Closing the Tritium Fuel Cycle

Abstract

Tritium is fusion's smallest mass flow and one of its hardest infrastructure constraints. A gigawatt-scale fusion source burns only a fraction of a kilogram per day, yet a plant can hold kilograms across breeder, coolant, extraction, separation, storage, injection, vacuum pumps, walls, dust, cleanup systems, and waste. Tritium decays, permeates hot metal, exchanges with water, hides on surfaces, and has limited external supply. I organize the fuel cycle around low residence time, small isolatable inventories, continuous assay, diverse confinement, and rapid recycle. The central metric is not nominal breeding ratio. It is the measured surplus available after burn, decay, processing loss, uncertainty, reserve, and fleet growth. A first plant that cannot explain each gram cannot establish scalable fusion.

A small burn rate can require a large working inventory

A $1\ \mathrm{GW}$ deuterium-tritium source burns about $77\ \mathrm{g}$ of tritium per day. The exact value follows reaction energy and isotope mass. This rate seems modest until the full loop is considered.

If a subsystem processes tritium at mass flow $\dot m$ with mean residence time $\tau$, its approximate inventory is

$$M=\dot m\tau.$$

The circulating flow can exceed the burn rate because most injected fuel leaves unburned. A large throughput multiplied by hours or days in pumps, separation columns, beds, tanks, blanket material, or coolant can create kilogram-scale holdup.

The loop contains distinct chemical jobs

The blanket produces tritium within ceramic or liquid breeder. Extraction moves it into a controlled process stream. Cleanup removes water, methane, ammonia, noble gases, dust, corrosion products, and seeded plasma impurities. Isotope separation divides protium, deuterium, and tritium. Storage buffers inventory. Metering and fueling return selected mixtures to the plasma. Exhaust processing closes the fast loop.

Detritiation systems recover tritium from room air, water, coolant, gloveboxes, and waste streams. These paths are not secondary. A small continuous permeation flow can accumulate significant inventory or release over a plant year.

Fuel-cycle segment	Desired output	Hidden inventory mechanism
Blanket extraction	Concentrated tritium-bearing stream	Diffusion, solubility, stagnant breeder zones
Coolant cleanup	Low coolant activity	Permeation, oxide films, corrosion traps
Exhaust processing	Recyclable hydrogen isotopes	Cryopump loading and regeneration queue
Isotope separation	Qualified D-T composition	Column or bed holdup
Storage and delivery	Metered fuel on demand	Reserve vessels and transfer-line volume
Detritiation and waste	Bounded releasable inventory	Water, molecular sieves, filters, discarded components

Decay sets a clock

Tritium half-life is about $12.3$ years. Its decay constant is

$$\lambda=\frac{\ln2}{T_{1/2}},$$

and inventory follows

$$N(t)=N_0e^{-\lambda t}.$$

Decay removes roughly $5.5\%$ of an isolated inventory each year. The daughter product is helium-3, which can alter gas purity and must be managed. Long construction delays and strategic storage therefore consume scarce startup fuel.

The decay heat per kilogram is modest relative to plant thermal power but relevant to concentrated storage and measurement. More important is that decay is an irreversible loss from the fuel ledger.

Permeation creates paths without leaks

Hydrogen isotopes dissolve and diffuse through metals, especially at high temperature. Tritium can cross blanket tubes into coolant, cross heat exchangers into a secondary loop, or enter building water without a pressure-boundary defect.

A simplified permeation flux follows

$$J\propto\frac{P}{\ell}\left(\sqrt{p_1}-\sqrt{p_2}\right),$$

where $P$ is permeability, $\ell$ thickness, and $p_1,p_2$ partial pressures. Temperature strongly affects $P$.

Barriers, low partial pressure, purge gas, cold traps, getters, cleanup, and intermediate loops reduce transfer. Coatings must remain effective after thermal cycling, irradiation, corrosion, and defects. The plant monitors both intended process flow and permeated flow.

Isotope separation is central machinery

Exhaust contains unburned deuterium and tritium plus protium and impurities. Cryogenic distillation is one established route for hydrogen-isotope separation. Other catalytic exchange, membrane, and getter processes address water and mixed chemical forms.

Separation equipment needs low inventory, high throughput, remote maintenance, criticality-independent but radiologically controlled design, and safe response to loss of cooling or power. Parallel trains permit regeneration and maintenance. Samples and online analyzers verify product composition before reinjection.

The process should recycle fuel rapidly enough that large storage buffers are unnecessary. Storage remains for startup, shutdown, assay, upset recovery, and emergency transfer.

Accountancy needs uncertainty bounds

I write the plant balance over interval $\Delta t$ as

$$\Delta M_{\mathrm{book}} =M_{\mathrm{bred}}+M_{\mathrm{received}}-M_{\mathrm{burned}}-M_{\mathrm{shipped}}-M_{\mathrm{decayed}}-M_{\mathrm{released}}.$$

Physical measurements of inventory change should agree within propagated uncertainty. Flow meters, calorimetry, pressure-volume-temperature methods, mass spectrometry, beta measurement, sampling, and component assay cover different forms and ranges.

A discrepancy threshold triggers investigation. The threshold is not allowed to grow merely because process equipment is difficult to measure. Instead, vessels are designed with assay geometry, calibration, drainability, and known dead volume.

Material unaccounted for can represent measurement error, surface retention, an unmodeled chemical reaction, leakage, or transfer into waste. Each possibility has operational and safety significance.

Fleet growth creates a dynamic constraint

A breeding plant may eventually produce surplus tritium, but the first plant needs startup inventory before its blanket operates. It also needs inventory during commissioning when burn and breeding performance are uncertain. A second plant may begin before the first has accumulated enough surplus.

For plant $i$, fleet inventory evolves as

$$\frac{dM_i}{dt}=B_i-C_i-L_i-\lambda M_i-S_i,$$

where $B_i$ is recovered breeding, $C_i$ consumption, $L_i$ process loss, and $S_i$ transfer to new plants. Growth is possible only when the fleet sum remains above operating and safety reserves.

This favors low process inventory, short commissioning, high blanket availability, and conservative build schedules. A concept that needs a large initial charge may scale slowly even if its eventual TBR is high.

Water deserves special control

Tritiated water is biologically mobile and can spread through ordinary drainage, humidity, concrete, clothing, and cooling systems. The plant separates tritiated and clean water systems, minimizes open water near fuel processing, controls condensation, and monitors discharge paths.

Water detritiation can use catalytic exchange and electrolysis or other isotope-transfer methods. The process is energy-intensive, so preventing dilution into large water volumes is preferable to recovering it later.

CreateBreed and extract rapidly

Limit residence in blanket, coolant, structures, and stagnant process zones.

PurifySeparate isotopes and impurities

Operate parallel low-inventory trains with verified output composition.

UseMeter, inject, exhaust, and recycle

Close the fast loop while measuring wall and pump retention.

GuardConfine, assay, and recover losses

Control permeation, air, water, waste, reserve, and fleet transfer.

Demonstration requires an end-to-end run

Individual pumps, columns, beds, membranes, and monitors can work while the integrated loop accumulates hidden inventory or creates bottlenecks. A full demonstration must operate with representative throughput, mixed exhaust, blanket extraction, regeneration, maintenance, outage, and reconciliation.

BaselineCalibrate every process inventory CirculateRun breeding and exhaust loops together DisturbInject faults, outages, and impurities ReconcileRecover and explain total mass

When the fuel loop, maintenance system, and safety case are physical, the plant can finally be evaluated as an economic asset. The next report treats cost and availability as consequences of component life, outage operations, internal power, construction duration, financing, and fleet learning.