I've been spending quite a bit of time digging in the world of moving very small things at very high speeds. This the first in a series of research pieces around applications of high energy physics devices and how we might go about building a new industry around them.
There are more than thirty thousand particle accelerators operating in the world today.
You've probably heard of the biggest: the Large Hadron Collider, a 27-kilometre ring beneath the Swiss-French border that smashes particles together at relativistic speeds. It’s the largest and most expensive machine ever built.
But the LHC’s headlines have probably given you the wrong impression. Particle accelerators are not just tools for fundamental physics. They treat cancer, manufacture semiconductors, and radiation-test components for spacecraft. The Louvre houses one that helps detect art fakes. In 1945, one produced the enriched uranium for the bomb dropped on Hiroshima. Quietly, they have become load-bearing infrastructure for the modern world.
Almost a century since their invention, there are now almost as many particle accelerators globally as there are McDonald's.
This is a brief history of particle accelerators: why we built them, what they've revealed about sub-atomic reality, and the myriad ways in which they've become essential to modern life.
I: The Birth of Particle Acceleration
On November 30, 1927, Ernest Rutherford – now widely regarded as the father of nuclear physics – stood before the Royal Society and made an impassioned plea for "a copious supply of atoms and electrons which have an individual energy far transcending that of the α and β-particles."
For three decades, Rutherford had been using nature's own projectiles to probe the inside of atoms. Alpha particles, emitted by radioactive elements, fired at targets to see what happened. In 1909, he aimed them at gold foil. Most passed straight through. A tiny fraction bounced back like they'd hit a wall. Turns out they had hit a wall: the nucleus. Nearly all of an atom's mass, packed into a core 100,000 times smaller than the atom itself. In 1917, he fired alpha particles at nitrogen gas and knocked loose individual protons. It was the first time anyone had deliberately broken apart a nucleus.
Still, Rutherford’s tools were of limited use for sub-atomic exploration. A lump of radium emits alpha particles in every direction, at one fixed energy, at a rate you can't control. You can discover new particles that way, but you can't map the internal structure of a nucleus. For that you need to fire projectiles at chosen energies and see how they scatter. You need a beam that you can aim, tune, and turn on at will. You need a particle accelerator.
On April 14, 1932, At Rutherford's Cavendish Laboratory, John Cockcroft and Ernest Walton stacked capacitors and rectifiers into a voltage multiplier that crammed 400,000 volts into a vacuum tube and fired protons at lithium. Physics by brute force. The lithium nucleus broke apart into two helium nuclei, proving the engineering worked. The Nobel beckoned for Cockcroft and Walton in 1951.
But the design had a fundamental constraint on its scalability. The particle passed through the machine once, in a straight line, and whatever energy it picked up in that single pass was all it got. Want more energy? You need more voltage. And there's only so much voltage you can give.
Over in Berkeley, California, Ernest Lawrence and his graduate student M. Stanley Livingston had already figured out a better approach: instead of firing one enormous voltage in a straight line, bend the particle's path into a spiral and give it a small kick on each lap. The particle gets pulled across a gap between two electrodes, gets a repulsive push from behind as the field switches polarity, then curves back around for another pass. Each lap adds more energy. In principle, the machine could reach higher energies just by widening the spiral. No need for ever-higher voltages.
Lawrence called it a "magnetic resonance accelerator". The physics world gave it a much better moniker: ‘the cyclotron’.
The cyclotron scaled fast. Livingston got the first proof-of-concept running in January 1931 and accelerated protons to 80 keV. By 1932, 1.2 MeV. By 1937, 16 MeV. Natural alpha particles top out around 9 MeV; the cyclotron had left nature's own projectiles behind in less than a decade. Lawrence won his own Nobel in 1939.
The fundamental engineering challenges had been overcome. The golden age of particle physics was about to begin.
II: Revelations
Accelerators enabled breakthroughs in nuclear physics from the start. Cockcroft and Walton hadn't just split the atom. They'd shown that the mass lost in the reaction matched the energy released, exactly as Einstein's E=mc² predicted. Rutherford and Oliphant discovered deuterium fusion at the Cavendish in 1934. Lawrence's cyclotrons produced entirely new elements: technetium, neptunium, plutonium.
But these were pieces of a much bigger puzzle. The holy grail of physics is a single theory that accounts for everything in the universe — every star, every cell, every grain of sand — in terms of a handful of particles and the rules governing their interactions.
