A linear accelerator, commonly abbreviated linac, is a particle accelerator in which charged particles gain energy while traveling along a substantially straight trajectory. Most modern linacs use oscillating electric fields in a succession of resonant structures, although electrostatic and induction machines also provide linear acceleration. The accelerated particles may be electrons, protons, or heavier ions; positrons and other charged particles can also be accelerated when suitable sources are available.
Linacs range from compact machines installed in hospitals and factories to kilometer-scale research facilities. They generate therapeutic radiation, produce intense neutron and X-ray beams, inject particles into circular accelerators, and support experiments on matter at nuclear and subnuclear scales. Their defining engineering challenge is to keep particles synchronized with accelerating fields while preserving a tightly controlled beam. Particle velocity, accelerating gradient, beam current, and acceptable energy spread determine the machine’s length, structure, power system, and operating mode.
Physical principles and synchronization
The accelerating action of a linac follows from the Lorentz force. For a particle of electric charge , velocity , and total energy , the rate of energy transfer is
An electric field component parallel to the particle’s motion changes its energy. A static magnetic field can steer or focus the particle but does no work on it, because the magnetic force is perpendicular to its velocity. Along a straight reference trajectory, the energy gain is consequently determined by the longitudinal electric field encountered at the particle’s arrival time:
An electrostatic accelerator obtains this energy from a potential difference. A singly charged particle gains one electronvolt when accelerated through one volt. The maximum practical energy is limited by insulation, electrical breakdown, and the difficulty of maintaining very high terminal voltages. Radio-frequency linacs instead reuse oscillating fields in a sequence of gaps or cells, allowing their individual energy gains to accumulate without requiring the entire machine to sit at an enormous static potential.
Synchronization is essential because an oscillating field alternately accelerates and decelerates a particle of a given charge. The geometry and timing must ensure that each bunch reaches successive gaps during an accelerating part of the cycle. Writing particle speed as , where is the speed of light, a common standing-wave structure has successive accelerating cells separated by approximately
where is the radio-frequency wavelength. This relationship applies to particular phase arrangements, including many drift-tube and -mode structures; it is not a universal spacing rule for every linac.
The difference between electrons and heavy particles follows from special relativity. Their kinetic energy is , with . An electron with 5 MeV of kinetic energy already travels at about , whereas a 5 MeV proton travels at about . Electron structures can therefore use nearly constant cell dimensions after a short initial section. Proton and heavy-ion machines generally require geometries that change as the particles accelerate.
Real particles also spend a finite time crossing each gap. The field changes during that transit, reducing the effective accelerating voltage relative to an ideal instantaneous crossing. Designers describe this through a transit-time factor and select a synchronous phase that balances energy gain, bunch stability, and energy spread rather than maximizing voltage gain alone.
Historical development
Early accelerator development addressed the limitations of high-voltage equipment. Experiments in nuclear physics required controllable projectiles with energies beyond those conveniently available from radioactive sources or ordinary electrical discharges. In 1924, the Swedish physicist Gustaf Ising proposed using a sequence of timed electric pulses to accelerate particles through successive gaps. His proposal established the principle that repeated, synchronized energy increments could replace one exceptionally large potential difference.
Rolf Widerøe demonstrated radio-frequency linear acceleration in 1928. His apparatus used an oscillating voltage and a drift tube to accelerate positive potassium ions in successive gaps, producing energies of approximately 50 keV from an applied voltage of roughly 25 kV. Inside the conducting tube, particles were shielded while the field reversed. They emerged when the field orientation again favored acceleration. Although modest in energy, this experiment established a practical route to repeated acceleration and influenced subsequent accelerator designs.
The technology changed substantially during and after World War II. Radar development produced powerful microwave sources, improved waveguides, and techniques for manufacturing resonant cavities. These capabilities made electron linacs much more practical. At Stanford, William Hansen developed resonant-cavity concepts and worked with Russell and Sigurd Varian on the klystron, an important source of high-power microwave radiation. Electron linac development subsequently combined microwave engineering with increasingly precise control of beam injection and transport.
For protons, Luis Alvarez and colleagues at Berkeley developed a resonant drift-tube linac during the 1940s. Their 32 MeV proton machine began operation in 1947. Unlike the earlier Widerøe arrangement, the Alvarez design placed many drift tubes inside one resonant tank, making efficient use of electromagnetic modes at higher frequencies. Drift tubes also provided locations for focusing elements, an increasingly important requirement as beam intensity rose.
