Radioactive disequilibrium and geochemical modelling as evidence of uranium leaching from gold tailings dumps in the Witwatersrand Basin

Abstract

Gold tailings dams from the Witwatersrand Basin usually contain elevated amounts of heavy metals and radionuclides. Uranium, in the form of uraninite (UO₂) and brannerite (UTi₂O₆), is normally associated with gold-bearing ores in the basin. As a result of acid mine drainage (AMD), uranium is released into groundwater and fluvial systems. Its transport, retardation and immobilisation depend strongly on the uranium species and prevailing geochemical conditions. This study was aimed at the quantitative assessment of the distribution of uranium based on measurement of its radioactivity and modelling of its geochemical speciation. Analyses of tailings, water and sediment in areas of previous mining were performed. The results indicate that there is active leaching of uranium from the tailings, transport of soluble uranium species through water systems, with subsequent deposition of insoluble uranium species in sediments of fluvial systems. Analysis of tailings material indicated that mobilisation and transportation of uranium from the tailings resulted in its decoupling from its progeny which remained largely unaffected by the weathering effects. Mobilisation occurs as uranium is oxidised to the U(VI) state which dominates aqueous chemistry, particularly via complexation with most ligands. The U(VI) is reduced to U(IV) which is immobile and is subsequently deposited in the wetland sediments downstream from the primary acid mine drainage. Geochemical modelling of uranium speciation revealed the two most influential hydrogeochemical facies in uranium mobility, namely a sulphate-dominated AMD system and a lime-neutralised carbonate-dominated system. In both cases, the uranium was shown to be soluble throughout a very wide pH regime, thus yielding important information for risk assessment considerations.

Keywords:

• radioactive disequilibrium
• Witwatersrand Basin
• geochemical modelling

Acknowledgements

The authors would like to thank the National Research Foundation for financial support and Mr. A. Kwelilanga of iThemba LABS for technical assistance.