Technical Report NTB 87-16

Trace Element and Microbiological Studies of Alkaline Groundwaters in Oman, Arabian Gulf: A Natural Analogue for Cement Pore-Waters

The microbiology and trace element chemistry of some highly alkaline groundwaters in Oman in the Arabian Gulf have been examined. The significance of the study to the understanding of processes in and around a radioactive waste repository lies in its value as a natural analogue of the high-pH sustained by cement in its pore-waters (it has been suggested that pH 11 – 13 might be maintained for up to a million years) and in the near-field aqueous environment. Cement and concrete may occur in the repository as components of waste, immobilisation matrix and backfill and as engineering materials. In addition to being highly alkaline, the near-field of a sealed repository will also become chemically reducing due to early-stage oxidation/corrosion reactions in the repository. Thus the safety analysis must, amongst other things, consider the chemical and microbiological influences on radionuclide behaviour in this environment.

Microorganisms will inevitably be introduced into a repository, and could participate in reactions which cause difficulties in predicting performance: e.g. deterioration of metal and of concrete which are accelerated by sulphur and iron bacteria, and production of gases by biodegradation of waste material, notably methane generation in which methanogenic bacteria are active. However, it has been thought that microbial proliferation is strongly inhibited in a highly alkaline environment such as that predicted for a cement-dominated repository, and is further constrained by the rapid development of strongly reducing conditions. This natural analogue study has sought more evidence of the extent of microbial viability and growth in alkaline groundwaters.

Equilibrium chemical thermodynamics provides the usual approach for evaluating elemental solubilities and speciation in aqueous solution. However, application to the radionuclide trace elements of importance in nuclear waste disposal (i.e. actinides and their daughters and radionuclides which are produced by fission or activation) is limited by the availability of relevant data and uncertainty over the character of aqueous ionic species. The natural analogue approach has been applied here in an attempt to check some of these predictions against observed elemental abundances in a natural environment.

There are few natural environments, or artificial situations sustained over long time-scales, in which evidence of the effects of high pH on chemical and microbiological behaviour of the system can be obtained. Aged concrete structures obviously present one possibility, whilst alkaline lakes, both natural and artificial, could also be studied. The present study has investigated the natural environments created around the discharge points of hyperalkaline groundwaters which occur only under rare geological conditions. This type of 'natural analogue' study provides evidence which is complementary to that gained in laboratory-based studies, in that the natural environment might mimic the long-term development of chemical and microbial conditions which cannot be achieved in short-term experiments.

The alkaline springs in Oman were selected for the present investigation because the hydrochemistry and some associated mineral precipitates closely resemble the inferred conditions in cement pore-waters, even to the extent of portlandite (Ca(OH)2) precipitate having been observed. Highly reducing conditions also occur in these groundwaters and hydrogen gas is evolved at some of the springs, thus increasing their value as a natural analogue for the repository environment. Samples were collected from several of these alkaline springs in order to isolate and identify the bacterial populations as far as possible and to test some of the predicted constraints on solubilities of radionuclides in the highly alkaline environment.

The microbial genera found in the spring-waters are similar to those in less extreme soil and water environments. The wide range of both aerobic and anaerobic bacteria isolated was enhanced by the local variations in conditions at specific sampling points, particularly with respect to redox conditions and nutrient availability. Indeed, at least some of the heterotrophic microorganisms could be attributed to contamination of the spring-waters by animals. Several heterotrophic bacteria were isolated which were able to grow beyond pH10 and some beyond pH11. Obtaining indubitable proof of viability above these values, however, presents considerable experimental difficulties. Most strains were found to be alkalotolerant. By standard criteria, the two alkalophilic strains isolated were shown to be strict aerobes and thus may be of little importance to the anaerobic repository environment.

Of particular interest is the demonstration of the presence of sulphate-reducing bacteria (SRB) in particular samples of both water and sediment. SRB are of relevance to waste disposal because of their participation in reactions which corrode steel and which may affect concrete integrity. They may not be present in large numbers but their existence shows that they are able to colonise, albeit in a dormant or near-dormant state, environments with a pH in excess of 11. The SRB are unable to grow at high pH in nutrient-rich culture, and this suggests that pH is the growth-limiting factor under experimental conditions. However, activity measurement data suggest that the high pH is not the principal constraint on microbial growth in the alkaline spring-waters. The hydrochemical data on the spring-waters indicate that microbial populations may be limited by supply of nutrients, especially carbon, nitrogen and phosphorus. It should be borne in mind that the phosphorus content of most Portland cements is much higher than was encountered in the spring-waters. The low levels of organic carbon present in the spring-waters are similar to those in many natural environments which, although at the lower limit for uptake and utilisation by most bacteria and fungi, can support oligotrophic activity. Heterotrophic bacteria, which predominate in the alkaline spring-waters and which could be present in a waste repository, can lead to production of organic acids by hydrolysis of cellulose materials in the waste inventory. These organic acids could speed up neutralisation of the high alkalinity and the organic chelating/complexing species formed may solubilize radionuclides or may act as further substrates for microbes.

The trace elements investigated for comparison of measured solution concentrations with equilibrium values predicted by thermodynamic modelling were Se, Pd, Sn, Zr, Ni, U and Th. Predicted equilibrium solution concentrations under these highly alkaline, reducing conditions are considerably lower than state-of-the-art analytical detection limits for Pd and Sn, and close to detection limits for Ni, Th and U (though there are discrepancies between predictions for U made using databases from different sources). The observed solution concentrations of these trace elements were mostly below detection limits, with significant values only for some Zr and U determinations. Data for Pd and Ni are considered to be unreliable due to analytical interferences. The measured data for Se, Zr and possibly Ni are significantly lower than predicted equilibrium concentrations, and could indicate that appropriate solid phases or thermodynamic data have been omitted from the model. However, the interpretation and significance of these observations is dependent on there being natural sources of the elements which buffer concentrations at solubility limits. For example, geochemical evidence suggests that Zr may be particularly depleted in the source rocks. Ni is perhaps the most interesting in this respect since the known rock geochemistry suggests a reasonable supply of Ni. Below-detection-limit analyses for Th are at least consistent with model predictions of low concentrations. Comparison of similarly low U analyses with predicted values varies widely, depending on the database used and the redox state; this example highlights some significant discrepancies between the databases which would benefit from further investigation and renewed attempts at validation.

The abundances and isotope compositions of U and Th in alkaline groundwater samples passed through different filter sizes suggest that, in one case, U and daughter nuclides may be associated with colloidal material. Colloids are acknowledged to be of potential importance in repository safety analysis, but tend not to be stable at high pH or high salinity. Therefore the indication of U and daughter Th being present in stable colloidal form here, though based on limited data, is potentially interesting and worthy of further investigation.

Therefore positive quantitative or semi-quantitative 'validation' of the predictive models has not been achieved in general, although the attempt has highlighted some potentially significant points of comparison between natural solute abundances and the predictions. It has also shown that this comparison of concentrations and species distributions may be feasible analytically for some elements such as Se, Zr, Ni, U and Th, though natural elemental sources and their reaction rates will inherently limit the extent to which models can be validated.