The Kronos 2026 design series — 44 open papers spanning the spherical-tokamak breeder, the D–³He tandem-mirror burner, magnets, direct energy conversion, AI & quantum control, and the physics de-risking programme — each paired with a reproducible, openly-archived Zenodo deposit, and available as one complete volume.
Every headline number regenerates from a named script and archived data under a fixed random seed; requirement-class assumptions and open risks are carried openly rather than absorbed into a single optimum. Each paper below has its own reading edition (the editorial), linking the manuscript to its data-and-code deposit, the interactive 3D model, and the live verification simulator.
All studies in one bound volume: the compact fusion isotope-and-energy platform, front to back — breeder, generator, magnets, direct energy conversion, AI/quantum control and the full de-risking programme — with the shared methodology, notation and reproducibility manifest that ties them together.
The hub of the series: a public, crawlable register that states every binding physics gate for the breeder and the burner, the campaign breakthroughs, the honest closure statement, and the companion-deposit DOI table that ties all 44 papers together.
The five foundational studies: the breeder, the generator, the magnets, direct energy conversion, and the control stack.
Hyperion is a compact negative-triangularity spherical-tokamak breeder that earns on fusion gain alone producing tritium, decay helium-3, and 14 MeV neutrons so it delivers value without waiting on the net-electricity milestone the rest of the field is chasing. At the frozen reference point it reaches Q = 3.0763 at 85.04 MW and 9.66 MA, with negative triangularity cutting ion-scale transport by 40–45% into a quiet subcritical core. The tritium output…
Aegis and MetroVolt are two application housings of one deuterium helium-3 tandem mirror. Its principal new result: synchrotron reabsorption is not a fixed fraction of fusion power but scales with machine size. At the design point the engineering gain is Q E 1.318 at a low neutron fraction of 5.44% low-neutron, never aneutronic. Closure is requirement-class, not demonstrated: it needs an end-plug density ratio of 16, stated as the design s largest open…
The magnet paper does one thing carefully and applies it twice: bore-resolved hoop stress, which scales with the winding bore exactly what worst-case derating hides. Applied to the two machines, the binding constraint flips: the breeder s centre-post is stress-comfortable, while the burner s small-bore plug coil, once feared infeasible as a bare winding, is resolved by a stress-managed structural shield at a healthy strain margin. Two claims are kept…
For a deuterium helium-3 magnetic-mirror burner, the directed axial power splits 55.5% into the electron channel and 44.5% into the ion channel so the electron stream, not the ion stream, is the larger prize, and the inversion holds across 20 200 keV. No published converter recovers directed electrons. A quasineutral two-species expander (field ratio ~20) lifts axial electric recovery by ~13% (~97 MW e ) at converter efficiency 0.7. The upside is…
Real-time control of a nuclear-regulated fusion generator is treated as a layered digital-twin architecture in which physics-informed machine learning is deployable now but only inside a deterministically clamped safe set. A control-barrier filter permits 0 safety-set escapes across 5000 trajectories against 5000 unfiltered, backed by an analytic proof. Quantum sensing is a near-term diagnostic tier; quantum computation is an offline calibration tier…
Companion studies that pin each binding gate individually — from the consumable centrepost and the blanket breeding ladder to disruptions, the helium-3 cycle, licensing and buildability.
The inboard centrepost is the defining vulnerability of a compact spherical-tokamak breeder, and it is handled here as a consumable, scheduled-replacement cartridge rather than a lifetime component. Four coupled analyses on the frozen design point close it: a 3-D coverage-versus-protection neutronics scan, a coil structural finite-element check against the analytic Lam stress, a fused strain-rate-plus-magnetization quench detector, and a…
The tritium breeding ratio of a compact spherical-tokamak breeder is a design lever, not a fixed number. Three-dimensional OpenMC neutronics on ENDF/B-VIII.0 place it: a bare lithium blanket gives net TBR about 0.86; a solid-beryllium multiplier raises it into the low-1.1s; and an advanced blanket reaches net TBR 1.42, the confirmed per-unit ceiling, reproducing across an independent nuclear-data library to under 0.3%.
Coupled OpenMC neutronics book the centrepost fast fluence, displacement damage, and consumable-cartridge life against a tungsten-carbide shield whose thickness trades directly against machine size and breeding coverage. The design damage rate is 18.85 dpa per full-power-year; the mid-current operating point extends the life. The centrepost is a scheduled-replacement cartridge, stated as such.
