AGING OF MEDIEVAL-LIKE MODEL GLASS

L. DE FERRI1, D. BERSANI2, Ph. COLOMBAN3, P.P. LOTTICI2, G. SIMON3, G. VEZZALINI1

1Dipartimento

Introduction

di Scienze della Terra, Universit di Modena e Reggio Emilia, 41121 Modena, Italy lavinia.deferri@unimore.it 2Dipartimento di Fisica, Universit di Parma, 43124 Parma, Italy 3UPMC Universit Paris 06, UMR 7075, Laboratoire de Dynamique, Interactions et Ractivit (LADIR), F-75005, Paris, France

Ash-based glass (potash-lime-silica, PLS glass), where K is the main fluxing agent and Ca the main stabilizer, is typical of the cathedral medieval windows in Northern Europe since 1000 A.D. [1,2]. It is very sensitive to alteration phenomena due to the attack of atmospheric pollutants conveyed by water [3,4]. In the alteration process the former leaching step gives a de-alkalinized layer, enriched in silica and H+, then the dissolution of the silica network may occur through an hydrolysis reaction due to the pH increase of the water film. Weathering crystalline products can be identified on the surface after the water evaporation. Three glass samples, representative of the PLS medieval glass, with about 12, 17 and 23% of K2O in V1, V2 and V3, samples, respectively [5] were prepared. Aim of this work is the determination of the influence of K amount on the glass alteration degree.

Stained glass windows from the St. Denis Abbey in Paris (details)

Chemical composition
SiO2 CaO MgO K2 O P2O5 Na2O Al2O3 TiO2 MnO Fe2O3 V1 61.11 18.50 3.82 12.29 1.67 0.38 1.99 0.09 0.03 0.15 V2 57.07 18.64 3.57 16.74 1.62 0.67 1.38 0.06 0.02 0.22 V3 53.10 17.53 3.24 23.49 1.52 0.26 0.73 0.03 0.02 0.09

Glass characterization
FT-IR ATR Raman
Raman spectra analyzed on the basis of the Qn model [6] V1 Sample cm-1 V1 774 V2 767 V3 756 Sample cm-1 V1 874 V2 884 V3 893

Sample cm-1 V2 1004 V3 1051

Q1 and Q2 configurations increase at the expense of Q3 and Q4 with increasing K content

Ip is nearly equal in the three samples ~ 0.3

Greater effect of the melting temperature than the K content

With increasing K content Weakening of the Si-O bonds Network depolymerization

In boiling concentrated sulfuric acid (from an hour to two weeks)

120 m V3-60 min V3-300 min The alteration rate for V2 after 4 After 6h hours aging is nearly constant at ~ exposure V3 21 m/h sample was The alteration rate for V3 completely gradually increases for the first 3 altered hours up to ~ 227 m/h 750 m

Aging experiments: H2SO4 ion exchange

Aging experiments: H2O attack

Bidistilled water in autoclave: 300C 80 bars V1-V2 = 2 weeks V3 = 1 week

Alteration layer Alteration rate Glass weight Extracted Final water Sample thickness (m) (m/h) loss (%) K2O (%) pH V1 357 1.06 53 3.8 10.0 V2 586 1.74 58 9.0 10.1 V3 523 3.11 67 13.6 10.7 K release increases with the glass K content. The pH increase causes the breakdown of the network forming Si-O bonds and the glass dissolution The alteration rate increases with the glass Kcontent increases

V3-360 min

Raman linear map on the V3-180 min cross section

60000

Ca-Sulphates (gypsum + bassanite) were identified as alteration products

50000

30000

Intensity (a.u.)

40000

20000

10000

0 500 1000

Wavenumber(cm-1) 1500

2000

Ca-sulphates crystals aggregate Points of the V2-H2O Raman map

Wavenumber (cm-1)

Time (s )

Time (s )

Time (s )

AAS analysis of H2SO4: K release increases with the glass K content .

Increase of Ip value in aged glass [6]. During aging the less bound tetrahedra (Q0 and Q1) are released together with the modifier ions. The area of the stretching band decreases giving higher Ip value. The increase of polymerization is only apparent.

Conclusions
In both acid and water aged glass the alteration degree is directly dependent on the glass K content. The alteration rate (modifier ions release, thickening of alteration layer, weight loss) is higher for K-based glass than Na-based ones [7] Studies on medieval glasses artificially aged following similar procedures in H2SO4 found an average alteration rate of 70 m/h for potassium-based samples [8]: in our case for V2 the alteration rate appears quite constant during the exchange and lower than the literature values. On the other hand, V3 samples show higher alteration rates confirming the degradation dependence on the glass composition. In the acid attacked glass, crystallization of gypsum and bassanite is observed, while in water attacked glass gyrolite is found.
References
[1] O. Schalm, K. Janssens, H. Wouters, D. Caluw, Composition of 1218th century window glass in Belgium: Non-figurative windows in secular buildings and stained-glass windows in religious buildings,Spectrochimica Acta Part B 62 (2007) 663668 [2] O. Schalm, I. De Raedt, J. Caen, K. Janssens, A methodology for the identification of glass panes of different origin in a single stained glass window: Application on two 13th century windows, J. Cult. Herit. 11 (2010) 487492 [3] M. Melcher, M. Schreiner, Leaching studies on naturally weathered potash-lime-silica glasses, J. Non-Cryst. Solids 352 (2006), 368-379 [4] M. Melcher, R. Wiesinger, M. Schreiner, Degradation of glass artifacts: application of modern surface analytical techniques, Acc. Chem. Res. 43 (2010) 916-926 [5] L. De Ferri, D. Bersani, A. Lorenzi, P.P. Lottici, A. Montenero, S. Quartieri, G. Vezzalini, Conservazione e restauro di vetrate antiche: dati preliminari sulla riproduzione di vetri medievali, Proceedings VI National Congress of Archaeometry (Aiar)Scienza e Beni culturali, Pavia, 2010 February 15-18, accepted. [6] Ph. Colomban, M.P. Etcheverry, M. Asquier, M. Bounichou, A. Tourni, Raman identification of ancient stained glass and their degree of deterioration, J. Raman Spectrosc. Vol. 37 (2006) 614-626 [7] J. Sterpenich, G. Liburel, Using stained glass windows to understand the durability of toxic waste matrices, Chem. Geol. 174 (2001) 181.193 [8] A. Tourni, P. Ricciardi, Ph. Colomban, Glass corrosion mechanism: a multiscale analysis, Solid State Ionics 179 (2006) 2142-2154

Intensity (a.u.)

Ip = 0.6