See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/233678980

Influence of Calcium on Sodium Aluminosilicate Glass Leaching Behaviour

Article  in  Physics and Chemistry of Glasses · August 2001

CITATIONS READS

22 285

6 authors, including:

Frédéric Angeli Stephane Gin

Atomic Energy and Alternative Energies Commission Atomic Energy and Alternative Energies Commission
54 PUBLICATIONS   1,539 CITATIONS    179 PUBLICATIONS   4,354 CITATIONS   

SEE PROFILE SEE PROFILE

Bruno Boizot
Atomic Energy and Alternative Energies Commission
105 PUBLICATIONS   1,484 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Archeological materials View project

Nuclear Waste Materials View project

All content following this page was uploaded by Stephane Gin on 15 January 2015.

The user has requested enhancement of the downloaded file.

 Journal of Non-Crystalline Solids 358 (2012) 2951–2960

Contents lists available at SciVerse ScienceDirect

Journal of Non-Crystalline Solids

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

Impact of soda-lime borosilicate glass composition on water penetration and water

structure at the first time of alteration
D. Rébiscoul a,⁎, F. Bruguier a, V. Magnin b, S. Gin a
a
CEA, DEN, (DTCD/SECM/LCLT)—Marcoule, F-30207 Bagnols-sur-Cèze Cedex, France
b
CNRS, ICSM—UMR 5257 CEA-CNRS-UMII-ENSCM, LDD, 30207 Bagnols-sur-Cèze Cedex, France

a r t i c l e i n f o a b s t r a c t

Article history: In this study, the impact of soda-lime borosilicate glass composition and particularly the effect of charge com-
Received 21 March 2012 pensators such Ca and Na and, of network formers such Si and Zr, on water penetration and water structure at
Received in revised form 6 July 2012 the first time of alteration were investigated. Two non-destructive techniques were combined: the Fourier
Available online 28 August 2012
transform infrared spectroscopy in attenuated total reflection geometry to precise the predominant alter-
ation mechanisms and assess the water structure in altered zone and the grazing incidence X-ray reflectom-
Keywords:
Borosilicate glass;
etry to determine the thickness of the altered glass zone allowing to calculate the water diffusion coefficients
Water penetration; through the glasses. The results of glass alteration at pH = 3 and 30 °C have shown that hydrolysis was the
X-ray reflectivity; predominant mechanism after few seconds for glass having a high amount of non-binding oxygen. For the
Attenuated total reflection infrared other glasses, which for the diffusion was the limiting reaction, the calculated water diffusion coefficients
spectroscopy; were comprised between 10−21 and 10 −19 m².s −1 and vary as a function of glass composition. An activation
Water diffusion coefficient energy of 76.9 kJ.mol −1 was calculated and appears to be higher than inert gas diffusion through the glass
highlighting that water molecules strongly interact with the glass matrix.
© 2012 Elsevier B.V. All rights reserved.

1. Introduction
Understanding the interactions of water with glass surface is of
great interest to improve glass alteration models and more particular-
ly the models of the long term behavior of the nuclear borosilicate
glass in deep geological repository, as water is expected to corrode
glass and allow the migration of soluble radionuclides [1].
When borosilicate glasses are in contact with water, several mecha-
nisms occur simultaneously: (i) the hydration of the glass and the inter-
diffusion process corresponding to an exchange between the protons of
the solution and the alkaline species of the glass [2]; (ii) the hydrolysis
of the ionic-covalent bonds of the matrix [3]; (iii) the condensation of
species dissolved in solution such Si, Al and Ca leading to the formation
of an hydrated and porous material called gel [3,4]; (iii) the precipita-
tion of crystalline phases [5,6]. Then, at high reaction progress, when
the glass is altered at a so-called residual rate [7], the glass alteration
layer is stratified and consists in a de-alkalized and hydrated glass
zone, a porous gel with crystalline phases on top and, in some cases, em-
bedded in the porosity of this gel. At this high reaction progress
Grambow et al. [8] and McGrail et al. [9], have proposed hydration
and ionic exchange as the underlying mechanisms at the origin of the
residual rate. This hydrated and de-alkalined glass zone was recently
brought to light experimentally by Gin et al. [10] as a result of the char-
acterization of SON68 altered during nearly 26 years at 90 °C in a
⁎ Corresponding author. Tel.: +33 4 66 39 79 40; fax: +33 4 66 79 66 20.
E-mail address: diane.rebiscoul@cea.fr (D. Rébiscoul).
confined granitic medium. Because these two reactions have to be in-
cluded in the model used to predict the long-term rates of nuclear
glass in disposal environment, we have to study these mechanisms in
detail and particularly understand the influence of parameters such
the pH, the temperature, the ionic strength, the glass composition.
Most of the studies about the glass hydration were essentially
performed on simplified [11,13,14], natural [15,16] or archeological
[17] glasses. In this study, we have chosen soda-lime borosilicate
glasses containing zirconium and having composition close to the
French R7T7 reference nuclear glass which for the structure [18]
and the alteration were already studied in details (dissolution rates
and morphologies of the alteration layers at pH higher than 7)
[19–21]. The effect of network formers, replacing Zr to Si, and the ef-
fect of the charge compensators, substituting Na to Ca, on the water
penetration through the glasses were investigated at pH of 3 and at
the first time of the alteration.
Glass alteration is mainly studied using solution analysis and the al-
tered glass is generally characterized using volume techniques (infrared
spectroscopy, small angle X-ray scattering, X-ray absorption…), de-
structive techniques requiring specific preparation (scanning and trans-
mission electron microscopy, secondary ion mass spectroscopy…).
Meanwhile, at the first time of alteration, the elements content in solu-
tion are too low to be detected and the altered zone is too thin to be
characterized by the major part of the techniques previously men-
tioned. Consequently, the most commonly used technique to study
glass hydration is the Fourier transform infrared spectroscopy because
it is highly sensitive, it offers the possibility to distinguish between

