Particle Accelerator Magnets — Deep Reference

Particle accelerators steer, focus, and analyse charged-particle beams with magnet systems that range from a 2 kg permanent-magnet quadrupole on a benchtop electron-microscope column to the 28-km ring of 1232 main dipole magnets in the LHC at CERN, each NbTi-superconducting and operating at 1.9 K in superfluid helium. This note covers the magnet families — dipole (bending), quadrupole (focusing), sextupole (chromaticity), octupole (Landau damping), combined-function, solenoid, wiggler / undulator — across the four production categories (resistive, superconducting LTS, superconducting HTS, permanent-magnet), and traces the engineering of the major machines from the Tevatron through the LHC to the High-Luminosity LHC, FCC-ee/FCC-hh, the muon-collider proposal, the medical proton-therapy gantries (Varian / IBA / Mevion), and the synchrotron / FEL light-source facilities (LCLS-II, XFEL, ESRF-EBS, Diamond-II, NSLS-II, APS-U, MAX IV). The cryogenic infrastructure that supports superconducting machines — superfluid helium-II at 1.9 K, magnet quench protection, energy extraction — is treated alongside the magnets themselves.

See also

• electromagnetics-engineering
• mri-magnets-and-coils-deep
• electric-motors
• power-electronics
• photonics
• quantum-materials-and-topological-phases
• magnetic-and-optical-materials

0. Historical timeline

• 1929 — Lawrence (UC Berkeley) invents the cyclotron, using a single large electromagnet and an RF dee.
• 1944 — Veksler / McMillan independently discover phase stability, enabling the synchrotron.
• 1952 — Courant, Livingston, and Snyder propose strong focusing (alternating-gradient quadrupole pairs); allows ring magnets of practical aperture.
• 1954 — Cosmotron at BNL — first GeV-scale proton synchrotron.
• 1959 — CERN Proton Synchrotron and BNL AGS reach 30 GeV using strong focusing.
• 1971 — ISR (Intersecting Storage Rings, CERN) — first proton-proton collider.
• 1972 — Kunzler’s NbTi conductor enters magnet R&D programs.
• 1983 — Tevatron commissioning, world’s first superconducting hadron synchrotron.
• 1989 — LEP-I starts e+e- physics at $\sqrt{s}=91\mathrm{GeV}$ (warm-magnet ring).
• 1992 — HERA superconducting magnet ring commissioned.
• 2000 — RHIC starts heavy-ion collisions.
• 2008 — LHC first beams. Sector-3/4 splice failure delays physics commissioning by a year.
• 2010 — LHC reaches 3.5 TeV per beam.
• 2012 — Higgs boson discovery at $\sqrt{s}=8\mathrm{TeV}$.
• 2015 — Run 2 starts at 6.5 TeV per beam.
• 2017 — Mevion HYPERSCAN — first compact superconducting proton-therapy gantry FDA-cleared.
• 2018 — European XFEL operational.
• 2024 — LCLS-II superconducting upgrade operational; HEPS (China 4th-gen synchrotron) first beam.
• 2026–2029 — HL-LHC Nb${}_3$Sn inner triplet installation.

1. Magnet families and multipole expansion

The transverse magnetic field at radius $r$ inside a 2D accelerator magnet expands as:

$B_y+i{B}_x={\sum }_{n=0}^{\infty }(b_n+i{a}_n){\left(\frac{x+iy}{R_{\mathrm{ref}}}\right)}^n$

with $b_n$ “normal” and $a_n$ “skew” multipole components and $R_{\mathrm{ref}}$ a reference radius (1 cm for LHC). $n=0$ is the dipole, $n=1$ quadrupole, $n=2$ sextupole, $n=3$ octupole, $n=4$ decapole, etc.

1.1 Function in a synchrotron

• Dipole ($B_y$ constant) — bends the beam along the circular orbit. The integral $\int Bdl$ around the ring equals $2\pi p/q$ — the rigidity-times-bending-radius required for a given momentum.
• Quadrupole ($B_y=Gx$, $B_x=Gy$) — focuses in one transverse plane, defocuses in the other. Alternating F-D quadrupole pairs give net focusing (the FODO lattice, the basis of every modern synchrotron).
• Sextupole ($B_y=S(x^2-y^2)$) — corrects chromaticity (the variation of focusing with particle momentum), allowing higher-momentum-spread beams to be stored.
• Octupole — provides Landau damping by introducing amplitude-dependent tune shift, stabilising coherent instabilities.
• Combined-function magnet — dipole + quadrupole + sometimes sextupole in a single iron core. Cheaper and more compact than separate magnets; standard on older machines (CERN PS, FNAL Booster, Tevatron arcs) and re-emerging for ultra-compact light sources.

1.2 Conductor choices

Type	Conductor	$B$ peak	Use
Resistive (copper / aluminum)	water-cooled hollow Cu	$\le$ 2 T	low-energy, fast-cycling, all dipoles in proton therapy
Permanent (NdFeB / SmCo)	n/a	$\le$ 1.3 T equivalent	undulators, beamline correctors, compact ion-beam
LTS NbTi	Cu-stabilised NbTi multifilament	$\le$ 8.5 T (4.2 K), 9.5 T (1.9 K)	LHC main, HERA, Tevatron, RHIC, MRI
LTS Nb${}_3$Sn	bronze-route / internal-Sn / RRP	$\le$ 16 T	HL-LHC, FCC-hh main, MRI 7 T+
LTS Nb${}_3$Al	conductor-by-reaction	$\le$ 12 T	research, strain-sensitive replacement
HTS REBCO (YBCO coated conductor)	flat tape, 4 mm wide	$\le$ 25+ T at 4.2 K, 10+ T at 50 K	research; future FCC-hh, fusion, muon collider
HTS Bi-2212	round wire	$\le$ 20+ T at 4.2 K	research
HTS Bi-2223	flat tape	$\le$ 5 T at 77 K	power transmission, fault current limiters

