Composition effects on synthetic glass alteration mechanisms: Part 1. Experiments

Chemical Geology 279 (2010) 106–119

Contents lists available at ScienceDirect

Chemical Geology j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / c h e m g e o

Composition effects on synthetic glass alteration mechanisms: Part 1. Experiments Natarajan Rajmohan, Pierre Frugier, Stephane Gin ⁎ CEA Marcoule, DTCD/SECM/LCLT, BP 17171, 30207 Bagnols-sur-Cèze Cedex, France

a r t i c l e

i n f o

Article history: Received 21 January 2010 Received in revised form 29 September 2010 Accepted 13 October 2010 Editor: J.D. Blum Keywords: Glass Nuclear glass Alteration Dissolution Kinetics Gel Passivating layer GRAAL

a b s t r a c t Alteration of nuclear waste glasses and silicate minerals is governed by complex processes regulated by several coupled mechanisms. Among these processes is reactive mass transfer through the amorphous gel layer (known as the passivating reactive interphase (PRI) in case of a rate-limiting effect) located between the pristine glass and the bulk solution. In order to assess the influence of the glass composition and the pH on the properties of the PRI, and thus on the nuclear glass durability, an experimental leaching study was performed on borosilicate glass samples with or without Ca, Al, and Zr. Experiments were conducted to understand the influence of the pH and glass composition on the solvated cation diffusion coefficient within the PRI and to generate data for calibration of a PRI solubility model (not presented here). All the experiments were carried out at high S/V ratios so that silicon rapidly reached apparent saturated conditions and the PRI could form: in such conditions glass alteration is controlled only by diffusion of water and dissolved species through the PRI and by precipitation of crystallized secondary phases. The constituents in the PRI and the crystallized secondary phases depend to a large extent on the glass composition and pH. Alkali metal (Na) or preferentially alkaline earth (Ca) elements are retained in the PRI for charge compensation of Al and Zr. The apparent diffusion coefficient calculated from the release of boron, a good tracer, varies with the pH from less than 4 × 10−22 to 9 × 10-18 m2 s−1 in the studied glasses. These very low diffusion coefficients decrease as the pH increases. Concerning the PRI composition we show that Si, Al, Ca and Zr have strong interactions and thus major consequences on the glass durability. Our findings indicate that the SiO2aq activity is relatively constant and independent of the pH below pH 9, followed by a drop at pH 10. In addition, the activity of SiO2aq is affected by the glass composition, and especially by aluminum and zirconium. As soon as dissolved silicon reaches steady state in solution the aluminum and zirconium concentrations start to decrease, probably due to silicon, aluminum and zirconium interactions with retention in the PRI. The formation of crystallized secondary phases is observed at pH 10 for aluminum-free glasses, which diminishes the saturation state of amorphous silica in solution. In these glasses the saturation index indicates that the solution is oversaturated with respect to calcium silicate hydrates (ex: tobermorite, gyrolite). Moreover, the formation of crystallized secondary phases causes dissolution of the PRI and the glass, which sustains renewed alteration. This study leads to the conclusion that modeling nuclear glass dissolution kinetics over a wide pH range (typically from pH 7 to pH 10) must take into account (1) PRI composition variations and relations between the PRI composition and properties (solubility, diffusion coefficient); and (2) crystallized secondary phases that can consume elements from the PRI. Applying PRI modeling concepts to other kinds of natural glasses or even multi-oxide minerals might prove useful for enhancing our understanding of alteration mechanisms. © 2010 Elsevier B.V. All rights reserved.

1. Introduction Many fields of research are concerned by water–rock interactions ranging from global cycling of elements to the transport of groundwater nutrients and contaminants. Mineral surface descriptions are required to understand the mechanisms involved. In France, one reference option for the management of vitrified high-level waste packages is deep geological disposal (waste management act of June 28, 2006). The current disposal concept studied by ANDRA (French

⁎ Corresponding author. E-mail address: [email protected] (S. Gin). 0009-2541/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.chemgeo.2010.10.010

National Radioactive Waste Management Agency) is based on three containment barriers: the glass package (inside its stainless steel canister), a 55 mm thick carbon steel overpack, and finally the host rock which is a 130 m thick Callovo-Oxfordian argillite layer (ANDRA, 2005). Based on the calculations for the safety assessment of a geological repository, the most important phenomena that must be thoroughly investigated concern radionuclide release from the glass matrix due to alteration by groundwater and migration through the host rock. In either normal or accident scenarios, these phenomena will take hundreds of thousands of years before the radionuclides reach near-surface aquifers. This is why a comprehensive methodology is required, including laboratory experiments to investigate chemical processes at different scales, a mechanistic model to predict

N. Rajmohan et al. / Chemical Geology 279 (2010) 106–119

the radionuclide source term, and model validation using natural or archeological analogs, integrated mockups or in situ tests. Dissolution of silicate glasses is a complex process controlled by several mechanisms (Conradt, 2008; Frugier et al., 2008; Frizon et al., 2009). The alteration kinetics of silicate glasses determined from laboratory studies involve several characteristic process steps that are detailed below (Advocat et al., 1991; Berger et al., 1987; Byers et al., 1985; Crovisier et al., 1985, 1989, 1992; Guy, 1989; Vernaz and Dussossoy, 1992; Verney-Carron et al., 2007; Van Iseghem et al., 2007). Recently, Frugier et al. (2008) proposed a new mechanistic model called GRAAL (Glass Reactivity with Allowance for the Alteration Layer), which highlights the key mechanisms in the glass alteration process. This model has been applied successfully to SON68 glass (i.e. the French inactive reference glass) altered in various conditions (Frugier et al., 2009). In the GRAAL paradigm, the glass alteration process is a combination of the following steps: (1) exchange and hydrolysis reactions first involve mobile glass components (alkali metals, boron, etc.) (Geneste et al., 2006; Rebiscoul et al., 2007); (2) hydrolysis, especially of silicon, results in the existence of an initial glass dissolution rate; (3) the difference between these two kinetics causes an amorphous alteration layer to form at the glass/solution interface regardless of the alteration conditions; (4) the amorphous alteration layer progressively creates a barrier limiting the transport of water toward the glass and of solvated glass ions into solution; (5) some glass constituents precipitate as crystallized secondary phases which are present on the external surface or dispersed in solution. Depending on their composition, the formation of these crystallized secondary phases may accelerate or maintain a pseudo-constant glass alteration rate. The “final” or “residual” alteration rate can be attributed to the mechanisms of steps (4) and (5). Overall, glass alteration produces an amorphous alteration layer (often called the “gel”) as well as crystallized secondary phases, and these two phenomena are of a completely different nature (Thomassin, 1984; Jercinovic et al., 1990, Advocat et al., 1991, Vernaz and Dussossoy, 1992; Verney-Carron et al., 2007). According to basic thermodynamics, the amorphous layer is stable only when the solution is saturated with respect to its constituent elements such as silicon, zirconium, aluminum, calcium, etc. Furthermore, the transport-inhibiting effect of the amorphous layer between the glass and solution controls glass alteration (i.e. the release of glass alteration tracers like boron or alkalis). The amorphous layer is therefore termed a “passivating reactive interphase” (PRI) in accordance with its properties (Chave et al., 2007; Frugier et al., 2008). However, a low-density amorphous layer can form far from saturation thanks to low-solubility elements such as Zr. Under these conditions, it has no or negligible passivating properties compared with the PRI that is eventually formed at saturation. It is then called a depleted gel. The nature of the amorphous layer varies with the reaction conditions (glass-surface-area-to-solution-volume (S/V) ratio, flow rate, pH, temperature, etc.) and with the reaction progress (forward rate, rate drop or residual rate regimes). At high S/V ratios the system reaches saturation very rapidly, hence only the PRI is observable and the depleted gel is not significant (b1 nm) (Frugier et al., 2008). Moreover, Rebiscoul et al. (2005) carried out a detailed investigation of the amorphous layer and its properties in different reaction conditions, reporting that the amorphous layer has two different parts: a porous layer in contact with water and a dense layer at the gel-glass interface. Furthermore, the dense zone is identified as the one controlling the residual rate, which is expected to be the most important regime in a geological repository. Numerous studies have reported the role of the amorphous layer and its protective properties on the alteration of glass or silicate minerals (Angeli et al., 2001; Ledieu et al., 2005; Hellmann et al., 2003, 2004; Lee et al., 2007; Rebiscoul et al., 2005). Some studies investigated the amorphous layer characteristics and its structural rearrangement during alteration, porosity, diffusion properties, retention of elements in the amorphous layer, etc. (Arab et al.,

