Operating since 1903, Rio Tinto’s Kennecott Mine near Salt Lake City remains one of the most productive mines in the world, where workers pulled 134,000 metric tons of copper from the earth last year, along with significant amounts of gold, silver and molybdenum.

That doesn’t come close to keeping up with demand, though. Prices of copper and other critical minerals surged to record highs last year, driven by supply shortages and aggravated by trade wars. The shortages show no signs of easing. According to J. P. Morgan, the global refined-copper shortfall will hit 330,000 tons this year and could widen to as much as eight million tons by 2035. The United Nations predicts that demand for critical minerals could triple by 2030. To meet its Net Zero 2050 goals, the International Energy Agency estimates that annual production of these minerals will need to increase sixfold.

A top-tier copper mine can be productive for many decades, but longevity comes at a cost. The Kennecott Mine, the deepest open-pit mine in the world, is the biggest human excavation ever—more than four kilometers wide and more than a kilometer deep. Its impact on the environment and landscape is massive. And the ore coming out of the mine today is lower grade than it used to be, meaning miners must extract a lot more waste rock to get the same amount of processed ore. Meanwhile the search for new deposits has stagnated; for every 1,000 precious-metal prospects worldwide, it’s estimated that fewer than five will ever become productive mines.

On supporting science journalism

If you're enjoying this article, consider supporting our award-winning journalism by subscribing. By purchasing a subscription you are helping to ensure the future of impactful stories about the discoveries and ideas shaping our world today.

Meeting current and future supply shortfalls requires progress on two fronts simultaneously: finding new deposits and extracting material more efficiently from existing ones. Underlying both efforts is the need for better tools to tell us what we’ll find when we start digging—so-called subsurface intelligence. If it works, the tech could lead to less trial-and-error drilling. The risk is that better intelligence could just accelerate extraction in an industry whose impacts remain enormous.

Mineral exploration has always been an exercise in inference. Geologists work from surface clues—mineral outcroppings, soil samples, magnetic anomalies—to make educated guesses about what lies underneath. To improve success rates in finding new “greenfield” mineral reserves—previously unexplored sites where no mining infrastructure yet exists—companies such as Earth AI and KoBold Metals are using novel artificial-intelligence models to find patterns in decades’ worth of existing geological and survey data. The companies say early results are promising, yielding new discoveries of copper and palladium. But any large deposit discovered today will take years—often decades—to become an active mine. A 2023 S&P Global analysis found that from 2002 to 2023, new mines took an average of 15.7 years to develop from discovery to production; in the U.S., the average is 29 years.

Those massive timelines are driving mining companies to expand older “brownfield” surface mines by going underground, using a method called block caving—a brute-force technique that makes the need for subsurface intelligence more urgent than ever. Widely used in copper and gold mining, block caving is suited to lower-grade ore deposits that are more or less vertically oriented. It works a little like open-pit mining in reverse. Engineers dig underground tunnels, then blast an undercut below the ore body, forming an artificial cavern. Large rock funnels called drawbells are built below the undercut to channel rubble into loaders. Once the setup is complete, the undercut removes the ore body’s support, and the rock above starts to fracture and cave in under its own weight, crushing itself as it funnels into the drawbells.

In theory, once a block cave is established, no additional blasting or construction should be required. As the funnels empty, broken ore continues to drop away from the “roof” above the undercut—picture coffee beans pouring out of a clear bulk dispenser when you open a slot at the bottom. As material is removed, the collapse progresses upward through the ore body until it is exhausted.

Block caving is far cheaper than other underground mining methods, in part because gravity helps to crush and move the ore. That makes the lower-grade ore that mining companies are going after these days economically viable. Block caving also keeps much of the disturbance underground, generating far less surface waste rock than open-pit operations. The method carries significant risk, however. Developing a large block caving operation requires building a massive underground infrastructure of tunnels, roads, railways, shafts and conveyor belts, which can cost upward of $10 billion. Block cave mines are also vulnerable to catastrophic failures—collapses, air blasts, and inrushes of water, mud and rock. To keep a mine from collapsing prematurely or trapping workers, engineers need to know how the rock is fracturing deep out of sight. That requires continuous, high-resolution mapping of an environment that is constantly destroying itself.

Mapping that subterranean chaos is where cosmic rays come in. British Columbia start-up Ideon Technologies, a spin-off from TRIUMF—Canada’s national particle-accelerator center—has built its business around muon tomography. Gary Agnew, the company’s co-founder and CEO, describes the approach as “the first net new geophysical technique in literally decades.”

Muons are subatomic particles produced when cosmic rays from supernova explosions interact with matter in Earth’s upper atmosphere. They rain down continuously, traveling at nearly the speed of light and penetrating up to 1.5 kilometers into Earth’s surface. Crucially, each muon carries information about its direction of travel and the density of the material it has passed through. Placing detectors underground and measuring arriving muons makes it possible to create a high-resolution, cone-shaped, three-dimensional map of the surrounding rock.