The two-mile Stanford linear accelerator, which began operating in 1966, demonstrated the scale achievable with normal-conducting microwave technology. It initially produced electron energies around 20 GeV and was later upgraded and adapted for different scientific programs. Work using its beams helped establish the internal structure of protons and neutrons. Other developments broadened the field: superconducting cavities enabled efficient high-duty operation; radio-frequency quadrupoles improved low-energy ion capture; and compact medical electron linacs became widely established. Accelerator history therefore reflects several parallel lineages rather than a single design becoming progressively larger.
Radio-frequency structures and accelerator types
The accelerating structure determines how radio-frequency power becomes a useful longitudinal field. In a drift-tube linac, conducting tubes shield particles during portions of the radio-frequency cycle that would otherwise decelerate them. Acceleration occurs mainly in the gaps between tubes. As nonrelativistic particles become faster, successive tubes generally become longer so that the transit time remains synchronized with the field. Alvarez-type structures are especially important for proton and ion acceleration at intermediate velocities.
At still lower energies, a radio-frequency quadrupole (RFQ) can combine acceleration, bunching, and transverse focusing. Four electrodes or vanes create an alternating quadrupole electric field. Longitudinal modulation of their surfaces introduces an accelerating component. An RFQ can accept an approximately continuous ion beam and progressively form bunches suitable for downstream radio-frequency structures. This integration is valuable because slow, intense ion beams are particularly susceptible to mutual electrical repulsion, while magnetic focusing becomes less effective as particle speed decreases.
Electron linacs commonly employ chains of microwave cavities operating in standing-wave or traveling-wave modes. In a standing-wave system, resonant fields occupy coupled cells with a prescribed phase relationship. In a traveling-wave structure, radio-frequency energy propagates along a loaded waveguide, transferring some of its energy to the beam before the remainder reaches a terminating load. Disks or other periodic features reduce the wave’s phase velocity so that an accelerating phase can remain synchronized with particles moving close to .
Structure dimensions scale broadly with wavelength, although detailed electromagnetic design introduces many additional constraints. Frequencies near 2.856 GHz, conventionally classified as S band, have been widely used in electron linacs. C-band and X-band structures operate at higher frequencies and can support compact, high-gradient designs, but their smaller apertures tighten alignment and beam-quality requirements. Lower frequencies are often advantageous for intense ion beams because they permit larger apertures and longer radio-frequency periods.
Not every linac uses resonant radio-frequency cavities. An induction accelerator generates voltage pulses through changing magnetic flux in induction cores. Successive accelerating cells add energy to an intense beam without requiring bunches to occupy a single microwave phase. Such machines are used particularly for high-current electron pulses and radiographic applications. Electrostatic devices, including tandem accelerators, also provide straight accelerating sections, although technical usage sometimes reserves “linac” primarily for repeated-gap radio-frequency machines.
These categories overlap in complete facilities. A proton accelerator may begin with an electrostatic extraction system, continue through an RFQ and drift-tube section, and finish with superconducting cavities. The resulting architecture follows the changing requirements of the particles rather than adherence to one structure throughout the machine.
Sources, power systems, and beamline equipment
A linac begins with a particle source matched to its intended beam. Electron guns commonly use heated cathodes, from which electrons escape by thermionic emission, or photocathodes illuminated by a laser. Photocathode guns offer precise timing and can produce short bunches with high brightness. Ion sources create a plasma and extract selected charged particles through electrodes. Some proton facilities initially accelerate negative hydrogen ions, allowing later removal of their electrons to simplify injection into a storage or accumulator ring.
The source beam normally passes through a low-energy transport system before entering the main accelerating structures. This section may contain focusing lenses, steering magnets, apertures, and devices that match the beam’s size and angular distribution to the accelerator’s acceptance. Bunchers impose an energy modulation: particles at different positions acquire slightly different velocities and subsequently converge into bunches. Choppers can remove selected time intervals to create a pulse pattern needed by downstream systems or experiments.
High-power radio-frequency systems supply the accelerating cavities. Klystrons convert the energy of an electron beam into microwave power through velocity modulation and bunching. Magnetrons are common in compact applications, including many medical machines, while solid-state amplifiers are increasingly used in suitable frequency and power ranges. Pulsed machines require modulators to deliver controlled high-voltage pulses. Waveguides, coaxial lines, couplers, circulators, and loads distribute power and manage reflected energy during cavity filling, changes in beam loading, or faults.
