Knocking on Heaven's Door (26 page)

Everyone involved with the LHC describes September 10, 2008, as a day they will never forget. When I visited a month afterward, I heard many stories of the day’s euphoria. People followed the trajectory of two spots of light on a computer screen with unbelievable excitement. The first beam almost returned successfully on its first go-round, and with minor tweaking followed the exact path that was intended within the first hour of its being turned on. The beam at first went around the ring for a few turns. Then each successive burst of protons was adjusted slightly so that the beam was soon circulating hundreds of times. Not long after this, the second beam did the same—taking about one and a half hours to get exactly on track.

Lyn was just as happy that he didn’t know about the live video feed at the time from the control room, where the engineers were following the project, to the Internet, where the events were being broadcast for anyone to see. So many people watched those two dots on their screens that the sites were shut down for breaking capacity. People all over Europe—the CERN press office claims a couple of million—sat mesmerized as engineers modified the protons’ path to make them successfully circulate around the full circumference of the ring. Meanwhile, inside CERN, the thrill was palpable as physicists and engineers gathered in auditoriums to watch the same thing. At this point, the LHC outlook seemed more than extremely promising. The day was a wonderful success.

But a mere nine days later, euphoria transformed into despair. At the time, two important new features were to be tested. First, the beams were to be accelerated inside the LHC ring to higher energy than they had been during the first test, which used only the beam injection energy that protons have when first entering the LHC ring. The second part of the plan was to collide those beams, which would of course have been a huge milestone in LHC development.

However, at the last moment—on September 19—despite the engineers’ many considerations and precautions, the test failed. And when it did, it did so catastrophically. A simple soldering error in the copper casing connecting two magnets combined with too few functioning helium release valves caused a yearlong delay before protons would first collide.

The problem was that as scientists tried to ramp up the current and energy of the eighth and final sector, a joint between two magnets along the busbar that connects them broke. A
busbar
is a superconducting joint that connects a pair of superconducting magnets. (See Figure 27.) The splice that holds together a joint between two magnets was the culprit. The faulty connection created an electrical arc that punctured the helium enclosure and caused six metric tons of liquid helium—that would ordinarily be warmed up slowly—to be suddenly released. Superconductivity was lost in the quenching that occurred when the liquid helium heated up and reverted to gas.

[
FIGURE 27
]
A busbar connects different magnets together. A faulty solder in one was responsible for the unfortunate incident in 2008.

The enormous amount of helium released created a huge pressure wave that effectively caused an explosion. In less than 30 seconds, its energy displaced some magnets and destroyed the vacuum in the beam pipe, damaged the insulation, and contaminated 2,000 feet of beam pipe with soot. Ten dipoles were totally destroyed and 29 more were so damaged they needed to be replaced. Needless to say, this was not exactly what we had been hoping for. And this was also something no one in the control rooms had any inkling of until someone noticed that a stop button in the tunnel for one of the computers had been triggered by the escaping helium. Soon afterward, they realized the beam had been lost.

I learned more about the backstory during a visit to CERN a few weeks after the mishap. Keep in mind that the ultimate goal for collisions is a center of mass energy of 14 TeV, or 14 trillion electron volts. The decision was made to keep the energy down to only about 2 TeV for the first run in order to ensure that everything functioned properly. Later the engineers planned to increase it to 10 TeV (5 TeV per beam) for the first actual data runs.

However, the plan became more ambitious following a small delay due to a transformer that broke on September 12. Scientists continued testing the tunnel’s eight sectors up to 5.5 TeV during the interval afforded by the short delay and had time to test seven out of the eight sectors. They verified those could run properly at higher energy, but they didn’t have the opportunity to test the eighth. They nonetheless decided to charge ahead and attempt higher-energy collisions since there didn’t seem to be any problem.

Everything worked fine until the engineers attempted to raise the energy of the last untested sector. The crippling accident occurred when its energy was being raised from about 4 to 5.5 TeV—which required between 7,000 and 9,300 amps of current. This was the last moment for something to go wrong, and it did.

During the year of the delay, everything was repaired at a cost of about $40 million. Although repairing the magnets and the beam took time, they were not impossible tasks. Enough spare magnets were on hand to replace the 39 dipole magnets that were beyond repair. In total, 53 magnets (14 quadrupole and 39 dipole) were replaced in the sector of the tunnel where the incident occurred. In addition, more than four kilometers of the vacuum beam tube were cleaned, a new restraining system for 100 quadrupole magnets was installed, and 900 new helium pressure release ports were added. In addition, 6,500 new detectors were added to the magnet protection system.

