Fusion Neutron-Driven Transmutation

• Fusion neutron-driven transmutation is the modification of nuclide inventories using 14.1 MeV D–T fusion neutrons to induce reactions inaccessible to lower-energy sources.
• It couples neutron transport, depletion modeling, and fine spatial resolution to capture multiscale effects in materials, with tungsten serving as a benchmark.
• The process enables applications in waste reduction, isotope synthesis, and elemental conversion while addressing challenges like radiation damage and nuclear-data uncertainties.

Fusion neutron-driven transmutation is the modification of nuclide inventories by fusion-produced neutrons, most commonly the $14.1\ \mathrm{MeV}$ neutrons from deuterium–tritium fusion, $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$. In fusion systems these neutrons simultaneously drive unwanted composition changes in structural materials, such as tungsten plasma-facing armor, and intentional conversion of feedstocks into transuranic burn products, stable metals, medical radioisotopes, and $\beta^{-}$ emitters. Because the birth spectrum is sufficiently hard to open threshold reactions such as $(n,2n)$, $(n,p)$, and $(n,\alpha)$, fusion transmutation is treated through coupled neutron transport, depletion, blanket design, and radiation-effects materials analysis (Gilbert et al., 2016, Parisi et al., 4 Nov 2025).

1. Reaction physics and transmutation kinetics

The basic physical principle is that a target nuclide in a neutron field undergoes reactions at a rate determined by the local spectral flux and the energy-dependent microscopic cross section. In the continuous-energy form used across transport and depletion studies, a reaction channel is written as

$R_i(r)=\int_0^\infty \phi(r,E)\,\sigma_i(E)\,dE,$

and the isotopic inventory evolves through Bateman-type balance equations with both decay and neutron-induced source and sink terms (Tanner et al., 2021, Gilbert et al., 2016). In a monoenergetic or one-group approximation the same idea is often written as $R=\Phi\,\sigma\,N$, or $dC_i/d\Phi=\sigma_iN_W$ for tungsten transmutation products (Zhao et al., 2017, Parisi et al., 10 Dec 2025).

The hard D–T spectrum opens reaction pathways that are weak or inaccessible in thermal or typical fission spectra. Representative examples include $$^{198}\mathrm{Hg}(n,2n){}^{197}\mathrm{Hg}\rightarrow{}^{197}\mathrm{Au}$$, with a threshold $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$0; $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$1; $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$2; and $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$3 (Rutkowski et al., 17 Jul 2025, Parisi et al., 4 Nov 2025, Parisi, 18 May 2026). In tungsten, neutron capture and $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$4 reactions produce tantalum and rhenium, while further capture on Re isotopes builds up osmium (Zhao et al., 2017).

For isotope-production studies, the produced activity under constant flux is commonly written as

$$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$5

with saturation activity $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$6 (Parisi et al., 4 Nov 2025). For multistep chains, such as mercury-to-gold transmutation, population dynamics are described by chained depletion equations with neutron-driven transfers between adjacent isotopes and radioactive decay of the terminal precursor (Parisi et al., 3 Apr 2026).

2. Transport, depletion, and spatially resolved modeling

Fusion transmutation calculations are typically performed by coupling neutron transport to inventory evolution. A detailed tungsten study computes $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$7 with continuous-energy Monte Carlo using MCNP6 and then evolves local nuclide densities with FISPACT-II in each spatial cell (Gilbert et al., 2016). In that implementation, FISPACT-II represents $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$8 and $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$9 on a 660-group structure up to $\beta^{-}$0 (UKAEA-709) and applies probability-table self-shielding corrections through

$\beta^{-}$1

where $\beta^{-}$2 corrects resonance dilution within a group (Gilbert et al., 2016). The same study resolves a $\beta^{-}$3 tungsten armor shell into $\beta^{-}$4 radial layers, with ADVANTG biasing used to achieve $\beta^{-}$5 statistical precision in most energy bins (Gilbert et al., 2016).

