Smudged volcanic crystals offer clues to past eruptions

Between the jungle and the rice paddies, Fidel Costa struggled to find bare rock on the slopes of Mount Gede, a towering volcano near the western tip of the Indonesian island of Java. But an abandoned quarry hewn into the mountainside offered a rare chance to nab a few samples. So on a muggy day in 2011, Costa, a volcanologist at the Earth Observatory of Singapore, scrambled up the steep wall to some rocks, marbled like rye bread, which he pried loose with a hammer. Four thousand years ago, they erupted from Gede and fell out of a cloud of hot ash.

There are no accounts of that eruption, let alone records of any seismic tremors or burps of gas leading up to it—the clues scientists now use to infer what's brewing deep beneath a volcano. Indeed, the volcano's last outburst occurred in 1957, long before modern monitoring efforts began, so scientists know little about its temperament. What signs portend Gede's eruptions, and how much warning do they give? For the millions of people living on its flanks, as well as in the nearby cities of Jakarta and Bandung, the answers are critical. There's no indication that Gede will erupt anytime soon. But when it does, Costa says, "anything that happens there is going to be a big mess."

From the rocks released by that 4000-year-old eruption, however, Costa and his colleagues at the Center for Volcanology and Geological Hazard Mitigation of Indonesia in Bandung were able to glean some crucial clues about Gede's behavior. The clues were locked in crystals, most smaller than lentils, embedded in the rocks. Each crystal grew in a soup of magma deep underground, accreting layers that bore witness to the events that preceded the eruption, and—most importantly—how fast they unfolded.

It’s one of those techniques that is about to explode in popularity.

Tom Sisson, U.S. Geological Survey

These crystal clocks told Costa's team that Gede's 4000-year-old eruption came roughly 4 weeks after the injection of a fresh batch of magma beneath the volcano. Crystals from four more ancient eruptions gave similar answers. The pattern gives planners an idea of what to expect in the future: When sensors detect signs of magma stirring below the slumbering giant, an eruption may follow within weeks. "It might be uncertain, but it's much better than not knowing anything," Costa says.

Costa has spent years learning to coax such stories out of tiny volcanic crystals with a technique he helped develop, known as diffusion chronometry. And it's catching on. "It's one of those techniques that is about to explode in popularity," says Tom Sisson, a volcanologist at the U.S. Geological Survey in Menlo Park, California.

Already, the few researchers adept at using the technique have found that magma can tear through the crust at searing velocities, and that volcanoes can gurgle to life in a geologic instant. Instead of taking centuries or millennia, these processes can unfold in a matter of decades or years, sometimes even months, says Kari Cooper, a volcano geochemist at the University of California, Davis. The results help explain why geophysicists haven't found simmering magma chambers under volcanoes like Yellowstone, and why some eruptions are more violent than others. "This is something that has the potential to really be a game changer in a lot of ways," she says.

The clock is ticking

Researchers are using a technique called diffusion chronometry to help them decipher the clocks hidden inside volcanic crystals. The crystals tell them, for example, how soon a volcano erupts after a last pulse of magma. The research suggests that many volcanic processes happen much faster than previously thought. It may also help researchers understand the eruptive tendencies of a volcano, even where there are no historical records.

C. Bickel/Science

Back in his lab in Singapore, Costa gleans his volcano histories from cellophane-thin slices of rock. Backlit under a microscope, minerals in the slices—including plagioclase, olivine, and pyroxene—burst into focus: polygonal islands swimming in a dark sea of rock. Many have concentric bands like tree rings, which formed as the crystals grew in an ever-changing bath of liquid magma.

The chemistry of each new band records the evolving composition of the magma, or changes in its temperature or pressure. Costa uses an instrument called an electron microprobe to map the chemical variations along the crystal faces, making a measurement every few microns. In the Gede crystals, the microprobe revealed higher concentrations of magnesium and iron in the outermost layers, which suggested that a fresh burst of magma rich in these elements bubbled up beneath Gede shortly before the eruption—potentially triggering it. But how shortly? That's where Costa and Daniel Krimler, a graduate student at the observatory, turned to diffusion chronometry, which converts the chemical smudging between the rim and the crystal's core into an estimate of time.

Researchers first developed the technique, originally dubbed geospeedometry, in the 1960s. They used it to estimate cooling rates in meteorites and rocks that have been subjected to extreme heat and pressure. The method relies on the premise that nature rarely abides sharp gradients. Just as a few drops of food coloring will diffuse throughout a glass of water—no stirring required—so, too, will diffusion shuffle atoms from areas of high concentration to low concentration within a solid crystal lattice.

