Technical Report NTB 22-07

Nuclear Waste Consortium Research Programme on Corrosion and Gas Generation: Phase 1 Final Report

The Nuclear Waste Consortium project was co-sponsored by the National Cooperative for the Disposal of Radioactive Waste (Nagra, Switzerland), the Nuclear Waste Management Orga­ni­zation (NWMO, Canada) and the National Agency for Radioactive Waste and Enriched Fissile Material (ONDRAF/NIRAS, Belgium). Research was performed at CanmetMATERIALS (Hamilton, Canada).

The objective of the research was to better understand the corrosion and hydrogen generation (or oxygen consumption) behaviours of several metallic materials under simulated deep geological repository environments (both for high-level and low-/intermediate-level waste). The two primary materials of interest were copper and carbon steel, although cements have also been studied and a number of other metallic materials were introduced toward the end of the project for which only preliminary results are presented. Whilst this report documents the findings of the concluded study, the research remains in progress.

The method for monitoring material degradation was primarily to accumulate hydrogen, an end product of the corrosion reaction, in the test cell. Hydrogen was measured by either changes in system pressure, which were directly monitored using a transducer, or by purging the hydrogen from the test cell, measuring it with a solid-state sensor. This latter approach has excellent sensitivity, permitting monitoring of corrosion rates as low as 0.01 nm/year – a fraction of a monolayer per year. However, both techniques are limited to calculating averaged and uniform corrosion rates. For the majority of materials, in particular copper, this assumption is visually justified but will be confirmed upon decommissioning of the test cells through microscopic exami­nation of the test specimens.

Copper, a proposed protective cladding for used fuel containers, was studied in pH-neutral environments representative of anticipated deep geological repository conditions as well as under more extreme conditions (i.e. mildly acidic environments) to stimulate a measurable response. Under conditions of near-neutral pH and at 75 °C, the uniform corrosion rate of copper wire was always found to be significantly less than 1 nm/year, often declining to below the limit of detection (0.01 nm/year) over the course of several years. Both chloride and low concentrations of hydrogen sulphide, a chemical species of microbial origin that may be introduced to the repository, were found to stimulate corrosion on the previously-inert surface, but in a short-lived manner and with a peak corrosion rate of only 0.4 nm/year. This was speculated to be due to a surface ripening effect, with the adsorbed anionic species enabling relaxation of copper atoms from the surface facilitating both oxidation (resulting in the monitored hydrogen) and surface diffusion. This latter process ultimately resulted in a more thermodynamically stable surface that did not yield measurable corrosion upon further addition of chloride or sulphide. This process is two-dimensional, suggesting that bulk corrosion cannot occur under similar conditions, as anticipated by classical thermodynamic calculations.

Several carbon steels were studied, primarily under conditions dominated by cement chemistry, representative of various periods in the anticipated life of a deep geological repository. Experiments were performed under unsaturated and water-saturated conditions. Temperatures ranged from 50 °C to 80 °C, and electrolyte pH (for simulated cement pore waters) ranged from 13.5 for young cement water, to 12.5, for portlandite-saturated water in degraded cement. In many cases, steels were embedded within cements directly. Tests were predominantly anoxic, but a limited number of initially-aerated tests were also conducted.

Cement alone was found to not only generate significant hydrogen over the course of several years (and in specific cases, in quantities directly comparable to the steel), but also to rapidly consume oxygen if initially present.

Where overlap between experimental conditions permitted comparison, it was generally observed that the annealed wire corroded significantly faster than steel rod, itself corroding faster than the SA516 grade 70 steel plate, the material closest in composition and microstructure to the overpack of the Belgian “supercontainer”. Whether this trend is due to composition and/or microstructure is not known.

Steels in 100% relative humidity at 50 °C corroded at 10 nm/year (wire) or less than 0.5 nm/year (rod) and, after several years, did not have a full surface coverage of corrosion product. Steel wire in young cement water was found to corrode steadily at a rate equivalent to 400 nm/year, regardless of the presence of a pre-formed corrosion product, but at 80 °C, following an initial period of corrosion of the order µm/year, the rate rapidly transitioned to a continuous 10 nm/year. Pre-corroded steel plate, also in young cement water at 80 °C, corroded at a uniform rate of 0.3 nm/year, significantly less than the wire. In portlandite-saturated water at 50 °C, steel wire corroded at a declining rate of between 1 and 10 nm/year with steel rod under comparable circumstances, at 0.1 nm/year. For steel rods embedded within cement at 50 °C, uniform corrosion rates of less than 0.5 nm/year were calculated in 100% relative humidity, decreasing further to 0.1 nm/year when immersed in saturated portlandite.

Overall, it was observed that in many cases, the corrosion behaviour of steel was not at steady state after five years, and with the exceptions of wire in young cement water, all steel in alkaline environments were trending toward a sub-1 nm/year uniform corrosion rate