107

2008; Jollivet et al., 2008; Houston et al., 2008; Angeli et al., 2001b, 2008; Chave et al., 2007). Nevertheless, the composition of the amorphous layer has still not been fully defined. Generally, the amorphous layer is depleted in highly soluble elements and enriched in sparingly soluble elements. The constituents of the amorphous layer vary with glass composition, pH, temperature, etc. (Angeli et al., 2006, 2008). However, silicon, aluminum and zirconium are the predominant elements in the amorphous layer when SON68-type nuclear waste glasses undergo alteration. Element interactions within the amorphous layer are also an important factor; for example, in aluminosilicate glasses, silicon and aluminum strongly interact in the amorphous layer. Houston et al. (2008) studied the interaction between aluminum and amorphous silica using bulk solution chemistry experiments with NMR techniques. They reported three reaction pathways for aluminum–silica interactions such as adsorption of Al at surface silanol sites, surface-enhanced precipitation of aluminum hydroxides and aluminosilicate secondary phases. Several researchers have documented the role of dissolved aluminum on aluminosilicate mineral dissolution reactions (Oelkers and Schott, 1994; Bickmore et al., 2006; Chou and Wollast, 1985; Samson et al., 2005; Jones and Handreck, 1963; Hingston and Raupach, 1967; Ballou et al., 1973; Iler, 1973; Van Bennekom et al., 1991; Van Cappellen and Qiu, 1997). They concluded that Al species diminish the silica or silicate mineral dissolution rates in different reaction conditions. Because the time scale considered for geological disposal is inaccessible to laboratory experimentation, modeling is the principal method used to assess the overall glass alteration behavior. Geochemical models are widely employed to simulate crystallized secondary phase formation during glass alteration (Advocat et al., 2001; Berger et al., 1987, 1994; Crovisier et al., 1985, 1989, 1992; Gislason et al., 1993; Gong et al., 1998; Grambow et al., 1985; Michaux et al., 1992,). These studies examined and simulated the dissolution kinetics and the formation of crystallized secondary phases, but the simulation did not take into account the formation of an amorphous layer at the glass/solution interface. Munier et al. (2004) modeled the amorphous layer composition of simple borosilicate glasses by the precipitation of an ideal solid solution. They considered three types of solid solutions: oxides, hydroxides and metasilicates. The major limitation of this work concerns the phases considered for the amorphous layer. Except for amorphous silica, silicates cannot form an amorphous layer. Furthermore, the present work will show that sodium is not retained as sodium oxide or hydroxide or silicates, but only as a charge compensator of Al and Zr in the amorphous layer. Moreover, if calcium is present in the glass constituents, most of the sodium will not remain in the amorphous layer. In addition, the chemical interactions between elements such as Si, Al, Ca, Zr within the amorphous layer were not considered in this study. Likewise, in the GRAAL model (Frugier et al., 2009), the amorphous layer (PRI) consists of simple phases which represent its forming elements such as Si, Al, Ca and Zr, and does not take into account the chemical interactions between these elements. Hence, the present study was carried out to model the PRI composition of French nuclear waste glasses (stoichiometry similar to SON68) using simple and complex phases which represent PRIforming elements and their chemical interactions. This study focuses on the main PRI-forming constituents of SON68 type glass: Si, Al, Zr and Ca. The study combines experiments and modeling. Experiments were carried out to understand the influence of the pH and glass composition on the diffusion coefficient and glass alteration phenomena, and to generate data for model calibration and validation. The pH range was chosen to cover alteration conditions imposed by the groundwater (pH near neutrality in most cases) up to conditions imposed by the glass dissolution (pH around 9.5). As the PRI is by definition (Frugier et al., 2008) the most passivating amorphous layer eventually formed in silicon-saturated condition, all the experiments were carried out at a high S/V ratio (2000 m−1) and at 90 °C. In these conditions, saturation is achieved in a few hours; consequently, the

108

N. Rajmohan et al. / Chemical Geology 279 (2010) 106–119

amorphous layer contains negligible amounts of depleted gel. Moreover, the activity of its constituents (Si, Al, Ca, Zr) will remain almost constant after a few hours so that we may consider the amorphous layer is homogenous PRI for modeling purpose.

Table 1 Glass composition (oxide wt.%/mol%).

a

Glass

SiO2

CJ2 CJ3 CJ4 CJ7 CJ8 CJ9 SON68a

61.2 58.1 56.2 59.1 62 59.8 45.5

B2O3 64.9 61.2 60.1 63.8 63.7 62.5 50.5

18.9 17.9 17.3 18.2 19.1 18.5 14.0

Na2O 17.3 16.3 16.0 17.0 16.9 16.7 13.4

13.3 12.6 12.2 12.8 13.4 13.0 9.9

SON68 contains other minor elements.

Al2O3 13.6 12.9 12.6 13.4 13.3 13.2 10.6

6.6 6.3 6.1 6.4

4.9

CaO 4.1 3.9 3.8 4.1

3.2

ZrO2

5.2 5.0

5.8 5.7

5.5 5.3 4.0

6.1 5.9 4.8

3.3 3.5

1.7 1.8

3.5 2.7

1.8 1.5

2. Methodology 2.1. Experimental procedure 2.1.1. Glass details Table 1 indicates the glasses used in this study and their composition. The detailed procedure for glass specimen preparation

Fig. 1. Equivalent thicknesses of elements versus time for CJ2, CJ3 and CJ8 glasses at pH 7. S/V = 2000 m−1, temperature 90 °C. Similar trends are observed in other glasses (CJ2 = CJ7, CJ3 = CJ4 and CJ8 = CJ9).

N. Rajmohan et al. / Chemical Geology 279 (2010) 106–119

is given in Jégou et al. (2000) and Gin and Jégou (2001). The glasses were fabricated in an induction-heated furnace at 1500 °C. Rayleigh– Brillouin light scattering was used to measure the glass homogeneity. The 100 − 125 μm and 20 − 40 μm size fractions were obtained by crushing glass rods in a ball mill followed by sieving. The powder samples were washed in acetone, alcohol, and finally in ultrapure water. The specific surface area of the sample was determined by BET method using Kr. The stoichiometry of these glasses is based on the SON68 glass composition. In order to understand the effect of Ca, Al and Zr, glasses

109

were prepared by adding their oxides to the three major oxides Si, B and Na (Table 1). 2.1.2. Alteration methods Experiments were carried out in a static system using PTFE reactors (Savillex) at 90 °C. Glass samples with size fractions 100 − 125 μm and 20− 40 μm were selected for this study. Static leaching experiments were carried out for about 150 days at a glass surface-area-to-solutionvolume (S/V) ratio of 2000 m−1 except for CJ9 (4800 m−1) and with various imposed pH values (7, 8, 10) at 90 °C. This covers the pH range

Fig. 2. Equivalent thicknesses of elements versus time for CJ2, CJ3 and CJ8 glasses at pH 10. S/V = 2000 m−1, temperature 90 °C. Similar trends are observed in other glasses (CJ2 = CJ7, CJ3 = CJ4 and CJ8 = CJ9).

