On September 10, 2008, engineers at CERN steered a beam of protons around a 17-mile ring buried beneath the French-Swiss border — the most expensive machine physics had built, disabled nine days later by one failed solder joint.

Lyn Evans had spent fourteen years building the machine. What he had not yet done, on the morning of September 10, was turn it on.

At 10:28 a.m. local time, a beam of protons injected from a smaller ring completed one full lap of a 17-mile loop buried under the fields and villages northwest of Geneva. It took less than an hour from the first attempt. Evans, the project leader who had shepherded the accelerator from proposal to concrete, called it "a fantastic moment." CERN's director general, Robert Aymar, went further, telling reporters the machine had "the potential to change our view of the Universe profoundly."

What CERN built, and why it took so long

The European Organization for Nuclear Research had been planning some version of this since the early 1980s, and construction proper ran for about a decade. The result was a ring of superconducting magnets, cooled with liquid helium to 1.9 kelvin (colder than the vacuum of deep space) so that streams of protons could be bent around a circle and smashed into each other at nearly the speed of light. The tunnel itself was inherited: it had held CERN's previous accelerator, the Large Electron-Positron Collider, and runs at a depth that varies from about 165 to 575 feet, angled to keep most of the ring out of the harder rock under the Jura Mountains. Four times, the tunnel crosses from France into Switzerland and back.

The bill for all of this ran into the billions of dollars, split among CERN's twenty member states and outside partners including the United States, Japan, Russia, and India. Physicists from more than a hundred countries had a stake in what came out of it.

What they wanted, mainly, was the Higgs boson: a particle theorized in 1964 to explain why other particles have mass at all, and never observed. Finding it, or ruling it out, was the accelerator's headline job.

Why some people wanted it stopped

Not everyone was celebrating that week. A retired nuclear safety officer named Walter Wagner and a Spanish journalist, Luis Sancho, had sued in federal court in Hawaii months earlier, arguing that colliding protons at such energy could spawn a microscopic black hole that would swallow the Earth from the inside out, or seed a runaway "strangelet" that converted ordinary matter into something else entirely.

It read like fringe science because it was; CERN's own safety review, along with independent physicists, put the odds at effectively zero, noting that cosmic rays routinely strike the atmosphere at far higher energies without incident and the planet is still here. A federal judge eventually dismissed the suit for lack of jurisdiction rather than ruling on the physics, and an appeals panel later turned away Wagner's attempt to revive it. But for a few weeks in 2008, "will the LHC destroy the world" was a genuine wire-service question, and CERN's press office fielded it alongside the real news.

The real news, that first day, was almost anticlimactic by comparison: a thin stream of protons, guided by magnets, went around a tunnel once. No collisions were scheduled. No black holes, real or imagined, were on the agenda. It was closer to a test flight than a launch.

What went wrong nine days later

It did not stay quiet for long.

On September 19, during a routine test to ramp the magnets toward full operating current, an electrical connection between two magnets failed. The joint, literally a soldered splice between two cables, had a higher resistance than it should have, and at high current it began to melt. The resulting arc punctured the vessel holding the liquid helium that keeps the magnets superconducting.

About six tons of helium vented into the tunnel, faster than the pressure-relief valves could handle it. The force of the escaping gas damaged supports and displaced magnets along a stretch of the ring; when engineers finally inspected the sector, fifty-three of the collider's 1,624 main magnets, some fifty feet long, had to be pulled out and either repaired or replaced.

Repairing a machine that size, in a tunnel that size, is not fast work. CERN put the direct cost at roughly $12.5 million, with spare parts running the total closer to $29 million: a rounding error against the multi-billion-dollar accelerator, but a bill nonetheless. The timing was worse than the cost: the damage happened just as CERN was heading into its scheduled winter shutdown, when the accelerator's cryogenic system is normally powered down anyway to save on electricity. Combined with the repair work, the outage stretched into a full year. The collider did not see beams again until November 2009.

Was the risk ever real

It is worth separating the two failures, because they got conflated in the press at the time. The black-hole lawsuit was about physics that, by the consensus of the people who study it, was never going to happen: proton collisions at the LHC's energies happen naturally, constantly, when cosmic rays hit the upper atmosphere, and nothing catastrophic has followed in the roughly four billion years Earth has been absorbing them. The helium leak was not exotic physics at all. It was a bad solder joint, the kind of thing that fails in ordinary electrical work, on a scale where failure meant tons of pressurized gas with nowhere quiet to go.

CERN published its own internal review of the incident in the weeks that followed, naming the specific joint and the electrical fault in detail rather than describing it in general terms — an unusually candid postmortem for an organization that had just taken a very public black eye. The fix included redesigned splices and new quench-detection and pressure-relief systems throughout the ring, meant to catch the next bad joint before it could do the same damage.

What the machine has done since

The Large Hadron Collider came back online in November 2009 and, within weeks, had already out-accelerated every collider before it. By March 2010 it was colliding protons at energies no machine had reached. On July 4, 2012, physicists at two of its detectors announced they had found a particle consistent with the long-sought Higgs boson, confirming the piece of the Standard Model that had gone unconfirmed since 1964. Peter Higgs, who had proposed it nearly fifty years earlier, was in the room.

The collider has since been upgraded twice, running at higher energy and higher collision rates than the machine that first circulated a beam in 2008. A "high-luminosity" version, designed to produce far more collisions per second, is due to come online later this decade. None of it looks much like doomsday. Mostly it looks like a very large, very expensive instrument, still buried under the same French and Swiss farmland, still looking for whatever comes after the Higgs.

Sources

• CERN, "CERN announces start-up date for LHC," press release, August 7, 2008 (home.cern)
• CERN, "First beam in the LHC – accelerating science," retrospective account of September 10, 2008 (home.cern)
• CERN Courier, "Incident in sector 3-4 of the LHC," December 2008
• NBC News, "'Big Bang' collider repairs: $29 million," 2008
• Discover Magazine, "Judge: Man Can't Sue Over LHC's Potential 'Destruction of the Earth'," September 2010