The theft of gold from primary deposits, primary producers and legal title holders is a major problem in many countries around the world. The identification of the source of gold and the means by which it was beneficiated are therefore crucial in determining ownership and legality, especially in the South African context, where large quantities of gold are stolen and illicitly mined, with major economic losses, predominantly from the mines in the Witwatersrand Basin.
Gold usually occurs in its metallic form, often alloyed with a limited range of elements which include Ag, Hg and Cu, in addition to a wide range of trace elements. The concentrations and associated elements are typical for specific genetic types of the gold deposits from which they are mined, and this elemental fingerprint can be used for identifying deposits.
Most stolen gold is melted down into small pieces of gold, or consolidated into small bars, which are then traded by criminal syndicates, and finally laundered via refineries. Before it gets to the refinery, however, it is possible to identify (via the elemental distribution) possible sources and processes employed to recover the gold. The processes used to produce illicit gold are discussed, as well as their effects on the gold metal produced, which may have an effect in masking the origin of the gold.
The legislation in South Africa regarding the mining and possession of precious metals is discussed, as well as the establishment of the South African Gold Database as a means for combatting the theft of gold from the major goldmining companies in South Africa and the reason for the type of sample selected to be representative of the gold produced at each mine – unrefined d’ore gold.
The different analytical techniques utilised for the analysis of gold grains, and their development and improvement of the past few decades are then discussed. It is important that a method to analyse stolen gold be fairly straight-forward and repeatable, so the different types of instrumention with their strengths and weaknesses are discussed. It is shown that the best method for the analysis of all types of gold material, from a comparison point of view and not for absolute analysis, is laser-ablation inductively-coupled-plasma mass-spectrometry (LA-ICP-MS). The development of this technique and its applications for gold are evaluated. The composition of the d’ore samples across the Witwatersrand Basin was analysed in order to determine compositional trends in the extracted gold between mines and goldfields, focussing on three goldfields in particular – Free State, Klerksdorp and West Wits. It is shown that there are two major groups of gold samples, spatially defined, within these three goldfields, which are clearly distinguished on the basis of Pb isotopic systematics and Ni content. It is also shown that the different goldfields can be distinguished on the basis of the d’ore composition, and that to a lesser extent discrimination between individual gold mines in the West Wits and Free State goldfields is possible.
Illicit gold recovered from the vicinity of the Driefontein mine was analysed and compared to the samples in the database in order to determine whether the use of d’ore gold samples, produced with the carbon-in-pulp method, would be sufficient to identify illicit gold recovered using mercury amalgamation. It is shown that depending on the elements used, such identification can be made. In addition it is shown that, due to differences in beneficiation methods and the stage within the beneficiation process, the specific type of material processed to extract the gold can be identified. This has important intelligence benefits when investigating the theft and the place from where the material was stolen.
The identification of two different types of gold mineralisation within the Witwatersrand Basin, and the use of the database to identify stolen gold from the basin, showed the applicability of the methodology. A second scenario was therefore examined, dealing with gold from two different genetic sources from northern South America, in which illicit gold bars were compared to artisanally recovered gold from two countries, and gold samples from the suspected source as a control. It is shown that in this second case, the two ore deposits could be very clearly distinguished, and even though the artisanal gold was recovered by amalgamation, a clear identification could be made with the control sample. The discrimination made here was much more distinct than that made within the Witwatersrand basin, and this is considered to be due to a single source of gold compared to possible multiple sources, as is the case in the Witwatersrand Basin.
Each mine within the Witwatersrand basin extracts gold from one or more reefs, so the variation within and between reefs at the Driefontein mine was studied, using gold prills as samples. The identified variation is discussed in terms of depositional and post-depositional processes, and it is shown that there appear to be two populations of gold within the Ventersdorp Contact Reef (VCR), i.e. one alluvial and the other hydrothermal. For the Carbon Leader Reef (CLR), however, the prill compositions differ from those from the VCR, and show that the gold mineralisation in this reef could have arisen from another, different process or combination of processes.
Individual gold grains from different reefs were then analysed, in order to determine to what degree the d’ore and prill results correlate with the gold grain compositions. Four distinct populations were found, each linked to a specific and different mineralisation process. Within the Vaal Reef, alluvial gold of epithermal vein gold from a granite-greenstone terrane was identified, as well as a second population of gold with an anomalous composition (containing U and Pb) which was interpreted as being of a remobilised origin, being formed during the first stages of diagenesis, and a third population of gold typical of low temperature hydrothermal mineralisation. Gold-bearing carbon from the CLR was also analysed, which revealed gold mineralisation attributed to precipitation of colloidal organometallic compounds. The uranium mineralisation associated with the CLR is shown to be co-eval with the gold on microbial mats, and this mineralisation was the source of the Au (and U and Pb) which was remobilised in an initial post-depositional event. The two populations identified within the VCR are confirmed as being of hydrothermal and alluvial origins.
The variation within the Witwatersrand Basin, due to differences between source areas of original detrital material and hydrothermal events post-deposition, provide sufficient discrimination to enable the identification of the origin of seized illicit gold. In addition, the use of LA-ICP-MS enables the discrimination of different gold materials of natural origin to a level not seen before, with trace element resolution at micron scales, resulting in a much better and more precise identification of genetic processes.