In the Gede crystals, diffusion moved atoms of magnesium and iron from the crystal rims to the cores, and shuttled other elements in the opposite direction to fill the vacancies left by these atoms. It transformed an abrupt, steplike change in chemical composition into a more gradual curve. By knowing how fast magnesium and iron diffuse through specific minerals, Costa and Krimler could calculate how long diffusion went on after the magma injection and before the volcano erupted, freezing the chemical profile in place. They had a stopwatch.

It's not quite that simple, of course. Diffusion rates depend not only on the element and mineral in question, but on the temperature, pressure, and oxidation state the crystal experienced, which researchers estimate from other clues in the crystals. In the 1980s, pioneers like Sumit Chakraborty, a geologist at Ruhr University in Bochum, Germany, began the tedious work of pinning down these diffusion parameters across a range of conditions. That meant long hours in the lab torturing natural and synthetic crystals with heat and pressure and then watching diffusion proceed. At first, single experiments could take weeks, but the results gave diffusion chronometry teeth.

Crystal clocks have helped reveal the multichambered plumbing under Italy's Mount Etna, shown erupting in 2012.

Martin Rietze/Science Source

Curiously though, the technique didn't catch on with volcanologists until the turn of the century. By the time Costa published his first paper on the topic in 2003, applying diffusion techniques to volcanic crystals from the San Pedro volcano in Chile, several other researchers were having similar epiphanies. It was an idea whose time had come.

One appeal of diffusion chronometry lies in its ability to track a wide range of volcanic processes. Any time a new zone forms within a crystal, diffusion chronometry can theoretically exploit it. And that allows scientists to target many stages leading up to an eruption, including when magma rises from the mantle, when it collects in crustal reservoirs and mixes with other magmas, and when it barrels up through the plumbing of the volcano toward the surface.

Take the initial ascent of magma from the mantle. Terry Plank, a geochemist at Columbia University's Lamont-Doherty Earth Observatory in Palisades, New York, and a former postdoctoral researcher in her lab, Philipp Ruprecht, wanted to understand the origin of magma in a famous eruption of the Irazù volcano in Costa Rica that lasted from 1963 to 1965. In certain olivine crystals, the researchers noted variations in nickel concentrations inside the crystal cores that appeared to have formed in the mantle.

The fact that the chemical variations survived the ascent through the crust implied that diffusion had little time to smear them out. Plank and Ruprecht concluded that the magma must have risen through roughly 35 kilometers of crust in just months, or at most, a few years. "That was a surprise," Plank says. The results—suggesting the possibility of a direct connection between the mantle and the surface—contradicted the widespread idea that magma follows a tortuous path upward, pooling in magma chambers along the way before finally erupting.

In many cases, though, it appears that magma does spend millennia loitering a few kilometers beneath the surface, only to mobilize rapidly before an eruption. At Mount Hood in Oregon, for example, Cooper examined rocks from the volcano's last two eruptions, 1500 and 220 years ago. She focused on plagioclase crystals that formed in shallow magma chambers in the crust. By measuring the concentrations of uranium and its radioactive daughter elements, she found that these crystals were born at least 20,000 years ago.

Many of these plagioclase crystals also had numerous layers with different chemical compositions, which they acquired over their long lifetimes. When Cooper looked closely at the boundaries between these layers, she found something startling: They were only slightly smudged, suggesting that in spite of their age, the crystals had only a brief sojourn in hot, liquid magma. In 2014, she and Adam Kent of Oregon State University in Corvallis explained their discovery by proposing that the magma spent as much as 99% of its time in storage at temperatures too cool to erupt and too cool for diffusion. Instead, it existed in a mostly solid crystalline mush beneath Mount Hood.

We can go pick up the rocks, study the minerals, and basically get timescale information about an eruption that happened, let’s say, 100,000 years ago.

Maren Kahl, petrologist, University of Iceland in Reykjavik

The results support the so-called "mush model," which has gained traction in the last decade or so. Cooper's work suggests that magma may liquefy and erupt even more quickly than many researchers thought. "It's a bit of a subtle shift, but it's very important, because all of a sudden, everything you want to do"—mixing and assembling the final pot of magma that erupts—"all that has to happen really quickly," Cooper says. Crystal clocks from a 3600-year-old eruption of the Greek volcano Santorini, which had been dormant for 18,000 years, suggest that it awakened in a century or less. If other volcanoes behave similarly, it would explain why researchers have struggled to find evidence for large molten magma chambers on Earth today—such vats of liquid magma may only exist immediately prior to an eruption.

When an eruption finally happens, magma races from its subterranean source to the surface. Plank's current quest is to understand whether the speed of that ascent influences how explosively a volcano erupts—causing Hawaii's Kilauea, for example, to erupt gently today in spite of explosive outbursts in the past. Plank and others suspect that, all else being equal, a slowly rising magma has more time to lose dissolved gases—and therefore erupts less violently—than magma that gushes toward the surface. "It's not the gas in the seltzer bottle, it's how fast you open the seltzer bottle," she says.