2. The LHC magnet system

2.1 Main dipole

The LHC has 1232 main dipole magnets, each 14.3 m long, $\sim$28 tonnes, producing 8.33 T at the 7 TeV design energy. The conductor is NbTi multifilament Rutherford cable: 6500 NbTi filaments (~6 µm each) embedded in a Cu matrix, wound as 28- and 36-strand Rutherford cables, two layers per aperture, two apertures per magnet (the LHC dipoles are “twin-aperture” — two beams sharing one cryostat). Operating temperature 1.9 K in pressurised superfluid helium-II — the lowest operating temperature of any production engineering system at this scale.

Key dimensions and parameters:

• Coil inner radius: 28 mm.
• Operating current: 11 850 A.
• Stored magnetic energy per dipole: 7 MJ.
• Total stored magnetic energy in the LHC ring: $\sim$11 GJ — equivalent to a fully-loaded 747 at landing speed.
• Field quality $\Delta B/B$ at 17 mm reference radius: $<$ 1 ppm (10 units = $10^{-4}$).

The dipoles were built by three European suppliers — Babcock-Noell (Germany), Alstom-MSA (France), and ASG Superconductors (Italy) — to a joint CERN design. Production ran 2000–2006.

2.2 Other LHC magnets

• Main quadrupoles: 392 superconducting NbTi quads, 3.15 m long, gradient 223 T/m at 7 TeV. Built by CEA Saclay + CERN.
• Inner triplets at IP1/IP2/IP5/IP8 (low-beta focusing at the four interaction points): 8 m long NbTi quadrupoles, gradient 205 T/m. Currently being replaced by HL-LHC Nb${}_3$Sn triplets.
• Sextupoles, octupoles, decapoles, correctors: $\sim$8000 small superconducting magnets per ring for orbit, chromaticity, and high-order correction.
• Warm magnets (resistive): injection, extraction, dump-line magnets in the LSS (long-straight sections) where the beam is not in the superconducting arc.

2.3 Cryogenic infrastructure

The LHC cryogenic plant is the largest helium refrigerator in the world: 130 tonnes of liquid helium at 1.9 K, plus 60 tonnes at 4.5 K for HV current leads and intermediate-stage cooling. Eight refrigerator units (Air Liquide / Linde) provide $\sim$1.8 kW of cooling at 1.9 K each, plus the higher-temperature stages.

Why 1.9 K (superfluid He-II)?

• $T_{\mathrm{c}}$ of NbTi is 9.2 K at zero field, falls to 4.5 K at 8.3 T. To reach 8.33 T with safe margin, operating temperature must be below 4.2 K.
• He-II below the lambda point ($T_{\lambda }=2.17$ K) has effectively infinite thermal conductivity for moderate heat fluxes — the cold mass is bathed in a continuous heat-sink.
• The $J_c$ of NbTi nearly doubles between 4.5 K and 1.9 K, allowing the LHC’s 8.3 T field; at 4.5 K NbTi would saturate at $\sim$5–6 T.

2.4 Quench protection

A 1 ms hot-spot in the LHC dipole conductor (driven by a tiny mechanical disturbance or proton beam loss into the coil) can propagate at $\sim$30 m/s, vaporising the helium and over-heating the conductor. Protection scheme:

• Detection — voltage tap pairs along each coil section, comparing inductive against resistive voltage drop. A resistive voltage component $>$ 100 mV indicates quench. Detection time $<$ 10 ms.
• Quench heaters — strip heaters in the coil collar fire at $\sim$300 V / 75 A, spreading the normal zone to multiple coil sections. Each heater pulse delivers $\sim$10 kJ.
• Energy extraction — once detected, the magnet circuit is opened and the stored energy is dumped into external resistors. The 8 main dipole sectors (154 magnets each) have a 0.04 Ω dump resistor; the magnet current decays with $\tau =L/R\approx 100$ s.
• Internal current sharing — for HL-LHC Nb${}_3$Sn, the CLIQ (Coupling-Loss-Induced Quench) system inductively fires the entire coil into normal state in $\sim$10 ms.

A quench costs CERN $\sim$1–2 days of downtime per magnet and $\sim$0.5–1 % of the helium inventory of the affected sector. Major incidents (the 2008 sector 3-4 splice failure that delayed LHC startup by a year, with 6 GJ of stored energy released catastrophically into the cryostat) drive continual upgrades to the protection scheme.

2.5 LHC cost summary

LHC construction 1992–2008: $\sim$$5B (CHF 4.3B nominal, including injectors and detectors). Magnet system alone: $\sim$$2B. Annual operating cost: $\sim$$1B including labour, electricity ($\sim$120 MW peak when operating, equivalent to a small city), helium, and detector maintenance.

2.6 Field quality and “good field region”

Accelerator magnet field quality is specified as $\Delta B/B_0$ at a reference radius (typically 17 mm for LHC main dipole) in “units” of $10^{-4}$ relative. Specification of the LHC main dipole:

• $b_3$ (sextupole) at injection: $-3.0\pm 0.5$ units (correction by sextupole spool-piece on the magnet).
• $b_5$ (decapole): $0.5\pm 0.2$ units.
• Random component: $\sigma <0.3$ units per multipole.