The Standard Model
From the 1950s, accelerators became the primary tool for chasing this holy grail. And the particles kept coming. Protons and neutrons, long assumed to be fundamental, turned out to be made of smaller things called quarks. Neutrinos appeared in multiple flavours. Antimatter — mirror-image particles with opposite charge — turned out to be a universal feature of sub-atomic reality. Each new machine revealed another layer of structure nobody had known was there.
The flood of new particles soon felt out of control. Fermi reportedly quipped, "If I could remember the names of all these particles, I'd have been a botanist." Willis Lamb opened his 1955 Nobel lecture by suggesting that the discoverer of each new particle should be fined rather than awarded a prize.
The particle zoo, as it came to be known, was badly in need of sorting. And gradually, it sorted itself. The particles weren't random; they fell into families with precise mathematical relationships between them. Relationships that could predict new particles before anyone found them. The omega-minus baryon was proposed on paper in 1962 and found in a bubble chamber two years later, exactly where the mathematics said it would be. Piece by piece, the zoo assembled into a single theoretical framework: the Standard Model (SM).
The SM holds that all matter is built from twelve particles (six quarks and six leptons). The forces that act on matter are, the SM says, carried by particles called bosons. Imagine two people on ice throwing a heavy ball back and forth: each throw pushes the thrower backward, each catch pushes the catcher backward. From a distance it looks like a repulsive force. All that's actually passing between them is a ball. That, roughly, is how bosons work.
In 1964, three groups of theorists independently proposed one more piece of what would become the Standard Model: a 'Higgs field' persistent through all of space. The basic idea is that particles acquire mass by interacting with this field. The more strongly a particle couples to the Higgs field, the more mass it acquires; the less strongly, the lighter it remains. Think of it like moving through a cocktail party: a nobody – or, correspondingly, a particle with weak coupling to the field – walks straight through the room unimpeded, but a celebrity gets mobbed, slowed down, made harder to accelerate. In quantum field theory, every fundamental field comes with a particle, the smallest possible ripple in that field. If the Higgs field existed, it would have a corresponding boson: the seventeenth and final particle the SM predicted.
How well does the framework actually work? Its prediction of the electron's magnetic moment matches experimental measurement to better than one part in a trillion. That's like predicting the distance from London to New York to within the width of a human hair. And, by 2000, sixteen of its seventeen particles had been found in experiments. But for over forty years, the seventeenth – the Higgs boson – never showed up.
The prodigal boson
To see smaller structures you need higher energies, and higher energies mean a bigger ring.
Enter the Large Hadron Collider. CERN shut down three antimatter machines in 1996 just to free up the resources to build it. Over 1,200 dipole magnets, each 15 metres long, bend the proton beams around the ring. To generate fields strong enough to steer protons at near-light speed, the magnets are superconducting: cooled by superfluid helium to 1.9 kelvin, colder than outer space. Its first collisions came in 2010.
Over two years, two independent detector teams – ATLAS and CMS, each comprising thousands of scientists – sifted signal from noise. The threshold for discovery in particle physics is five sigma: a statistical standard meaning there's roughly a one-in-3.5-million chance the result is a fluke. In July 2012, both teams announced they'd crossed it. Independently, and for the same particle. A new boson at 125 GeV, with exactly the properties predicted half a century earlier.
A cause for conCERN? Going bigger
The Higgs boson was found, but its measured properties left room for a possibility the Standard Model alone doesn't predict: the Higgs boson we discovered might not be the only one. Some extensions of the Standard Model require a whole family of Higgs bosons, each appearing at higher energies.
The problem is that discovering heavier particles requires colliding already-discovered lighter particles with them at higher energies. And higher energies demand larger, more powerful accelerators. Even the LHC, enormous as it is, has a ceiling. Subject to a decision in 2028, a new Future Circular Collider will be built at CERN with a circumference of 91 km. It will be built to reach energies where, if additional Higgs bosons exist, they should have nowhere left to hide.
III: Applications
The LHC makes the headlines, but fewer than 1% of the world's accelerators are high-energy research machines. The rest keep the modern world running. Three industries in particular show how.
Medicine
By 1936, Lawrence's physician brother, John, had begun work at the Berkeley Rad Lab. That December, he treated his first leukaemia patient with phosphorus-32 produced in a cyclotron. Over the next few years, the list of biologically useful isotopes grew: radioactive iodine for thyroid disease, sodium-24 for tracing circulation, iron-59 for studying blood, produced virtually to order.