Blanket-scale transmutation studies use analogous workflows with different transport and depletion engines. An ARC-class mercury blanket study uses Paramak for geometry generation, DAGMC export, OpenMC v0.14 transport on a $\beta^{-}$6 toroidal sector with reflective boundary conditions, JENDL-5 nuclear data, and OpenMC’s CoupledOperator over $\beta^{-}$7 days in $\beta^{-}$8 timesteps while “harvesting” gold at each step (Rutkowski et al., 17 Jul 2025). Molten-salt transmutation studies solve fixed-source, $\beta^{-}$9-eigenvalue, and subcritical problems with MCNP 6.2, SCALE/ORIGEN-S, and OpenMC, characterizing source amplification by

$(n,2n)$0

(Tanner et al., 2021).

A recurring result is that spatial resolution is not optional. In DEMO neutronics, the total neutron flux varies from $(n,2n)$1 at the first-wall armor to $(n,2n)$2 at the blanket backplate, and both dpa and gas-production rates follow that depth dependence (Gilbert et al., 2013). In tungsten and blanket feedstocks alike, local moderation, backscatter, self-shielding, and geometrical interfaces set the actual transmutation field (Gilbert et al., 2016).

3. Tungsten as the benchmark fusion transmutation problem

Tungsten is widely treated as the reference case because it is the leading plasma-facing material candidate and because its transmutation products strongly affect service behavior. In spatially heterogeneous models of fusion armor, low-$(n,2n)$3 materials behind or within tungsten—especially water and steel—moderate and back-scatter neutrons, strongly increasing the eV–keV flux near interfaces (Gilbert et al., 2016). In one representative configuration, the MeV flux falls only modestly, from $(n,2n)$4 at the plasma side to $(n,2n)$5 at the moderator side, yet the local $(n,2n)$6 rate can rise by up to a factor of $(n,2n)$7 from front to back because moderated neutrons repopulate the giant capture resonances around $(n,2n)$8–$(n,2n)$9, with a dominant resonance at $(n,p)$0 (Gilbert et al., 2016).

After $(n,p)$1 of D–T irradiation in that scenario, the local rhenium concentration varies from $(n,p)$2 $(n,p)$3 at the plasma-facing surface to $(n,p)$4 $(n,p)$5 at mid-thickness and $(n,p)$6 $(n,p)$7 at the back, a full-profile variation of $(n,p)$8 (Gilbert et al., 2016). A $(n,p)$9 coolant layer directly behind tungsten boosts back-of-W Re by $(n,\alpha)$0 over the reference case, while embedded $(n,\alpha)$1 internal water channels create sharp local Re peaks at the water–W interfaces (Gilbert et al., 2016). By comparison, neglecting self-shielding factors over-predicts Re by only $(n,\alpha)$2–$(n,\alpha)$3, which is smaller than the spatially induced variation (Gilbert et al., 2016).

Temperature broadening is present but weaker. Recomputed Re-vs-depth profiles at $(n,\alpha)$4 and $(n,\alpha)$5 show only a modest upward shift of $(n,\alpha)$6–$(n,\alpha)$7 relative to room temperature, implying that Doppler broadening is subdominant to spatial heterogeneity in these geometries (Gilbert et al., 2016). Experimentally, this matters because homogenized or blanket-averaged spectra can be badly misleading: comparison with HFR irradiation of tungsten at Petten shows that only a detailed geometry-plus-flux model reproduces the measured $(n,\alpha)$8 Re, whereas a blanket spectrum predicts $(n,\alpha)$9 (Gilbert et al., 2016).

The metallurgical consequences extend beyond composition. A first-principles matrix-Hamiltonian and thermodynamic-integration study of W–Re–Os–Vac and related systems shows that vacancies are decorated by Re and Os, but not by Ta, with the distinction traced to the sign of the Warren–Cowley short-range-order parameter $R_i(r)=\int_0^\infty \phi(r,E)\,\sigma_i(E)\,dE,$0 and to vacancy–solute binding energetics (Nguyen-Manh et al., 2021). At $R_i(r)=\int_0^\infty \phi(r,E)\,\sigma_i(E)\,dE,$1 and $R_i(r)=\int_0^\infty \phi(r,E)\,\sigma_i(E)\,dE,$2, Monte Carlo snapshots show Re and especially Os clustering around vacancy clusters, whereas Ta remains randomly mixed (Nguyen-Manh et al., 2021). Under HFR-relevant conditions, the same framework predicts voids containing concentrated Re and Os, consistent with TEM and APT observations of Re “clouds,” Os “rods,” and core–shell clustering in irradiated tungsten (Nguyen-Manh et al., 2021).