110

N. Rajmohan et al. / Chemical Geology 279 (2010) 106–119

expected in an argillaceous geological repository (Bildstein et al., 2007). Experimental data for these glasses at pH 9 were taken from Jégou (1998) and Gin and Jégou (2001). For pH 7 and 8, 0.2 mol of tris (hydroxymethyl)aminomethane (TRIS) buffer solution was prepared in 0.1 mol/L and 0.01 mol/L HNO3 solutions, respectively. Potassium hydroxide (KOH 0.028 mol/L) solution was used as a buffer for pH 10 experiments. In order to avoid the large pH adjustment, buffer solutions were directly used as the solvent for all the leaching experiments. The quantity of buffer solutions needed for each reaction was calculated with the JCHESS code (van der Lee et al., 2003). Buffer solutions were heated to 90± 1 °C before addition to the reactor to avoid the initial delay in the reaction. After adding the glass powder and solvent, reactors

were kept in 1000 mL PTFE reactors with a small quantity of water to prevent evaporation loss of the reaction solution. Solutions were not stirred during the experiment.

2.1.3. Sampling and chemical analysis Two milliliters of solution was sampled from the reactor and ultrafiltered to 10 000 Da, and then acidified with 3 mL of 0.5 N HNO3 to prevent the formation of colloids. The solution samples were refrigerated at 4 °C pending analysis. Analysis was carried out using inductively-coupled plasma atomic emission spectrometry (ICP-AES) for cations (Si, B, Na, Al, Ca, K) and ion chromatography for chloride.

Fig. 3. Boron equivalent thickness versus time for different glasses and pH. Data for initially pure water leading to a pH close to 9 collected from Gin and Jégou (2001) and Jégou et al. (2000).

N. Rajmohan et al. / Chemical Geology 279 (2010) 106–119

The uncertainty on these data is generally 3%, although potentially higher for low Al concentrations. 2.2. Solid characterization Altered glass samples were collected from the reactor and characterized by SEM (JEOL JSM6330F, 15 kV, EDS analysis with PGT system). We carried out direct observations of the altered grain surface to view the crystallized secondary phases. Samples were carbon coated prior to analysis. The altered glass powder was also characterized by X-ray diffraction (Philips X'Pert). 2.3. Data analysis Elemental analysis results were employed to calculate the altered glass percentage, altered glass equivalent thickness, diffusion coefficient, and retention factor. The altered glass percentage (AGB%) was estimated from the boron solution analysis by iteratively calculating the leached fraction without omitting all mass losses due to sampling. Boron is generally selected for such calculations because it is not retained in secondary phases and is one of the first leached components (Scheetz et al., 1985). AGB% was calculated at each sampling interval using the following relation:

mleached mleached boron = = AGB % = boron m0boron m0glass ⋅xB

t

t

t−1

i

i

CB ⋅V + ∑ CB ⋅ VSV i=1

m0glass ⋅ xB

ð1Þ

where CBi is the boron concentration (g/m3) at time i in solution, CBt is the boron concentration (g/m3) at time t in solution, V t is the solution t volume (m3) remaining in the reactor at time t, VSV is the solution 3 0 volume (m ) sampled at time t, mglass is the initial mass of glass powder (g), and xB is the boron mass fraction in the glass (g/g). The altered glass equivalent thickness Eq(B) (m−1) calculated from mobile element (B) is given by ! dEqðBÞ d Ct
t B t = dt dt xB S = V ρ

ð2Þ

111

is mainly an amorphous phase that grows within the volume of altered glass. The retention factor (RF) for element i in the glass alteration products is calculated using the following relation:

RFi = 1−

EqðiÞ EqðBÞ

ð5Þ

where Eq(i) is the equivalent thickness of element i. The diffusion coefficient of boron through the PRI was determined using Fick's second law by the following relation (Chave et al., 2007): rffiffiffiffiffiffiffiffi DB t π

EqðBÞ−EqðSiÞ = 2

ð6Þ

where Eq(B) − Eq(Si) is the equivalent gel thickness, DB is expressed in (m2/s) and t is the time (s). During glass alteration, once saturation with respect to PRI is achieved, the boron concentration increases only due to the diffusion process (Frugier et al., 2008). As long as saturation with respect to the PRI is not reached, boron also enters solution due to the PRI dissolution process. Consequently, the equivalent thickness calculated from silica must be subtracted from the equivalent thickness calculated from boron to obtain a better estimate of the diffusion coefficient. This equation is a valid approximation as long as the amount of silica in secondary phases is negligible compared to the amount of silica in solution, and as long as the amount of silica in solution is negligible compared to the amount of silica in the PRI. The chemical analysis results were employed to calculate the activity of individual species and the saturation index of minerals in solution using the JCHESS code (van der Lee et al., 2003; Wolery, 1992). The saturation index shows whether the water will tend to dissolve or precipitate a particular mineral. Its value is negative when the mineral may be dissolved, positive when it may be precipitated, and zero when the water and mineral are at chemical equilibrium. The saturation index (SI) is calculated by comparing the chemical activities of the dissolved ions of the mineral (ion activity product, IAP) with their solubility product (Ksp). In equation form, SI = log(IAP/Ksp).

where ρ is the glass density (g/m3). Eq. (2) defines the altered glass equivalent thickness based on the mass balance of elements between the solid and the solution. Another equation (Eq. (3)) can be derived for the same purpose. It is assumed that glass grains are perfect spheres and have the same specific surface area, Sp (m2/g). Based on this assumption, the altered glass equivalent thickness (Eq(i)) is obtained from the altered glass percentage (AGi%) (Chave et al., submitted for publication).   1=3 R0 EqðiÞ = 1−ð1−AGi %Þ R0 =

3 ρSp

ð3Þ ð4Þ

As mentioned earlier, boron is a good tracer of glass alteration because it is not retained in the alteration products (PRI and crystallized secondary phases) (Scheetz et al., 1985). Based on solubility data, condensation/precipitation of boron may only be observed at very high boron concentrations in solution to precipitate phases such as borax (Na2B4O7 •10H2O) or colemanite (Ca2B6O11•5H2O). Such concentrations are rarely obtained during glass alteration processes and have never been observed in our experiments. The consequence is that equivalent thicknesses calculated for boron do in fact represent the amount of glass that has been altered whatever the underlying mechanism. It is a good approximation of the thickness of the amorphous layer when the latter

Fig. 4. Altered glass percentage at the 150th day, calculated for boron, for studied glasses and various pHs.