Each magnet is magnetically measured at the manufacturer (Bruker BioSpin / Babcock-Noell test stations) using a rotating-coil harmonic analyser, then again at CERN before installation. Magnets that fall outside spec are paired with compensating “spool-piece” correctors so the cumulative field quality across the ring is uniform.

The “good field region” — the cross-sectional area within which the field meets the multipole spec — is typically 70 % of the geometric aperture (28 mm radius for LHC main dipoles → ~20 mm good-field radius).

3. High-Luminosity LHC (HL-LHC)

HL-LHC is the LHC upgrade that pushes peak luminosity from $10^{34}$ to $5\times 10^{34}{\mathrm{cm}}^{-2}{\mathrm{s}}^{-1}$ via stronger focusing at IP1 (ATLAS) and IP5 (CMS). The headline component:

• Nb${}_3$Sn inner triplet quadrupoles — 11–12 T peak field, 150 T/m gradient, 70 mm aperture. Replaces the LHC’s NbTi inner triplets. The first production magnet, MQXFA in the US-LARP program, achieved spec in 2017; serial production at Fermilab / BNL / LBNL for the four IPs is ongoing, scheduled install 2027–2029.
• 11 T Nb${}_3$Sn dipoles — replace MB.B11 and B11/B13 dipoles to gain straight-section space for cryogenic collimators. Built at CERN with industry (ASG, Bilfinger).
• Crab cavities (not magnets but worth noting) — 800 MHz superconducting RF cavities tilting the bunch before collision.

HL-LHC cost: $\sim$$1.4B (CHF 1.2B). Beam commissioning planned 2029.

3.1 LARP and MQXFA development

The HL-LHC inner-triplet quadrupole MQXFA is the product of a 15-year R&D program. The LHC Accelerator Research Program (LARP) at FNAL, BNL, and LBNL (2004–2018) developed the Nb${}_3$Sn cable, coil winding, and pre-stress techniques. Five 1.5 m short-model magnets (MQXFS series) reached 220 T/m gradient and 12 T peak field by 2017. Full-scale 4.5 m MQXFA pre-production magnets reached spec by 2019. Production of 20 MQXFA units (10 for installation + 10 spares) by FNAL + BNL is underway, with CERN producing the MQXFB longer cousins. Total HL-LHC inner-triplet cost: $\sim$$400M for the magnet system alone.

4. Other major operational and historical machines

4.1 Tevatron (Fermilab, decommissioned 2011)

The Tevatron was the world’s first superconducting hadron collider (operational 1983–2011). 6.3 km ring, 4.2 K bath-cooled NbTi dipoles at 4.2 T. 774 dipoles, 200 quadrupoles. Discovered the top quark (1995). Decommissioned to make way for the Main Injector neutrino program and FNAL’s intensity-frontier focus.

4.2 HERA (DESY Hamburg, decommissioned 2007)

HERA collided electrons / positrons on protons (920 GeV protons + 27.5 GeV electrons). 6.3 km ring with 416 superconducting NbTi proton-ring dipoles at 4.7 T and a separate warm electron-ring magnet system.

4.3 RHIC (Brookhaven, operational since 2000)

RHIC collides heavy ions (Au-Au, Cu-Cu, p-p) at top energy 100 GeV/u Au or 250 GeV polarised p. 3.8 km ring with $\sim$1700 superconducting NbTi magnets (dipoles 3.5 T, quadrupoles, sextupoles, correctors). Two independent rings sharing a cryostat. RHIC will retire $\sim$2025 to make way for the Electron-Ion Collider (EIC) at the same site, with new SC magnets in the existing tunnel.

4.4 FAIR (GSI Darmstadt)

FAIR is the heavy-ion accelerator complex under construction at GSI (CRYRING since 2014, SIS100 main ring expected 2027). SIS100 is a 1.1 km synchrotron with 108 superconducting (NbTi, fast-cycling at 4 T/s ramp-rate) dipoles. Fast-cycling magnets demand a different conductor design: low-AC-loss cable with thin copper matrix, full magnetic-field-quality data at all ramp rates.

4.5 J-PARC (Tokai, Japan)

J-PARC’s Main Ring (50 GeV proton synchrotron, beam to T2K neutrino experiment) uses warm-iron magnets; the FX (fast extraction) magnets cycle at 0.5 Hz to feed neutrino target. The J-PARC Linac uses superconducting Nb${}_3$Sn solenoids in the final ACS section.

4.6 Electron-Ion Collider (EIC)

The EIC at Brookhaven (under construction since 2020, planned operation 2032) reuses the RHIC tunnel and many existing RHIC superconducting magnets, but adds new high-field interaction-region quadrupoles based on Nb${}_3$Sn or HTS. Cost: ~$1.7–2.8B. The EIC will collide polarised electrons (5–18 GeV) on polarised protons (41–275 GeV) or heavy ions, for QCD spin and gluon-saturation physics.

5. Future collider proposals

5.1 FCC-ee (electron-positron) and FCC-hh (hadron)

The Future Circular Collider at CERN is the proposed successor to the LHC, with a 91 km ring tunnel running primarily under the Léman Basin. Two operational scenarios:

• FCC-ee — Higgs / Z / W / top factory, $\sqrt{s}$ = 91 / 161 / 240 / 365 GeV. Operates with low-field (0.05 T main dipoles) warm magnets. Cost $15–20B; construction 2030s, operation 2045–2065.
• FCC-hh — proton collider at $\sqrt{s}$ = 100 TeV. Requires 16 T dipoles — pushing Nb${}_3$Sn close to its theoretical limit, or REBCO HTS for some sections. Cost $30–40B; operation 2070+.

