The Particle at the End of the Universe: How the Hunt for the Higgs Boson Leads Us to the Edge of a New World

by Sean Carroll

  Basically a proton is a floppy bag of quarks, antiquarks, and gluons, moving around the LHC beam pipe near the speed of light. Richard Feynman dubbed all these constituent particles “partons.” According to relativity, objects moving near light speed are contracted along their direction of motion. So two protons colliding inside the detector resemble pancake-shaped collections of partons, flying at each other face-on. When one proton interacts with another one, it’s actually just one of the partons in one proton that interacts with a parton inside the other proton. As a result, it’s hard to know exactly how much energy is involved in a collision, because we don’t know which of the partons did the interacting.

  Conditions inside an LHC experiment can get pretty intense. There are about fourteen hundred bunches of protons in each beam, and a bunch moving in one direction passes inside the detector by one moving in the other direction about 20 million times per second. Each bunch carries more than 100 billion protons, so that’s a lot of particles ready to interact. However, even though the bunches are quite small (a thousandth of an inch across), they are still huge compared with the size of a proton; the overwhelming volume of a bunch is empty space. Each time bunches cross, maybe twenty or so interactions will occur between the billions of protons.

  Cartoon of two protons approaching each other in an LHC experiment. The protons, ordinarily spherical, are squeezed into pancake shapes by the effects of relativity, due to moving near the speed of light. Inside the protons are partons, which include quarks (filled circles), antiquarks (empty circles), and gluons (squiggles). There are three more quarks than antiquarks; these are the “valence quarks,” while all the other partons are virtual particles.

  Twenty interactions is still a lot. A single collision of two protons often gives off a messy spray of particles, as many as a hundred hadrons in a single event. We’re therefore faced with the danger of “pileup”—many events occurring inside the detector at the same time, making it difficult to distinguish what happened where. This is one of the many reasons why CMS and ATLAS have to strain technology and computing power to the limits of what is currently possible. More collisions are good because it means more data; but too many collisions at once means you can’t tell what’s going on.

  Particles in the chamber

  There is a logic to the construction of a particle detector, and it is dictated by the particles themselves. What can possibly come out of a collision? Only the various Standard Model particles we know and love—the six quarks, the six leptons, and the various force-carrying bosons. (We hope to produce completely new species, but those will typically decay into Standard Model particles.) So all we have to do is consider these possibilities and ask how we can best detect and correctly identify them. Let’s go through the list.

  Quarks

  We can lump all the quarks together, because we don’t ever see them isolated—they are confined inside hadrons. But you can create a quark-antiquark pair in a collision, and have the two particles move quickly in opposite directions. In that case what happens is that the strong force surrounding the quarks asserts itself, and a spray of hadrons coalesces around the original particle. This shows up in your detector as the “jets” mentioned above. Our job then is to detect the resulting hadrons, which is a relatively easy task, and reconstruct the individual jets, which can be a pain. It can be hard to tell what kind of quark was produced, although there are tricks we can use. For example, bottom quarks last just long enough that they travel a tiny distance before decaying. The particles resulting from the decay therefore emerge slightly offset from the main collision, which can be used to identify bottoms even if their own tracks aren’t directly seen.

  Gluons

  Although they are bosons rather than fermions, gluons are still strongly interacting, so they show up in your detector in a similar way: as a jet of hadrons. One difference is that it’s possible to make a single gluon—a quark could spit one off, for example—while newly produced quarks always come paired with antiquarks. So if you see three jets in an event, it means you’ve made a quark-antiquark pair and a gluon. Events like that are what Sau Lan Wu and her collaborators used to first establish that gluons are real.

  W bosons, Z bosons, tau leptons, Higgs bosons

  These quite different particles are all grouped together for a simple reason: They are very heavy and therefore short-lived, decaying quickly into other particles—so quickly that they will never be seen directly in your detector. You have to infer their existence by looking at what they decayed into. From this list, tau leptons have the longest lifetime, and in the right circumstances they can last just long enough to be identified.

  Electrons and photons

  These are the easiest particles to detect and precisely measure. They don’t fragment into messy jets like quarks and gluons, but they readily interact with charged particles in a material, creating electrical currents that are simple to identify. It’s also straightforward to tell the two apart, because electrons (and positrons, their antiparticles) are electrically charged and therefore swayed by a magnetic field, while photons are neutral and move unimpeded in a straight line.

  Neutrinos and gravitons

  These are the particles that don’t feel either the strong force or the electromagnetic force. As a result, there’s no practical way to capture them in a detector, and they just escape unnoticed. Gravitons are only produced by the gravitational interaction, which is so weak that essentially no gravitons are made in a collider and we don’t have to worry about them. (In some exotic theories, gravity is effectively strong at high energies and gravitons are produced; physicists certainly take this possibility into account.) Neutrinos, however, are produced by the weak interactions, so they occur all the time. Fortunately they are the only Standard Model particles that can be produced but not detected. So there is a simple rule: Everything that is not detected is probably a neutrino.

