New analytical developments have made radiogenic helium (4He) applicable to archeological gold artifacts for age determinations. Here we report the application of the U/Th–4He method to the direct dating of gold from the historically important gold deposit in Diamantina, Minas Gerais, Brazil. The U/Th–4He age of 515 ± 55 Ma for the Diamantina gold is corroborated by a new U/Pb age of 524 ± 16 Ma for rutile recovered from auriferous pockets. These ages tie the Diamantina gold mineralization to the Brasiliano orogenic event, in the context of the Gondwana amalgamation. Our results indicate that U/Th–4He dating of gold is possible, opening new perspectives for the dating of gold deposits without assuming contemporaneity between gold and datable hydrothermal minerals.


The radiometric dating of gold deposits almost invariably requires the assumption that datable minerals, e.g., monazite and arsenopyrite, are coeval with the gold with which they are spatially associated. Only exceptionally, natural gold accumulates enough Re to allow the direct dating of gold using the decay of 187Re to 187Os (Kirk et al., 2002). Another possibility of direct dating of gold is by means of the radiogenic helium (4He) produced by α-decay of U and Th, which are incorporated in trace amounts in gold. The gold lattice has a particularly strong retentivity for 4He, which is released only near the melting temperature of gold, ∼1000 °C (Shukolyukov et al., 2010, 2012). Early attempts to date natural gold using the U/Th–4He method gave geologically reasonable results (Eugster et al., 1992; Niedermann et al., 1993), but had problems with excess He from fluid inclusions (Eugster et al., 1995; Pettke et al., 1997). However, a new step-heating technique and recent advances in high-sensitivity mass spectrometry have allowed successful application of the U/Th–4He method to age determinations of archaeological artifacts of gold (Eugster et al., 2009). Here we report the application of this method to dating of gold mineralization, through the case study of an auriferous deposit in Diamantina in Brazil.

A gold rush that took place ∼300 yr ago in Diamantina, Minas Gerais, southeastern Brazil (Fig. 1), exploited gold with a distinctive palladiferous signature. The emplacement of the Diamantina gold is constrained by field geology, which indicates a Brasiliano age (Cabral et al., 2009), i.e., ca. 600 Ma, that is related to the Pan-African–Brasiliano orogenic event. A previous attempt to date rutile, which is spatially associated with the gold, by the U/Pb method failed to yield an isochron age. Nevertheless, a Brasiliano-related age of 650 ± 130 Ma was suggested (de Abreu, 1991). We present new measurements using laser ablation–inductively coupled plasma–mass spectrometry (LA-ICP-MS) that define a Brasiliano age for the rutile coexisting with the gold, as an independent estimate of the validity of the gold dating.

Diamantina is historically and economically relevant, and offers geological and age constraints on the validity of the direct dating of gold by 4He. The direct dating of gold not only adds information to the geological history of regions, but also has the potential of establishing new prospecting strategies for the discovery of auriferous deposits. For example, the confirmation of a Brasiliano-related age for the Diamantina gold by the direct dating of gold implies that exploration campaigns for gold mineralization in Minas Gerais need to take the Brasiliano tectonic overprint into account.


The Diamantina gold was recovered by panning of friable veins that crosscut the Brasiliano regional tectonic foliation of a hematitic phyllite in the São João da Chapada Formation of the Espinhaço Supergroup (de Abreu, 1991; Cabral et al., 2009). A Statherian age (ca. 1703 Ma) for the São João da Chapada Formation is based on the U/Pb dating of zircon from the hematitic phyllite (Chemale et al., 2010). This rock was likely derived from two rock types, quartz porphyry and diabase (Correns, 1932), which were extensively metasomatized by metaevaporitic fluids (Cabral et al., 2012). The vein system superimposed on the hematitic phyllite consists of rutile and specular hematite in kaolinite-quartz pockets, where the gold occurs as visible grains (de Abreu, 1991; Dossin et al., 1990). Compositionally, the gold is argentiferous, but has a distinctive Pd component (Cabral et al., 2009).