112

N. Rajmohan et al. / Chemical Geology 279 (2010) 106–119

3. Experimental results 3.1. Glass alteration at various pH values Fig. 1 illustrates the evolution curve for equivalent thicknesses of elements (see Eqs. (1) and (3)) as a function of time for three of the six glasses studied. Only glasses CJ2, CJ3 and CJ8 are shown in Fig. 1

because the element leaching trends in CJ2 and CJ7, CJ3 and CJ4, and CJ8 and CJ9 are similar. Furthermore, the element leaching trends are similar at pH 7 and pH 8. In CJ2, at pH 7, the boron and sodium trends are similar, continuously increasing but at a decreasing rate (Fig. 1, CJ2). A similar observation can be made for CJ3 glass. In CJ8 and CJ9, the curves for boron and sodium are completely different from the other glasses. Early dissolution of these elements is high: the

Fig. 5. Scanning electron microscope images (15 kV) of altered glass surface (CJ8 (a), CJ9 (b)) and XRD patterns (CJ8 and CJ9 (c)) at pH 10, S/V = 2000 m−1, temperature 90 °C, duration 150 days. Secondary phases are likely calcium silicate hydrates (CSH).

N. Rajmohan et al. / Chemical Geology 279 (2010) 106–119

equivalent thickness is around 400 nm at pH 7 and 500 nm at pH 8. However, after reaching these values, the boron and sodium equivalent thicknesses remain stable without fluctuations. In CJ3 and CJ8, the equivalent thicknesses of boron and sodium are identical, which indicates that boron and sodium are leached congruently. However, this is not observed in calcium-free glasses. Indeed, in CJ2 the difference between the boron and sodium equivalent thicknesses is significant. This difference is consistent with sodium being retained in the amorphous layer (Houston et al., 2008; Angeli et al., 2001a, 2008). The silicon concentration in solution reaches an apparent steady state quickly (b 15 days) and consequently the equivalent thickness remains almost constant (Fig. 1, CJ2 and CJ3). In other words, silicon retention in the amorphous layer increases regularly over time. This observation is common to both CJ2 and CJ3. In contrast, silicon reaches steady state more rapidly in CJ8 than in CJ2 and CJ3, and then remains stable throughout the experimental duration. The equivalent thickness of silicon is greater than 90 nm in CJ8, which shows that dissolution is higher than for the other glasses (40 nm for CJ2, 60 nm for CJ3). Aluminum reaches a maximum within a few days, then begins to decrease in the case of CJ2. It is important to note that once silicon is stabilized or has reached steady-state conditions, aluminum starts to diminish. This behavior has been reported elsewhere (Ribet and Gin, 2004). The concentrations of aluminum and zirconium in solution are generally below or near the detection limits (Al: 0.01 mg/L; Zr: 0.005 mg/L). In CJ3, aluminum is below the detection limit. Fig. 2 shows the evolution curve for Si, Al, B, Na and Ca at pH 10 for CJ2, CJ3 and CJ8 glasses. At pH 10 for CJ2 the equivalent thickness of boron is less than at pH 7 and 8. In contrast, the equivalent thickness of silicon exhibits the opposite trend and is higher at pH 10 (Figs. 1 and 2). For example, the equivalent thickness of silicon at pH 10 for CJ2 is 200 nm, whereas it is 40 nm at pH 7 and pH 8. In CJ3, all elements except Ca are initially congruent (b30 days), after which silica and aluminum stabilize. In CJ8, Si, B and Na are congruent until 60 days, when a sudden increase in both B and Na is observed. The evolution curve of Ca shows that it decreases progressively throughout the reaction.

113

Table 2 Elements in glass and PRI. Glass

Elements in glass

Elements in PRI

Crystallized secondary phases

CJ2 CJ7 CJ3 CJ4 CJ8 CJ9

Si, Si, Si, Si, Si, Si,

Si, Al, Na Si, Al, Na, Zr Si, Al, Ca Si, Al, Ca, Zr Si (Ca)a Si, Zr (Ca)a

No No No No Only at pH 9 and pH 10 Only at pH 10

Al, Na, B Al, Na, B, Zr Al, Ca, Na, B Al, Ca, Na, B, Zr Ca, Na, B Ca, Na, B, Zr

a pH 8 for CJ8 and pH 7-9 for CJ9. Elements retention in the PRI varies with pH. See the text for more detail.

glass experiments which prevents direct comparison based on altered glass fractions: same equivalent thicknesses correspond to larger altered glass fractions if the grains are smaller. Fig. 4 suggests that glass alteration is highly pH-dependent, and minimum alteration (b 1%, CJ3) is observed at pH 10. The alteration is very high at pH 7. This was the case for all the studied glasses except CJ8 and CJ9. Aluminum-free glasses (CJ8 and CJ9) show high alteration (~50%) at pH 10 (see equivalent thicknesses for comparison with CJ9). 3.3. Solid characterization The formation of secondary phases can lead to a resumption of alteration by dissolution of the amorphous layer. The altered grain surfaces are therefore observed by SEM (Fig. 5a, b and c); XRD is also performed when crystallized secondary phases are observed by SEM. This was the case for CJ8 and CJ9 at pH 10: Fig. 5a and b show SEM images for CJ8 and CJ9 after 150 days of alteration. As there is no aluminum in these glasses, the possible crystallized secondary phases are calcium silicate hydrates (CSH). X-ray diffraction patterns on CJ8 and CJ9 seem to confirm the formation of CSH even if the clear identification of the precipitated minerals is not possible due to the large amount of amorphous material in the samples (Fig. 5c). Moreover, the calculated saturation index confirms solution saturation state with respect to CSH. 4. Discussion

3.2. Effect of glass composition and pH on glass alteration The experimental results suggest that the glass alteration process is significantly affected by the glass composition, but these elemental effects strongly depend on the solution pH. Fig. 3 shows the equivalent thickness of altered glass, calculated for boron, as a function of time for different glasses and at various pH values. The curves show that the glasses behave quite distinctly in the studied pH range (Fig. 3). Among the glasses, CJ2 exhibits the greatest alteration, except at pH 10. In CJ2, aluminum is added to the basic three-oxide composition (Si, B and Na). Substitution of Ca to CJ2 decreases the degree of alteration significantly at high pH (see CJ3 glass), and slightly at pH 7. Like Ca, the incorporation of Zr in CJ2 and CJ3 also the reduces alteration kinetics (see CJ7, CJ4). Nevertheless, glasses CJ2 and CJ7 show more or less identical properties and are more altered at pH 7 than at pH 10. Furthermore, comparison of glass pairs e.g. CJ2 and CJ3 (effect of Ca), CJ2 and CJ7 (effect of Zr without Ca), CJ3 and CJ4 (effect of Zr with Ca) clearly reveals that the addition of Ca and Zr diminishes glass alteration below pH 10. These observations are consistent with the literature (Gin and Jégou, 2001; Sicard et al., 2004; Spalla et al., 2004; Angeli et al., 2001a; Cailleteau et al., 2008). CJ8 and CJ9 show different trends compared to other glasses: at pH 7 to 9 they initially undergo significant alteration, then the reaction ceases. However, at pH 10 CJ8 and CJ9 behave differently and show constant alteration progress, which suggests a specific mechanism. Fig. 4 shows the altered glass fraction for the longest duration for five of the six glasses. CJ9 alteration is not included because its specific surface area and S/V ratio are significantly different from the other

Experimental results demonstrated that the glass alteration process strongly depends on the glass composition and pH. These

Fig. 6. Na and Ca retention properties at various pH values in the studied glasses. Alteration duration 150 days, S/V = 2000 m− 1, temperature 90 °C.

