How Particle Accelerators Work: From Cyclotrons to the Large Hadron Collider

Step 1: Generate the Particles

Every accelerator begins with a particle source. For proton accelerators like the LHC, the source is a small cylinder of hydrogen gas. An electric field strips the electrons from hydrogen atoms, leaving bare protons. For electron accelerators, a heated filament emits electrons through thermionic emission, similar to how old cathode ray tubes worked. Heavy-ion accelerators start with heavier elements, ionizing atoms of lead, gold, or other metals to produce beams of charged nuclei. The source produces particles in bunches, discrete packets containing billions of particles that are then fed into the first stage of acceleration.

At the LHC, the proton source produces bunches of roughly 100 billion protons. These initially move at low speed and pass through a series of pre-accelerators before reaching the main ring. CERN's injector chain consists of Linac4 (a linear accelerator that brings protons to 160 MeV), the Proton Synchrotron Booster (to 2 GeV), the Proton Synchrotron (to 26 GeV), and the Super Proton Synchrotron (to 450 GeV). Only after passing through all four stages are the protons injected into the LHC for final acceleration to 6.5 TeV per beam. This staged approach is more practical and cost-effective than trying to do all the acceleration in a single machine.

Step 2: Accelerate with Electric Fields

The workhorse of particle acceleration is the radiofrequency (RF) cavity, a hollow metallic structure in which an oscillating electric field pushes particles forward. The key is timing: the electric field must switch direction at exactly the right moment so that it always pushes the particles, never slows them. In a circular accelerator, the RF frequency is synchronized to the particles' orbital frequency, so that every time a particle passes through the cavity, it receives another kick of energy. In a linear accelerator, a series of cavities with progressively longer spacing (or higher frequency) keeps pace with the particles as they speed up.

The LHC uses 16 superconducting RF cavities (8 per beam), each operating at 400 MHz and providing an accelerating voltage of about 2 MV per cavity. Each time a proton bunch passes through a cavity, it gains about 485 keV of energy. At 6.5 TeV, a proton circles the 27-kilometer ring 11,245 times per second, receiving a total energy boost of about 485 keV x 8 cavities per revolution. Ramping the beam from 450 GeV (injection energy) to 6.5 TeV takes about 20 minutes, during which each proton makes roughly 13 million revolutions and passes through the cavities over 100 million times. The RF system also shapes the beam into tight bunches, keeping particles clustered together and preventing them from spreading out longitudinally.

Step 3: Steer and Focus with Magnets

Charged particles naturally travel in straight lines. To keep them on a circular path, accelerators use dipole magnets that produce a uniform magnetic field perpendicular to the beam direction. A charged particle in a magnetic field experiences a force that bends its trajectory without changing its speed, just as gravity bends a satellite's orbit without speeding it up or slowing it down. The stronger the magnetic field and the larger the ring, the higher the energy of particles that can be confined. This is the fundamental relationship that determines accelerator design: a more energetic particle requires either a stronger magnet or a larger ring to keep it on course.

The LHC uses 1,232 dipole magnets, each 14.3 meters long and weighing 35 tonnes. These magnets use niobium-titanium superconducting wire cooled to 1.9 Kelvin (colder than outer space) by superfluid helium. At this temperature, the wire carries current with zero electrical resistance, enabling magnetic fields of 8.3 Tesla, roughly 100,000 times the Earth's magnetic field. The entire 27-kilometer ring contains about 96 tonnes of superfluid helium, making the LHC the largest cryogenic installation in the world. If these magnets were made with conventional copper wire instead of superconductors, they would require about 120 MW of electrical power just for the magnets, an impractical energy consumption.

Quadrupole magnets focus the beam, squeezing it in one direction while allowing it to spread in the perpendicular direction. By alternating the orientation of quadrupole magnets along the beam line (a technique called strong focusing or alternating gradient focusing, invented in 1952), the beam is kept tightly focused in both transverse directions simultaneously. The LHC uses 392 quadrupole magnets in addition to numerous sextupole, octupole, and higher-order corrector magnets that fine-tune the beam orbit and compensate for tiny imperfections in the main magnets. At the collision points, special quadrupole magnets called inner triplets squeeze the beam to a width of just 16 micrometers, about one-fifth the width of a human hair, to maximize the probability of collisions.

