Anal Bioanal Chem 375(4):567–573

Doremus RH (1975) Interdiffusion of hydrogen and alkali ions in aglass surface. J Non-Cryst Solids 19:137–144

Doremus RH (1979) In: Tomozawa M, Doremus RH (eds) Treatise onmaterials science and technology, vol 17. Academic, New York,pp 41–69

Douglas RW, El-Shamy TM (1967) Reactions of glasses with aqueoussolution. J Am Ceram Soc 50:1–8

Ferm M (2005) Nitric acid measurement in connection with corrosionstudies. Atmos Environ 39:6664–6672

Fuchs D, Römich H, Schmidt H (1991) Glass sensors: assessment ofcomplex corrosive stresses in conservation research. Mat Res SocSymp Proc 239–251

Gillies KJS, Cox A (1988) Decay of medieval stained glass at York,Canterbury and Carlise: Part 1. Composition of the glass and itsweathering products. Glastech Ber 61:75–84

Hamilton RS (1995) Sulphur and nitrogen particulate pollutantdeposition on to building surface. Sci Total Environ 167:57–66

Hench LL (1975) Characterisation of glass corrosion and durability. JNon-Cryst Solids 19:27–39

ISO 8565 (1992) Metals and alloys—atmospheric corrosion testing—general requirements for field tests 90.92 TC 156 vegetation. IndJ Radio Space Phys 37:272–278

Kumar R (2005) Atmospheric dry deposition to marble and red stone.J Atmos Chem 50:243–261

Kumar R, Srivastava S, Maharaj Kumari K (2008) Modellingdeposition of S and N compounds on vegetation. Indian J RadioSpace Phys 37:272–278

Lefèvre RA, Grégoire M, Derbez M, Ausset P (1998) Origin ofsulphated grey crusts on glass in polluted urban atmosphere:stained glass windows of Tours Cathedral (France). Glastech BerGlass Sci Technol 71:75–80

Lefèvre RA, Ausset P (2002) Atmospheric pollution and buildingmaterials: stone and glass. In Natural stone, weathering phenomena,conservation strategies and case studies. Geol Soc London SpecialPubl 205:329–345


Lombardo T, Ionescu A, Lefevre RA, Chabas A, Ausset P, Cachier H(2005) Soiling of silica–soda–lime float glass in urban environment:measurements and modelling. Atmos Environ 39:989–997

Melcher MS (2007) Quantification of the influence of the atmosphericpollution on weathering of low-durability potash–lime–silicaglasses. Pollution atmosphérique. Octobre:32–36

Melcher MS, Schreiner M (2004) Statistical evaluation of potash–lime–silica glass weathering. Anal Bioanal Chem 379:628–639

Munier I, Lefèvre RA, Geotti-Bianchini F, Verità M (2002a) Influenceof polluted urban atmosphere on the weathering of low durabilityglasses. Glass Technol 43(6):225–237

Munier I, Lefevre RA, Losno R (2002b) Atmospheric factorsinfluencing the formation of neocrystallisations on low durabilityglass exposed to urban atmosphere. Glass Technol 43(C):114–124

Newton RG (1984) The durability of glass—a review. Glass Technol26(1):21–38

Newton RG, Bettembourg JM (1976) Effects of repeated condensation.CMVA News Lett 22:9–10

Paul A (1977) Chemical durability of glasses; a thermodynamicapproach. J Mater Sci 12:2246–2268

Perez y Jorba M, Dallas JP, Collongues R, Bahezre C, Martin JC(1978) Etude de l’altération des vitraux anciens par microscopieélectronique à balayage et microsonde. Silicate Industriels 89–99

Römich H (2000) Simulation of corrosion phenomena of historicalglasses on model glasses: an update. Riv StazSper Vetro 6:9–14

Schreiner M (1988) Deterioration of stained medieval glass byatmospheric attack: Part 1. Scanning electron microscopicinvestigation of the weathering phenomena. Glastech Ber GlassSci Technol 61(7):197–204


early stage of weathering of medieval-like potash–lime model glass: evaluation of key factors

Documents