The 16 T Nb${}_3$Sn R&D program (EuroCirCol, US-MDP) has produced 14.5 T magnet demonstrators by 2024 (FRESCA2 at CERN, MDPCT1 at FNAL); production of $>$ 5000 16 T dipoles for FCC-hh is the open engineering challenge of the next 20 years.

5.2 Muon collider

A circular 10–14 TeV muon collider has been proposed since 2010 as a higher-energy reach than FCC-hh with smaller footprint (3–10 km ring vs 91 km). Magnet challenge: very-high-field (15–20 T HTS) solenoids in the muon cooling channel and dipoles + quadrupoles in the collider ring tolerant to the heat load from muon decay. Active R&D at CERN, BNL, FNAL since 2022 under the International Muon Collider Collaboration (IMCC).

5.3 CEPC / SPPC (China)

Equivalent Chinese proposal: 100 km ring for CEPC (e+e- Higgs factory) → SPPC (75–125 TeV hadron). Same Nb${}_3$Sn / HTS magnet requirements as FCC-hh. Construction proposal for 2030s.

5.4 ILC and CLIC (linear colliders)

Linear electron-positron colliders (proposed Japan ILC and CERN CLIC) avoid synchrotron-radiation losses but require very long machines (30–50 km). Superconducting RF cavities dominate; magnets are warm or modest LTS (focusing quads in the linac, beam-dump-line magnets). Status: ILC site selection in Japan pending; CLIC included in CERN strategic roadmap.

5.5 ProtoDUNE / DUNE near-detector magnetic spectrometer

The DUNE long-baseline neutrino experiment (LBNF/DUNE at Fermilab) includes a near-detector magnetic spectrometer with a large-aperture warm dipole for charged-pion / kaon momentum analysis. Construction at FNAL, with operations from 2031.

6. Medical proton-therapy gantries

A clinical proton-therapy gantry is a 100–200 tonne rotating structure that aims a 70–250 MeV proton beam at the patient from any angle. The bending magnets in the gantry are the dominant mass.

6.1 Resistive gantries (most installed systems)

• Varian ProBeam — 360° rotating gantry, resistive water-cooled iron-yoke dipoles. Beam energy continuously variable 70–250 MeV via accelerator energy selection (a synchrocyclotron upstream).
• IBA Proteus PLUS / Proteus ONE — 360° (PLUS) or 220° (ONE) gantries, resistive. ONE uses a compact “ion-beam-applied-from-above” geometry with a single half-arc, reducing footprint $\sim$50 %.
• Hitachi Probeat — resistive, isocentric.
• Sumitomo SHI proton therapy — resistive.
• Mevion HYPERSCAN S250i — single-room compact system; the bending magnet is superconducting (Nb${}_3$Sn) on a small synchrocyclotron mounted directly on the gantry, eliminating the long beam line. Smallest footprint of any proton-therapy room (~6 m × 6 m).

6.2 Superconducting gantries

• Varian ProBeam Superconducting Gantry — the first commercial superconducting rotating gantry; NbTi dipoles cooled by a closed-cycle cryocooler. Reduces gantry mass from ~200 tonnes to ~60 tonnes. First clinical install Mayo Clinic 2023.
• Heidelberg HIT gantry — first carbon-ion gantry (2012), resistive but at the limit of practical iron-yoke design (carbon ions need 4× the bending-rigidity of protons, magnet weighs $\sim$600 tonnes — the largest medical magnet ever built).
• Maya-1 (proposal) — fully ironless superconducting gantry; conceptual at CERN and Maya Medical.

6.3 Compact systems

• Mevion S250i with HYPERSCAN — gantry-mounted synchrocyclotron, FDA 2018.
• ProTom Radiance 330 — synchrotron-based, smallest room footprint.
• PMB IBA Proteus ONE — single-room offering.

Proton-therapy capital cost: $30–60M per single-room install, up to $150M for a 3-room multi-room centre.

6.4 Gantry rotation engineering

A proton-therapy gantry rotates 200 tonnes of structure around a horizontal axis, with the patient on a couch at the centre. Engineering constraints:

• Isocentre stability: $\pm$0.5 mm at any rotation angle. Achieved with custom-machined slewing bearings (Rothe-Erde, IMO-Antriebstechnik), $\sim$3 m diameter.
• Cable / hose management: cooling water, power, beam pipe, and instrumentation cables route around a counter-rotating energy chain.
• Beam transport from accelerator to gantry: typically a fixed beam line through a $\sim$135° bending section into the rotating section.
• Mass balancing: counterweights or active CG-shift mechanisms keep the rotational moment constant against gantry angle.

The Heidelberg HIT carbon-ion gantry (2012) at 600 tonnes is the largest medical magnet structure ever built. The first 90° rotation took 90 seconds at commissioning; modern systems rotate at 1 rpm.