A low-level radio-frequency control system measures cavity amplitude and phase and corrects departures from their specified values. Feedback suppresses disturbances, while feedforward applies anticipated corrections for predictable effects such as the arrival of a beam pulse. Timing systems coordinate the source laser or gun, radio-frequency fields, pulsed magnets, diagnostics, and experimental equipment. In short-bunch facilities, timing errors that are negligible on an everyday scale can materially alter beam energy or arrival time.
The beamline operates under high or ultrahigh vacuum to limit collisions with residual gas. Such collisions can scatter particles, change ion charge states, or generate unwanted radiation. Pumps, vacuum gauges, valves, and clean assembly procedures are therefore integral accelerator components rather than auxiliary conveniences. Magnets provide focusing and steering; dipole magnets may also separate beams according to momentum.
Diagnostics include current transformers, Faraday cups, beam-position monitors, profile screens, and energy spectrometers. Some intercept the beam and are used chiefly during setup; others provide continuous measurements with little disturbance. Their calibration and interpretation matter because a displayed beam position, charge, or size is an instrument-dependent measurement, not a complete description of the particle distribution.
Beam dynamics, stability, and beam quality
A useful beam is not characterized by energy alone. Its current, pulse duration, transverse size, angular divergence, energy spread, and timing distribution determine whether it can perform the intended task. Accelerator physicists describe these properties in phase space, where particle positions are paired with momenta or related angular coordinates. The area occupied by a transverse distribution is quantified through beam emittance, with precise definitions depending on the statistical convention and coordinate system.
A small emittance allows a beam to be focused to a small spot without excessive divergence. For a relativistic beam, normalized emittance is commonly written
where is geometric emittance. In ideal linear, conservative transport, normalized emittance remains approximately constant as particles accelerate. Actual machines experience emittance growth through nonlinear forces, mismatched focusing, scattering, coupling between directions, and collective interactions. Beam quality must therefore be protected from the source onward; increasing energy does not automatically repair a degraded distribution.
Longitudinal dynamics concern bunch position relative to the radio-frequency phase and deviations from the reference energy. For suitable synchronous phases, early and late particles receive different energy increments that help keep them grouped around a stable trajectory. The details depend on particle velocity and accelerating structure. A phase choice that improves longitudinal capture may reduce average energy gain, and radio-frequency defocusing may have to be counteracted by external lenses. These trade-offs are especially important in low-energy ion linacs.
Mutual repulsion, known as space charge, strongly affects slow or intense beams. It can expand the beam, distort its distribution, and create a low-density halo around the core. At high relativistic energies, other collective effects often become dominant. A passing bunch excites electromagnetic fields in cavities and beam pipes; these wakefields act on later particles or bunches. Longitudinal wakes change energy distribution, while transverse wakes can amplify offsets and cause beam breakup. Cavity shape, higher-order-mode damping, alignment, and bunch spacing influence their severity.
Beam loading is another collective effect: the beam extracts energy from the accelerating field and changes its amplitude and phase. In high-current machines, this interaction can be a major part of the power balance. Compensation may involve changing the incident radio-frequency waveform, adjusting coupling, or controlling the temporal current profile.
Very short electron bunches are often produced with magnetic bunch compressors. An imposed correlation between energy and longitudinal position allows a sequence of bending magnets to shorten the bunch. Compression can increase peak current dramatically, but it can also amplify small density variations or generate coherent radiation that degrades emittance. The operating goal is consequently not the shortest possible bunch in isolation, but the distribution best suited to the experiment.
Superconducting linacs and energy recovery
A superconducting radio-frequency cavity uses a material with very low surface resistance at cryogenic temperature. Niobium is the dominant material in established large installations. Although superconductivity eliminates direct-current resistance under appropriate conditions, microwave cavities retain a finite surface resistance and dissipate some power. Their losses can nevertheless be far below those of room-temperature copper cavities, making them attractive for long pulses or continuous-wave operation.
Superconducting cavities are assembled into cryomodules containing thermal insulation, magnetic shielding, mechanical supports, tuners, and connections to a refrigeration system. Bulk-niobium systems commonly operate with helium near 2 K or 4.5 K, depending on cavity frequency and design. Producing refrigeration at these temperatures requires substantial electrical power. The benefit therefore depends on total system performance, including cavity losses, static heat leaks, radio-frequency efficiency, beam loading, and operating duty factor.
