Earth’s oldest minerals face a test from a younger crater

Only one of 276 zircon grains older than 3.2 billion years was classified as Sudbury-like impact zircon in an August 2026 machine-learning study. The comparison used Canada’s 1.85-billion-year-old Sudbury impact melt sheet as an analogue for events whose rocks may have vanished.[1]

Why temperature left a question open

A 2016 conference report by Gavin Kenny and colleagues helps explain why this was worth testing. They measured titanium in zircon along a transect through Sudbury’s different rock layers. Titanium provides an estimate of the temperature at which a crystal formed. Their inferred temperatures overlapped those of ancient grains, keeping impact melts in contention as a possible birthplace.[2]

Sampling through the layers mattered: the most evolved rocks contained crystals with lower titanium concentrations. The team argued that earlier, more restricted sampling and differences in measurement technique had obscured that range. A match in crystallisation temperature could therefore arise from more than one geological history.[2]

A test across four chemical features

Researchers trained a support vector machine on four trace-element features to distinguish Sudbury zircon from zircon produced by ordinary magmatism: magma generated inside Earth. In its held-out analogue data, it correctly classified 79% of impact analyses and 98% of ordinary analyses.[1]

Applied to ancient grains from Jack Hills, Barberton and the Acasta gneiss, the model overwhelmingly favoured ordinary magmatism. Given its errors, the single impact-like prediction is weak positive evidence.[1]

The conclusion concerns Sudbury-like melting of continental crust in these samples. Impacts into different target rocks could leave different chemistry. The classifier narrows the proposed birthplace of surviving minerals; it does not show that early Earth escaped bombardment.[1]

Report an error or suggest a correction