7. Synchrotron light sources

A synchrotron storage ring produces X-rays from relativistic electrons forced through bending magnets, wigglers, or undulators. The 4th-generation diffraction-limited storage rings (DLSR) have replaced classical 3rd-generation machines worldwide:

Facility	Location	Energy	Ring	Status
MAX IV	Lund, Sweden	3 GeV	528 m	operational since 2016, first DLSR
Sirius	Campinas, Brazil	3 GeV	518 m	operational since 2020
ESRF-EBS	Grenoble, France	6 GeV	844 m	upgraded 2020
APS-U	Argonne, USA	6 GeV	1104 m	upgrade completing 2024
HEPS	Beijing, China	6 GeV	1361 m	operational since 2024
Diamond-II	Didcot, UK	3.5 GeV	562 m	upgrade in progress, $\sim$2028
ALS-U	Berkeley, USA	2 GeV	196 m	upgrade in progress
NSLS-II	Brookhaven, USA	3 GeV	792 m	operational since 2014, pre-DLSR
SPring-8-II	Hyogo, Japan	6 GeV	1436 m	upgrade in progress

DLSR magnet lattices are based on multi-bend achromats (MBA): each cell has 5–7 small dipoles separated by quadrupoles + sextupoles, producing a small horizontal emittance ($\sim$100 pm·rad, an order of magnitude below 3rd-gen). The magnets are typically combined-function permanent + electromagnet hybrids, or compact resistive units assembled on a single “girder” with sub-µm alignment.

7.1 Wigglers and undulators

A wiggler is a periodic permanent-magnet array that forces the electron beam into a sinusoidal trajectory, producing broadband synchrotron radiation. An undulator is the same physical structure but tuned (period + field strength) so that radiation from adjacent peaks interferes constructively, producing narrow-band coherent X-rays at the fundamental and harmonics. The K-parameter $K=0.934\cdot {\lambda }_u[\mathrm{cm}]\cdot B[\mathrm{T}]$ distinguishes wiggler ($K\gg 1$) from undulator ($K\lesssim 1$).

Undulator technologies:

• PPM (pure permanent magnet) — NdFeB blocks alternating polarity. Period 15–50 mm, max field 1.0 T at minimum gap.
• Hybrid — NdFeB + vanadium-permendur iron poles; concentrates flux. Used at most 3rd-gen rings.
• In-vacuum — entire undulator inside the storage-ring vacuum, allowing minimum gap (5–7 mm vs 20 mm with external vacuum chamber). 1.5–2× field gain. Standard at modern light sources.
• Cryo-cooled (CPMU) — cool NdFeB to 80–150 K; remanence rises 5–10 %. Operational at ESRF, Diamond.
• Superconducting undulators (SCU) — NbTi at 4 K. Peak field 1.5–2 T at small period (15 mm). Operational at APS-U, in research at PSI / KIT.
• REBCO HTS undulators — at 20–40 K, would give 2.5+ T at small period. Pre-commercial.

Vendors of permanent-magnet undulators: Hitachi Metals Engineering, Kyma, Bruker EAS, Adelphi Technology. Cryomodule and SCU integration: Argonne ASD, KIT, Hitachi Cryogenics.

7.2 Multi-bend achromat lattice engineering

MAX IV’s 7-bend achromat (7BA) was the pioneering 4th-gen lattice. The ESRF-EBS hybrid 7BA with high-gradient quadrupoles squeezes the natural emittance from 4 nm·rad (3rd-gen ESRF) to 130 pm·rad. The magnets are extremely compact:

• Combined-function dipole-quadrupole magnet — 0.5 m long, 0.4 T bend with embedded gradient. Built by SigmaPhi (France), Tesla Engineering (UK), Hitachi Industrial Equipment.
• Compact quadrupole — 120 T/m gradient at 30 mm bore radius, only 200 mm magnetic length.
• Octupole — 7000 T/m${}^3$ for Landau damping.

The magnets sit on common steel girders aligned to $\pm$30 µm tolerance over 5 m girder length. The girder + magnet assembly is the unit of replacement — at MAX IV and ESRF-EBS the upgrade campaigns replaced the entire ring tunnel of girders in $\sim$18 months of dark time.

8. Free-electron lasers (FELs)

FELs use a high-quality electron beam from a linac driving a long undulator to produce coherent X-rays via SASE (self-amplified spontaneous emission) or seeded amplification:

• European XFEL (Hamburg) — 17.5 GeV superconducting linac, 1.5 km, $\sim$3000 SC RF cavities; six undulator beamlines. Operational 2017.
• LCLS / LCLS-II (SLAC) — LCLS operational 2009 (warm linac); LCLS-II upgrade replaced warm linac with superconducting 4 GeV section, operational 2024 with 1 MHz repetition rate.
• PAL-XFEL (Pohang, Korea) — 10 GeV warm linac, operational 2017.
• SACLA (RIKEN SPring-8, Japan) — 8.5 GeV warm linac, operational 2012, soft + hard X-ray.
• SwissFEL (PSI Switzerland) — 3 GeV warm linac, operational 2018.
• FLASH / FLASH-2 (DESY Hamburg) — 1.25 GeV soft-X-ray FEL.

SC linacs (XFEL, LCLS-II, FLASH) use 1.3 GHz nine-cell Nb cavities at 2 K, with surrounding helium vessels and magnetic-shielding tanks. Bunch-compressor chicanes in the linac use warm dipoles arranged in 4-magnet “C-chicane” or “S-chicane” geometry.

9. Spectrometer and beam-diagnostic magnets

9.1 Sector magnets and mass spectrometers

Sector magnets bend ions by an angle dependent on $m/q$. Used in:

• JEOL JMS-T200GC — gas-chromatograph mass spectrometer with a 60° magnetic sector.
• Thermo MAT 253 IRMS — isotope-ratio mass spectrometer (multi-collector 90° sector).
• Bruker AutoFlex MALDI-TOF — uses electrostatic rather than magnetic separation (TOF, not sector).

Standard sector magnet: H-shape or C-shape iron yoke with flat parallel pole faces, field 0.5–1.6 T. Built by AMETEK Advanced Material Sciences, Bruker BioSpin, JEOL custom.

