Novel approach to study diffusion of hydrogen bearing species in silicate glasses at low temperatures

Abstract

Diffusion of hydrogen bearing species in glasses plays a significant role in numerous applications in commercial as well as scientific domains. The investigation of diffusion of water in glasses at low temperatures led to experimental and analytical difficulties in the past. We present a new approach that lets us overcome these complications. Diffusion couples of An50Di50 glass (mol %, NBO/T = 0.67) were produced by coating anhydrous glass substrates with thin films of hydrated glass (~200 nm, ~2 wt% H₂O) using pulsed laser deposition (PLD). Bonding the diffusant to the glass matrix of the thin film instead of using free water at the interface during experiments precludes other glass altering processes such as dissolution and precipitation. This allows us to confidently interpret the measured profiles to be a result of diffusion only. Nanoscale concentration profiles that result from diffusion at low temperatures on experimentally feasible time scales were measured with the Nuclear Resonance Reaction Analysis (NRRA, ¹H(¹⁵N,αγ)¹²C). The non-destructive nature of NRRA enables us to observe and better understand the evolution of diffusion profiles with time within one sample. Evaluation of the sample quality by EPMA, SEM, optical microscopy, Rutherford backscattering spectroscopy (RBS), and NRRA was performed and confirmed the suitability of the samples for diffusion studies. Experiments at 1 atm in a box furnace and at 2 kbar in a CSPV (pressure medium = water) and an IHPV (pressure medium = Argon) prove that the diffusion couples can be used under various experimental conditions. We present diffusion profiles that were measured in experiments carried out in these devices and discuss the distinct features of each that result from different boundary conditions in the experiments.

Introduction

The incorporation of water into the structure of silicate glass or melt can lead to critical changes in the materials properties (e.g. viscosity and density), with major implications for phase relations as well as mechanical behavior. The time-dependence of such incorporation is controlled by diffusion rates of H-bearing species in silicate glasses and melts. Diffusion also controls the atomistic mechanisms of processes such as viscous flow. Thus, a range of geological processes, such as bubble nucleation in magmas (Allabar and Nowak, 2018), seafloor weathering (Berger et al., 1987), or aqueous alteration of amorphous material in asteroids (Dobrică et al., 2019), depend on the rates of diffusion of H-bearing species in natural amorphous / molten materials. Similarly, diffusion of H-bearing species is of importance in many material science problems in the areas of electronic devices (Kostinski et al., 2012) and fiberglass material (Lezzi et al., 2015; Glazneva et al., 2019). The relevant processes occur over a range of conditions (temperature, oxidation state) and in many different compositions and therefore it is important to obtain a quantitative understanding of diffusion rates as a function of temperature, composition, and other variables. Consequently, a large body of data for H-diffusion in amorphous materials are available from the geological as well as the glass-science literature (Zhang et al., 1991; Zhang et al., 1997; Zhang and Behrens, 2000; Doremus, 2000; Hepburn and Tomozawa, 2001; Berger and Tomozawa, 2003; Ferrand et al., 2006). Nevertheless, significant open questions remain to be addressed. Two of these are: (i) Are diffusion coefficients measured from experiments where water is placed in direct contact with a glass surface compromised by reaction between the glass and water? and (ii) Is the speciation of H in glasses, obtained from extrapolation of results from higher temperatures, valid at lower temperatures where many of the applications occur? The second question is relevant because the net flux of H depends on the speciation and the relative rates of diffusion of different species.

The questions remain open issues because of experimental and analytical difficulties. For example, using a hydrated glass rather than free water as a source of H helps to overcome the reaction-diffusion problem, but it is difficult to obtain tight physical contact between a hydrated and an anhydrous glass at low temperatures. As a result, most data at low temperatures (e.g. Tomozawa and Tomozawa, 1989; Davis and Tomozawa, 1995; Dersch et al., 1997; Tomozawa et al., 2001; Berger and Tomozawa, 2003; Sato and Tomozawa, 2004; Oehler and Tomozawa, 2004; Ferrand et al., 2006; Kuroda et al., 2018; Glazneva et al., 2019) have been obtained using a free water phase as a source of H. Analytically, diffusion rates are slow at low temperatures and therefore it is essential to have analytical tools that are capable of measuring concentration gradients of H over very small spatial scales.Two tools have been used conventionally for this purpose, each with its own strengths and weaknesses: Infrared spectroscopy (Zhang and Stolper, 1991; Behrens and Nowak, 1997; Freda et al., 2003; Behrens et al., 2004; Sato and Tomozawa, 2004; Liu et al., 2004; Okumura and Nakashima, 2006; Anovitz et al., 2008; Behrens and Zhang, 2009; Ni et al., 2009; Fanara et al., 2013) has a spatial resolution of 10's of microns, but allows the concentrations of (OH) species as well as molecular H₂O to be determined separately. There are issues related to the calibration of the tool to obtain absolute concentrations. The second tool, secondary ion mass spectrometry, SIMS, has also been used in a number of studies (Lapham et al., 1984; Anovitz et al., 1999, Anovitz et al., 2009; Liritzis and Diakostamtiou, 2002; Riciputi et al., 2002; Gin et al., 2015; Kuroda et al., 2018). It offers a higher spatial resolution (in submicron region), but determines only the total concentration of H and requires extensive efforts for calibration. Notably, both of these tools require calibrations that are specific to particular compositions (matrix effects).

In this study we present a new approach for determining the diffusion rates of H in glasses that allows one to address the above issues. We use the tool of pulsed laser deposition (PLD) to produce a tight contact between a water-rich source and an anhydrous glass that not only gets around the problem of reactions with a free water phase, but also helps to address some other experimental issues (e.g. the undesired presence of concentration gradient of elements other than H). The diffusion couples are miniaturized to nanoscales, so that experiments can be carried out over short durations. For the analysis of H concentration profiles we make use of Nuclear resonance reaction analysis (NRRA) carried out at a particle accelerator. This has been used in the past in materials science as well as geosciences for the study of H-diffusion (Laursen and Lanford, 1978; March and Rauch, 1986; Xiong et al., 1987; Bell et al., 2003; Tomozawa et al., 2012; Le Guillou et al., 2015; Cherniak et al., 2018; Hoffmann et al., 2018; Mai et al., 2018). The tool has a major advantage in that it requires no calibration, is relatively insensitive to matrix effects, and provides absolute concentrations of H over nanometer scales. The measurement of total H without differentiating between individual species of H that may be present is the reason we discuss our results in terms of H diffusion or diffusion of H-bearing species throughout this paper. However, plotted profiles are presented in units of H₂O wt%, because this unit is likely more familiar to the reader. This is done by converting H at% to H₂O wt% assuming that all H is present as H₂O (see supplementary data). In practice of course, H would be partitioned between different species such as molecular H₂O and (OH)- and modelling the profiles using different speciation models may provide insights regarding such speciation. Such modelling on a series of experiments will be the topic of a future paper.