This paper states the verification-and-validation discipline used across the Kronos 2026 series: frozen, evidence-tagged anchors regenerable from deposited code, a two-tier reproducibility standard, and an independent re-run on separate hardware and an independent nuclear-data library. The load-bearing three-dimensional tritium breeding ratio reproduces across ENDF/B-VIII.0 and FENDL-3.2 to within 0.3 percent.
In a deuterium-helium-3 mirror the fusion-born fast ions carry most of the released energy. RF alpha-channeling resonant waves that extract energy from the fast ions and hand it to the fuel while expelling the ash raises the recoverable charged fraction from a collisional baseline near 8 percent to about 80 percent in the worked embodiment, easing the recirculating-power burden that sets the engineering gain. It is booked as a bounded lever; the plug…
A flute-mode electrostatic dispersion solver, cross-checked electromagnetically, maps the exact marginal-stability boundary of the deuterium-helium-3 tandem-mirror plug. The design s warm-ion fill sits exactly on the stabilization knee: it clears the drift-cyclotron loss-cone mode only for sufficiently broad plug-edge gradients, with a weak residual cyclotron mode. Marginal but bounded, and quantified.
A deterministic, control-barrier-certified safety clamp provides a provable zero-escape guarantee for the compact spherical-tokamak breeder operating envelope. A robust, disturbance-tightened control-barrier-function filter with a closed-form separable solution wraps an arbitrary controller: 0 of 5000 Monte-Carlo trajectories escape the safe set with the clamp versus 5000 of 5000 without it, backed by an analytic forward-invariance proof.
Landau damping is reproduced on a 5-qubit Hamiltonian-simulation circuit; adaptive variational quantum chemistry reaches chemical accuracy on candidate fusion-alloy fragments; and a resource frontier separates the tractable linear kinetic operator (early fault-tolerant, 11-19 logical qubits) from the nonlinear gyrokinetic problem (fault-tolerant horizon). Noisy hardware shows no advantage today; classical gyrokinetics remains cheaper at every…
A reduced-order red-team of the binding gates of the compact breeder and the tandem-mirror burner. For each honest negative, alternatives are brainstormed, bounded, ranked, and only the promising ones promoted to high-fidelity validation. The current-drive burden falls 58 percent below fusion power inside unused pressure headroom; tritium closure is restored independent of the beryllium multiplier; and the burner plug-density requirement is attacked by…
Nonlinear CGYRO with kinetic electrons at the frozen negative-triangularity design point shows negative triangularity cutting ion-scale ITG+TEM transport by 40–45 percent versus a positive-triangularity twin, with the operating point quiet and subcritical; electron-scale runs show 79–80 percent ETG suppression. The direction is corroborated against the TCV and DIII-D negative-triangularity campaigns; reduced-mass and design-target caveats are stated…
A reduced-order assessment of disruptions, runaway electrons, and electromagnetic loads on a NIMROD-verified equilibrium. The current-quench time is 7-19 ms; the runaway-electron avalanche is severe (E/E_c about 1128, a beam of order 70 percent of plasma current); the halo current is 35 percent of plasma current. The conclusion is a hard design requirement: shattered-pellet-class disruption mitigation triggered within the current-quench window. The…
Fast-alpha and fusion-product confinement in the deuterium-helium-3 tandem-mirror burner. The mirror-trapped charged products and their loss-cone dynamics set the ash accumulation and the axial power that the direct-energy-conversion and alpha-channeling systems must handle. Reported with its confinement caveats.
The helium-3 half of the breeder fuel cycle: helium-3 bred as a co-product of surplus-tritium decay at equilibrium, distinct from bulk fuel supply. The 12.3-year tritium half-life makes the breeder s decay-sourced helium-3 far too slow to fuel a burner fleet (~192/105/45 breeders per burner), so burner-scale helium-3 is lunar-gated; the breeder s helium-3 is a strategic co-product with national-security, cryogenic, and quantum value.
Ideal-MHD stability of the compact negative-triangularity spherical-tokamak breeder at the frozen design point. The equilibrium is ideal-MHD-stable with quantified margins against the binding kink and ballooning limits, carried openly.
A portfolio assessment of divertor and heat-exhaust options for the compact breeder, spanning conventional, advanced-magnetic, and liquid-metal targets, ranked against the concentrated exhaust power of a compact machine. The flowing-lithium vapor-box option is developed in a dedicated companion. Screening bounds, honest gates carried.