The bigger risk was the presence of 10,000 joints between magnets that could potentially cause the same problem. The danger had been identified, but how could anyone trust that this problem would not reemerge elsewhere in the ring? Mechanisms were needed to detect any similar problem before it could cause any harm. The engineers once again rose to the challenge. Their updated system now looks for minuscule voltage drops that might signal the presence of resistive joints, signaling a break in the closed system that houses the cryogenics that keeps the machine cold. Caution also dictated some delays to improve the helium release valve system and to further study the joints as well as the copper casings of the magnets themselves—which meant a delay in achieving the highest energies at which the LHC is designed to operate. Nonetheless, with all the new systems to monitor and stabilize the LHC, Lyn and others were confident that the kind of pressure buildups that caused the damage will be avoided.

In some sense, we are lucky that engineers and physicists were able to fix things before true operations began and filled the experiments with radiation. The explosion cost the LHC a year before they could even begin to test beams and aim for collisions again. That was a long time, but not so long on the scale of a quest for the underlying theory of matter that we have had for the last 40 years, and in many respects for thousands of years.

On October 21, 2008, the CERN administration did, however, stick to one piece of their initial plan. On that day, I joined 1,500 other physicists and world leaders outside Geneva to celebrate the official LHC inauguration, which had been optimistically planned well in advance—before anyone could have predicted the disastrous events that occurred a mere few weeks before. The day was filled with speeches, music, and—as is important at any European cultural event—good food. It was enjoyable and informative even with the premature timing. Despite anxieties about the September incident, everyone was filled with hope that these experiments would shed light on some of the mysteries surrounding mass, the weakness of gravity, dark matter, and the forces of nature.

Although many CERN scientists were unhappy about the infelicitous timing of the event, I saw the celebration more as a contemplation of this triumph of international cooperation. The day’s events did not yet honor discovery but instead recognized the potential of the LHC and the enthusiasm of the many countries participating in its creation. A few of the speeches were truly encouraging and inspirational. The French prime minister, Frangois Fillon, spoke of the importance of basic research and how the world financial crisis should not impede scientific progress. The Swiss president, Pascal Couchepin, spoke of the merit of public service. Professor Jose Mariano Gago, Portugal’s minister for science, technology, and higher education, spoke about valuing science over bureaucracy and the importance of stability for creating important science projects. Many of the foreign partners visited CERN for the first time for the day’s celebration. The person seated next to me during the ceremony worked for the European Union in Geneva—but had never set foot inside CERN. Having seen it, he enthusiastically informed me of his intention to return soon with his colleagues and friends.

NOVEMBER 2009: VICTORY AT LAST

The LHC finally came back online on November 20, 2009, and this time, it was a stunning success. Not only did proton beams circulate for the first time in a year, but a few days later, they finally collided, creating sprays of particles that would enter the experiments. Lyn enthusiastically described how the LHC worked better than he had expected—a remark that I found encouraging but a bit peculiar in light of his being in charge of making the machine run as successfully as it had.

What I hadn’t understood was how much more quickly all the pieces had fallen into place than would have been anticipated based on the experience with past machines. Maurizio Pierini, a young Italian CMS experimenter, explained to me what Lyn had meant. Tests that took 25 days in the 1980s for the LEP beams of electrons and positrons in the same tunnel were now completed in less than a week. The proton beams were remarkably on target and stable. And the protons stayed in line—very few stray particles were detected. The optics worked, the stability tests worked, realignments worked. The actual beams matched precisely the computer programs that simulated what should occur.

In fact, the experimenters were taken by surprise when they were told Sunday at 5:00
P.M
., only a couple of days after the renewed beams began circulation, to expect collisions the next day. They had anticipated a little bit of time between first beams after the shutdown and the first actual collisions they could record and measure. This was now to be their first opportunity to test their experiment with actual proton beams, rather than the cosmic rays they had used while waiting for the machine to run. The short notice meant, however, that they had very little time to reconfigure their computer
triggers
that tell computers which collisions to record. Maurizio described the anxiety they all felt, since they didn’t want to foolishly fumble this opportunity. At the Tevatron, the first test had been mangled by an unfortunate resonance of the beam circulation with the readout system. No one wanted to see this happen again. Of course, in addition to unease, an enormous amount of excitement was shared by everyone involved.

[
FIGURE 28
]
Brief outline of the LHC’s history.

On November 23, the LHC at long last had its first collision. Millions of protons collided with the injection energy of 900 GeV. These events meant that after years of waiting, experiments could begin taking data—recording the results of the first proton collisions in the LHC ring. Scientists from ALICE, one of the smaller experiments, even submitted a preprint (a paper before publication) on November 28.

Not too long afterward, a modest acceleration was applied to create 1.18 TeV proton beams, the highest-energy circulating beams ever. Only a week after the first LHC collisions, on November 30, these higher-energy protons collided. The net center of mass energy of 2.36 TeV exceeded the highest energies ever achieved before, breaking Fermilab’s eight-year-old record.

Three LHC experiments registered beam collisions and tens of thousands of such collisions occurred over the next few weeks. Those collisions won’t be used to discover new physical theories, but they were incredibly useful for determining that the experiments in fact worked and could be used to study Standard Model
backgrounds
—events that don’t indicate anything new, but could potentially interfere with real discoveries.