9.2 Wien filters

A Wien filter combines crossed $\mathbf{E}$ and $\mathbf{B}$ fields such that only particles of one velocity pass undeflected ($v=E/B$). Used in:

• Cassini SOI ion-mass spectrometer — historical example.
• TEM monochromator (FEI Titan, JEOL Grand-Arm${}^2$) — narrows the electron-beam energy spread.
• Charged-particle deflectometry in semiconductor metrology.

9.3 Beam-position monitors (BPMs)

Most accelerator BPMs are non-magnetic (button-electrode or stripline), but some specialised “magnetic” BPMs use saddle-coil pickups around the beam pipe for high-current proton machines (e.g. SNS at ORNL). The “current transformer” beam-current monitor (Bergoz ICT) is a magnetic device — a wound toroid around the beam pipe that integrates the time-domain beam pulse.

10. Cryogenic conductors in detail

10.1 NbTi multifilament Rutherford cable

Production by Bruker EAS, Furukawa Electric, Western Superconducting Technologies (WST), Bochvar Institute (Russia, sanctioned):

• Filament diameter: 5–7 µm (driven by AC-loss requirements; smaller filaments → lower hysteretic loss).
• Cu:NbTi ratio: 1.6–2.0 (for stabilisation against transient quench).
• Twist pitch: 18–25 mm (decouples filaments at low frequency).
• Cable: 28–36 strands, keystoned cross-section.

10.2 Nb${}_3$Sn — wind-and-react

Nb${}_3$Sn is brittle in its reacted A15 form (cannot tolerate $>$ 0.3 % strain). Standard fabrication:

1. Wind the magnet from precursor copper-clad Nb-Sn-Cu wire.
2. React at 600–650 °C for $\sim$200 hours under inert atmosphere. The Sn diffuses into Nb forming Nb${}_3$Sn intermetallic.
3. Impregnate with epoxy / CTD-101 to lock the coil mechanically.

Three Nb${}_3$Sn precursor architectures:

• Bronze route — Nb in Cu-Sn bronze matrix; oldest, lower $J_c$, low-AC-loss for NMR / MRI magnets.
• Internal Sn (IT) — discrete Sn cores in Nb matrix; higher $J_c$.
• RRP (Restacked-Rod Process) — sub-elements with Sn at the centre; highest $J_c$, used in MQXFA and FRESCA2.

Producers: Bruker EAS (RRP), Furukawa, JASTEC, Luvata (legacy), WST.

10.3 REBCO coated conductor

REBa${}_2$Cu${}_3$O${}_{7-\delta }$ thin film grown on Hastelloy substrate with buffer layers (MgO, LMO, CeO${}_2$) and capping (Cu, Ag). 4 mm or 12 mm wide tape, 75–100 µm total thickness. Producers:

• SuperPower Inc. (USA) — IBAD-MOCVD, premium high-current product.
• American Superconductor (AMSC) — RABiTS-MOD, high-volume.
• Fujikura (Japan) — IBAD-PLD.
• Bruker HTS (Germany).
• Faraday Factory Japan — large-scale production for tokamak fusion + AMB market.
• Theva (Germany) — ISD substrate.
• SuperOX (Russia / Japan) — used in fusion (RFX-mod) and accelerator R&D.

REBCO currently dominates the fusion magnet ramp-up (SPARC at Commonwealth Fusion Systems, Tokamak Energy ST40); accelerator deployment is at the R&D stage.

11. Worked example — beam rigidity and dipole field

A 7 TeV proton has momentum $p\approx 7\mathrm{TeV}/c$. Beam rigidity $B\rho =p/q$ (in T·m if $p$ in eV/c, divided by $c$):

$B\rho =\frac{p}{q}=\frac{7\times 10^{12}\mathrm{eV}/c}{e}=7\mathrm{TV}\cdot \frac{1}{c}=23300\mathrm{T}\cdot \mathrm{m}$

For a circular ring of circumference 26.7 km with $\sim$2/3 of it filled by dipoles (the rest is straight sections), bending radius $\rho \approx 2800\mathrm{m}$, so dipole field $B=(B\rho )/\rho =23300/2800=8.33\mathrm{T}$ — exactly the LHC main dipole design.

For 50 TeV (FCC-hh half-energy commissioning), $B=16\mathrm{T}$ at the same $\rho$ → Nb${}_3$Sn or HTS required.

11.1 Strong focusing in a FODO cell

The standard FODO (“focusing-defocusing-O” empty drift-O empty drift) cell consists of one focusing and one defocusing quadrupole per period $L$. Phase advance per cell:

$\mu =2\arcsin \left(\frac{L}{2f}\right)$

with $f$ the quadrupole focal length. For a typical LHC arc cell with $L=53.45\mathrm{m}$, $f=22.9\mathrm{m}$, $\mu \approx 90^{\circ }$. The 23 FODO cells per arc plus dispersion-suppressor regions give a horizontal tune $Q_x=64.31$ and vertical $Q_y=59.32$ — far from major resonances. Tune is adjusted in operation by trimming the quadrupole strengths through dedicated “trim quadrupole” magnets.

12. Power electronics for accelerator magnets

A 50 MVA accelerator power converter must deliver ramp-controlled DC at very tight tolerance (10${}^{-6}$ relative current stability). Standard topology:

• 12-pulse thyristor rectifier — legacy machines (LEP, Tevatron, RHIC).
• 6- or 12-pulse PWM IGBT rectifier — modern machines (LHC, SLS, ESRF-EBS).
• Active-front-end PWM with output regulator — fast-cycling machines (FAIR SIS100 at 4 T/s ramp).

