The production magnet papers are fifteen IEEE Transactions on Applied Superconductivity publications from 2023 to 2026 recording what came after the model coil: the shift from building one magnet to qualifying, manufacturing and testing them in quantity.
📝The TFMC Engineering Papers stop at the end of 2021, with a single 📝Toroidal Field Model Coil (TFMC) taken to 20 📝tesla and then deliberately quenched. This cluster is their successor, and the problem it addresses is a different one. A model coil is qualified by testing the coil; a production line is qualified by testing what it consumes and the processes it runs — tape lots, solder soaks, joints, instrumentation — long before a finished magnet exists. Almost all of it is measured rather than projected: cables were built and energised, samples irradiated, fibres bleached, quenches detected on hardware that exists. 📝SPARC's assembled magnet set in service is still ahead.
Two papers treat the 📝coated conductor as a purchased commodity that must be accepted or rejected at volume. Aliya Greenberg and six co-authors correlated 📝critical current measured at 44–77 kelvin below 5 tesla against the same tape at 20 kelvin and 6–20 tesla across several manufacturers, so that a cheap liquid-nitrogen test predicts behaviour in the regime the magnets actually operate in (10.1109/TASC.2024.3360937). J. L. Cheng and eleven co-authors used X-ray diffraction rocking curves on production quantities of 📝REBCO tape from two vendors to measure the small tilt between the crystal ab-plane and the plane of the tape; that offset is vendor-dependent, and it moves magnet critical current in untwisted windings such as SPARC's 📝toroidal field coils (10.1109/TASC.2025.3545023).
Manufacture and joining follow. 📝Amy Watterson and thirteen co-authors held PIT 📝VIPER cable in flowing tin–lead solder for 2.5 hours during vacuum pressure impregnation — the first test at a duration relevant to large-scale manufacture — then measured critical current, n-value and resistance on a 2.5-metre straight cable and a 20-metre coiled one in liquid nitrogen (10.1109/TASC.2025.3527947). Nine experiments compared indium wire against indium foil in joints between PIT VIPER cables: foil joints came out roughly 40 % more resistive, and the indium oxide layer proved negligible beside voids and incomplete plastic deformation (Uscuplic et al., 10.1109/TASC.2025.3630150, with 📝Charlie Sanabria third author).
Both of those tests ran in facilities that got papers of their own. Alexey Kaplan and sixteen co-authors describe the liquid nitrogen test facility at the 📝MIT Plasma Science and Fusion Center, operational since 2020, which has run quality-assurance tests on sub-elements of both the TFMC and the 📝central solenoid model coil (10.1109/TASC.2025.3530389). 📝Philip C. Michael and eleven co-authors added per-layer mass flow meters and calibration heaters to the supercritical helium circuit at the MIT Superconducting Magnet Test Facility, turning it into a calorimeter that cross-checks dissipation in layer-to-layer joints, measures 📝AC loss during swept current, and quantifies the energy deposited inside a magnet during a 📝quench (10.1109/TASC.2024.3508664).
Four papers attack quench detection, on the premise that voltage measurement is difficult on insulated windings and impossible on 📝no-insulation coils, which pushes detection onto optics. Bartholomew Ludbrook and eight co-authors at 📝Paihau–Robinson Research Institute mapped the sensitivity of quasi-continuous 📝fiber Bragg grating arrays of up to 1000 identical gratings, showing that broadband attenuation down the array — not where along it a perturbation sits — sets the limit (10.1109/TASC.2023.3244762). Owen Duke's group measured radiation-induced attenuation in fluorine-doped fibre irradiated to 1.2 megagray of gamma at 77 kelvin, and showed that continuous optical annealing at 970 nanometres raised the survivable dose by a factor of 2000 — with the gain falling as sample length went from 45 to 205 metres (10.1109/TASC.2023.3347369). That length dependence is the follow-up's subject: a length-resolved attenuation measurement out to 250 metres, 📝Erica Salazar third author (10.1109/TASC.2025.3543808). Together they are what 📝radiation hardening means for a 📝distributed fiber-optic sensing system inside a fusion magnet. 📝Alexey Radovinsky, Nicolai Martovetsky and Sergey Kuznetsov take the opposite line, proposing a voltage-based method they argue works on insulated and no-insulation magnets alike, with 📝quench protection explicitly left out of scope (10.1109/TASC.2024.3520084).
The remaining five are calculation, three of them anchored to hardware. OpenSc2, an open-source thermal-hydraulic solver, reproduced the measured temperature traces from VIPER samples tested at the 📝SULTAN facility in 2019 once tape-stack twist was accounted for (Placido et al., 10.1109/TASC.2024.3387880). Michael and eight co-authors model transport-current loss in a twisted-stacked-tape cable under fast ramps, where inductance rather than terminal resistance decides how current distributes (10.1109/TASC.2025.3630611). Kuznetsov and four co-authors model how to separate quench strain from ordinary operational strain on fibres inside SPARC central solenoid PIT-VIPER cables (10.1109/TASC.2024.3391228). Radovinsky, Kuznetsov and 📝Dan Brunner give an analytical method for force-balanced tokamak design, in which non-planar coil orientation cancels the 📝Lorentz force locally and relieves the stress that caps achievable field (10.1109/TASC.2024.3365087). The fifth is not a CFS paper: led from Brookhaven National Laboratory and the National High Magnetic Field Laboratory, with Dylan Kolb-Bond the single CFS author, it revisits the quench of the 32-tesla prototype magnet and derives hoop strain from 📝screening current alone in solenoids built from REBCO double pancakes, the 📝pancake winding geometry common across HTS magnet design (10.1109/TASC.2025.3570868).
Twenty tesla once was the model coil. These fifteen papers are our answer to "and again, and again" — the tape acceptance, the solder soak, the joint, the fibre — for every coil SPARC needs.