Measurements for 4He were performed at the Universität Bern, Switzerland, with a high-sensitivity mass spectrometer. The instrument was designed for determining low concentrations of helium. In contrast to archeological gold artifacts, geological samples of gold accumulate 4He at a million-year scale; therefore, no more than 2 mg of gold were used for the measurements. Two samples of the Diamantina gold were measured for 4He. The samples were placed in a Re crucible and heated in temperature steps of 40–70 °C to ∼900 °C, i.e., below the melting point of gold. At each temperature step, the released gases were cleaned with two titanium getters to adsorb gases, such as H2, N2, O2, H2O, CO2, and hydrocarbons. Helium was then introduced into the mass spectrometer. Electron bombardment was used to ionize the He atoms, which were registered with a secondary electron multiplier (Channeltron Magnum produced by Burle Electro-Optics, Inc.). The total He release at the aforementioned temperatures was verified by means of a repeated heating following the same temperature steps performed in the first release experiment. No additional He release was observed in this reextraction procedure. The complete procedural blank determined by heating the Re crucible to 900 °C corresponds to ∼100,000 atoms of He, whereas the detection limit of the instrument is ∼30,000 atoms of He. The experimental procedures for 4He and the method of age calculation were detailed in Eugster et al. (2009).

Uranium and Th were determined by quadrupole ICP spectrometry at the Curt-Engelhorn-Zentrum Archäometrie, Mannheim, Germany. Previously outgassed gold samples were transferred into screw-top polytetrafluoroethylene beakers. Freshly prepared aqua regia (500 μL, 1:5 HNO3 14.5N: HCl 11N) was then added and the samples were left for 1 day at 100 °C. After that, the solutions were free from any residues. All solutions were evaporated, and freshly prepared aqua regia was added again to the dried samples and left for 1 day at 100 °C. Each sample was dissolved in HCl 1N; U and Th purification was accomplished by an electrolytic separation. Repeated control yield for both elements is better than 98%, and a measured blank for the procedure is better than 0.2 pg for U and Th. The blank assessment was always done before the electrolytic separation; we performed in each separation cell (5 cells) a blank electrolysis with only 1 mL HCl 1N and the typical current used for the gold separation (1.3 V). After the blank electrolysis, the sample was processed with the same separation cell to provide the best possible blank correction for each separation cell used for the electrolytic separation. The liquids from each blank electrolysis were measured and did not exceed 20 cps for Th and 10 cps for U. Because the blank for each reagent is known and, for this study, negligible (not even with 50 g of water, nitric acid, or hydrochloric acid was it possible to acquire a measurable signal for U or Th), we used, in addition to the blank measurements, repeated analyses on a synthetic gold standard solution with known U and Th concentrations (13.1 ppt Th, 19.3 ppt U). We used equivalents of 3.9 pg to 15.4 pg U for analysis in order to evaluate the influence of a potential blank on the U and Th concentration (see Table DR1 in the GSA Data Repository1). These measurements demonstrated that, even with low U and Th concentrations (<10 pg U and Th total), the overall U and Th blanks are negligible.

The gold samples had U and Th concentrations measured simultaneously out of 0.5 mL nitric acid at 2%. Mass 234 was also measured, as it monitors the AuCl molecular interference, which affects mass 232. Due to the effective gold separation, it was not necessary to apply any correction for 197Au35Cl.

Rutile grains recovered from auriferous pockets were measured by LA-ICP-MS at the Institut für Geowissenschaften, Universität Mainz, Germany, for trace element concentrations and U/Pb ages, employing an Argilent 7500ce, coupled with a 213 nm New Wave UP213 Nd:YAG laser ablation system. The laser ablation system features a large format cell, allowing investigation of several samples together with a range of different standard mounts under nearly identical conditions at very fast wash out (1 order of magnitude drop in signal in 0.3 s). Trace element concentration and U/Pb age measurements were carried out with 80 μm and 50 μm spot diameters, respectively, at 10 Hz repetition rate. Ablation time was set at 30 s, resulting in pits ∼30 μm deep. The glass NIST SRM 610 (Jochum et al., 2011) and the rutile R10 (Zack et al., 2011) were used as trace element standards. R10 was also a primary standard for correction of 207Pb/235U and 206Pb/238U ratios. Off-line data reduction was performed by an in-house Microsoft Excel spreadsheet, employing the 208Pb correction method for common Pb (Zack et al., 2011). Pooled ages were plotted and calculated using Isoplot Excel version 3.7 (Ludwig, 2008). Instrument parameters and further analytical details on U/Pb dating in Mainz were described by Zack et al. (2011).