0022-3093/$ – see front matter © 2012 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.jnoncrysol.2012.07.014
2952 D. Rébiscoul et al. / Journal of Non-Crystalline Solids 358 (2012) 2951–2960

different water-related species and it allows determining the water dif- Table 2
fusion coefficients [22,23]. More recently, X-ray reflectometry was used Mass ρm and electron ρe densities of glasses.

to study water penetration through simplified and complex nuclear-
type glasses [24–27]. For these reasons, we propose a method combining
these two non-destructive glass surface characterizations: the Fourier
transform infrared spectroscopy in attenuated total reflection geometry
(FTIR-ATR) to precise the predominant alteration mechanisms and assess
the water structure in altered zone and the grazing incidence X-ray re-
flectometry (XRR) to determine the thickness of the altered glass zone
allowing to calculate the water diffusion coefficients through the glasses
with the method already used in [27].
2. Materials and methods
2.1. Materials
2.1.1. Samples preparation
Two series of borosilicate glasses already studied [18] were elaborat-
ed. The glasses references and the molar compositions are presented in
Table 1. The first series is referred xCa and has a variable calcium con-
tent ranging from 0 to 9.1 mol% substituted by Na while conserving a
constant zirconium content. The second series is referred xZr and has
a variable zirconium oxide ranging from 0 to 7.9 mol% substituted
for silicon oxide while conserving a constant calcium concentration.
One additional glass sample (8ZrCa) was fabricated with lower sodium
content.
Glasses were elaborated mixing SiO2 powder (Millisil), H3BO3,
Na2CO3, CaO and ZrO2 (Prolabo) and melting in a platinum–gold cru-
cible at temperature ranging from 1300 °C to 1400 °C for 3 h, then
poured into graphite crucibles, and annealed for 1 h at 600 °C. Mono-
liths of these glasses measuring (25 × 25 × 2.5 mm 3), were cut from
the large glass bares and polished in ethanol to grade 4000 with SiC
paper. Monoliths were then ultrasonically cleaned in acetone and in
ethanol.
2.1.2. Glass alteration
Glass monoliths were altered at glass surface area-to-solution vol-
ume ratio (S/V) of 1 cm −1 in solution of ultrapure water having a pH
adjusted using HCl (Prolabo ‘Normapur’) at pH = 3 ± 0.1 measured at
the temperature of the experiment, i.e. 12 °C, 30 °C, 50 °C and 60 °C
in a PTFE (PolyTetraFluoroEthylene) reactor. Before alteration, the so-
lution was placed in an oven at the temperature of the experiment
during 24 h. Alteration durations were adjusted as a function of the
glass and shorter than 2 h. For each alteration, the same glass mono-
lith was taken from the reactor, rinsed with ultrapure water, dried at
the laboratory atmosphere, analyzed using X-ray reflectometry and
FTIR-ATR and then reimmersed in its leachate. This method allows
to analyze the same glass sample and to alter it in the same leachate.
2.1.3. Glass surface characterization
Altered glass surfaces were analyzed using a FTIR spectrometer
Bruker VERTEX 70 in ATR mode equipped with a DTGS/KBR detector.
The samples were placed at the surface of the diamond and pressed
Composition 0Ca 3Ca 6Ca 9Ca 0Zr 1Zr 4Zr 8Zr 8ZrCa
(mol%)
ρm (g.cm−3) 2.568 2.554 2.555 2.554 2.542 2.569 2.569 2.740 2.746
ρe (g.Å−3) 0.762 0.759 0.759 0.759 0.757 0.764 0.779 0.805 0.808
adding 32 scans with a 2 cm −1 of resolution correcting from the
background spectrum for each substrate. Three different locations
were analyzed on the monoliths. The pressure on the piston was ad-
justed until the stabilization of the spectra. In our experimental con-
dition, the incident IR beam through the diamond has a limited
penetration between 500 and 5000 nm through the glass which al-
lows us to characterize the totality of the altered zone. Baseline ad-
justments, normalization of spectrum and bands decomposition
were performed using the OPUS software. The O\H stretching band
deconvolutions were performed using Gaussian functions adjusting
the intensity and the width of the fitting curves. The “best fit” was
considered when the statistical parameter R was the lowest.
X-Ray reflectometry (XRR) was used to determine the thickness
and the density of the glass altered zone. Measurements were carried
out using a Bruker D8 diffractometer using CuKα1 (k = 0.154056 nm)
radiation and standard θ–2θ scan for the data collections. Step sizes
and count times were variable and chosen according to the intensity
measured during fast prescans in small angular intervals. Reflectivity
curves are generally presented as the evolution of the logarithmic of
intensity received by the detector as a function of the wave vector
transfer q in Eq. (1).
4π sinθ
q¼ ð1Þ
λ
with θ the incident angle.
When the sample is composed of several flat and homogeneous
layers on a substrate with sufficient density contrast, the reflectivity
curve presents some fringes due to constructive and destructive in-
terferences of waves reflected at the various interfaces. The distance
between two consecutive fringe minima is directly related to the
thickness of the layers, which can be calculated using a simple formu-
λ
la e ¼ 2Δθ [28]. But when the sample is composed of several layers
having rough interfaces and/or layers presenting a density gradient,
the fringes are attenuated or disappeared, and this simple analysis is
not applicable. Because of the complexity of the stacks characterized
1.6
1.4
1.2
1.0
Intensity (a.u.)