114

N. Rajmohan et al. / Chemical Geology 279 (2010) 106–119

factors influence element retention and diffusion processes in the amorphous layer, and the formation of crystallized secondary phases. 4.1. Nature of the PRI and crystallized secondary phases As mentioned earlier, alteration of nuclear waste glasses produces a PRI and crystallized secondary phases as the main alteration products. The PRI constituents vary with the glass composition and pH. Based on the experimental results, elements retained in the PRI of each test glass are indicated qualitatively in Table 2. Except for aluminum-free glasses (CJ8 and CJ9) and the highest pH, secondary phases are not observed in these experiments. Therefore, Table 2 assumes that elements leached congruently with boron (same equivalent thicknesses) will not be retained in the PRI. For CJ9, calcium is retained in the PRI for charge compensation of zirconium at lower pH (7, 8). In the case of CJ8, calcium is not retained at pH 7 and the PRI is mostly amorphous silica. However, calcium is partly

retained at pH 8 in CJ8 due to the presence of negatively charged species of silicon. Due to the formation of secondary phases at pH 10 in the aluminum-free glasses, the PRI is mostly amorphous silica (CJ8) or amorphous silica with zirconium (CJ9). Alkali (Na) or alkaline-earth (Ca) elements can be retained in the PRI for charge compensation of Al and Zr (Houston et al., 2008; Angeli et al., 2001a, 2008). In this study, in order to understand the role of Na and Ca in charge compensation in the PRI at various pH values, retention factors were calculated relative to boron (Eq. (5)) and plotted for the studied glasses (Fig. 6). In CJ2 and CJ7, there is no calcium in the glass composition and Na is retained in the PRI. Fig. 6 shows that Na retention is close to 30%, the same at pH 7 and pH 8 for a given glass; there is no retention at pH 10. The opposite is the case for Ca: calcium is retained more at pH 10 (in the PRI and/or CSH) than at pH 7. Moreover, Na retention is almost zero or very low when the glasses contain Ca. It is apparently observed in glasses CJ3, CJ4, CJ8 and CJ9 (Table 2, Fig. 6). This observation strongly suggests that the PRI

Fig. 7. Equivalent thickness and retention factor of Na and Ca in CJ2 and CJ3 glasses versus time. Alteration duration 300 days, S/V = 2000 m− 1, temperature 90 °C.

N. Rajmohan et al. / Chemical Geology 279 (2010) 106–119

115

Fig. 8. Boron and silicon equivalent thicknesses versus time for different glasses at various pH values. S/V = 2000 m− 1, temperature 90 °C.

prefers Ca rather than Na. This is well correlated with published experimental data (Angeli et al., 2006, 2008). Fig. 6 also illustrates that Ca retention in CJ4 and CJ9 is high at all pH values compared to CJ3 and CJ8, respectively. The only difference between CJ3 and CJ4 (and between CJ8 and CJ9) is zirconium (Table 1). This shows that glasses with zirconium retain more calcium in the PRI (Angeli et al., 2008). In other words, zirconium enhances the retention of charge compensators in the PRI. As mentioned earlier, Na is not retained in the PRI at pH 10 (Fig. 6). At pH 10, KOH is used as a buffer, and K may be retained in the PRI for charge compensation. In order to verify this assumption, two experiments were selected (CJ2 and CJ3 at pH 7) and a known quantity of KCl solution was added to these experiments. Fig. 7 displays the results of leach test experiments. Fig. 7 shows that for CJ2, Na is congruent with boron after adding KCl. Before adding KCl, around 25% to 30% of Na is retained in the PRI (Fig. 7c). After adding KCl, the Na retention factor drops to zero. In CJ3, Na is congruent with boron and consequently the Na concentration remains unchanged after adding KCl. However, calcium behaves differently after adding KCl. Fig. 7d indicates that around 50% to 60% of calcium is retained in the PRI before adding KCl. After adding KCl, a rapid drop is observed in Ca retention. These results evidence that PRI may use whatever cation available in solution depending on its concentration and reactivity. The PRI probably retains Ca rather than Na for charge compensation.

Fig. 8 can be interpreted considering that the glass alteration is initially governed by dissolution. Once silicon reaches a steady state (after about 20 days in most cases), PRI dissolution ceases and the glass dissolution rate drops by several orders of magnitude. In such conditions glass alteration remains controlled only by a diffusion process. This mechanism is clearly observed in CJ2 glass and dissolution is active for the first 20 days (Fig. 8, CJ2). Conversely, in the case of CJ9, initial dissolution is fastest at pH 7 and 8, and then the glass alteration reaction ceases. Alteration of CJ9 strongly suggests that dissolution is the predominant process that regulates overall glass alteration and that diffusion through the PRI is very slow. Previous studies have shown that in CJ8 and CJ9 the PRI formed at pH 7 and 8 is very protective due to the porosity clogging by efficient silicon condensation reaction within the PRI because of the absence of

4.2. Influence of pH and glass composition on the diffusion kinetics Nuclear waste glass alteration is governed by dissolution and diffusion phenomena (Grambow and Muller, 2001; Rebiscoul et al., 2004, 2007; Conradt, 2008; Gin et al., 2008; Frugier et al., 2008; Frizon et al., 2009). Boron is a good diffusion tracer in silicon saturation conditions. Silicon represents the dissolution front in glass alteration. Fig. 8 shows how the dissolution and diffusion phenomena drive glass alteration at various pH. Fig. 8 shows only CJ2 and CJ9, although a similar trend is observed in other glasses. CJ3, CJ4 and CJ7 behave like CJ2, while CJ9 represents CJ8.

Fig. 9. Apparent diffusion coefficient calculated for studied glasses at various pH using boron. Alteration duration 150 days, S/V = 2000 m− 1, temperature 90 °C.

116

N. Rajmohan et al. / Chemical Geology 279 (2010) 106–119

aluminum (Arab et al., 2008; Jollivet et al., 2008). Arab et al. (2008) studied the alteration of aluminum-free five-oxide silicate glasses with varying percentages of zirconium, and characterized the PRI using SAXS. They reported that the reorganization of PRI can occur very early, leading to porosity clogging. Likewise, Jollivet et al. (2008) used 29Si isotopic tracing to investigate gel (PRI) porosity clogging for the same glasses (Arab et al., 2008) and suggested that porosity clogging occurs in the external part of the PRI (solution/PRI interface) after densification of the layer by silicon condensation. In these studies the glass composition is similar to CJ8 and CJ9, and the conclusion is consistent with our study. In addition, the CJ9 alteration curve at pH 10 shows different trends in comparison with pH 7 and 8 (Fig. 8, CJ9). At pH 10, glass alteration is driven by precipitation of crystallized secondary phases, calcium silicate hydrates (CSH), followed by renewed alteration (e.g. after 90 days in case of CJ9, Fig. 8, CJ9) (Gin et al., 2001; Ribet and Gin, 2004). During CSH precipitation, glass alteration is governed only by dissolution of the PRI and not by a diffusion process: CSH precipitation decreases the activities of Si and Ca in solution, just as pure water renewal would, leading to PRI dissolution to keep these activities high. Fig. 8 also illustrates that in the studied pH range, high diffusion is observed at pH 7 followed by pH 8 and 10: the greater the difference between the equivalent thicknesses of boron and silicon, the more important the diffusion mechanism in the glass alteration. This observation is common to all the studied glasses except CJ8 and CJ9, for which the residual diffusion process is too low to be measurable whatever the pH. Fig. 9 shows how the apparent diffusion coefficient of boron varies with the pH. The diffusion coefficient, DB (see Eq. (6)), is influenced by the pH, temperature, PRI composition, etc. (Chave et al., 2007). In the studied glasses, the diffusion coefficient varies from less than 4 × 10−22 to 10−18 m2 s−1. The diffusion coefficient decreases with increasing pH (Ojovan et al., 2006; Verney-Carron et al., 2010) (Fig. 9). Note that for CJ9 the value of 4 × 10−22 m2 s−1 was obtained by simply applying Eq. (6) for comparison with other glass, but it has no real meaning: the boron equivalent thicknesses minus the silicon equivalent thicknesses do not follow a square root of time evolution for these two glasses. This suggests a strong decrease over time of the apparent diffusion coefficient to an unmeasurable value. In such a case, it makes no sense