Table 1 summarizes the results of measurements for 4He, U, Th and the resulting ages. Two gold samples from Diamantina show ages of 560 ± 80 Ma and 480 ± 70 Ma (2σ), and we obtain an average age of 515 ± 55 Ma. The analytical errors of the measurements for He and U/Th were quadratically added to obtain the experimental error of the U/Th–4He age.

Rutile grains from Diamantina recovered from auriferous pockets are characterized by high and variable amounts of common Pb (206Pb/208Pb ratios from 1.5 to 32) and low U contents (6–15 ppm) (see Table 2). Concordia ages for rutile are therefore only calculated with grains having favorable 206Pb/208Pb ratios (here >5). A pooled age of five rutile grains with the highest 206Pb/208Pb ratios (of 12 analyses) gives a U/Pb Concordia age of 524 ± 16 Ma (2σ; see Fig. 2). We note that all 12 analyses have an intercept age of 516 ± 18 Ma in a Tera-Wasserburg plot (Fig. DR1 in the Data Repository), as well as a pooled 206Pb/238U age of 530 ± 12 Ma (Fig. DR2), pointing to a robust and unbiased age based on the subset of the best analyses.

The U/Th–4He age for the Diamantina gold correlates with the information from field observations and the Brasiliano-related Concordia age for the rutile (Fig. 2). The new rutile age broadly agrees, within error, with the previous errorchron age of 650 ± 130 Ma (de Abreu, 1991). The hydrothermal Brasiliano overprint on rocks of the Espinhaço Supergroup has further been recorded by U/Pb ages of monazite from quartz veins, which span between 490 and 440 Ma (Chaves et al., 2010).

Many economic geologists working on cratonic areas tend to ignore relatively young overprints as important gold-mineralizing events. Where hydrothermal minerals are spatially associated with auriferous veins, their ages are considered to have been reset by the overprint. This is especially the case for K-Ar age determinations on biotite and white mica; for example, these minerals gave a Brasiliano age for the auriferous mineralization at the Passagem de Mariana deposit, Quadrilátero Ferrífero of Minas Gerais (e.g., Chauvet et al., 2001), but their age was interpreted to result from isotopic resetting (e.g., Vial et al., 2007). The direct dating of gold by 4He in principle rules out the problem of isotopic resetting of minerals associated with gold. The gold lattice effectively retains 4He until temperatures near the melting temperature of gold are reached (Shukolyukov et al., 2010, 2012). The strong retentivity of the gold lattice makes it impervious to helium release after the deposition of gold. We note that the dated rutile may have formed above the lowest possible closure temperature for Pb in rutile (∼400 °C; Mezger et al., 1989). Zirconium-in-rutile thermometry from the dated samples averages ∼525 °C (see Table DR2); however, other rutile grains give temperatures from ∼500 to 350 °C (Cabral et al., 2011), so rutile data from this study are interpreted as cooling ages only marginally younger than formation ages.

The U/Th–4He age for the Diamantina gold and U/Pb ages for the associated rutile unambiguously relate to the Brasiliano orogenic event. The Diamantina gold mineralization can be understood as an oxidized variant, characterized by abundant specular hematite, within the spectrum of orogenic gold deposits (e.g., Groves et al., 1998). The orogenic affiliation of the Diamantina gold mineralization is manifested in the belt-like distribution of a number of alluvial and lode deposits, which follow the traces of the Brasiliano thrust faults (Fig. 1).


Our results signalize that the U/Th–4He dating of gold is feasible, suggesting a new dating tool directed to gold, i.e., without the assumption of temporal equivalence between gold and datable hydrothermal minerals. The feasibility of the method is supported by the Brasiliano age for the Diamantina gold, which matches the field geology and the new U/Pb dating of the gold-associated rutile. The direct dating of gold shows that the Diamantina gold represents mineralization related to the amalgamation of Gondwana.

An earlier version of the manuscript benefited from thoughtful comments by Rob Chapman and two anonymous reviewers. Another anonymous referee further improved the manuscript. We thank Patience Cowie for the editorial handling. Cabral gratefully acknowledges the DFG (Deutsche Forschungsgemeinschaft, project CA 737/1-1) for financing his stay in Germany.

1GSA Data Repository item 2013040, rutile geochronology, trace elements in rutile, and measurements for U and Th in synthethic Au standard, is available online at www.geosociety.org/pubs/ft2013.htm, or on request from editing@geosociety.org or Documents Secretary, GSA, P.O. Box 9140, Boulder, CO 80301, USA.