Long term low latitude and high elevation cosmogenic 3He production rate inferred from a 107 ka-old lava flow in northern Chile; 22°S-3400 m a.s.l.

Available geological calibration sites used to estimate the rate at which cosmogenic 3He is produced at the Earth’s surface are mostly clustered in medium to high latitudes. Moreover, most of them have exposure histories shorter than tens of thousands of years. This lack of sites prevents a qualitative assessment of available production models used to convert cosmogenic 3He concentrations into exposure ages and/or denudation rates. It thus limits our ability to take into account the atmospheric, geomagnetic and solar modulation conditions that might have affected the production of cosmogenic nuclides in the past for longer exposure histories and in low latitude regions. We present the cosmogenic 3He production rate inferred from a new geological calibration site located in northern Chile. Five samples were collected on the surface of the largest and best-preserved lava flow of the San Pedro volcano (21.934°S–68.510°W-3390 m a.s.l.), which displays pristine crease-structure features. 40Ar/39Ar dating yields a reliable plateau age of 107 ± 12 ka for the eruption of this lava flow. Eight pyroxene aliquots separated from the surface samples yield a weighted average cosmogenic 3He concentration of 99.3 ± 1.2 Mat g−1 from which a local cosmogenic 3He production rate of 928 ± 101 at g−1 yr−1 is calculated. The local production rate is then scaled to a sea level high latitude (SLHL) reference position using different combinations of geographic spatialization schemes, atmosphere models and geomagnetic field reconstructions, yielding SLHL production rates between 103 ± 11 and 130 ± 14 at g−1 yr−1 consistent with the most recent estimates available from the literature. Finally, we use the same scaling frameworks to re-evaluate the mean global-scale cosmogenic 3He production rate in olivine and pyroxene minerals at 120 ± 16 at g−1 yr−1 from the compilation of previously published calibration datasets.