Cavity quality is often described by the intrinsic quality factor,
where is angular frequency, is stored electromagnetic energy, and is power dissipated at the cavity surface. A high reduces the refrigeration required for a given stored energy. Surface preparation is crucial: chemical treatment, electropolishing, high-pressure rinsing, cleanroom assembly, and controlled heat treatments can influence field emission, residual losses, and the maximum usable field.
Mechanical effects also matter. External vibrations, collectively called microphonics, shift the resonant frequency. Electromagnetic pressure deforms cavity walls and produces Lorentz-force detuning. Because superconducting cavities can have narrow operating bandwidths, even small shifts may demand significant corrective radio-frequency power. Mechanical stiffening, slow tuners, piezoelectric actuators, and active controls help maintain resonance. Excessive local heating can trigger a quench, in which part of the surface loses superconductivity and the accelerating field collapses.
Superconducting technology also supports energy-recovery linacs. After performing useful work, such as generating light in an insertion device, an electron beam is returned to the accelerating cavities approximately half a radio-frequency period out of the accelerating phase. It is then decelerated, transferring much of its kinetic energy back to the electromagnetic field. Fresh bunches can use that recovered energy, reducing the external power otherwise required to sustain a high-energy, high-current beam.
Energy recovery does not reclaim all facility power. Cryogenic losses, source power, radiation emitted by the beam, and the residual energy delivered to the dump remain. Recirculation also introduces demanding beam-optics and stability problems, including interactions between returning bunches and higher-order cavity modes. Nevertheless, it provides a route to beam powers that would be costly to produce and dispose of in a simple single-pass machine.
Scientific facilities and research applications
Linacs serve both as complete scientific instruments and as the first stage of larger accelerator complexes. Their beams probe atomic nuclei, elementary particles, materials, and chemical processes. As injectors, they establish the energy, emittance, intensity, and timing needed by downstream synchrotrons. At CERN, for example, Linac4 accelerates negative hydrogen ions to 160 MeV before injection into the Proton Synchrotron Booster, where their electrons are removed. It became part of the operational injector chain in 2020.
High-energy electron linacs have played an important role in particle physics. In the late 1960s, deep-inelastic scattering experiments at Stanford revealed pointlike constituents inside the proton, contributing decisively to the development of the quark model. Electron beams were especially useful because their electromagnetic interaction with the target could be described with high precision. The resulting measurements connected accelerator engineering directly to the emerging understanding of hadronic structure.
A linear collider accelerates opposing beams in separate linacs and brings them into collision. Unlike particles in a storage ring, the beams generally encounter the interaction region only once before disposal. This avoids the severe bending-radiation penalty encountered by very high-energy electrons in circular machines, but requires extremely small beam spots, stable alignment, and efficient production of fresh bunches. The Stanford Linear Collider, which began physics operation in 1989, demonstrated electron–positron collisions using a linear accelerator and separate transport arcs.
In a free-electron laser, a high-brightness electron beam passes through an undulator, a periodic magnetic structure that makes the particles oscillate transversely. Interaction between the electrons and the emitted radiation can organize the bunch into density modulations at the radiation wavelength, producing intense, coherent light. The Linac Coherent Light Source at SLAC began producing hard X-ray laser pulses in 2009. Such pulses support studies of atomic motion, electronic dynamics, and samples that change or are damaged during exposure.
High-power proton linacs also drive spallation neutron sources. Protons striking a heavy target release neutrons through nuclear cascades and subsequent de-excitation. At the Spallation Neutron Source in Oak Ridge, a linac accelerates negative hydrogen ions to about 1 GeV before an accumulator ring forms short proton pulses for the target. Neutrons then investigate magnetic order, molecular motions, crystal structures, and engineering materials.
Other research uses include isotope production, nuclear astrophysics, and irradiation studies relevant to reactor materials. Linacs can also generate radiation through bremsstrahlung or inverse Compton scattering. These facilities prioritize different beam properties: a light source may require exceptionally low emittance, whereas a neutron facility may emphasize average beam power, reliability, and stringent control of particle losses.
Medical and industrial applications
Medical electron linacs are a principal source of external-beam radiotherapy for cancer. They accelerate electrons to energies typically ranging from several MeV to a few tens of MeV. The electrons may be used directly or directed onto a high-atomic-number target to produce bremsstrahlung X-rays. Clinical photon beams are conventionally labeled in megavolts, but they contain a broad photon-energy spectrum rather than photons all possessing one energy equal to the nominal machine setting.