  When two protons collide, they are both moving along the beam pipe, so the total momentum in directions perpendicular to the beam adds up to zero. (The momentum of one particle is the amount of oomph it carries along its direction of motion. For several particles, we just add the separate momenta, but they can combine to zero when the particles are moving in opposite directions.) Momentum is conserved, so it should also add up to zero after the collision. Therefore, we can measure the actual momentum of the particles we detect, and if the answer is not zero, we know there must be neutrinos moving the other way to compensate. This is known as the method of “missing transverse momentum,” or just “missing energy.” We might not know how many neutrinos carried off the missing momentum, but that can often be deduced from knowing what other particles were produced. (A weak interaction that produces a muon will also make a muon neutrino, and so on.)

  Muons

  This leaves the muon, which is one of the most intriguing particles from the perspective of an LHC experiment. Like electrons, they leave an easily detectable electrical track, and curve within a magnetic field. But they are two hundred times heavier than an electron. That means they can decay into lighter particles, but their lifetime is still pretty long; unlike the even heavier tau lepton, muons will generally last long enough to survive to the edge of your detector. And they will make it, because muons tend to bash through materials rather than be captured. That’s the benefit of being much heavier than electrons but not strongly interacting. A muon will lumber through all the layers of the experiment like a Jeep driving through a field of wheat, leaving an easily identifiable trail in its wake.

  Muons act like super X-rays, penetrating deeply through ordinary stuff. This property was put to good use years ago by Luis Alvarez, who won the Nobel Prize for finding all those hadrons at the Bevatron. Alvarez was intrigued by the pyramids of Egypt, and in particular the large pyramids of the pharaoh Khufu (“Cheops” in Greek) and his son Khafre (“Chephren”), which sit next to each other in Giza, outside Cairo. Khufu’s is the “Great Pyramid,” and it was originally slightly larger, alt
hough external wear has left it a bit smaller than Khafre’s these days. Inside Khufu’s pyramid are three chambers, while Khafre’s pyramid seems solid save for one burial chamber at ground level. This difference has puzzled archaeologists for years, and many have theorized that Khafre’s pyramid contains undiscovered chambers.

  Alvarez, a brilliant physicist with a penchant for puzzles, hit on an idea: Use muons coming from the skies in the form of cosmic rays to peek inside the rock of Khafre’s pyramid. It would be a crude experiment, but able to distinguish between solid rock and an empty chamber. Alvarez’s team of Egyptian and American physicists assembled a muon detector in the single known chamber at the lower level of the pyramid, looking to count the number of muons coming from different directions. This was 1967, and the project suffered a delay when the Arab-Israeli war broke out the day before they were scheduled to first take data. But eventually they got up and running, and discovered . . . nothing. All directions of the pyramid appeared to be equally good at stopping muons, in contrast to the hope that some directions would let more of them through because they contained an empty chamber. It remains a puzzle why the son’s pyramid is noticeably less complicated than the father’s.

  Layers of detectors

  The ATLAS and CMS experiments settled on a strategy for squeezing as much information as possible out of the particle collisions they observe. Both detectors are constructed in layers, with four different pieces of apparatus serving very specific purposes: an inner detector, surrounded by an electromagnetic calorimeter, which is in turn surrounded by a hadronic calorimeter, and finally a muon detector on the outside. Any particles produced in a collision will radiate outward from the collision point, passing through different layers until they are finally captured or they escape to the external world.

  A cartoon depiction of a general-purpose particle experiment, such as ATLAS or CMS. The central region contains an inner detector that measures the paths of charged particles. Next is the electromagnetic calorimeter that captures photons and electrons, then the hadronic calorimeter that captures hadrons. Finally, the muon detector that tracks the muons.

  The job of the inner detector, the innermost layer of the onion, is to act as a tracker that provides pinpoint information about the trajectories of charged particles that emerge from the collision point. It’s not an easy job; every square centimeter of the instrument is bombarded with tens of millions of particles per second. Anything you put there has to do its job while surviving an unheard-of amount of radiation exposure. Indeed, the very first design drawings for CMS simply left this region of their detector empty, since physicists didn’t think they could build a precision instrument that could take the heat. Fortunately, they were encouraged to keep trying by rumors that the military had solved the problem of making electronic readouts that could function effectively in this kind of harsh environment. They ultimately succeeded by figuring out how to “harden” very fine commercial electronics that weren’t originally intended to withstand such conditions.

  The inner detectors are complicated multicomponent machines with slightly different features between the two experiments. The ATLAS inner detector, for example, consists of three different instruments: a pixel detector with incredibly fine resolution; a semiconductor tracker made of silicon strips; and a transition radiation tracker made of gold-plated tungsten wire inside thin tubes known as “straws.” The job of the inner detector is to record the paths of emerging particles as precisely as possible, allowing physicists to reconstruct the interaction points from which they originate.