4. Blanket-scale objectives: waste burning, isotope production, and elemental conversion

Fusion neutron-driven transmutation has been proposed for four distinct blanket-scale objectives: reducing minor actinides, producing radionuclides, manufacturing stable high-value products, and coupling these outputs to tritium-breeding blanket design (Tanner et al., 2021, Rutkowski et al., 17 Jul 2025, Parisi et al., 4 Nov 2025, Parisi, 18 May 2026).

Objective	Representative system or pathway	Reported result
Minor actinide transmutation	CAIL-driven molten salt transmutator, $R_i(r)=\int_0^\infty \phi(r,E)\,\sigma_i(E)\,dE,$3, $R_i(r)=\int_0^\infty \phi(r,E)\,\sigma_i(E)\,dE,$4	$R_i(r)=\int_0^\infty \phi(r,E)\,\sigma_i(E)\,dE,$5 MA/year; $R_i(r)=\int_0^\infty \phi(r,E)\,\sigma_i(E)\,dE,$6 reduction of Am and Cm over $R_i(r)=\int_0^\infty \phi(r,E)\,\sigma_i(E)\,dE,$7–$R_i(r)=\int_0^\infty \phi(r,E)\,\sigma_i(E)\,dE,$8 years
Gold synthesis	ARC-class tokamak with $R_i(r)=\int_0^\infty \phi(r,E)\,\sigma_i(E)\,dE,$9 Hg/Li channel	$R=\Phi\,\sigma\,N$0 net yield $R=\Phi\,\sigma\,N$1 at $R=\Phi\,\sigma\,N$2; scales to $R=\Phi\,\sigma\,N$3; TBR $R=\Phi\,\sigma\,N$4
Medical isotopes	Stable feedstocks irradiated by D–T neutrons	$R=\Phi\,\sigma\,N$5: $R=\Phi\,\sigma\,N$6, $R=\Phi\,\sigma\,N$7: $R=\Phi\,\sigma\,N$8, $R=\Phi\,\sigma\,N$9: $dC_i/d\Phi=\sigma_iN_W$0 g/(MW$dC_i/d\Phi=\sigma_iN_W$1 yr) for natural/enriched feedstock
Betavoltaic isotopes	$dC_i/d\Phi=\sigma_iN_W$2 blanket	$dC_i/d\Phi=\sigma_iN_W$3 and $dC_i/d\Phi=\sigma_iN_W$4 per GW$dC_i/d\Phi=\sigma_iN_W$5

In molten-salt waste transmutation, the source is a distributed array of beam–target D–T neutron heads combined with a well-mixed FLiBe or FLiNaK core carrying Pu, Am, and Cm fluorides (Tanner et al., 2021). For $dC_i/d\Phi=\sigma_iN_W$6 and $dC_i/d\Phi=\sigma_iN_W$7 $dC_i/d\Phi=\sigma_iN_W$8, the simulated thermal power is $dC_i/d\Phi=\sigma_iN_W$9 and the transmutation rate is $$^{198}\mathrm{Hg}(n,2n){}^{197}\mathrm{Hg}\rightarrow{}^{197}\mathrm{Au}$$0 MA/year in the static phase, while periodic in-line batch processing yields $$^{198}\mathrm{Hg}(n,2n){}^{197}\mathrm{Hg}\rightarrow{}^{197}\mathrm{Au}$$1 reduction of Am and Cm over $$^{198}\mathrm{Hg}(n,2n){}^{197}\mathrm{Hg}\rightarrow{}^{197}\mathrm{Au}$$2–$$^{198}\mathrm{Hg}(n,2n){}^{197}\mathrm{Hg}\rightarrow{}^{197}\mathrm{Au}$$3 years (Tanner et al., 2021).