with a piston. The spectra were recorded from 400 to 4000 cm −1 0.8
0.6
0.4
Table 1 0.2
References and molar compositions of soda-lime borosilicate glasses. xZr series
0.0
Composition 0Ca 3Ca 6Ca 9Ca 0Zr 1Zr 4Zr 8Zr 8ZrCa
(mol%) -0.2
xCa series
SiO2 62.6 62.6 61.9 62.4 61.4 60.4 56.5 51.3 51.7 -0.4
B2O3 16.5 16.4 17.1 16.7 16.9 16.8 17.5 17.7 17.8 400 800 1200 1600 2000
Na2O 18.4 15.8 12.9 9.8 18.1 18.2 18.3 18.9 14.5 k (cm-1)
CaO 0.4 3.1 6.1 9.1 3.6 3.6 3.8 3.9 7.9
ZrO2 2.1 2.1 2.0 2.0 0 1.0 3.9 8.2 8.1
Fig. 1. ATR-FTIR spectra of all glasses.
 D. Rébiscoul et al. / Journal of Non-Crystalline Solids 358 (2012) 2951–2960 2953

in this study, experimental reflectivity curves were fitted using the with cj the element amount j in the material, Zj the atomic number, Aj
Firefx4c_6 software [29] based on the Parratt algorithm [30] adjusting the atomic mass and ρm (g. Å−3) the mass density. The mass and elec-
the electron densities ρe (e. Å−3) of n layers on a substrate. For our sam- tron densities of the glasses are presented in Table 2. Regarding the nu-
ples, we have used the model already proposed in [26] fitting correctly merous fitting results obtained in this study, we will only present the
the glass altered zone. This model takes into account three zones: an ex- total thickness of the altered zone, i.e. the sum of the thicknesses of
ternal zone having a low density and high roughness leading to a de- the three zones described previously.
crease of the global reflectivity intensity, a zone of low density and
small thickness and a zone having a density close to the density of the
pristine glass. The electron density of the substrate, i.e. the glass, is 3. Results
fixed and calculated from Eq. (2):
3.1. Structure of the glass network
∑ cj :Z j
j Comparing the FTIR-ATR spectra between 400 and 2000 cm − 1 of
ρe ¼ ρm ð2Þ
∑ cj :Aj two series of glasses presented in Fig. 1, it appears that the peak be-
j tween 600 and 1200 cm − 1 corresponding to the overlay of Si\O

1.6 1.6
0.2

9Ca
0.2

0Ca 1.4
Absorbance (au)

1.4

Absorbance (a.u)
0.1 0.1
1.2 ref, 10 to 40 min 1.2 ref, 10 min to 1h20

Absorbance (a.u)
Absorbance (au)

ref, 10 min to 1h20

(ref, 10 to 40 min)
1.0 0.04 1.0 0.0
0.0 0.04 2600 2800 3000 3200 3400 3600 3800
0.03
2600 2800 3000 3200 3400 3600 3800 -1
k (cm )
Absorbance (au)

0.8
-1
k (cm ) 0.8 0.03

Absorbance (a.u)
0.02 (a) 0.02
0.6 0.01 0.6
0.00
(ref, 10 to 40 min) 0.01
ref, 10 min to 1h20 (d)
0.4 1550 1600 1650 1700
k (cm-1)
1750 0.4 0.00
1550 1600 1650 1700 1750
-1
k (cm )
0.2 0.2

0.0 0.0
400 800 1200 1600 2000 2400 2800 3200 3600 4000 400 800 1200 1600 2000 2400 2800 3200 3600 4000
k (cm-1) k (cm-1)
1.6 0.2 1.6 0.2

3Ca 0Zr 20 min

Absorbance (a.u)

1.4

absorbance (a.u)
1.4 ref 10 min
0.1
1min30 0.1
1.2 ref, 20min to 2h 1.2 2min30
Absorbance (a.u)

ref, 20 min to 2h
absorbance (a.u)

2 min 30 1 min 30
1.0 0.04 0.0
2600 2800 3000 3200 3400 3600 3800 1.0 10min 0.0
ref
-1 2600 2800 3000 3200 3400 3600 3800
k (cm ) 0.04 -1
k (cm )
0.03
Absorbance (a.u)

20min 20min
0.8 0.8
absorbance (a.u)

0.02 (b) 0.03

0.6 0.01 ref, 20 min to 2h 0.6 0.02 10min

(e)
0.01
0.00 2min30 1min30
0.4 1550 1600 1650
-1
1700 1750 0.4 0.00 ref
k (cm )
1550 1600 1650 1700 1750
-1
k (cm )
0.2 0.2
0.0 0.0
400 800 1200 1600 2000 2400 2800 3200 3600 4000 400 800 1200 1600 2000 2400 2800 3200 3600 4000
k (cm-1) k (cm-1)

1.6 1.8 0.2

0.2

6Ca 1Zr
absorbance (a.u)

1.6
Absorbance (a.u)

1.4 ref
0.1
1.2 1.4 10 min 0.1 20 min
10 min
ref, 10 min to 1h38 ref, 10 min to 1h38
20 min
absorbance (a.u)
Absorbance (a.u)

1.2 ref
1.0 0.04
0.0
2600 2800 3000 3200 3400 3600 3800 0.0
2600 2800 3000 3200-1 3400 3600 3800
-1 0.04
k (cm ) 1.0 k (cm )
Absorbance (a.u)

0.8 0.03
absorbance (a.u)

0.03
0.02 0.8 20 min
0.02
0.6 0.01
(c) 0.6 10 min

0.4 0.00
ref, 10 min to 1h38 0.01
ref
(f)
1550 1600 1650 1700 1750
-1
k (cm )
0.4 0.00
1550 1600 1650-1 1700 1750
k (cm )
0.2 0.2
0.0 0.0
400 800 1200 1600 2000 2400 2800 3200 3600 4000 400 800 1200 1600 2000 2400 2800 3200 3600 4000
k (cm-1) k (cm-1)