Vendors: Bruker BioSpin (small magnets), TDK-Lambda (mid-range), Ocem Energy Technology (Italian large-magnet PCs), Powerex (legacy), CERN in-house. The LHC has $\sim$1700 power converters across the ring.

13. Quench protection of superconducting magnet strings

A circuit of N magnets in series must protect against:

• Single-magnet quench — fast detection + heater firing in the affected magnet, energy extraction across the entire string.
• Quench-back propagation — once one magnet quenches, the inductive voltage transient can quench others; the detection system must distinguish primary from cascaded quenches.
• Heater-induced quench — modern CLIQ system (LHC, HL-LHC) electrically induces a quench in all magnets simultaneously to dump the stored energy distributedly rather than concentrating it in dump resistors.

The standard reference is the IEEE Power Engineering volume on superconducting-magnet protection (Wilson 1983, Iwasa 2009).

14. Beam-loss-induced quench

A primary loss mechanism: a misdirected beam halo deposits 100 mJ–10 J into the coil over $\mu$s. The LHC collimation system intercepts $>$ 99.9 % of beam halo upstream; the remainder is targeted by tertiary collimators to avoid coil contact. The Phase-II collimator upgrade (HL-LHC) adds new MoC and ARMCO-iron jaws.

14.1 Beam dumps

The LHC has two beam dumps (TDE, target dump external) at IP6 — graphite-block targets in 30 m steel-clad concrete vessels. The 360 MJ beam (worst-case, both rings full energy) is extracted in a single 89 µs sweep by 15 fast-pulsed kicker magnets (MKD, 1.4 T pulse-field, 100 µs rise time) into a dilution-magnet quartet that paints the beam over a 30 cm radius on the graphite face. Each year LHC performs $\sim$10 000 dump cycles; the dump survives by spreading the deposition.

14.2 Collimators

The LHC primary collimator system (TCP) intercepts beam halo with movable C-C composite jaws closed to 5-7 sigma transverse aperture. The secondary collimators (TCS) catch scattered halo at the next interaction region. HL-LHC upgrades replace some collimators with MoC and ARMCO-iron jaws for higher absorption. Collimator settings change shot-by-shot during operation as the beam emittance varies.

15. Cost summary

Project	Cost (USD)	Period
LHC construction	$5B	1992–2008
LHC operating cost	$1B / year	2008–
HL-LHC upgrade	$1.4B	2017–2029
FCC-ee (proposal)	$15–20B	2030s–2065
FCC-hh (proposal)	$30–40B	2050s–2090
Muon collider 10 TeV (proposal)	$10–20B	2040s
Single proton-therapy room (Mevion)	$30–40M	per install
Multi-room proton centre	$120–180M	per install
Synchrotron light source (3 GeV class)	$0.6–1.2B	per facility
FEL hard X-ray (LCLS-II class)	$1.0–1.5B	per facility

16. Standards and qualification

• IEC 60044 — instrument transformer accuracy class; relevant for accelerator current monitoring.
• API for high-energy-physics magnet thermal-mechanical stability — site-specific (CERN-EDMS-LHC magnet specs; FNAL technical division).
• ASME Boiler and Pressure Vessel Code Section VIII — cryostat vessel certification (most accelerator cryostats use ASME-stamped He vessels).
• NFPA 99 / NFPA 50 — bulk-helium safety in medical and research facilities.
• DOE 10 CFR 851 — US national-lab worker safety in accelerator facilities.

16.1 Cosine-theta and block-coil designs

Two layouts dominate superconducting dipole construction:

• Cosine-theta — the current distribution around the aperture follows $J(\theta )=J_0\cos \theta$, which produces an ideal uniform dipole field by Ampere’s law. Practical implementation uses 2 or 4 layers of Rutherford cable wedged into a circular arrangement with stainless or aluminum keystoned spacers. LHC, HERA, Tevatron, RHIC all use cos-theta.
• Block-coil — racetracks of cable arranged as rectangular blocks. Lower precision than cos-theta but tolerates larger conductor and easier reaction for Nb${}_3$Sn. The FRESCA2 (CERN) 14.6 T demonstrator and HL-LHC 11 T MBH dipole use block-coil; planned FCC-hh 16 T magnets weigh both options.
• Canted-cosine-theta (CCT) — tilted double-helix winding; emerging research design with intrinsic stress management, in development at LBNL, CERN, Paul Scherrer Institute.

17. Pitfalls

• Field-quality drift with cycle history. Superconducting magnets have hysteresis in the field at low fields (the “snapback” effect at LHC injection). Calibration tables are run-history-dependent and updated continuously by the beam-based feedback system.
• Persistent-current decay. Even in persistent operation, screening currents in superconducting filaments decay over hours; the field drifts $\sim$10${}^{-4}$ over 10 hours after a ramp. Feedback from beam-position monitors closes the loop.
• Mechanical alignment. A 14 m LHC dipole must be aligned to $\sim$0.1 mm over its length. Survey using laser trackers (Leica AT960, FARO Vantage) is done in tunnel + repeated annually.
• Radiation damage to NdFeB undulators. Permanent-magnet undulators in high-radiation environments lose remanence over months from beam-loss halos. Cryo-cooled and HTS alternatives are more radiation-tolerant.
• Helium-leak detection. Even pinhole helium leaks at 1.9 K accumulate over years and degrade insulation vacuum. Annual leak surveys with mass-spec sniffers + warm-up campaigns.

