Helium in diamonds unravels over a billion years of craton metasomatism

Chemical events involving deep carbon- and water-rich fluids impact the continental lithosphere over its history. Diamonds are a by-product of such episodic fluid infiltrations, and entrapment of these fluids as microinclusions in lithospheric diamonds provide unique opportunities to investigate their nature. However, until now, direct constraints on the timing of such events have not been available. Here we report three alteration events in the southwest Kaapvaal lithosphere using U-Th-He geochronology of fluid-bearing diamonds, and constrain the upper limit of He diffusivity (to D ≈ 1.8 × 10−19 cm2 s−1), thus providing a means to directly place both upper and lower age limits on these alteration episodes. The youngest, during the Cretaceous, involved highly saline fluids, indicating a relationship with late-Mesozoic kimberlite eruptions. Remnants of two preceding events, by a Paleozoic silicic fluid and a Proterozoic carbonatitic fluid, are also encapsulated in Kaapvaal diamonds and are likely coeval with major surface tectonic events (e.g. the Damara and Namaqua–Natal orogenies).


The budget of He in HDF-bearing diamonds
Possible sources contributing to the budget of He in an HDF-bearing diamond are: (1) initial He trapped in the diamond lattice during formation; (2) implantation of 4 He from the surrounding host rock; (3) 3 He produced within the diamond lattice from cosmic ray impact; (4) in situ radiogenic production of 4 He from U and Th decay and 3 He from 6 Li(n,α) 3 H → 3 He, and (5) 4 He and 3 He trapped within inclusions. 4 He can be lost from the diamond due to recoil from the outermost <25 μm of a diamond, and both 4 He and 3 He can be lost by diffusion. The budget of He in a diamond can therefore be written as: 4 He Total = 4 He Diamond lattice initial + 4 He Implanted + 4  The following paragraphs discuss the possible processes that can add He to or remove He from diamonds, and their significance for determining the He content, isotopic compositions and (U-Th)/He age determinations of the De Beers Pool and Finsch HDF-bearing diamonds that were analyzed in the present study.

Contributions of He from the diamond lattice:
Release of He (both 3 He and 4 He) from the diamond lattice is possible upon diamond graphitization at ~2000 °C. Kurz et al. 6 showed that the amount of 4 He released from gem-quality monocrystalline diamonds by crushing is ~2 orders of magnitude lower than by burning the same diamond. To evaluate the contribution of He released from the diamond lattice by our sequential crushing method, we crushed and analyzed gem-quality diamonds and found that the amounts of He released are within blank levels, and 2-3 orders of magnitude lower compared to the De Beers Pool and Finsch HDF-bearing diamonds analyzed.
Thus, He trapped in the diamond lattice has a minimal contribution to the budget of He released by crushing HDF-bearing diamond.
Implantation of 4 He from the surrounding host rocks: Implanted 4 He (α-particles) produced from radioactive decay of U and Th in the surrounding rock can penetrate <25 µm into a diamond 7 .
The damage to the diamond lattice is expressed as a unique green and brown surficial coloration 8,9 .
The De Beers Pool and Finsch HDF-bearing diamonds have no visual radiation color-change.
Moreover, in the present study, inner fragments of the diamond samples were selected for analyses to avoid the possible contributions of implanted 4 He from the outermost 25 µm. Cosmogenic 3 He: Spallation reactions that produce 3 He in rocks occur predominantly within a few meters of the Earth's surface 10,11 . 3 He production by spallation in diamonds is therefore significant only within the upper few meters of the kimberlite diatreme or in alluvial diamonds.
On this basis, extremely high 3 He/ 4 He ratios (>100 Ra) obtained in alluvial diamonds have been interpreted to result from in situ cosmogenic 3 He production at the surface 12,13 ; a connection that was later confirmed 14 . The studied De Beers Pool and Finsch HDF-bearing diamonds are from deep mining and are therefore unaffected by cosmic ray spallation.
In situ production of 4 He and 3 He in the diamond lattice: 4 He can be produced in situ from U and Th decay within the diamond lattice. For a U content of 1 ppb and Th/U ratio of 3.5 in a diamond, only ~10 -7 ccSTP g -1 4 He will be produced over ~4.5 Ga 15 , an amount which is 1-2 orders of magnitude lower than the measured concentrations in HDF-bearing diamonds. Moreover, given more realistic U and Th concentrations of 0.15 and 0.45 ppb, respectively, in an inclusion-free diamond 16 , the 4 He contribution is negligible. 3 He can be produced within the diamond by 6 Li(n,α) 3 H → 3 He. Kurz et al. 6 calculated that over 90 Ma, an amount of 3 He=3×10 -13 cc g -1 will be produced within a diamond containing 1 ppm Li, from the production of 2.5×10 -6 neutrons by Th and U decay within the kimberlite (assuming 5 ppm U and Th/U=3.5 in the kimberlite). Such an amount of nucleogenic 3 He is ~2 orders of magnitude lower than most of our measured 3 He contents in De Beers Pool and Finsch HDF-bearing diamonds. Moreover, the amount of Li in diamonds rarely exceeds ~200 ppb 17 , thus a concentration of 1 ppm Li in a diamond is unrealistically high, and the amount of 3 He produced by 6 Li(n,α) 3 H → 3 He will be much smaller over much longer periods of time. 4 He produced by U-Th radioactive decay are expected to move <25 µm within a diamond 7 . Thus, radiogenic 4 He atoms produced within µm-size C-O-H microinclusions can be implanted into the surrounding diamond matrix.