Low-activation structural-material selection for the compact fusion platform: reduced-activation steels and their activation, decay-heat, and waste-classification behaviour under the 14 MeV spectrum, targeting a low-level-waste class and short-cooldown maintenance. Material-selection targets pending qualification.
A breeder s product is tritium, and its credibility rests on an honest mass balance. The per-unit burn rate is fixed by the fusion power at 4.76 kg-T per full-power-year, and the net surplus is a confirmed ladder : net TBR ~1.06 (literal recipe), ~1.19 (solid-beryllium layer), and 1.42 (advanced blanket, ~2.0 kg-T/yr per unit) the last the confirmed per-unit ceiling, above which the spherical-tokamak centrepost caps the inboard coverage. A single…
A flowing-lithium vapor-box divertor for the compact breeder that couples heat exhaust to the fuel cycle. Coupled FESTIM/TMAP permeation-retention modelling ranks four liquid metals by tritium loss: lithium is optimal by three to six orders of magnitude over lithium-tin, lead-lithium, and tin, so the best power-handling surface is also the best inventory-control surface. An internal-platform concept, not a built system.
The unsolved protection problem of high-temperature-superconductor fusion magnets is that REBCO quenches propagate too slowly for voltage detection to act before local damage. A fused-detector scheme on two prompt precursors local strain rate and magnetization change fires at 0.089 s against a voltage-threshold 3.024 s, a lead of 2.935 s at signal-to-noise ratio 322.5, cutting the MIITs the winding must survive. An in-winding nitrogen-vacancy…
Fusion s safety advantages are real but are often asserted rather than bounded. The dominant radiological hazard is the mobilizable tritium inventory, minimized by direct internal recycling rather than a large buffered store; the standing-inventory value that sets the source term is computed in the fuel-cycle model, not asserted. There is no fission-product inventory and no chain reaction, so an unmitigated loss of control cannot produce a runaway or a…
Following the 2023 NRC determination and the ADVANCE Act, fusion energy systems in the United States are regulated as byproduct material (10 CFR Part 30), not under the reactor framework. That framework is materials-oriented and Agreement-State-administrable, giving a graded, hazard-proportionate path suited to a machine whose dominant hazard is a bounded tritium inventory and which has no fission-product source term or criticality mechanism. A…
A steady-state breeder must sustain its plasma current non-inductively, and the recirculating power to do so is a real charge. At the frozen operating point the pressure-driven bootstrap current self-generates about 15.5 percent of the plasma current and external non-inductive drive supplies the remaining 8.16 MA, its wall-plug power a named line in the recirculating budget. The bootstrap fraction is a lever, stated modestly.
A spherical tokamak has almost no room for a central solenoid, so it cannot start its plasma current the way a conventional tokamak does. The current is initiated by outboard induction, electron-cyclotron heating, and a helicity or beam seed, then handed off to the non-inductive and bootstrap currents that hold the flat-top. The ramp auxiliary power and profile-control hand-off are flagged as carried risks. A shorter synthesis paper.
A breeder is a strong 14 MeV neutron source, and that source has value beyond tritium. The design delivers a full-rate fast-neutron materials-qualification capability about 59 times the reference dedicated facility, in a true fusion spectrum; and it breeds shortage-prone medical isotopes such as molybdenum-99 without a fission reactor. Strategic-material co-products booked as capability, not revenue.
Kronos is not one machine but a platform: a compact negative-triangularity spherical-tokamak breeder that supplies tritium, helium-3, 14 MeV neutrons, and co-produced isotopes, and a deuterium helium-3 burner in two housings that supplies power at Q E 1.318 with a neutron fraction of only 5.44%. The two share a REBCO magnet basis and a single physics de-risking record, and they close a strategic loop. We state closure in two tiers throughout…
Fuel choice is the most consequential decision in a fusion program, and it is often made emotionally toward the aneutronic ideal rather than on the physics. We grade the four candidate fuels on one ruler: reactivity, temperature, neutron fraction, tritium logistics, availability, and demonstrated gain. No fuel wins on every axis. The Kronos architecture follows the scorecard, not the ideal: run deuterium tritium where the product is the neutron and the…
The kinetic distribution function is highly compressible in a low-rank tensor-network representation, and that yields a practical classical solver, not a storage trick. On a 1D1V BGK test, a matrix-product-state truncation reaches relative-L² error 2 10 at rank 8 using ~0.19 the dense storage, with the error falling exponentially in rank. Unlike the fault-tolerant-horizon quantum route, this runs today.