  The next layers are the calorimeters, electronic and hadronic. “Calorimeter” is a fancy word for “device that measures energy,” just as “calories” are used to quantify the energy in the foods we eat. The electromagnetic calorimeter is able to capture electrons and photons via their interactions with nuclei and electrons in the calorimeter itself. Strongly interacting particles generally pass right through the electromagnetic calorimeter, only to be captured by the hadronic calorimeter. This component consists of layers of dense metal that interact with the hadrons, alternating with scintillators that measure the amount of energy deposited. Measuring the energies of the particles is a crucial step in identifying what they are, and often the mass of whatever particle decayed to create them.

  A cross-section of an experiment, showing the behavior of different particles. Neutral particles like photons and neutral hadrons are invisible to the inner detector, but charged particles leave curved tracks. Photons and electrons are captured by the electromagnetic calorimeter, while hadrons are captured by the hadron calorimeter. Muons make it to the outer detector, and neutrinos escape detection entirely. In the CMS experiment, muons curve in the opposite direction in the outer detector because the magnetic field points the opposite way.

  The final layers of the experiments are the muon detectors. Muons have enough momentum to punch through the calorimeters, but can be precisely measured by the giant magnetic chambers that surround them. This is important because muons are not created by the strong interactions (since they are leptons, not quarks), and only rarely by the electromagnetic interactions (because they are so heavy and it’s easier just to make electrons). Therefore, muons generally come about from the weak interactions, or something brand-new. Either alternative is interesting, and muons play an important role in the search for the Higgs.

  We now see why the design of the ATLAS and CMS experiments takes the form that it does. The inner detectors provide precision information about the trajectories of all charged particles leaving the collisions. Electrons and photons are captured, and their energies measured, by the electromagnetic calorimeter, while strongly interacting particles suffer the same fate in the hadronic calorimeter. Muons escape the calorimeters but are carefully studied in the muon detector. Among the known particles, only neutrinos escape undetected, and we can infer their existence by looking for missing momentum. All in all, an ingenious scheme to squeeze out all the information we can from the colliding protons produced by the LHC.

  Information overload

  At the LHC, bunches of protons come into collision 20 million times per second. Every crossing produces dozens of collisions, so we have hundreds of millions of collisions a second. Every collision is like fireworks going off inside the detector, creating multiple particles, up to a hundred or more. And the finely calibrated instruments inside the experiments collect precise information about what every one of those particles does.

  That’s a lot of information. A single collision event at the LHC results in about one megabyte of data. (The raw data is more than twenty megabytes, but clever compression brings it close to a single megabyte.) That’s the size of the text of a large book, or the total amount of RAM in a space shuttle’s operating system. Decent home-computer hard drives these days can store a terabyte of data, or a million megabytes, which is huge—all the text in all the books of the Library of Congress amounts to only about twenty terabytes. You could store a million LHC events on one of these ordinary hard drives, which sounds good—except when you remember that there are hundreds of millions of events per second. So you would be filling up a thousand hard drives per second. Not really feasible, even given that CERN can afford better hard drives than you have on your laptop.

  Outside the LHC, the largest single database in the world belongs to the World Data Center for Climate in Germany. It contains about six petabytes of climate data, or six thousand terabytes. If we recorded all the data created at the LHC, we would overflow a database of that size in a couple of seconds. Welcome to the world of Big Data.

  Clearly, data storage (and transmission and analysis) at the LHC is a major challenge, one that is met by a combination of many different techniques. The most important one, however, is the most basic: not recording the data in the first place. This is worth emphasizing: The overwhelming majority of data collected by the LHC is instantly thrown away. We have no choice; there is no feasible way to record it all.

  You
might think that a more cost-effective strategy would simply be to not produce so much data in the first place, for example by lowering the luminosity of the machine. But particle physics doesn’t work that way—every collision is important, even if we don’t record its data to disk. That’s because quantum mechanics, which is ultimately responsible for the interactions that create these particles, only predicts the probability of certain outcomes. We can’t pick and choose what comes out when we collide two protons; we have to take what nature gives us. A large majority of the time, what nature gives us is pretty boring, at least in the sense that it’s stuff we already understand. To create a small number of interesting events, we have to produce an enormous number of pedestrian events, and swiftly pick out the interesting nuggets.

  This raises a different problem, of course: how to figure out whether an event is “interesting,” and to do so extremely quickly, so that we can decide whether this is data worth keeping. That’s the job of the trigger, one of the most crucial aspects of an LHC experiment.

  The trigger itself is a combination of hardware and software. The first-level trigger brings the output of all the instruments in the experiment into an electronic buffer and performs an ultrarapid scan (in about a microsecond) to see if anything potentially interesting is going on. About ten thousand events out of a billion get a stamp of approval and move on. The second-level trigger is a sophisticated piece of software that looks at more precise characterizations of the events (much like an ER doctor making a preliminary rapid diagnosis, then homing in with more delicate tests) to get you down to the events that are actually recorded for later analysis. We end up keeping only several hundred events out of the many millions produced per second—but they’re the most interesting ones.