It takes more than plate tectonics to cause an earthquake, according to geologist Ze'ev Reches of the University of Oklahoma in Norman. The gradual buildup of stress in a fault as plates collide or slide past each other is necessary, he acknowledges. But if nothing else were going on, all the pent-up energy could be released via fault creep, a motion so sluggish it's virtually unnoticeable by human standards, and the temblors that create so much havoc across the globe wouldn't exist.
"There's no real reason that this [energy] must be released in the form of earthquakes," Reches says. For something catastrophic or even detectable to occur, the fault must abruptly get so much weaker that it can no longer cope with the stress by simply plodding along.
Scientists have uncovered several mechanisms that could make a fault lunge into earthquake mode. The heat from friction can melt its sides, or quartz grains can react with water to create a silica gel that slickens the rocks and makes them slide faster. Reches and his colleague David Lockner took a closer look at fault gouge, the fine powder fault slabs create as they grind against each other, and discovered something surprising: The same material that weakens the fault—making earthquakes much more likely to occur—can also strengthen it.
To figure out exactly how fault gouge forms and what it does, the researchers slid 5-centimeter-tall granite plugs against each other while tinkering with pressure and velocity. After hundreds of trials, they identified five "stages" as the slabs accelerated to the speed of an average earthquake: 1 meter per second. During the second stage, the rising velocity created fault gouge that resulted in increasingly steeper drops in friction, leaving the fault just one-half to one-third as strong as it had been before by stage three. But in the fourth stage, when the higher velocity raised temperatures to roughly 150˚C, the opposite happened: Friction increased, and the fault grew stronger. At that temperature, the thin layer of water coating the grains inside the fault could evaporate, leaving the newly dehydrated grains less slippery than they were before, Reches and Lockner speculate in a paper published  in the 23 September issue of Nature. Beyond stage four, however, the velocity became so high that the slabs fractured and failed and weakening took over again. The results suggest that gouge could temporarily shore up a real fault's resistance to an earthquake but would fail if the fault continued accelerating.
Structural geologist Giulio Di Toro of the University of Padua in Italy, who was not part of the research, says the study makes sense, and he agrees with the authors' interpretation about why the fault strength suddenly spiked. But he says he's concerned that the top pressures they studied—7 megapascals, barely one-tenth the pressure felt in even shallow quakes—were too low to reflect what's actually going on deep underground. "The problem is, how can we extrapolate something we produced in the lab in a small sample to a natural fault?" asks Di Toro, who studies earthquake mechanics.
But Reches thinks the pressure at greater depths only speeds up the process. For example, whereas in the lab a sample might have to travel 10 meters before a certain change in friction occurred, a real fault might feel the effects after moving just 1 meter. "It's a question of getting the same process, only shorter distance," he says.