A hot deuterium helium-3 mirror loses more power to electron-cyclotron emission than to any other channel, and that loss pins the optimal electron temperature. A first-principles transport calculation gives a reabsorbed synchrotron loss of 1803 MW against a 2546 MW radiated scale versus 919 MW (Trubnikov) and 66,677 MW optically-thin, a near-two-order model spread it collapses. The plasma is optically thick to n* 10.4, and the loss fraction falls with…
The survival question, computed on an independent code stack: displacement damage (~9.78 dpa/fpy behind armor), helium/hydrogen bubble swelling (accumulating but not life-limiting), plasma-material sputtering/erosion, and hydrogenic permeation. The governing life-limiter is displacement-driven embrittlement, and the wall is a scheduled-replacement component on a defined damage budget.
The credible nuclear-data uncertainty on the TBR is set by the light breeding channels Li(n,t), Li(n,n't), Be(n,2n) show a 55 66% TALYS-vs-ENDF model spread at 14 MeV, against ~15% for heavy structural channels. The load-bearing light channels must come from evaluated covariances, not model defaults.
The magnet's survival in a neutron environment is a materials question that changes the conductor specification. Four results: the un-doped REBCO tape outlasts the doped (artificial-pinning) tape for fusion fluence; a J c -onset-knee surrogate screens candidates; a graphene-reinforced aluminum stabilizer retains ~2 10 S/cm at cryogenic temperature in 30 35 T; and a V O self-protecting quench interlayer gives passive turn-to-turn protection.
The plug's electrostatic potential is the burner's binding gate. A thermal barrier a local density/potential depression sustained by ECH hot electrons and sloshing ions decouples the plug potential from the central-cell electrons: the ion-confining-potential requirement drops from 248.75 keV to 167.9 keV (a 33% gate cut), hot-electron heating removes ~1093 MW of central-cell ECH, and in the aggressive limit Q E rises from ~1.3 toward 3 5.
A fusion generator built around direct energy conversion is intrinsically a high-voltage DC source. About 54% of gross output leaves the converter as DC at 50 250 kV, reaching an 800 V rack bus in 2 3 power-electronic stages (vs 4 5 for an AC plant). Load-following belongs on the ms-class electronic path: a 4.14 gross/net amplifier makes plasma throttling expensive. Economics/siting held internal.
The compact breeder's buildability on engineering grounds (cost held internal): the longest-lead item is 90%-enriched lithium-6, but the neutronics meet the breeding target at 30 60 at% Li moving the hardest item off the critical path. The centrepost is a scheduled-replacement cartridge on a ~10 15 unit/yr line, converting a manufacturing risk into a production-rate spec.
A prognostics-and-health-management framework turns the centrepost's fluence-driven degradation into a scheduled-replacement decision with quantified confidence: a dual-channel remaining-useful-life estimator fuses a physics fluence model with an in-situ damage observable, guarded by an out-of-distribution backstop that defaults to the conservative physics bound when signals leave the trained regime.
A control-governance principle that makes trust quantitative: the actuation authority a controller or AI is permitted at runtime is a monotone function of the verification-and-validation maturity (PCMM level) of the underlying models. Machine-validated subsystems admit wide, fast authority; reduced-order ones are clamped to reversible actions an auditable rule where authority expands only as evidence does.
The method behind the low-activation alloy family: a three-stage GPU pipeline a universal ML interatomic potential (CHGNet) screens thousands of compositions by formation energy, density-functional theory confirms the survivors, and an activation-transport down-select (OpenMC FISPACT) rejects any whose waste class or decay heat is unacceptable. The novelty is making activation a first-class filter, early.
The operating window synthesized from two findings: a strict invariance source strength and wall loading trade one-for-one against centrepost life across the whole density temperature plane, so there is no free operating-point gain and a coupled admissible-current window whose floor is the q95/MHD edge and ceiling the centrepost fluence, with the tritium-startup requirement non-binding at ~1.7 margin.
Reproducibility. Each deposit is Open Access (CC BY 4.0) in the Kronos Fusion Energy Zenodo community and carries a run-all reproduction driver, a pinned computing environment, and per-file checksums. Reserved DOIs are private drafts that resolve unchanged at doi.org once published. The papers contain no financial or commercial information.