Fig. 2. (a) to (i). ATR-FTIR Spectra of all glasses during the alteration at pH = 3 and 30 °C.
2954 D. Rébiscoul et al. / Journal of Non-Crystalline Solids 358 (2012) 2951–2960

1.6 0.2 3.2. Predominant mechanism and water structure in the glass altered

4Zr zone

absorbance (a.u)
1.4 ref 20 min
0.1
1.2 10 min 10 min
First, to determine the predominant mechanism of glass alteration,
absorbance (a.u)

20 min ref
hydration of the glass or hydrolysis of the matrix, and the water struc-
1.0 0.0
2600 2800 3000 3200 3400 3600 3800 ture in the altered zone, we were focused on the evolution of four
0.04 -1
k (cm )
0.8 bands: the peak located between 600 and 1200 cm−1 corresponding
absorbance (a.u)

0.03
20 min
to an overlay of Si\O stretching bands [31] and B\O vibration bands
0.02
0.6 10 min
(g) [32], the band at 1390 cm−1 corresponding to the B\O stretching vi-
0.01
bration of the borate structural units [33], the OH bending of free
0.4 0.00
ref
1550 1600 1650
-1
1700 1750 water and water hydrogen-bonded to the proton of silanol groups at
k (cm )
0.2 1649 cm−1 and the large band at 3000–3700 cm−1 corresponding to
OH stretching vibrations [11]. Water hydration was considered as a pre-
0.0
400 800 1200 1600 2000 2400 2800 3200 3600 4000 dominant mechanism when the glass network is not modified and
k (cm-1) when water content increases in the matrix. This glass surface modifica-
tion during the alteration is traduced on the FTIR-ATR spectra only by a
1.8 slight increase of the OH bending at 1649 cm−1 and O\H stretching
0.2
bands at 3000–3700 cm−1. As shown in Fig. 2(a) to (d), (h) and (i),
8Zr
absorbance (a.u)

1.6
the water penetration without matrix modification is predominant for
0.1
1.4 glasses 0Ca, 3Ca, 6Ca, 9Ca, 8Zr and 8ZrCa. In case of glass matrix hydro-
ref, 10 min to 50 min
lysis, the Si\O/B\O bands at 600–1200 cm−1 are modified and the
absorbance (a.u)

ref, 10 min to 50 min

1.2 0.0
0.04 2600 2800 3000 3200 3400 3600 3800
-1
O\H bending and stretching bands strongly increase. On the other
k (cm )
1.0 hand, when glass hydrolysis is the predominant mechanism, the rough-
absorbance (a.u)

0.03

0.8 0.02
ness of the glass surface was becoming too high, X-ray reflectivity
0.01
curves do not present some fringes (Fig. 6(e) to (g)). Taking into ac-
0.6 ref, 10 min to 50 min
count these criteria, we can conclude that 0Zr, 1Zr and 4Zr glasses alter-
0.00
1550 1600 1650 1700 1750
(h) ation is dominated by hydrolysis.
0.4 -1
k (cm )
To better highlight the behavior of these two types of glasses, the
0.2
evolution of the O\H stretching peak areas vs. time was plotted and
0.0 was presented in Fig. 3. It appears that glasses of the xCa series, 8Zr
400 800 1200 1600 2000 2400 2800 3200 3600 4000 and 8ZrCa present a low alteration whereas glasses 0Zr, 1Zr and 4Zr
k (cm-1)
show an important alteration.
We have more precisely investigated the large band located be-
1.6 0.2 tween 3000 and 3700 cm −1 corresponding to the O\H stretching vi-
8ZrCa brations in the same way as Crupi et al. [34] and Davis and Tomozawa
absorbance (a.u)

1.4
0.1 [11] resolving the band into several components. Two specific com-
1.2 ref, 5 min to 35 min ponents have been included according to [11] and [34]: one is linked
absorbance (a.u)

ref, 5 min to 35 min

1.0 0.04 0.0
to the existence of strongly bonded water due to the confinement
2600 2800 3000 3200 3400 3600 3800
corresponding to the small size of the empty space in glass matrix
absorbance (a.u)

-1
0.03 k (cm )
0.8 and the other one is linked to the water bonded with the cations
0.02

0.6 0.01
existing in glass acting as charge compensators. Then, four subbands
ref, 5 min to 35 min
were used: a first subband around 3000 cm −1 attributed to the ion
0.00
0.4 1550 1600 1650
-1
k (cm )
1700 1750 (i) bounded water, a second subband of free H2O clusters around
0.2

0.0
400 800 1200 1600 2000 2400 2800 3200 3600 4000
k (cm-1)
100
Fig. 2. (continued).
0Ca
80 3Ca
νOH peak area (a.u)

6Ca
60 9Ca
8Zr
stretching bands [31] and B\O vibration bands [32] shifts towards
8ZrCa
40
high wave numbers for the xCa series and with the Ca content. This 0Zr
phenomenon can be due to the polymerization state of the silicate 1Zr
network which decreases with the Ca content and increases with 20
4Zr
the Zr content due to the formation of Si\O\Zr bond enhancing
the silicate network rigidity. For the xCa series, the shift of this
peak is also associated with the formation of bands around 0 20 40 60 80 100 120 140
780 cm − 1. Moreover, some shoulders are visible around 700 cm − 1 t (min)
when Ca/Na or Zr/Si increases. The decrease of tetrahedral boron
Fig. 3. Evolution of the OH stretching peak area as a function of time for all glasses at
proportion already observed in [18] could be at the origin of this pH = 3 and 30 °C. The uncertainties are about the same order as the data symbol
shoulder. (0.5 a.u.).
 D. Rébiscoul et al. / Journal of Non-Crystalline Solids 358 (2012) 2951–2960 2955