But isotopes were only half the story. In 1946, Robert Wilson, a Manhattan Project physicist, published a paper in Radiology on how protons move through matter. For most of their journey they are relatively gentle, depositing energy in small increments. Then, just before they stop depositing, they release a large fraction all at once: a sharp spike, at a depth set precisely by the proton's initial speed. The tissue on the way in is largely spared. The tissue where the beam stops is not. Tune the beam correctly and that spike sits inside a tumour. The first patients were treated with proton beams at Berkeley in 1954.
There are now over 120 proton therapy centres worldwide, with 50 more under construction. By the end of 2023, close to 350,000 patients had been treated with proton beams, some more recently with beams of heavier ions like carbon. That figure grows by tens of thousands each year.
Accelerators’ increased medical impact will not result only from existing treatments being scaled. Extraordinary technological innovations are still being developed. CERN is working with hospitals on compact machines that deliver an entire cancer treatment in under a second. And at FAU Erlangen and Stanford, researchers are developing dielectric laser accelerators — particle accelerators etched onto microchips no bigger than a grain of sand — that could one day be placed on an endoscope to deliver radiotherapy directly to a tumour, destroying cancerous cells while leaving surrounding healthy tissue untouched.
Pharmaceuticals
Not all the discoveries were deliberate. Every time a charged particle curves — bent by a magnetic field, as it constantly is inside a synchrotron — it emits radiation. At synchrotron velocities, that radiation is extraordinarily intense, tightly collimated, and spans wavelengths short enough to resolve individual molecules. For years it was treated as a nuisance. By the 1960s, someone thought to look more carefully at what was being thrown away.
The first dedicated synchrotron light sources were built in the 1970s. There are now over seventy worldwide. Over 70% of all known protein crystal structures have been solved using their beamlines. Much of modern drug design starts with a picture taken by a particle accelerator.
Semiconductors
The semiconductor industry came to accelerators through a problem it couldn't otherwise solve. Building a transistor requires altering the electrical properties of silicon at exact depths and concentrations. The older method, thermal diffusion, used heat. Heat spreads and cannot be aimed. As transistors shrank, approximate became inadequate.
Ion implantation replaced it in the 1970s: beams of charged atoms fired directly into the wafer, depth set by beam energy, dose set by exposure time. A modern chip may require up to sixty separate implant steps. Today, roughly 12,000 ion implanters – over 40% of the world’s accelerators – operate worldwide.
IV: An accelerator industry?
If particle accelerators are so important, why is there no particle accelerator industry?
The answer is in the history you've just read. But you have to look at the technology, not the timeline.
All accelerators do one thing: use electromagnetic fields to hurl charged particles to extraordinary speeds and shape them into beams. That is where the similarity ends. A 27-kilometre circular collider smashing protons together at near-light speed has almost nothing in common with a compact linear accelerator firing electrons at medical devices to sterilise them. Different particles, different energies, different geometries, different end uses.
So you can't draw a clean line around an accelerator industry the way you can with semiconductors. The technology is everywhere. The industry is not. It is fragmented, specialised, and in many areas it barely exists at all.
The companies that have scaled reflect this; Varian built its linacs into complete radiotherapy systems and was absorbed into Siemens Healthineers. IBA went deep on proton therapy. They picked one application and built around it. No one has yet built a business serving an accelerator market, because there isn't one.
Take space radiation testing. Every satellite launched must be radiation-hardened before it flies. Space is flooded with high-energy particles that damage and disrupt onboard electronics. The only way to know whether hardware will survive is to blast it with heavy-ion beams on the ground and find out. This is not optional. It is a prerequisite for flight.
So where does the satellite industry go for this? In the vast majority of cases, one of three facilities: Texas A&M University, Brookhaven National Laboratory, and the Lawrence Berkeley National Laboratory. They’re research labs, built for nuclear physics, not satellite qualification. For one of its most critical pre-flight requirements, an industry launching thousands of satellites a year depends on borrowed time at three university and government labs.
For decades this was fine. The demand was real but small. The research labs, however imperfectly, met the need. Nobody needed to make the capital case for a purpose-built commercial facility.
But as we enter the constellation age, that model is going to break. It’s time to start thinking about building commercial infrastructure.