OR/14/048 Summary

Rochelle, C A, Purser, G, and Milodowski, A E. 2014. Results of laboratory carbonation experiments on NRVB cement. British Geological Survey Internal Report, OR/14/048.

Some repository concepts envisage the use of large quantities of cementitious materials — both for repository construction and as a buffer/backfill. However, some wastes placed within a subsurface repository will contain a significant amount of organic material that may degrade to produce carbon dioxide. This will react with cement buffer/backfill to produce carbonate minerals such as calcite, which will reduce the ability of the buffer/backfill to maintain highly alkaline conditions and as a consequence its ability to limit radionuclide migration. The reaction may also alter the physical properties of the buffer/backfill. The work involved in this study investigates these processes through elevated pressure laboratory experiments conducted at a range of likely future in situ repository conditions. These will provide information on the reactions that occur, with results serving as examples with which to test predictive modelling codes. This report details a series of batch experiments to study carbonation of Nirex Reference Vault Backfil (NRVB) cement.

Thirty-two static batch experiments were pressurised with either CO₂, or with N₂ for ‘non- reacting’ comparison tests at 20°C or 40°C, and 40 or 80 bar. Twenty-six of these were left to react for durations of between 10–40 days, with six more left to react for a year. The aim of them was to help investigate mineralogical and fluid chemical changes due to the diffusional ingress of CO₂ into unconfined NRVB samples measuring 2.5 cm in diameter and 5 cm long.

All the cement samples showed rapid reaction with CO₂, manifested by a colour change from grey to light brown. Petrographic analysis of the reacted cement revealed that this colour change reflected the breakdown and dissolution of primary calcium ferrite and calcium alumina-ferrite (CAF) cement clinker phases (e.g. brownmillerite, Ca₂(Al,Fe)₂O₅ to form calcium carbonates and finely-disseminated free ferric oxide (probably hematite, Fe₂O₃), as a result of reaction with CO₂ to give a ‘rusty’ colour. It should be noted that his is not an oxidation reaction as the iron is present as Fe³⁺ in the original cement phases.

The cement blocks remained intact, even after prolonged exposure to CO₂-rich fluids. Carbonation was associated with an increase in weight of up to 8.5% during CO₂ uptake, though the samples did not change in overall size. There is potential therefore, for carbonation to immobilise ¹⁴CO₂ if that were present. Free-phase CO₂ gave slightly more reaction than dissolved CO₂, possibly because of its higher concentration and greater ability to penetrate the samples. In terms of major reactions during carbonation, these were the breakdown of portlandite, calcium silicate hydrate (CSH) phases, calcium aluminate (or calcium aluminate hydrate) phases, and ettringite-like phases, and the formation of carbonate phases and silica gel. Carbonation also revealed that heterogeneity within the cement samples had a major impact on migration pathways and extent of carbonation. This heterogeneity may have been a result of casting, and was only observed in some of the samples studied. It led to faster carbonation in some areas, and may account for some of the differences observed in the reacted cement samples. Such heterogeneity may be present within a repository, and should be taken into account when assessing repository performance.

Carbonation resulted in a series of reaction fronts that moved through the cement over time. These fronts separated several reaction zones: Zone 1 = minor carbonation with minimal apparent volume change, Zone 2 = partial carbonation and very localised shrinkage, Zone 3 = complete conversion of portlandite and CSH with localised shrinkage associated with the development of abundant calcium carbonate-sealed microfractures, Zone  4 = dissolution of initially-formed carbonate minerals in the outermost parts of the sample by the surrounding, slightly-acidic water. The shrinkage in Zone 2 was expressed as small fractures (typically several mm long), though these do not appear to extend beyond this zone. Zone 3 contained an intense anastomising ‘3D chickenwire’ meshwork of interconnected, higher-density, carbonate-filled microfractures (typically on a 10s–100s µm scale) that separated silica-rich areas having lower-density and high porosity, and sub-parallel ‘relic’ reaction fronts. The small fractures of Zone 2 appear to have filled with secondary precipitates in Zone 3. It would be useful to understand how these reaction zones evolve over longer timescales, and investigate whether they have the potential to become narrower (or even merge together), with more efficient sealing of porosity that might ‘armour’ the cement from further carbonation.

Appreciable amounts of two Cl-rich phases were formed at the boundary of Zones 2 and 3. At least one of these phases appears to have been enhanced by the presence of CO₂. The formation of Cl-rich phases within a repository could be beneficial as it might help to immobilise ³⁶Cl leaching from the waste. Further work is needed to precisely-identify these Cl-rich phases and ascertain the impact such phases could have on ³⁶Cl retardation and repository safety.

Carbonation features and secondary phases observed in these experiments using a relatively porous/permeable cement, bear many similarities to those found in far lower porosity/permeability borehole cements used in CO₂-storage operations. There are also similarities to samples of naturally-occurring CSH phases which have been naturally-carbonated over prolonged timescales. A number of common carbonation processes may be operating in all these systems.