Step 4: Collide the Beams

The LHC circulates two proton beams in opposite directions in separate beam pipes, bringing them together at four designated interaction points where the major experiments (ATLAS, CMS, ALICE, LHCb) are located. Each beam contains up to 2,808 bunches of roughly 100 billion protons, spaced 25 nanoseconds apart. When bunches from opposite beams cross at an interaction point, the combined collision energy reaches 13.6 TeV (the sum of both beams' energies). This is the energy that is available to create new particles, because the collision converts kinetic energy into mass.

Despite each bunch containing 100 billion protons, individual collisions are extremely rare because protons are incredibly small (about 10^-15 meters in diameter). When two bunches cross, typically only 20 to 50 proton-proton collisions occur out of the 200 billion possible pairings. The bunches cross at each interaction point 40 million times per second, producing roughly 600 million collisions per second per experiment. Over the course of a year, each experiment records data from trillions of collisions. The instantaneous luminosity of the LHC, a measure of its collision rate per unit cross-section, has exceeded 2 x 10^34 cm^-2 s^-1, roughly twice its original design specification.

Not all accelerators are colliders. Fixed-target experiments fire a beam at a stationary block of material. This approach provides less center-of-mass energy than head-on collisions (because much of the beam's energy goes into moving the collision products forward rather than creating new particles) but produces far more collisions per second and is essential for producing secondary beams of neutrinos, kaons, pions, muons, and antiprotons. Fermilab's neutrino program fires an intense proton beam into a graphite target to produce pions, which decay into the neutrino beams used by the NOvA and future DUNE experiments.

Step 5: Detect and Analyze the Products

Particle detectors are layered instruments surrounding the collision point, each layer designed to identify different types of particles. The innermost layer is the tracking detector, typically made of silicon pixel sensors that record the paths of charged particles with micrometer precision. A strong magnetic field (2 Tesla in CMS, 2 Tesla from a solenoid in ATLAS's inner region) bends charged particle tracks: the curvature reveals each particle's momentum, and the direction of curvature reveals its charge sign.

The electromagnetic calorimeter, the next layer out, stops and measures the energy of electrons and photons. These particles produce electromagnetic showers, cascades of electron-positron pairs and photons, in dense crystalline or liquid materials. CMS uses 75,848 lead tungstate crystals, each about the size of a brick, that emit scintillation light proportional to the deposited energy. ATLAS uses liquid argon as the active medium interleaved with lead absorber plates. The hadronic calorimeter, beyond the electromagnetic calorimeter, stops and measures the energy of hadrons (protons, neutrons, pions, kaons) using alternating layers of dense absorber material (steel, brass, or tungsten) and active sampling material (scintillator tiles or liquid argon).

The outermost layer is the muon detector. Muons are the only charged particles that pass through both calorimeters without being stopped, because they are 207 times heavier than electrons (so they radiate far less bremsstrahlung) and do not interact through the strong force (so they pass through hadronic material with little interaction). Muon detectors use large gas-filled chambers that record ionization tracks left by passing muons. CMS's muon system spans an area of 25,000 square meters, and ATLAS's covers even more. Neutrinos escape the detector entirely without leaving any signal, so they are inferred from "missing energy," the imbalance between the total measured momentum and what conservation of momentum requires.

The data challenge is enormous. Each bunch crossing produces roughly 1 megabyte of raw data, and at 40 million crossings per second, the raw data rate exceeds 40 terabytes per second. No storage system can record this volume, so a multi-level trigger system selects only the most interesting events in real time. The Level-1 trigger, implemented in custom electronics, reduces the rate from 40 MHz to about 100 kHz in less than 4 microseconds. The High-Level Trigger, running on a farm of thousands of processors, further reduces the rate to about 1,000-2,000 events per second, which are permanently stored. These stored events are distributed to computing centers worldwide through the Worldwide LHC Computing Grid, which links over 170 computing centers in 42 countries.