3280–3308 cm −1, a third subband of bonded water around 3470– the subbands and thus the various types of O\H. Knowing that the
3522 cm −1 and a fourth subband at 3585–3611 cm −1 representing intensity of the absorption band is directly related to the number of
the surface and internal silanols. The FTIR-ATR OH stretching spectra oscillators [35] it means that the intensity of the subbands corre-
with the best fit resulting of the deconvolution in four components sponds to the amount of OH. Consequently, it appears in Fig. 4(a) to
are presented in Fig. 4(a) to (i). In some cases, two to three peaks be- (i) that O\H of the free water clusters are the most numerous,
tween 2800 and 3000 cm −1 were observed. These peaks attributed to followed by the O\H belonging to bonded water and ion bonded
C\H vibrations come from the spectrometer diamond cleaning step water and then the surface and internal silanols. This result is more
with ethanol. In our fitting procedure we have not subtracted this obvious in Fig. 5 showing the evolution of the ratio of the subband
contribution to the spectra. Fig. 4 shows a continuous increase of peaks area, i.e. the proportion of the various types of O\H, vs time.

0.020 0.020
10 min to 40 min
0Ca 3Ca
0.016 0.016

Absorbance (a.u)
(a) (b)
Absorbance (au)

0.012 0.012
20 min to 2h

0.008 0.008

ref ref
0.004 0.004

0.000 0.000
2800 3000 3200 3400 3600 3800 2800 3000 3200 3400 3600 3800
k (cm-1) k (cm-1)

0.020
0.020

6Ca ref 10 min to 1h38

0.016 9Ca
3800 3000
0.016
0.020
Absorbance (a.u)

(c)
Absorbance (a.u)

0.012 0.012 (d)

ref
10 min to 1h20
0.008 0.008

ref
0.004 0.004 ref

0.000 0.000
2800 3000 3200 3400 3600 3800 2800 3000 3200 3400 3600 3800
k (cm-1) k (cm-1)
0.20 0.20

0Zr 20 min 1Zr

0.16 0.16
absorbance (a.u)
absorbance (a.u)

10 min
0.12 0.12
(e) (f)
20 min
0.08 0.08 10 min

0.04 2 min 30
1 min 30 0.04

ref ref
0.00 0.00
2800 3000 3200 3400 3600 3800 2800 3000 3200 3400 3600 3800
k (cm-1) k (cm-1)

Fig. 4. (a) to (i). Experimental and fitted ATR-FTIR spectra using the four components of the various types of O\H vibrations for all the glasses at pH = 3 and 30 °C.
2956 D. Rébiscoul et al. / Journal of Non-Crystalline Solids 358 (2012) 2951–2960

0.20 0.8 0Ca

-1 3Ca
4Zr 3280-3308 cm 6Ca

Ratio of δOHi peak area

0.16
0.6 9Ca
absorbance (a.u)

20 min 8Zr
0.12 (g) 8ZrCa
10 min 0.4 0Zr
-1 1Zr
0.08 3470-3522 cm -1
3000 cm 4Zr
0.2
0.04
-1
3585-3611 cm
ref
0.00 0.0
2800 3000 3200 3400 3600 3800 0 20 40 60 80 100 120
k (cm-1) time (min)
0.020 Fig. 5. Evolution of the ratio of the different types of O\H vibrations during the glass
10 min to 50 min alteration at pH = 3 and 30 °C. The uncertainties are about the same order as the
8Zr data symbol.
0.016

(h)
absorbance (a.u)

0.012
the X-ray reflectivity fitting results (Fig. 7). The altered glass thicknesses
0.008 obtained were plotted as a function of the square root of the time and
show linear evolution characteristics of a diffusive process (Fig. 8). For
ref most of the tests, the intercept points are different from zero, probably
0.004 due to the small diffusion layer existing before the experiment and cor-
related with the waiting time between the end of the polishing of the
0.000 glass monolith and the beginning of the alteration (between 6 months
2800 3000 3200 3400 3600 3800 and 2 years depending on the glass). Nevertheless these artifacts dot
not prevent the measurement of water diffusion within the pristine
k (cm-1)
glass.
0.020 The water diffusion coefficients through the glass DH2O were calcu-
lated from the equation based on the first Fick law assuming that dif-
8CaZr fusion is the limited reaction [36] (Eq. (3)).
0.016 (i)
rffiffiffiffiffiffiffiffiffiffiffiffiffiffi
DH2O :t
absorbance (a.u.)

ei ¼ 2 ð3Þ
0.012 π
5 min to 35 min
with t the time required to form by a diffusion process a layer thickness
0.008
ei. The results are presented in Table 3. These apparent coefficients are
about the same order of magnitude that the water diffusion coefficients
0.004
ref in the glasses containing boron (10 −20 b DH2O b 10−24 m².s −1) [12,37].

3.4. Temperature dependency of the water diffusion

0.000
2800 3000 3200 3400 3600 3800
Temperature dependency of the water diffusion coefficient was
k (cm-1) also measured for the 9Ca glass. Experimental and simulated X-ray
Fig. 4. (continued). reflectivity curves of the 9Ca glass during its alteration at 12 °C,
50 °C and 60 °C are presented in Fig. 9(a) to (c). The thickness deter-
mined from the models used to fit the X-ray reflectivity curves has
allowed to calculate the water diffusion coefficients (Table 3). It ap-
pears that water diffusion coefficients vary of two orders of magni-
The proportions of OH groups show approximately the same evolu- tude between 12 °C and 60 °C showing the strong impact of the
tion whatever the glass and whatever the predominant mechanism temperature. This dependence is comparable with data obtained in
which could evidence a structural equilibrium of the alteration layer [38] on the SON68. An Arrhenius behavior is observed (Fig. 10) with
(charge equilibrium, ions solvation…). an activation energy of 76.9 kJ.mol −1.