Loss of 4 He from the microinclusions due to α-recoil:
However, the fact that radiogenic 3 He/ 4 He signatures are observed in HDF-bearing diamond measurements by crushing ( Fig. 1; Supplementary Data 1) indicates that these radiogenic 4 He atoms find their way back to the microinclusions. A reasonable explanation for such return is fast diffusion of He through radiation-damaged diamond paths [18][19][20] , compared to the very slow diffusion in an ordered diamond lattice (diffusion of He in diamonds is discussed below). While radiogenic 4 He will be lost from microinclusions located in the outer <25 µm of the diamond, using inner fragments of the De Beers Pool and Finsch HDF-bearing diamonds for He analyses avoids such a loss.
Diffusive loss of He from diamonds: Luther and Moore 18 were the first to experimentally measure He diffusion in diamonds. Using He-irradiated synthetic diamonds they determined the diffusion coefficient (D) to range between 10 -5 -10 -7 cm 2 s -1 . These fast diffusion rates result from the radiation-induced damage to the diamond lattice 19,20 . A set of industrial-grade diamonds yielded slower diffusion rates (D=1.9×10 -16 cm 2 s - 1 19 , similar to diffusion rates determined in carbonado diamonds (D=3×10 -17 cm 2 s -1 at 1300 °C 21 ) and chemical vapor deposition (CVD) polycrystalline diamonds 22 ). Much smaller D-values were determined for He in monocrystalline diamonds at 1000-1300 °C (1-4×10 -21 cm 2 s -1 23 ). Diffusion coefficients in HDF-bearing diamonds have not been determined experimentally, however, considering the large range of possible He diffusivities, between D=10 -16 to 10 -21 cm 2 s -1 , this process may impact the diamond's potential to retain He over geological time scales and thus can influence its He content, isotopic compositions and resolution of calculated (U-Th)/He ages.
Considering the above, the He budget for the studied De Beers Pool and Finsch HDF-bearing diamonds, which are from deep mining, and whose inner fragments were crushed rather than burned for He analyses, can be therefore simplified to:

He Total = He Inclusions -He Diffusion
This He budget is consistent with previous conclusions that the prime source of He in HDF-bearing diamonds is the microinclusions [24][25][26] . Supplementary Fig. 1: Helium diffusion profiles in a diamond