For mercury-to-gold conversion, a $$^{198}\mathrm{Hg}(n,2n){}^{197}\mathrm{Hg}\rightarrow{}^{197}\mathrm{Au}$$4 ARC-class tokamak with a circulating liquid alloy of $$^{198}\mathrm{Hg}(n,2n){}^{197}\mathrm{Hg}\rightarrow{}^{197}\mathrm{Au}$$5 Hg $$^{198}\mathrm{Hg}(n,2n){}^{197}\mathrm{Hg}\rightarrow{}^{197}\mathrm{Au}$$6 and $$^{198}\mathrm{Hg}(n,2n){}^{197}\mathrm{Hg}\rightarrow{}^{197}\mathrm{Au}$$7 Li $$^{198}\mathrm{Hg}(n,2n){}^{197}\mathrm{Hg}\rightarrow{}^{197}\mathrm{Au}$$8 in a $$^{198}\mathrm{Hg}(n,2n){}^{197}\mathrm{Hg}\rightarrow{}^{197}\mathrm{Au}$$9 multiplier channel produces $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$00 at $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$01, while preserving a simulated total TBR of $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$02, with $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$03 from the Hg/Li channel and $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$04 from FLiBe (Rutkowski et al., 17 Jul 2025). Because $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$05 on Hg contributes neutron multiplication, the study reports essentially zero penalty on net thermal power and continuous delivery of $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$06 for a $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$07-efficient $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$08 plant (Rutkowski et al., 17 Jul 2025). A related remediation study extends the concept to successive $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$09 steps on all stable mercury isotopes until $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$10 or $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$11, emphasizing that at sufficiently high flux all stable Hg isotopes become eligible for conversion to gold (Parisi et al., 3 Apr 2026).

For radionuclides, high-energy fusion neutrons enable both same-element and proton-changing routes. OpenMC slab studies at flux $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$12 and $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$13 thickness report production rates in g/(MW$$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$14 yr) for multiple medical isotopes, including $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$15, $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$16, $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$17, $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$18, $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$19, and $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$20 (Parisi et al., 4 Nov 2025). A complementary economic study gives a specific low-gain example: a $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$21 fusion system transmuting $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$22 at $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$23 and $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$24 yields $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$25 and has $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$26 for $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$27 at $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$28 (Parisi et al., 10 Dec 2025).

For $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$29 emitters, blanket calculations identify $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$30, $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$31, $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$32, and $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$33 as prominent products (Parisi, 18 May 2026). In a tokamak blanket using enriched $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$34 with $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$35, OpenMC simulations show $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$36 production above $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$37 per GW$$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$38 with TBR $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$39, and the tabulated value for a fully enriched $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$40 case is $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$41, equivalent to $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$42 (Parisi, 18 May 2026).

5. Source concepts, control architectures, and flux shaping

The transmutation problem depends strongly on the neutron-source concept. For materials qualification and fusion-prototypical irradiation, the IFMIF-Lite compact-cyclotron concept accelerates $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$43 of deuterons to $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$44 onto a $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$45 flowing lithium jet, producing a D–Li neutron spectrum peaked near $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$46 with a representative yield of $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$47 and flux $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$48 at $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$49 (Snead et al., 2023). In that framework, specimen positions $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$50–$$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$51 downstream of the Li jet define a $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$52 irradiation cavity, and multi-beam incidence is used to reduce the flux gradient across the test volume to below $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$53 (Snead et al., 2023).

Distributed neutron production has been proposed for subcritical molten-salt transmutators. In the CAIL concept, a high-repetition-rate array of coherently combined femtosecond fiber lasers irradiates nanometric deuterated foils, producing deuterons that strike a tritium-loaded catcher; the quoted scaling is $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$54 at $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$55, or $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$56 per laser head, scalable to $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$57 heads and $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$58–$$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$59 (Tanner et al., 2021). Source placement is optimized through a phyllotactic pattern, while real-time spectroscopy and an Evolutionary Algorithm coupled with a Neural Network tune individual source powers, processing schedules, and salt flow-rates between sub-volumes (Tanner et al., 2021). After optimization, Monte Carlo tallies show fast-spectrum flux uniformity within $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$60 (Tanner et al., 2021).