A treatment head shapes and monitors the radiation. Its components may include a target, primary collimator, monitor ionization chambers, movable jaws, and a multileaf collimator whose individually positioned leaves define an irregular aperture. Many systems rotate around the patient on a gantry. Computer-controlled motion and beam delivery support techniques such as intensity-modulated radiotherapy and volumetric modulated arc therapy. Integrated imaging systems provide information about patient position and internal anatomy relative to the planned treatment geometry.
Electron and photon beams deposit energy differently. Therapeutic electrons have a finite useful penetration depth and are employed for relatively superficial targets. Megavoltage photons penetrate more deeply and build up their dose beneath the surface before attenuation dominates. Clinical dose calculations account for beam energy, field shape, tissue composition, scattering, and geometry. At sufficiently high photon energies, photonuclear reactions in accelerator components can also generate neutrons, affecting shielding and radiation-protection requirements.
Radiation delivery is quantified by absorbed dose, measured in grays, where . Machine output is controlled using monitor units, whose relationship to dose depends on calibration and irradiation conditions. Independent checks, dosimetry, mechanical tests, interlocks, and periodic quality assurance are integral to clinical operation. The accelerator is only one component of a treatment system that also includes imaging, planning software, immobilization equipment, and records of delivered irradiation.
Industrial electron linacs exploit radiation-induced chemical changes. Electron irradiation can cross-link polymers, modify insulation, cure coatings, and sterilize medical products. Direct electron processing is efficient for suitable product thicknesses because energy is deposited without first converting the beam to photons. Conversion to X-rays sacrifices efficiency but provides greater penetration for denser or more bulky loads. Product density, conveyor speed, beam scanning, and dose uniformity determine a processing installation’s design.
High-energy X-ray linacs also support nondestructive testing and inspection of large structures or cargo. Pulsed electron machines can produce radiographs of rapidly changing systems, while lower-energy specialized accelerators serve laboratory and manufacturing applications. In all these settings, the relevant specification is not simply maximum energy: throughput, radiation-field uniformity, uptime, shielding, and reproducible output can be equally important.
Performance limits, radiation protection, and emerging methods
The achievable accelerating gradient is constrained by material and electromagnetic effects. In normal-conducting structures, strong surface fields can initiate electrical breakdown, while radio-frequency currents produce heating and thermal stress. In superconducting cavities, local defects, field emission, excess surface losses, and quenching limit performance. A gradient quoted for an individual cavity or short test structure should not be confused with the average energy gain per meter of an entire facility, which includes spaces for focusing, diagnostics, couplers, and other equipment.
Beam power sets another practical limit. For particles gaining energy through an effective accelerating voltage , average beam power is
Thus, a 1 GeV proton beam with an average current of 1 mA carries 1 MW. Even a small fractional loss from such a beam can damage equipment or produce substantial activation. High-power machines therefore use conservative apertures, halo control, loss monitors, rapid machine-protection systems, and carefully engineered targets and beam dumps.
Radiation fields arise both from intentional beam interactions and from unwanted losses. Electrons generate bremsstrahlung in matter, while energetic hadrons produce complex secondary-particle cascades. Targets and accelerator components may become radioactive, requiring controlled access and, in some facilities, remote maintenance. Thick concrete shielding, localized metal shielding, interlocked enclosures, and monitored operating procedures address distinct parts of this problem. Electrical high voltage, stored radio-frequency energy, powerful lasers, and cryogenic systems introduce additional hazards independent of the beam itself.
An important alternative is plasma wakefield acceleration. An intense laser pulse or charged-particle bunch drives collective motion in a plasma, creating longitudinal electric fields that can reach tens of gigavolts per meter under suitable conditions. Because the medium is already ionized, the familiar breakdown limit of a metal cavity does not apply in the same way. Experiments have demonstrated substantial energy gains over short distances, motivating research on compact accelerators and radiation sources.
High gradient alone, however, does not establish a complete replacement for conventional linacs. Useful facilities also require reproducible injection, small energy spread, controlled emittance, high efficiency, acceptable repetition rate, and reliable transport between stages. Positron acceleration poses additional difficulties for many plasma configurations. Dielectric structures driven by microwaves or lasers provide another research direction, with their own constraints on aperture, phase acceptance, materials, and average power.
Conventional technology is also evolving through improved cavity treatments, new superconducting materials, more efficient radio-frequency sources, and advanced controls. Each development addresses a different part of the facility’s performance budget. For a hospital machine, compactness and reproducibility may dominate; for a collider, beam brightness and wall-plug efficiency may be decisive; for a high-power proton source, uncontrolled loss and maintainability can determine the usable operating point.