3.3. Determination of the water diffusion coefficients through the glass
The FTIR-ATR results have shown that glass hydration is the pre-
dominant mechanism for the xCa series, 8Zr and 8ZrCa glasses. For
these glasses, a linear correspondence between the O\H stretching
peak areas and the thicknesses of the altered glasses obtained from
the X-ray reflectivity curves (Fig. 6(a) to (d), (h) and (i)) confirmed
4. Discussion
The nature of the predominant mechanism and the various de-
grees of glass alteration have been related to the non-bridging oxygen
(NBO) content in glass calculated from experimental quantities of
four-coordinate boron from [18] as illustrated in Fig. 11. Indeed, a
high NBO content in glass, traducing a network depolymerisation,
makes the water diffusion of molecular water easier [14] and the
 D. Rébiscoul et al. / Journal of Non-Crystalline Solids 358 (2012) 2951–2960 2957

presence of a high amount of reactive site regarding water can lead
rapidly to a glass alteration dominated by hydrolysis. For glasses
0Zr, 1Zr and 4Zr presenting a high alteration dominated by a hydroly-
sis mechanism the NBO/Si ratio is comprised between 0.25 and 0.30
[18]. Despite its high NBO/Si ratio, the 9Ca glass has an alteration
driven by the hydration and interdiffusion. It is explained by the
role of the Ca. Indeed, in this glass, the low amount of Na is no suffi-
cient to compensate the boron and zirconium units and thus, it is re-
placed by a fraction of Ca. The presence of Ca can have two effects. It
can act as a charge compensator which strengthens the glass struc-
ture by the incorporation of the two positive charges leading to a de-
crease of water accessibility [14]. Another potential effect is related to
the formation of strong complexes of H2O molecules with Ca 2+ limit-
ing water penetration [39].
The glass composition will put an impact significantly on the
water penetration (Fig. 12(a) and (b)). Comparing the 8Zr and the
0Ca glass with the others glasses of the xCa series, it appears that
the water diffusion coefficients through theses glasses are about one

100 100
0Ca 3Ca
10-2 10-2

10-4 10-4
I/I0

I/I0
10-6 10-6
ref ref
10 min 20 min
10-8 10-8
20 min 41 min
60 min
10-10 30 min 10-10
40 min 80 min
(a) (b) 100 min
10-12 10-12
0.0 0.1 0.2 0.3 0.0 0.1 0.2 0.3
q (A-1) q (A-1)

100 100
6Ca 9Ca
10-2 10-2

10-4 10-4

10-6 10-6
I/I0

I/I0

ref
ref
10 min 10 min
10-8 10-8
15 min 20 min
38 min 40 min
10-10 10-10
1h08 1h
(c) 1h38
(d)
10-12 10-12
0.0 0.1 0.2 0.3 0.0 0.1 0.2 0.3
q (A-1) q (A-1)

100 100

0Zr 1Zr
10-2 10-2

10-4 10-4
I/I0

I/I0

10-6 10-6
ref ref

10-8 1min30 10-8 10 min

2min30 20 min
(e) (f)
10-10 10-10
0.0 0.1 0.2 0.3 0.0 0.1 0.2 0.3
q (A-1) q (A-1)

Fig. 6. (a) to (i). Experimental and simulated X-ray reflectivity curves of glasses during alteration at pH = 3 and 30 °C.
2958 D. Rébiscoul et al. / Journal of Non-Crystalline Solids 358 (2012) 2951–2960

100 100
4Zr 8Zr
10-2
10-2
10-4
I/I0

I/I0
10-4 10-6
ref
10 min
10-8
20 min
10-6
30 min
ref 10-10
40 min
(g) 5 min (h) 50 min
10-8 10-12
0.0 0.1 0.2 0.3 0.0 0.1 0.2 0.3
q (A-1) q (A-1)
100
8ZrCa
10-2

10-4

10-6
I/I0

ref
5 min
10-8 15 min
25 min
10-10
35 min
(i)
10-12
0.0 0.1 0.2 0.3
q (A-1)
Fig. 6. (continued).

order of magnitude higher in 8Zr and 0Ca glasses than in the 3Ca, 6Ca
and 9Ca glasses. The high amount of sodium oxide (around 18 mol%)
in the 8Zr and 0Ca glasses can increase the ion exchange process lead-
ing to an increase of water penetration in the matrix. The replace-
ment of Na2O by CaO in the glass does not modify the water
diffusion as the close values of DH2O in 8Zr and 8ZrCa show.
We have also observed that the water diffusion in glass is lower
than the neutral gas diffusion, such as helium, though others are
more complex glasses (10 −16 b DH2O b 10 −17 m².s −1 at 25 °C)
[40,41]. The associated activation energy of 76.9 kJ.mol −1 is higher
than the activation energy of helium diffusion in more complex boro-
silicate glass (57 kJ.mol −1) [39,41], even though the size of the two

12
250
y = 3.6766x + 28.87 6Ca
10 R2 = 0.9942
0Ca
0Ca y = 0.7638x + 144.51
vOH peak area (a.u.)