But how does one measure the speed of magma rising deep beneath the ground? Instead of studying diffusion in crystals, Plank has looked at how dissolved volatiles like water and carbon dioxide diffuse through melt tubes, tiny burrows in crystals that fill with liquid magma. As crystals rise toward the surface, the volatiles trapped in a melt tube diffuse toward its mouth, striving to stay in equilibrium with the dropping concentrations in the magma outside the crystal. This produces a diffusion profile along the tube that Plank and others can use as a clock. Because these gases move relatively fast, the technique allows them to time processes that unfold rapidly. They have found that slugs of magma can rise 10 kilometers in roughly 10 minutes. "It's like a freight train," she says.

A quarry on the flanks of Mount Gede, an Indonesian volcano, reveals deposits from past eruptions that are ripe for harvesting crystals.

F. Costa

Her preliminary results from a handful of volcanoes in Alaska, Hawaii, and Central America support the idea that ascent rates correlate with explosivity. She's working now to study more volcanoes, but, she says, "the problem is that most of those eruptions go unsampled." So she's getting creative; when the Pavlof Volcano erupted violently on the Alaska Peninsula in March of this year, she bartered a box of fresh fruit for a trashcan of ash collected by locals. Her team still had to search through it for crystals that meet their criteria—another challenge. "My students pick for hours under a microscope looking for the one olivine that has a weird tube in it," she says. "We published a paper on four of them, that's how rare they are."

Relying on a handful of tiny crystals to track an entire body of magma worries some outsiders. "People are making strong interpretations based on not a whole lot of results," Sisson says. Dan Morgan, a petrologist at the University of Leeds in the United Kingdom and an early practitioner of diffusion chronometry, shares that concern. Although there's no way around it in Plank's work, Morgan says researchers should be careful with small sample numbers. "If you find five very photogenic crystals, they will be anomalous," he says. One 2015 study, led by Thomas Shea, a volcanologist at the University of Hawaii in Honolulu, points out that researchers must analyze at least 20 olivine profiles to account for the fact that chemicals diffuse in three dimensions through a crystal.

So Morgan and others have been working on faster ways to measure and analyze diffusion profiles. One strategy is to skip the slow process of moving point by point across the crystal face with the microprobe, and instead use a technique called backscattered electron microscopy, which essentially snaps a chemical photo of the crystal. The brightness of the image can serve as a proxy for the concentration of iron and magnesium, and the process takes much less time.

Both Morgan and Costa are also developing user-friendly software to help researchers who aren't experts in diffusion modeling interpret their data. Morgan says that when he started out, he could model two or three chemical profiles in a day; his new daily record is 80. By speeding up the process, he hopes researchers can generate more data, "which then starts to tell you things at the scale of the whole magma mass rather than an individual crystal's story."

Others worry about uncertainties in the diffusion rates, especially for less common elements and minerals. But Costa says that it's important to keep perspective. Even if the uncertainties are 100% or more, the clock results can still be meaningful. "If I find 1 month, 100% uncertainty is a few months," he says. "It's still not 100 years."

The biggest challenge, according to experts like Costa and Morgan, isn't calibrating the stopwatch—it's knowing which volcanic processes the crystals are recording. That's why many researchers are studying crystals from eruptions of actively monitored volcanoes. Maren Kahl, a petrologist at the University of Iceland in Reykjavik, has used that approach at one of the best studied volcanoes on Earth, Mount Etna in Italy. She and her colleagues examined crystals from eight well-documented eruptive episodes between 1991 and 2008. The researchers were able to tie monitoring records of earthquakes, ground deformation, and gas emissions to pulses of magma recorded in crystal chemistry, which they dated using diffusion chronometry. The result was an unprecedented picture of the volcano's multichambered plumbing, with five different magma zones and three dominant pathways between them. The researchers were even able to create a model of how the volcano erupts based on realistic physics. "We've never been able to quite do that before," Plank says.

As researchers get better at linking the crystals' stories with observations of modern eruptions, Kahl says, they will gain confidence about applying the technique to ancient ones, as Costa is doing at Gede. Of the 1500 potentially active volcanoes on Earth, only a small fraction are actively monitored, and fewer still have erupted since scientists started watching.

With diffusion chronometry, however, researchers can use crystals to learn the histories and personalities of these hibernating volcanoes. "We can go pick up the rocks, study the minerals, and basically get timescale information about an eruption that happened, let's say, 100,000 years ago," Kahl says. And by diving deep into a volcano's past, scientists can gain a glimpse into its future.