Muon-catalyzed fusion and short-pulse laser sources extend the concept into high-flux or ultrashort-pulse regimes. In $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$61CF, the absence of external plasma heating reduces wall loading to the $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$62-heating component alone, leading to an estimated allowable first-wall neutron flux of $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$63 for a $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$64 heat-flux limit (Parisi et al., 26 Nov 2025). An example with $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$65 muons/s, $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$66, and a $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$67 $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$68 feedstock yields up to $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$69 of $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$70 per year (Parisi et al., 26 Nov 2025). Short-pulse DD fusion, although not a $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$71 D–T source, provides quasi-monoenergetic $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$72 neutrons with $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$73 pulse duration, $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$74 source size, and local peak flux $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$75 in $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$76 bins, while LWFA-driven photonuclear sources trade lower peak brightness for higher repetition rates and a $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$77 advantage in cumulative capture events for short-lived isomers (Labun et al., 18 May 2026). This suggests a division between blanket-scale continuous transmutation and pulsed fast-neutron transmutation experiments.

6. Materials consequences, qualification metrics, and unresolved constraints

Fusion transmutation cannot be separated from displacement damage and gas production. In DEMO first-wall conditions, integrated MCNP–FISPACT modeling gives for pure Fe at the outboard equatorial armor a helium production rate of $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$78, $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$79, and $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$80, whereas tungsten in first-wall or divertor armor shows $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$81–$$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$82 He and $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$83–$$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$84 (Gilbert et al., 2013). Using DFT-based grain-boundary helium insertion energies, the same study estimates $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$85 for Fe under first-wall conditions, but $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$86 for W and Ta (Gilbert et al., 2013).

At the atomistic level, transmutation products feed back on tungsten behavior. First-principles calculations for W–Re, W–Os, and W–Ta alloys report that all transmutation elements lower the Pugh ratio $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$87, with pure W at $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$88 and values around $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$89–$$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$90 at $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$91, implying increased ductility under the assumptions of the calculation (Zhao et al., 2017). Re and Os reduce the formation energy of H and He in W and attract He at substitutional sites, whereas Ta repels He; transmutation elements also alter migration barriers and can increase the diffusion rate of H and He in W (Zhao et al., 2017). Coupled with the vacancy-decoration results for Re and Os, this links transmutation chemistry directly to swelling, embrittlement, and bubble evolution (Nguyen-Manh et al., 2021, Zhao et al., 2017).

Several engineering constraints recur across blanket proposals. One is nuclear-data uncertainty: a $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$92 neutron-flux uncertainty or $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$93 cross-section error produces linear variation in the predicted Au yield in the Hg blanket study, and a separate $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$94-emitter analysis notes $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$95–$$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$96 uncertainties in $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$97 cross sections from ENDF/B-VIII.0 (Rutkowski et al., 17 Jul 2025, Parisi, 18 May 2026). Another is chemical and structural compatibility. Mercury has higher vapor pressure than Pb or Li, must satisfy strict containment because of toxicity and Minamata-Convention controls, and requires qualification of W, V-alloys, and steel in contact with Hg and Li (Rutkowski et al., 17 Jul 2025). The same study identifies $$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$98 ($$d+t\rightarrow \alpha(3.5\ \mathrm{MeV})+n(14.1\ \mathrm{MeV})$$99, $\beta^{-}$00) as the longest-lived radionuclide in the activation inventory, with the Au product cooling to $\beta^{-}$01Class A waste in $\beta^{-}$02 and to background-equivalent levels in $\beta^{-}$03 (Rutkowski et al., 17 Jul 2025).

Qualification sources introduce their own caveats. IFMIF-Lite’s D–Li neutron spectrum is slightly softer below $\beta^{-}$04 than pure D–T, so spectral weighting functions are required when equating damage and transmutation rates to reactor conditions (Snead et al., 2023). For high-activity isotope blankets, chemical separation in in-blanket streams must accommodate dose rates on the order of $\beta^{-}$05–$\beta^{-}$06, while tritium handling, permeation control, corrosion, swelling, and activation remain non-trivial (Parisi, 18 May 2026).

A persistent misconception is that average or homogenized neutron spectra are adequate for design. The tungsten literature shows the opposite: fine spatial heterogeneity in moderators adjacent to W armor can produce local Re variations of $\beta^{-}$07–$\beta^{-}$08, and only detailed geometry-plus-flux models reproduce experimental transmutation levels (Gilbert et al., 2016). A plausible implication is that fusion neutron-driven transmutation should be treated not as a single blanket-average quantity, but as a multiscale problem coupling local spectral shaping, resonance physics, self-shielding, isotope extraction, and defect-mediated microstructural evolution.