Predicting zeolites’ stability during the corrosion of nuclear waste immobilization glasses: Comparison with glass corrosion experiments

Abstract

During the long-term corrosion of nuclear waste glasses under nuclear waste disposal conditions, the precipitation of zeolitic phases has been linked to a delayed acceleration in glass corrosion (known as “Stage III”). Hence, predicting the thermodynamic propensity for zeolites to form upon the dissolution of nuclear waste glasses is key to ensure their long-term performance. Here, we compile a unified, internally-consistent thermodynamic database “clay20” to estimate the stability of clay and feldspar phases relevant to nuclear waste immobilization glasses, including beidellite(Mg, Ca, Na, K), kaolinite, montmorillonite(Mg, Ca, Na, K), nontronite(Mg, Ca, Na, K), saponite(Ca, Na, K), and albite. Based on this, we report a geochemical modeling method allowing us to predict the stability of secondary phases (including zeolites, calcium–silicate–hydrate gels, and clays) upon the dissolution of nuclear waste immobilization glasses. We show that this approach offers a realistic description of the stability of the secondary phases forming during the dissolution of two archetypical model nuclear glasses (namely, the International Simple Glass, ISG, and WVUTh-203) under conditions relevant to nuclear waste disposal (T = 90 °C, p = 1 bar) as a function of pH. We find that the formation of silica and clay secondary phases is thermodynamically favored at low pH (pH < 10), whereas, in contrast, zeolite (analcime) and calcium–silicate–hydrate phases are favored at high pH (pH > 10.5). This suggests that thermodynamics (i.e., not solely kinetics) might play a key role in determining the range of solution pH wherein stage III corrosion may occur, i.e., when zeolite formation is favored.

Introduction

For reasons of cost, efficiency, and safety, vitrification is the method of choice to immobilize high-level nuclear wastes, namely, by chemically incorporating radionuclides within glasses [1], [2], [3]. In many countries, produced nuclear waste immobilization glasses are (or are expected to be) stored in deep geological disposal sites, wherein radionuclides will decay over time in (fairly) isolated environments [4], [5], [6], [7]. During the disposal period, the contact between nuclear waste immobilization glasses and groundwater (and subsequent dissolution) is the most likely mechanism by which radionuclides may be released from the glass wasteforms [6,[8], [9], [10]]. As such, it is critical to understand and predict glass corrosion under disposal conditions to ensure the safety of nuclear waste immobilization operations.

Numerous experiments have been conducted to understand the behavior and reactivity of nuclear waste immobilization glasses under disposal conditions [11], [12], [13], [14]. Several mechanistic models have been developed to describe the mechanism of glass corrosion [8,[15], [16], [17], [18]]. Glass dissolution under disposal conditions is typically described as following three stages: (Stage I) initial fast dissolution far from saturation (forward rate), (Stage II) establishment of a low steady-stage, residual dissolution rate due to the solution saturation and the formation of alteration layers at the surface of the glass, and, in select cases, (Stage III) a delayed acceleration of glass alteration [6,19,14]. Although Stage III is often referred as a corrosion “resumption” in the literature, we adopt herein the more general terminology of “delayed acceleration.” This choice is motivated by the fact that the corrosion rate observed in Stage III is often not as fast as the initial one observed in Stage I—so that it is at this point unclear whether the underlying dissolution mechanisms associated with Stage I are really “resuming” during Stage III. Understanding and predicting the occurrence of Stage III is important as such acceleration of corrosion tends to increase by several orders of magnitude the amount of radionuclides that may be released to the environment [20], [21], [22], [23], [24].

Previous observations have suggested that the occurrence of Stage III strongly depends on the glass composition and solution pH, and is associated with the rapid precipitation of zeolitic secondary phases [20], [21], [22], [23], [24]. For instance, the formation of analcime (NaAlSi₂O₆⋅H₂O, a zeolite phase) is commonly observed for most nuclear waste glasses in highly basic conditions (pH₉₀ °_C > 10.5) [14,[25], [26], [27], [28], [29]] and/or at high temperatures (T > 95 °C) [30], [31], [32]. Such high pH conditions can be achieved under disposal conditions, especially when the glass is in contact with concrete—since cement paste pore solutions typically exhibit pH values ranging from 13 to 14 [33]. Besides analcime, other zeolitic phases such as phillipsite, merlinoite, and chabazite have been observed to form in cement pore water at pH 12.6 at a temperature of 50 °C [24]. In addition to zeolitic phases, cementitious hydrated phases (e.g., calcium–silicate–hydrate, C–S–H) typically constitute the vast majority of the secondary phases forming at high pH [34]. In contrast, non-zeolitic secondary phases (e.g., clays) are typically observed at lower pH (pH₉₀ °_C < 10) and/or lower temperature (T < 95 °C)—wherein the formation thereof typically does not result in Stage III corrosion [24,29,35]. Besides pH, the nature of the secondary phases forming upon glass corrosion depends on the composition of the solution and of the dissolving glass [24,14,26,29,31]. Importantly, it remains unclear whether the formation of zeolites secondary phases (and, hence, the occurrence of Stage III) at high pH is controlled by their thermodynamics or kinetics, or, conversely, whether the absence of zeolite formation at moderate pH is due to the fact that (i) zeolites formation is thermodynamically unstable or (ii) their nucleation and growth is extremely long in such pH conditions [11,12]. Answering these questions requires an accurate description of the relative thermodynamic stability of competing secondary phases (e.g., zeolites vs. clays) as a function of pH.

In this study, we introduce a new thermodynamic database to describe the stability of clays and feldspars. This database is combined with existing thermodynamic databases of relevant phases (namely, zeo19 for zeolites [36] and cement hydrate phases [37]) to predict the relative propensity for secondary phases to form upon glass corrosion with a unified Gibbs Energy Minimization (GEM) solver [38,39]. This modeling approach applies to Na₂O–CaO–MgO–Fe₂O₃–Al₂O₃–B₂O₃–ZrO₂–SiO₂–H₂O systems and simulates the precipitation of the most stable phases as a function of pH. This methodology is applied to simulate the dissolution of two model nuclear waste glasses, i.e., the International Simple Glass (ISG) and WVUTh-203 type glass, over a wide range of solution compositions (with glass-to-liquid ratios ranging from 0 to 1) and pH values (from 4 to 14) under disposal conditions (T < 95 °C, p = 1 bar). Predictions are compared with available glass leaching experimental data for validation.