210 R2 = 0.965 3Ca

8
6Ca
thickness (A)

8Zr y = 2.009x + 55.24

6 170 R2 = 0.9964
9Ca
3Ca y = 0.8228x + 91.999
R2 = 0.9975
8Zr
4 130 9Ca y = 0.7767x + 82.353
R2 = 0.9921
8ZrCa

2 y=0.059x-2.694 90
R2=0.979 8ZrCa y = 1.4767x + 39.314
0 R2 = 0.996
40 80 120 160 200 240 50
thickness (A) 0 20 40 60 80 100 120
t0.5 (s0.5)
Fig. 7. O\H stretching peak areas as a function of the thickness of the altered zone de-
termined by XRR for all glasses at pH= 3 and 30 °C. The thickness uncertainties are Fig. 8. Evolution of the thickness of the glass altered zone obtained from the fit as a
about the same order of the data symbol. function of the square root for the glass alteration performed at pH = 3 and 30 °C.
 D. Rébiscoul et al. / Journal of Non-Crystalline Solids 358 (2012) 2951–2960 2959

Table 3 1/RT (Kmol.J-1)

Water diffusion coefficient DH20 obtained by XRR in glasses. 0.34 0.36 0.38 0.4 0.42 0.44
-40
DH2O (m².s−1) 12 °C 30 °C 50 °C 60 °C
y = -76.9/RT - 15.754
-42 R2 = 0.9673

LnDH2O(m².s-1)
0Ca (1.1 ± 0.5) 10−19
3Ca (5.3 ± 2) 10−21 -44
6Ca (4.6 ± 2) 10−21
−21
9Ca (1.5±1) 10 (4.7 ± 2) 10−21 (7.2±2) 10−20 (1.2±2) 10−19
8Zr (3.2 ± 1) 10−20
-46
8ZrCa (1.7 ± 1) 10−20
-48

-50

molecules is similar. This is probably because inert gases do not inter-
act with the glass network and occupy empty interstices within the
network like what Doremus has already suggested [42]. Indeed, the
water diffusion through the glass can be slowed down by both the
water diffusion onto non-reactive sites and the hydrolysis of
reactive sites which could explain the existence of X\OH in the
volume of the altered zone as we observed from our FTIR data [43,44].
5. Conclusion
In this study, we have combined IRFT-ATR and XRR to determine
the water penetration and water structure at the first time of alter-
ation on soda-lime borosilicate glasses and investigate the effect of
the charge compensators, Ca and Na, and of network formers such
Fig. 10. Logarithm of the water diffusion coefficient DH2O for the 9Ca glass vs. the in-
verse temperature.
Si and Zr. The results have shown that glass composition have an im-
portant impact on the duration of the predominant mechanism, hy-
dration of glass or hydrolysis, and the water diffusion coefficients.
Now, other parameters have to be studied such as the pH, the ionic
strength and the cations either on the water diffusion or on the mech-
anisms duration.
Moreover, the decrease of water diffusion, compared to inert gas dif-
fusion through the glass matrix, highlights the strong interactions of
water molecules with glass matrix. To better understand the behavior
of the water in such confined media and more particularly to determine

100
9Ca 12°C 10-1 9Ca 50°C
10-2
10-3
10-4

10-5
I/I0

I/I0

10-6
ref
ref
10 min 10-7
10-8 10 min
20 min
30 min 15 min
10-10 10-9
40 min 20 min
(a) (b) 40 min
10-12 10-11
0.0 0.1 0.2 0.3 0.0 0.1 0.2 0.3
q (A-1) q (A-1)
100
9Ca 60°C
10-2

10-4
I/I0

10-6
ref
10 min
10-8
20 min
30 min
10-10
40 min
(c)
10-12
0.0 0.1 0.2 0.3
q (A-1)

Fig. 9. (a) to (c). Experimental and simulated X-ray reflectivity curves of the 9Ca glass altered at pH= 3 and 12 °C, 50 °C and 60 °C.
2960 D. Rébiscoul et al. / Journal of Non-Crystalline Solids 358 (2012) 2951–2960

xZr Series xCa Series References

0.34
[1] C. Poinssot, S. Gin, J. Nucl. Mater. 420 (1–3) (2012) 182.
0.3 [2] R.H. Doremus, J. Non-Cryst. Solids 19 (1975) 137.
hydrolysis [3] J.H. Thomassin, J.L. Nogues, J.C. Touray, C. R. Acad. Sci. II 297 (12) (1983) 857.
NBO/Si ratio

0.26 [4] T. Advocat, J.L. Corvisier, E. Vernaz, G. Ehret, H. Charpentier, Mater. Res. Soc.
Symp. Proc. 212 (1990) 57.
0.22 [5] J.L. Crovisier, T. Advocat, J.L. Dussossoy, J. Nucl. Mater. 321 (2003) 91.
[6] P. Frugier, S. Gin, Y. Minet, T. Chave, B. Bonin, N. Godon, J.-E. Lartigue, P. Jollivet, A.
0.18 Ayral, L. De Windt, G. Santarini, J. Nucl. Mater. 380 (1–3) (2008) 8.
water hydration [7] S. Gin, P. Frugier, Mater. Res. Soc. Symp. Proc. 757 (2003) 175.
0.14 [8] B. Grambow, R. Muller, J. Nucl. Mater. 298 (1–2) (2001) 112.
and interdiffusion [9] B.P. McGrail, J.P. Icenhower, D.K. shuh, P. Liu, J.G. Darab, D.R. Baer, S. Thevuthasen,
0.1 V. Shutthanandan, M.H. Engelhard, C.H. Booth, P. Nachimuthu, J. Non-Cryst. Solids
0 2 4 6 8 10 296 (2001) 10.
ZrO2 or CaO concentration (mol %) [10] S. Gin, C. Guittonneau, N. Godon, D. Neff, D. Rébiscoul, M. Cabié, S. Mostefaoui,
J. Phys. Chem. C 115 (2011) 18696.
[11] K.M. Davis, M. Tomozawa, J. Non-Cryst. Solids 185 (1995) 203.
Fig. 11. Evolution of the nonbridging oxygen per silicon NBO/Si as a function of the CaO
[12] N. Yangisawa, K. Fujimoto, S. Nakashima, Y. Kurata, A. Sanada, Geochim.
(mol%) for the xCa series and ZrO2 (mol%) for the xZr series from [18]. The predomi-
Cosmochim. Acta 61 (1997) 1165.
nant glass alteration mechanisms are indicated. [13] M.E. Lynch, D.C. Folz, D.E. Clark, J. Non-Cryst. Solids 353 (2007) 2667.
[14] F. Angeli, D. Boscarino, S. Gin, G. Della Mea, B. Boizot, J.-C. Petit, Phys. Chem.
Glasses 42 (4–5) (2001) 279.
[15] A. Verney-Carron, N. Vigier, R. Millot, Geochim. Cosmochim. Acta 75 (12) (2011)
(a) CaO/Na2O 3452.
0.0 0.2 0.4 0.6 0.8 1.0 [16] S. Okumura, S. Nakashima, Chem. Geol. 227 (1–2) (2006) 70.
10-18 [17] J. Sterpenich, G. Libourel, J. Non-Cryst. Solids 352 (2006) 5446.
Ca series [18] F. Angeli, T. Charpentier, D. De Ligny, C. Cailleteau, J. Am. Ceram. Soc. 93 (9)
Zr series (2010) 2693.
[19] C. Cailleteau, F. Angeli, F. Devreux, S. Gin, J. Jestin, P. Jollivet, O. Spalla, Nat. Mater.
7 (12) (2008) 983.
10-19
DH2O (m².s-1)