Agnew offers a medical analogy. “You go to the dentist. There’s an x-ray machine to the side of your mouth. They put a detector plate in your mouth, and your gums and teeth are blocking those x-rays from getting to the detector plate,” he says. “That’s exactly the way muon tomography works”—with two key differences: no radiation is involved, and instead of imaging part of the human body, Ideon’s sensors assess hundreds of millions of cubic meters of earth at a time.

The detectors themselves were once the size of a room, confined to government labs. Ideon has miniaturized its borehole sensors to roughly the diameter of a coffee cup and hardened them for field conditions. “We’ve kind of industrialized particle physics,” Agnew says. “The technology used to find hidden chambers in the pyramids is now working in mine sites a mile deep, under pressure, under temperature.”

Such hardware innovations are necessary for building better subsurface models, says Mengli Zhang, a research assistant professor and director of the Center for Geophysics, Energy and Minerals at the Colorado School of Mines. “And in this environment, even more than cost and high resolution, the key is smaller and faster sensors that can get enough information before a borehole collapses.”

Muon tomography offers resolution—from about 20 meters down to submeter scale—that competing techniques cannot match. Passive seismic sensing can go deeper than muons but generally offers resolution of only 50 to 100 meters. Many other subsurface-imaging techniques are limited to 2D outputs, showing a big blob on the surface where minerals might be. Critically, whereas other subsurface-imaging techniques are impacted by the operational noise of a working mine, “muons don’t care,” Agnew says.

Ideon integrates density measurements from its muon detectors with gravity, seismic, magnetic and drill-core data into a dynamic Earth model it calls the Reveal platform. “Density is the backbone, and then complementary datasets color the picture,” Agnew says. Traditionally a mining company might wait six to 12 months to update its resource model, but Reveal incorporates new data continuously.

To gather the data, Ideon adapts its hardware to the environment. For greenfield exploration, the company deploys borehole sensors down existing drill holes, where they survey a 120-degree field of view. Inside a working mine it uses flat-panel detectors mounted on tunnel walls, which can collect muons at four to five times the speed of borehole sensors because of their greater surface area.

For block cave mines, near-real-time imaging removes one of the industry’s most dangerous blind spots. A block cave is not a static structure but rather hundreds of millions of cubic meters of earth in continuous motion. Ore collapses, flows and settles over months and years. Understanding where the cave “back”—the upper surface of the developing void—is located, how material is flowing and whether dangerous air gaps are forming is essential to both productivity and safety. Irregular cave-back shapes can cause uneven caving that can jam up production; excessive air gaps behind the cave back can generate sudden, lethal air blasts.

The ability to observe cave collapse allows a mine operator to see when the ore body is almost exhausted, triggering the start of excavation in a new area. “Right now they have no idea,” says Jef Caers, a professor of earth and planetary science at Stanford University and director of the school’s Mineral-X program. “They’re poking with boreholes from the top, and they’re trying to find out where the air gap is.” Better imaging technology can also locate subsurface fault lines, helping to spot dangerous stress points before they trigger minor earthquakes, which can shut down mine operations for months at a time. “As you start changing the stress situation in the subsurface, it changes the stress situation on the fault surface,” Caers says. But in mineral mining, “they have absolutely no idea where any of the faults are. So you’re really just waiting.”

The consequences of uncertainty can be severe. Last September a mudslide at the Grasberg Block Cave mine in Papua, Indonesia—the world’s largest underground block cave and second-largest copper mine—killed seven workers. Phoenix-based Freeport-McMoRan, which operates the mine, blamed the disaster on an uneven collapse that unleashed a flood of mud and rock. Although Ideon’s muon technology was only in a pilot phase at the mine at the time, the tragedy showed the exact kind of unseen hazard the sensors are designed to catch. Freeport said it plans to use an expanded array of muon detectors going forward to map the true shape of the cave and verify that the rock has stabilized before workers return.

Last fall Ideon signed a five-year partnership with Rio Tinto to deploy muon tomography at six of the company’s largest operations, following a successful demonstration at the Kennecott Mine, where Rio Tinto is currently bringing new underground operations online. At Kennecott, Agnew says, Ideon’s work addressed two distinct problems: reducing errors in production estimates that can cause unexpected bottlenecks and stall the trucks and conveyor belts hauling ore to the surface; and mapping unknown subsurface voids left by more than a century of “artisanal” mining activity, allowing operators to plan for those voids rather than “stumble across [them] halfway through.”

This leap in visibility is overdue, Agnew argues. “When we look back at the oil and gas business in the late 1990s, it kind of looked like mining does now,” he says. “It was working under a lot of uncertainty. Fast-forward three decades, and the field-services companies brought probabilistic technologies to the table—and that completely transformed the way the oil and gas business explores and extracts resources.”

Historically the mining industry hasn’t faced the kind of surging demand that has driven the oil and gas sector to innovate. But now, Agnew says, the rush for critical minerals is doing exactly that. “We’re at record highs on most major commodities right now, and that’s only going to get higher.”

Better subsurface maps may change how mining happens—streamlining the discovery of new reserves, avoiding nasty surprises underground and minimizing disruptions on the surface compared with giant open pits. But these tools won’t slow down extraction, because that’s not where the incentive lies. “I believe Earth can handle a few surgical disturbances from mining because of what it enables,” Agnew says.