to take into account any diffusion process. Nevertheless it is interesting to know that, should the model remain applicable, the value would be below 10−24 m2 s−1. Furthermore, the studied glasses can be classified into 3 pairs: CJ2 and CJ7, CJ3 and CJ4, and CJ8 and CJ9, based on DB. The only difference within each pair is zirconium. Glasses without aluminum (CJ8, CJ9) have the lowest diffusion coefficient. A comparison of CJ2 and CJ3 suggests that incorporating calcium in the glass composition significantly diminishes the diffusion process. This highlights the roles of aluminum and calcium in glass alteration. 4.3. Influence of the PRI on the solution chemistry Speciation calculations of all these solutions were carried out with JCHESS using the chess.tdb database taken from EQ3/6 database for each glass, pH and sample time (Van der Lee et al., 2003; Wolery, 1992). As solution samples were ultrafiltered to 10000 Da, the speciation calculation represents only dissolved species in solution. Fig. 10 shows the activities of SiO2aq and Ca2+, and their relation with pH when the solution is at equilibrium with the PRI. Fig. 10 indicates that SiO2aq activity is relatively constant and independent of the pH below pH 9, followed by a drop at pH 10. At pH 10, silicon activity is diminished for all the glasses except those without calcium. In CJ8 and CJ9, SiO2aq can only be stabilized with calcium and significantly decreases at pH 10. Fig. 10 also indicates the influences of the glass composition on the SiO2aq activity. The activity of SiO2aq is the highest for aluminum-free glasses (CJ8, CJ9). In CJ8 and CJ9, silicon is mostly recondensed as pure silica below pH 9 (Fig. 11). In other glasses, silicon interacts with Al and Zr, which reduces silicon activity in solution. Calcium concentrations are high near neutral pH (7, 8). Calcium could be congruent with boron but a fraction of calcium is retained in the PRI for Al and Zr charge compensation. At pH 9 the calcium activity drops sharply: PRI silicon species have a major role in calcium condensation. At pH 10 the formation of secondary phases controls calcium activity. This is also true for CJ8 glass at pH 9 because secondary phases were observed in the long duration experiments (Gin and Jégou, 2001). Saturation indices were calculated for all the glasses in the studied pH range. The results reveal that the solution is saturated or near

Fig. 10. Activities of SiO2(aq) and Ca2+ versus pH for studied glasses. S/V = 2000 m−1, temperature 90 °C.

N. Rajmohan et al. / Chemical Geology 279 (2010) 106–119

117

Fig. 12. Saturation index of CSH phase versus pH. S/V = 2000 m−1, temperature 90 °C.

5. Conclusions The alteration of nuclear waste glasses and silicate minerals is governed by dissolution and diffusion phenomena. Alteration produces a silica-rich amorphous layer and sometimes crystallized phases as the main alteration products. Part of the amorphous layer can be passivating depending on the alteration conditions (in such cases this material is known as the passivating reactive interphase (PRI)). Modeling the PRI composition taking into account the chemical interactions between the main forming elements (Si, Al, Ca, Zr) is the goal of the present study. For this purpose the study combines two approaches: experimentation (this paper) and modeling. Experiments were carried out to understand the influence of the pH and glass composition on the diffusion coefficient and glass dissolution kinetics, and to generate data for model calibration. Our findings suggest that Si, Al, Ca and Zr have strong interactions that influence the PRI properties (solubility and diffusivity) and thus the glass durability.

Fig. 11. Evolution of saturation index of amorphous silica versus time and pH. S/V = 2000 m−1, temperature 90 °C. For pH variations a mean value has been plotted (based on values at 60, 90 and 150 days.

saturation with respect to amorphous silica and oversaturated with respect to chalcedony, quartz and other high-temperature silica phases. Amorphous gel is the only phase observed by solid characterization methods when the pH is less than 10. Fig. 11 shows the saturation index of amorphous silica for various glasses with respect to time and pH. Fig. 11 indicates that CJ8 and CJ9 leachates are close to saturation whereas CJ7 and CJ2 leachates are undersaturated. This observation suggests that silicon activity is reduced by aluminum. Neglecting the Al effect when modeling the Si concentrations would lead to a significant overestimation by a factor of two to five. At pH 10, secondary phase precipitations were observed for CJ8 and CJ9 glasses, which diminished the saturation state of amorphous silica in solution. In these glasses the saturation index indicates that the solution is oversaturated with respect to classical calcium silicate hydrates available in the data base such as gyrolite, tobermorites (11 Å, 14 Å, 9 Å), etc. (Fig. 12). CSH stoichiometry and solubility should be confirmed precisely. Nevertheless oversaturation is consistent with a kinetically limited precipitation. Slight differences are observed in pH and solution compositions between 10 and 60 days when alteration renewal is observed on CJ8 at pH 10.

• Al, and to a small extent Zr, have been shown to reduce the Si activity at saturation: an Al-rich glass will be more resistant to pure water renewal for this reason alone. Moreover, the lower the Si activity, the higher the pH required for precipitation of hydrated calcium silicates. • Ca interacts mainly with Al and Zr at pH below 8, but starts to interact with Si above pH 9: CSH are observed to form at pH 10 and sustain alteration but the most notable phenomenon is a Ca–Si interaction in the PRI at pH 9. This interaction strongly limits the calcium activity in solution and sharply decreases the reactive diffusion coefficient of water and solvated ions within the PRI. Up to now kinetic models have neglected interactions between elements such as those discussed in this paper (Grambow and Muller, 2001; Frugier et al., 2009; Verney-Carron et al., 2010). Except for the last reference, the reason is that such models were applied in media chemically controlled by glass dissolution, i.e. in a sufficiently narrow pH range (typically near pH 9 for borosilicate glasses) to be able to disregard pH-dependant element interactions within the PRI. In case of Roman archaeological glass, this simplification is acceptable because the PRI contains 85% of silica (Verney-Carron et al., 2008). For nuclear and also basaltic glass we think that this simplification must be revised to improve the models. Precipitation of crystallized secondary phases that consume elements from the PRI is another phenomenon that must be taken into account to model and predict the glass durability. Precise modeling of this phenomena requires a sophisticated PRI model. The next steps of