17.1 Mechanical pre-stress

Superconducting accelerator magnets are mechanically pre-stressed to compress the coils against Lorentz-force loads at full field. Without pre-stress, coil motion would cause friction quenches; with too much, the conductor sees damaging strain at room temperature. The LHC main dipole is pre-loaded with $\sim$50 MPa azimuthal stress (collar + yoke + shrinking-cylinder shrink-fit assembly). Modern Nb${}_3$Sn magnets use bladder-and-key construction (Caspi / LBNL) — inflatable hydraulic bladders pre-load the keys during assembly, then bake out leaving permanent stress.

17.2 Training quenches

Even after careful assembly, a new superconducting magnet “trains” through a series of small quenches: the first time it ramps to full field, conductor settles by 1-10 µm under Lorentz load, dissipates 1-10 mJ friction heat → quench. Each subsequent ramp pushes the quench current higher; after 5-30 quenches the magnet reaches its plateau (close to short-sample limit). Modern impregnation techniques (epoxy + ceramic-filled paraffin) reduce training to $\le$ 3 quenches per LHC magnet.

18. Cryogenic plant scale

For context on the helium economy of LHC-class machines:

• LHC inventory: 130 t LHe at 1.9 K + 60 t at 4.5 K → 190 t total, $\sim$$10–20M raw material at modern prices.
• ITER inventory (tokamak fusion): 24 t LHe at 4.5 K.
• HL-LHC inventory: $\sim$140 t (+10 t for new triplets).
• FCC-hh inventory (projected): 800–1000 t — the LHe market would have to expand to support it.

19. Software for accelerator-magnet design

The accelerator-physics community has developed dedicated tools:

• ROXIE (CERN) — the canonical 2D superconducting-magnet design tool. Coil layout, field harmonics, Lorentz-force calculation, quench protection, mechanical stress.
• OPERA (Vector Fields, now Dassault) — general-purpose 3D FEA for electromagnets including non-linear iron.
• CST EM Studio / ANSYS Maxwell — alternative 3D EM solvers.
• MAD-X (CERN) — accelerator-lattice tracking code; integrates magnet field maps with beam dynamics.
• Bmad (Cornell, SLAC) — alternative lattice / beam dynamics code.
• PETRA-IV / Sirius / Diamond-II in-house — facility-specific magnet design and tracking tools.
• OPAL (PSI) — particle-in-cell beam dynamics with space charge.

CERN’s ROXIE is open-source within the HEP community; commercial codes dominate the proton-therapy and synchrotron-magnet manufacturers.

Further reading

• Mess, K. H., Schmüser, P. & Wolff, S. (1996). Superconducting Accelerator Magnets. World Scientific. The standard textbook from the HERA era.
• Wilson, M. N. (1983). Superconducting Magnets. Oxford University Press. The reference on stability, training, quench protection.
• Iwasa, Y. (2009). Case Studies in Superconducting Magnets (2nd ed.). Springer. Modern engineering case studies.
• Devred, A. (2002). “Practical low-temperature superconductors for electromagnets.” CERN Yellow Report CERN-2004-006.
• Brüning, O. et al. (Eds.). (2004). LHC Design Report Vol. I-III. CERN-2004-003. The LHC bible.
• HL-LHC Project Team. (2020). HL-LHC Technical Design Report. CERN-2020-010.
• FCC Collaboration. (2019). “FCC Conceptual Design Report.” Various volumes in Eur. Phys. J. ST.
• Bottura, L. & Godeke, A. (2012). “Superconducting materials and conductors for accelerator magnets.” Reviews of Accelerator Science and Technology 5, 25–50.
• Schopper, H. (Ed.). (2008). Elementary Particles: Accelerators and Colliders (Landolt-Börnstein Vol. 21A, 21B, 21C). Springer.
• Wiedemann, H. (2015). Particle Accelerator Physics (4th ed.). Springer. Standard graduate textbook including magnet design from the beam-dynamics perspective.

20. Magnet manufacturer ecosystem

The accelerator-magnet vendor base is small and specialised:

• Babcock-Noell (Würzburg, Germany) — LHC main dipole, ITER cryostat fabrication.
• ASG Superconductors (La Spezia, Italy) — LHC dipoles, MRI magnets, ITER TF coil.
• Alstom-MSA → Sigmaphi / GE Power Conversion — historical LHC magnet supplier; now Sigmaphi (France) for synchrotron-class warm magnets.
• Bilfinger Noell (Germany) — HL-LHC 11 T dipole development.
• Hitachi Industrial Equipment Systems — synchrotron and FEL dipoles.
• Tesla Engineering (UK, no relation to Tesla Motors) — synchrotron multipole magnets, undulators, MRI.
• Buckley Systems (Auckland, NZ) — ion-beam dipoles for semiconductor implantation and synchrotron.
• Danfysik (Denmark) — accelerator and beamline magnets; standard supplier to European labs.
• Bruker BioSpin — MRI and NMR superconducting magnets; spin-out from former Oxford Instruments.

Adjacent

• electromagnetics-engineering — Maxwell + multipole expansion + Lorentz force; the field foundation.
• mri-magnets-and-coils-deep — sibling note: NbTi and Nb${}_3$Sn at clinical scale.
• electric-motors — accelerator magnets are stationary, but the LSM linear motor and accelerator dipole share electromagnetic-circuit principles.
• power-electronics — multi-MW DC and pulsed power supplies for accelerator magnets.
• photonics — synchrotron and FEL applications of accelerator magnets to coherent X-ray light sources.
• quantum-materials-and-topological-phases — NbTi, Nb${}_3$Sn, REBCO superconductor pinning physics and $J_c$ behaviour.
• magnetic-and-optical-materials — NdFeB and SmCo permanent magnets for undulators and accelerator correctors.