[20] P. Jollivet, F. Angeli, C. Cailleteau, F. Devreux, P. Frugier, S. Gin, J. Non-Cryst. Solids

354 (45–46) (2008) 4952.
[21] C. Cailleteau, F. Devreux, O. Spalla, F. Angeli, S. Gin, J. Phys. Chem. C 115 (2011)
5846.
10-20 [22] R.W. Hepburn, M. Tomozawa, J. Non-Cryst. Solids 281 (2001) 162.
[23] K. Ferrand, A. Abdelouas, B. Grambow, J. Nucl. Mater. 355 (1–3) (2006) 54.
[24] R.M. Richardson, R.M. Dalgliesh, T. Brennan, M.R. Lovell, A.C. Barnes, J. Non-Cryst.
Solids 293 (1–3) (2001) 93.
[25] T. Brennan, R.M. Dalgiesh, M.R. Lovell, R.M. Richardson, A.C. Barnes, S. Sergeant,
10-21 Langmuir 19 (2003) 7761.
[26] D. Rébiscoul, P. Frugier, S. Gin, A. Ayral, J. Nucl. Mater. 342 (1–3) (2005) 26.
[27] D. Rébiscoul, F. Rieutord, F. Ne', P. Frugier, R. Cubitt, S. Gin, J. Non-Cryst. Solids 353
(b) NBO/Si (2007) 2221.
14 18 22 26 30 [28] A. Van Der Lee, Solid State Sci. 2 (2000) 257.
10-18 [29] Ober, Firefx4c_6:raymond.ober@college-de-france.fr.
Ca series [30] L.G. Parratt, Phys. Rev. 95 (1954) 359.
[31] M. Nogami, J. Non-Cryst. Solids 69 (2-3) (1985) 415.
Zr series [32] L.D. Bogomolova, I.A. Ivanov, S.V. Stefanovskii, Y.G. Teplyakov, O.A. Trul, Glass
Phys. Chem. 19 (5) (1993) 413.
10-19
DH2O (m².s-1)

[33] V. Sudarsan, V.K. Shrikhande, G.P. Kothiyal, S.K. Kulshreshtha, J. Phys. Condens.
Matter 14 (2002) 6553.
[34] V. Curpi, F. Longo, D. Majolino, V. Venuti, J. Chem. Phys. 123 (2005) 145702.
[35] R. Janoschek, The Hydrogen Bond I, in: P. Schuster, G. Zundel, C. sandorfy (Eds.),
North-Holland, Amsterdam, 1976, p. 166.
10-20 [36] J. Crank, The mathematics of diffusion, in: Oxford University Press, New York,
1956, pp. 12–15.
[37] B. Grambow, W. Lutze, R. Muller, Mater. Res. Soc. Symp. Proc. 257 (1992) 143.
[38] T. Chave, P. Frugier, A. Ayral, S. Gin, J. Nucl. Mater. 362 (2007) 466.
10-21 [39] S. Indris, P. Heitjans, H. Behrens, R. Zorn, B. Frick, Phys. Rev. B 71 (2005) 06205-1.
[40] F. Chamssedine, T. Sauvage, S. Peuget, T. Fares, G. Martin, J. Nucl. Mater. 400
(2010) 175.
Fig. 12. (a) and (b). Evolution of the water diffusion coefficients DH2O with the ratios of
[41] T. Fares, S. Peuget, O. Bouty, X. Deschanels, M. Magnin, C. Jegou, J. Nucl. Mater. 416
CaO/Na2O and NBO/Si. (2011) 236.
[42] R.H. Doremus, J. Am. Ceram. Soc. 49 (9) (1966) 461.
[43] G. Geneste, F. Bouyer, S. Gin, J. Non-Cryst. Solids 352 (28-29) (2006) 3147.
the mechanisms causing water mobility to slow down, such as water [44] F. Bouyer, G. Geneste, S. Ispas, W. Kob, P. Ganster, J. Solid State Chem. 183 (2010)
sorption on surface, water dissociation and protonation. However, 2786.

some water characterizations and modeling are ongoing.

Acknowledgment

We thank Emilien Burger for the fruitful discussions.

View publication stats