118

N. Rajmohan et al. / Chemical Geology 279 (2010) 106–119

this work will be: (1) modeling the PRI composition and diffusivity within the GRAAL model framework and (2) validating the model for complex nuclear glass compositions in realistic conditions (e.g. alteration by groundwater). References Advocat, T., Crovisier, J.L., Vernaz, E.Y., Ehret, G., Charpentier, H., 1991. Hydrolysis of R7T7 nuclear waste glass in dilute media: mechanisms and rate as a function of pH. Materials Research Society Symposium Proceedings 212, 57–64. Advocat, T., Jollivet, P., Crovisier, J.L., Del Nero, M., 2001. Long-term alteration mechanisms in water for SON68 radioactive borosilicate glass. Journal of Nuclear Materials 298, 55–62. ANDRA, 2005. Evaluation of the feasibility of a geological repository in an argillaceous formation. ANDRA report available at http://www.ANDRA.fr. Angeli, F., Boscarino, D., Gin, S., Della Mea, G., Boizot, B., Petit, J.C., 2001a. Influence of calcium on sodium aluminosilicate glass leaching behavior. Physics and Chemistry of Glasses 42, 279–286. Angeli, F., Charpentier, T., Gin, S., Petit, J.C., 2001b. 17O 3Q-MAS NMR characterization of a sodium aluminoborosilicate glass and its alteration gel. Chemical Physics Letter 341, 23–28. Angeli, F., Gaillard, M., Jollivet, P., Charpentier, T., 2006. Influence of glass composition and alteration solution on leached silicate glass structure: a solid-state NMR investigation. Geochimica Cosmochimica Acta 70, 2577–2590. Angeli, F., Charpentier, T., Gaillard, M., Jollivet, P., 2008. Influence of zirconium on the structure of pristine and leached soda-lime borosilicate glasses: towards a quantitative approach by 17O MQMAS NMR. Journal of Non-Crystalline Solids 354, 3713–3722. Arab, M., Cailleteau, C., Angeli, F., Devreux, F., Girard, L., Spalla, O., 2008. Aqueous alteration of five-oxide silicate glasses: experimental approach and Monte Carlo modeling. Journal of Non-Crystalline Solids 354 (2–9), 155–161. Ballou, E.V., Leban, M.I., Wydeven, T., 1973. Solute rejection by porous glass membranes: III. Reduced silica dissolution and prolonged hyperfiltration service with feed additives. Journal of Applied Chemistry & Biotechnology 23, 119–130. Berger, G., Schott, J., Loubet, M., 1987. Fundamental processes controlling the first stage of alteration of a basalt glass by seawater: an experimental study between 200° and 320 °C. Earth and Planetary Science Letters 84, 431–445. Berger, G., Claparols, C., Guy, C., Daux, V., 1994. Dissolution rate of a basalt glass in silicarich solutions: implications for long-term alteration. Geochimica Cosmochimica Acta 58, 4875–4886. Bickmore, B.R., Nagy, K.L., Gray, A.K., Brinkerhoff, A.R., 2006. The effect of Al(OH)(4)(-) on the dissolution rate of quartz. Geochimica et Cosmochimica Acta 70 (2), 290–305. Bildstein, O., Trotignon, L., Pozo, C., Jullien, M., 2007. Modeling glass alteration in an altered argillaceous environment. Journal of Nuclear Materials 362, 493–501. Byers, C.D., Jercinovic, M.J., Ewing, R.C., Keil, K., 1985. Basalt glass: an analogue for the evaluation of the long-term stability of nuclear waste form glasses. Materials Research Society Symposium Proceedings 84, 67–83. Cailleteau, C., Angeli, F., Devreux, F., Gin, S., Jestic, J., Jollivet, P., Spalla, O., 2008. Insight into silicate glass corrosion mechanisms. Nature Materials 7, 978–983. Chave, T., Frugier, P., Ayral, A., Gin, S., 2007. Solid state diffusion during nuclear glass residual alteration in solution. Journal of Nuclear Materials 362, 466–473. Chave T., Frugier P., Gin S., Ayral A., submitted for publication. Glass-water interphase reactivity with calcium rich solutions, implications for glass long term modeling, Geochimica et Cosmochimica Acta. Chou, L., Wollast, R., 1985. Steady-state kinetics and dissolution mechanisms of albite. American Journal of Science 285, 963–993. Conradt, R., 2008. Chemical durability of oxide glasses in aqueous solutions: a review. Journal of the American Ceramic Society 91, 728–735. Crovisier, J.L., Fritz, B., Grambow, B., Eberhart, J.P., 1985. Dissolution of basaltic glass in seawater: experiments and thermodynamic modeling. Materials Research Society Symposium Proceedings 50, 273–280. Crovisier, J.L., Atassi, H., Daux, V., Honnorez, J., Petit, J.C., Eberhart, J.P., 1989. A new insight into the nature of the leached layers formed on basaltic glasses in relation to the choice of constraints for the long term modeling. Materials Research Society Symposium Proceedings 127, 41–48. Crovisier, J.L., Honnorez, J., Fritz, B., Petit, J.C., 1992. Dissolution of subglacial basaltic glasses from Iceland: laboratory study and modeling. Applied Geochemistry (1), 55–81. Frizon, F., Gin, S., Jégou, C., 2009. Mass transfer phenomena in nuclear waste packages. In: Wang, Liqiu (Ed.), Advance in Transport Phenomena, vol. 1. Springer Verlag, Berlin-Heidelberg, pp. 31–133. Frugier, P., Gin, S., Minet, Y., Chave, T., Bonin, B., Godon, N., Lartigue, J.E., Jollivet, P., Ayral, A., De Windt, L., Santarini, G., 2008. SON68 nuclear glass dissolution kinetics: current state of knowledge and basis of the new GRAAL model. Journal of Nuclear Materials 380, 8–21. Frugier, P., Chave, T., Gin, S., Lartigue, J.E., 2009. Application of the GRAAL model to leaching experiments with SON68 nuclear glass in initially pure water. Journal of Nuclear Materials 392, 552–567. Geneste, G., Bouyer, F., Gin, S., 2006. Hydrogen-sodium interdiffusion in borosilicate glasses investigated from first principles. Journal of Non-Crystalline Solids 352, 3147–3152. Gin, S., Jégou, C., 2001. Limiting mechanisms of borosilicate glass alteration kinetics: effect of glass composition. Water-Rock Interaction. A.A. Balkema, Villasimius, Italy, pp. 279–282.

