© John W. ValleyTiny crystals called zircons, harvested from the Australian outback, suggest plate tectonics may have begun more than 4 billion years ago.
After using a form of artificial intelligence (AI) to analyze some of the most ancient crystals on the planet, researchers have concluded that plate tectonics — Earth's geological machinery that floats giant slabs of crust across and sometimes into the mantle — began much earlier than many scientists had assumed.
The group's study finds evidence for its start more than 4 billion years ago, during the Hadean eon — just a few hundred million years after the planet's formation. The work suggests tectonics might have had an early hand in creating the first land and helping life begin, says Ross Mitchell, a geophysicist at the Chinese Academy of Sciences who co-authored the new study, which was published today in the Proceedings of the National Academy of Sciences. "Hadean Earth would have been habitable," he contends.
Plate tectonics is generally accepted to have begun more than 3 billion years ago, after Earth had cooled enough to form a stagnant lid of crust capping an ultrahot mantle. As slabs of ocean crust began to fall into the mantle, they triggered volcanic eruptions, which created chains of islands that eventually bunched up into the first continents. Evidence for this picture comes from the oldest surviving continental interiors, which date to about 3.5 billion years ago. But modeling suggests the crustal movement could have started earlier, says Qian Yuan, a geodynamicist at the California Institute of Technology. "More and more [scientists] are proposing that there could have been Hadean plate tectonics," he says.
Finding hard evidence of such activity has been difficult, however. The only surviving materials from the Hadean are zircons, nearly indestructible crystals the size of sand grains. The oldest zircons, discovered in Australia's Jack Hills, date back to 4.3 billion years ago. Zircons can crystallize out of fresh magma from the mantle, much like the ocean crust of today, but they can also form from sedimentary rocks on land that wash into the ocean, sink back into the mantle, and re-emerge in later bursts of igneous activity as granites. These resurfaced "S-type" zircons provide evidence of the existence of both continents and the subduction process, Mitchell says.
Researchers identify S-type zircons by looking for inclusions of mica, a flaky mineral often found in sedimentary rocks, or by measuring enriched levels of trace elements associated with magmas formed from weathered land rocks, such as aluminum or phosphorus. But no single element or mineral can provide a definitive origin story, and the older the samples are, the harder it gets.
Rather than looking at just a few diagnostic factors, Mitchell and colleagues instead turned to a basic AI approach known as machine learning to sort the zircons. They assembled a data set of the elemental abundances of 374 zircons with known origins. They used 80% of the zircons to train the algorithms, and 20% to test them. They varied the number of diagnostic factors, and found that nine was "just right," Mitchell says. Using fewer factors provided too little information for fingerprinting; using more limited the available zircon records for training. "It's a many-dimensional classification problem that seems perfectly suited" for such tools, says Roger Fu, a planetary scientist at Harvard University unaffiliated with the work.
Mitchell and colleagues then applied the trained algorithm to 971 of the very ancient Jack Hills zircons whose origins were uncertain. Overall, they found that more than one-third were S-type, including a few up to 4.2 billion years old. Mitchell and his team have "something that looks like it could be a really useful tool," says Beth Ann Bell, a geochemist at the University of California, Los Angeles. "This is significant not only for the early Earth, but for going throughout the geological record."
Mitchell's group also found that the percentage of S-type zircons changes through time, rising and falling in a pattern that resembles known cycles of supercontinent creation and destruction later in Earth's history. That conclusion is "pure speculation," says Jun Korenaga, a geophysicist at Yale University whose modeling and geochemistry work has suggested an early start for plate tectonics.
Another criticism comes from Chris Hawkesworth, a geochemist at the University of Bristol who says evidence for subduction more than 4 billion years ago does not necessarily mean the engine of plate tectonics had turned on completely by that point; he points out that there are other ways to get crust back into the mantle, such as with a giant asteroid impact.
Still, evidence for early plate tectonics is growing — at least based on the stories preserved in the Jack Hills zircons. In April, researchers reported that freshwater was involved in the formation of 4-billion-year-old zircons — another sign that continents, and presumably, plate tectonics, existed at the time.
What is needed now are additional ancient zircons from places other than Australia, Fu says. The Jack Hills zircons could record events from one exceptional region that's not representative of the globe. Still, given that plate tectonics is the best known way to subduct crust, and the lack of a convincing case against it occurring, it may be time to assume plate tectonics is nearly as old as the crust itself, Fu says — at least until it can be disproved.
Source link