Gin, S., Ribet, I., Couillard, M., 2001. Role and properties of the gel formed during nuclear glass alteration: importance of gel formation conditions. Journal of Nuclear Materials 298, 1–10. Gin, S., Jégou, C., Frugier, P., Minet, Y., 2008. Theoretical consideration on the AagaardHelgeson rate law to the dissolution of silicate minerals and glasses. Chemical Geology 255, 14–24. Gislason, S.R., Veblen, D.R., Livi, K.J.T., 1993. Experimental meteoric water-basalt interactions: characterization and interpretation of alteration products. Geochimica et Cosmochimica Acta 57, 1459–1471. Gong, W.L., Wang, L.M., Ewing, R.C., Vernaz, E., Bates, J.K., Ebert, W.L., 1998. Analytical electron microscopy study of surface layers formed on the French SON68 nuclear waste glass during vapor hydration at 200 °C. Journal of Nuclear Materials 254, 249–265. Grambow, B., Muller, R., 2001. First-order dissolution rate law and the role of surface layers in glass performance assessment. Journal of Nuclear Materials 298, 112–124. Grambow, B., Jercinovic, M.J., Ewing, R.C., Byers, C.D., 1985. Weathered basalt hyaloclastites: a natural analogue for the effects of saturation on nuclear waste glass durability. Materials Research Society Symposium Proceedings 50, 263. Guy C., 1989. Mécanismes de dissolution des solides dans les conditions hydrothermales déduits du comportement de verres basaltiques et de calcites déformées. Ph.D. thesis, Université de Toulouse III, France. Hellmann, R., Penisson, J.M., Hervig, R.L., Thomassin, J.H., Abrioux, M.F., 2003. An EFTEM/HRTEM high resolution study of the near surface of labradorite feldspar altered at acid Ph/evidence for interfacial dissolution-reprecipitation. Physics and Chemistry of Minerals 30, 192–197. Hellmann, R., Penisson, J.M., Hervig, R.L., Thomassin, J.H., Abrioux, M.F., 2004. Chemical alteration of feldspar: a comparative study using SIMS and HRTEM/EFTEM. In: Wanty, R.B., Seal, R.R. (Eds.), Water Rock Interaction. A.A. Balkema, Rotterdam, pp. 753–756. Hingston, F.J., Raupach, M., 1967. The reaction between monosilicic acid and aluminum hydroxide: I. Kinetics of adsorption of silicic acid by aluminum hydroxide. Australian Journal of Soil Research 5, 295–309. Houston, J.R., Herberg, J.L., Maxwell, R.S., Carroll, S.A., 2008. Association of dissolved aluminum with silica: connecting molecular structure to surface reactivity using NMR. Geochimica et Cosmochimica Acta 72, 3326–3337. Iler, R.K., 1973. Effect of adsorbed alumina on the solubility of amorphous silica in water. Journal of Colloid and Interface Science 43, 399–408. Jégou C., 1998. Mise en évidence expérimentale des mécanismes limitant l'altération du verre R7T7 en milieu aqueux. Critique et proposition d'évolution du formalisme cinétique. Ph.D. thesis, Université de Montpellier II. Jégou, C., Gin, S., Larche, F., 2000. Alteration kinetics of a simplified nuclear glass in an aqueous medium: effects of solution chemistry and of protective gel properties on diminishing the alteration rate. Journal of Nuclear Materials 280, 216–229. Jercinovic, M.J., Keil, K., Smith, M.R., Schmitt, R.A., 1990. Alteration of basaltic glasses from north-central British Columbia, Canada. Geochimica et Cosmochimica Acta 54, 2679–2696. Jollivet, P., Angeli, F., Cailleteau, C., Devreux, F., Frugier, P., Gin, S., 2008. Investigation of gel porosity clogging during glass leaching. Journal of Non-Crystalline Solids 354, 4952–4958. Jones, L.H.P., Handreck, K.A., 1963. Effects of iron and aluminum oxides on silica in solution in soils. Nature 198, 852–853. Ledieu, A., Devreux, F., Barboux, P., 2005. The role of aluminum in the durability of alumino-borosilicate glasses. Physics and Chemistry of Glasses 46, 12–20. Lee, M.R., Brown, C.L., Hodson, M.E., MacKenzie, M., Hellmann, R., 2007. Characterization of mineral surfaces using FIB and TEM: a case study of naturally-weathered alkali feldspars. American Mineralogist 92, 1383–1394. Michaux, L., Mouche, E., Petit, J.C., Fritz, B., 1992. Geochemical modeling of the longterm dissolution behavior of the French nuclear glass R7T7. Applied Geochemistry Suppl. 1, 41–54. Munier, I., Grambow, B., Fritz, B., Clement, A., 2004. Modeling the alteration gel composition of simplified borosilicate glasses by precipitation of an ideal solid solution in equilibrium with the leachant. Journal of Nuclear Materials 324 (2–3), 97–115. Oelkers, E.H., Schott, J., 1994. The effect of aluminum, pH, and chemical affinity on the rates of aluminosilicate dissolution reactions. Geochimica et Cosmochimica Acta 58 (9), 2011–2024. Ojovan, M.I., Pankov, A., Lee, W.E., 2006. The ion exchange phase in corrosion of nuclear waste glasses. Journal of Nuclear Materials 358 (1), 57–68. Rebiscoul, D., Van der Lee, A., Rieutord, F., Né, F., Spalla, O., El-Mensouri, A., Frugier, P., Ayral, A., Gin, S., 2004. Morphological evolution of alteration layers formed during nuclear glass alteration: new evidence of a gel as diffusive barrier. Journal of Nuclear Materials 326, 9–18. Rebiscoul, D., Frugier, P., Gin, S., Ayral, A., 2005. Protective properties and dissolution ability of the gel formed during nuclear glass alteration. Journal of Nuclear Materials 342, 26–34. Rebiscoul, D., Rieutord, F., Né, F., Frugier, P., Cubitt, C., Gin, S., 2007. Water penetration mechanisms in nuclear glasses by X-ray and neutron reflectometry. Journal of NonCrystalline Solids 353, 2221–2230. Ribet, S., Gin, S., 2004. Role of neoformed phases on the mechanisms controlling the resumption of SON68 glass alteration in alkaline media. Journal of Nuclear Materials 324, 152–164. Samson, S.D., Nagy, K.L., Cotton, W.B., 2005. Transient and quasi-steady-state dissolution of biotite at 22-25 °C in high pH, sodium, nitrate, and aluminate solutions. Geochimica et Cosmochimica Acta 69, 399–413. Scheetz, B.E., Freeborn, W.P.D.K., Anderson, C., Zolensky, M., White, W.B., 1985. Materials Research Society Symposium Proceedings 44, 129. Sicard, L., Spalla, O., Barboux, P., 2004. Study of the kinetics of glass alteration by smallangle X-ray scattering. Journal of Physical Chemistry B 108, 7702–7708.

N. Rajmohan et al. / Chemical Geology 279 (2010) 106–119 Spalla, O., Barboux, P., Sicard, L., Lyonnard, S., Bley, F., 2004. Influence of insoluble elements on the nanostructure of water altered glasses. Journal of Non-Crystalline Solids 347, 56–68. Thomassin J.H., 1984. Étude expérimentale de l'altération des verres silicates dans l'eau douce et en milieu océanique. Apport des techniques d'analyse de surface des solides. Ph.D. thesis, Universite d'Orléans, France. Van Bennekom, A.J., Buma, A.G.J., Nolting, R.F., 1991. Dissolved aluminum in the Weddell-Scotia Confluence and effect of Al on the dissolution kinetics of biogenic silica. Marine Chemistry 35, 423–434. Van Cappellen, P., Qiu, L., 1997. Biogenic silica dissolution in sediments of the Southern Ocean: I. Solubility. Deep-Sea Research II 44, 1109–1128. Van der Lee, J., De Windt, L., Lagneau, V., Goblet, P., 2003. Module-oriented modeling of reactive transport with HYTEC. Computers & Geosciences 29 (3), 265–275. Van Iseghem, P., Aertsens, M., Gin, S., Deneele, D., Grambow, B., McGrail, P., Strachan, D., Wicks, G.A., 2007. Critical Evaluation of the Dissolution Mechanisms of High-Level Waste Glasses. Conditions of Relevance for Geological Disposal (GLAMOR). SCK. CEN, CEA, Subatech, PNNL, and SRNL. EUR 23097, pp. 1–164.

119

Vernaz, E.Y., Dussossoy, J.L., 1992. Current state of knowledge of nuclear waste glass corrosion mechanisms: the case of R7T7 glass. Applied Geochemistry 1, 13–22. Verney-Carron, A., Gin, S., Libourel, G., 2007. Use of natural and archaeological analogs to validate long-term behavior of HLW glass in geological disposal conditions. In: Bullen, T.D., Wang, Y. (Eds.), Water-Rock-Interaction. Balkema, Rotterdam, pp. 659–662. Verney-Carron, A., Gin, S., Libourel, G., 2008. A fractured Roman glass block altered 1 800 years in seawater: analogy with nuclear waste glass in deep geological repository. Geochimica et Cosmochimica Acta 72, 5372–5385. Verney-Carron, A., Gin, S., Frugier, P., Libourel, G., 2010. Long-term modeling of alteration-transport coupling: application to a fractured Roman glass. Geochimica et Cosmochimica Acta 74 (8), 2291–2315. Wolery T.J., 1992. A software package for geochemical modeling of aqueous systems: package overview and installation guide (version 7.0), Lawrence Livermore National Laboratory (USA), UCRL-MA-110662 ed., 1992.