Vacuum as a Working Medium

Abstract

Vacuum is active infrastructure in a fusion plant. The torus vacuum allows a controlled plasma to form without overwhelming neutral contamination. The cryostat vacuum limits heat transfer into superconducting magnets. Neutral-beam lines require their own pumping environments, and the exhaust train turns a dilute, contaminated gas stream into accounted fuel and waste. I develop these regions as separate but coupled pressure domains. Gas load, effective pumping speed, conductance, outgassing, permeation, dust, virtual leaks, and maintenance determine recovery time and purity. Because access becomes restricted after deuterium-tritium operation, the architecture emphasizes sector isolation, diverse leak localization, remotely replaceable pumps, controlled regeneration, and a complete gas inventory.

A plasma begins by removing almost everything

Before useful fuel enters the chamber, air, water vapor, hydrocarbons, cleaning residues, and manufacturing gases must be reduced. Otherwise energetic electrons break those molecules apart and deliver oxygen, carbon, and other impurities directly to the plasma.

The gas balance of a vacuum volume can be approximated by

$$V\frac{dp}{dt}=Q_{\mathrm{in}}+Q_{\mathrm{outgas}}+Q_{\mathrm{leak}}-S_{\mathrm{eff}}p,$$

where $V$ is volume, $Q$ terms are gas throughputs, and $S_{\mathrm{eff}}$ is effective pumping speed at the vessel. At equilibrium,

$$p=\frac{Q_{\mathrm{total}}}{S_{\mathrm{eff}}}.$$

Large nominal pump speed does not guarantee low vessel pressure. Narrow ducts reduce conductance. Surfaces outgas. Elastomers permeate. Trapped volumes become virtual leaks. Plasma-facing materials retain and later release hydrogen isotopes.

ITER’s vacuum vessel is about $1{,}400\ \mathrm{m^3}$ and its cryostat about $8{,}500\ \mathrm{m^3}$. Its vacuum system includes hundreds of pumps and kilometers of lines. Those figures make the central lesson physical: reactor vacuum is a distributed plant, not a single pump attached to a chamber.

Four vacuum domains need different rules

Domain	Purpose	Dominant concern
Torus	Form plasma and receive fuel and exhaust	Purity, throughput, dust, and tritium inventory
Cryostat	Thermally insulate magnets and cold structures	Large-volume leak detection and cryogenic heat load
Heating lines	Permit neutral beams and radiofrequency windows to operate	Differential pumping and component protection
Fuel process	Move and separate hydrogen isotopes and impurities	Confinement, accountancy, detritiation, and regeneration

These domains should not share failures casually. A water leak into the torus, an air leak into the cryostat, and a tritium release in a process enclosure demand different isolation and recovery sequences. Pressure boundaries, valves, relief paths, and instrumentation must show the direction in which material can move under every credible pressure difference.

Conductance determines usable pumping speed

A pump connected through a duct sees an effective speed

$$\frac{1}{S_{\mathrm{eff}}}=\frac{1}{S_{\mathrm{pump}}}+\frac{1}{C},$$

where $C$ is duct conductance. If a $100\ \mathrm{m^3/s}$ pump is connected through a path with $C=10\ \mathrm{m^3/s}$, effective speed is only about $9.1\ \mathrm{m^3/s}$.

Blanket modules, divertor cassettes, ports, neutron shielding, and maintenance features compete for the same geometry that pumping requires. Exhaust design has to reserve conductance early. Adding a larger pump later cannot repair a throttled path.

The divertor deliberately channels particles toward pumping openings. Geometry, neutral pressure, surface temperature, and plasma detachment govern how much helium and unburned fuel reach the pumps. The vacuum system is therefore coupled to edge-plasma design.

Surfaces are temporary gas reservoirs

Stainless steel, tungsten, coatings, dust, and redeposited layers absorb, adsorb, trap, and release gas. Water dominates many initial pumpdowns. Baking raises surface temperature to accelerate desorption. Glow discharge or other conditioning methods modify the wall before plasma operation.

Hydrogen isotopes introduce a nuclear inventory question. Deuterium and tritium can implant, diffuse, permeate, and co-deposit. A material with acceptable mechanical behavior may retain too much tritium or release it unpredictably during heating. Temperature histories and plasma exposure matter to accountancy.

I treat every plasma-facing surface as part of the fuel cycle. The plant model includes injected fuel, pumped exhaust, measured vessel inventory, component retention, processing holdup, storage, decay, and uncertainty.

Pumps work in trains

Roughing pumps reduce pressure from atmosphere. High-vacuum pumps then reach operating conditions. Cryopumps provide high speed by capturing gases on cold surfaces, but they have finite capacity and require regeneration. Regeneration releases a concentrated gas inventory that must flow into a controlled processing system.

A power plant should permit one train to regenerate or undergo maintenance while another supports operation or safe shutdown. Isolation valves need position confirmation, leak-tight closure, and protection from dust and thermal distortion. Backstreaming and accidental cross-connection are analyzed as contamination events.

The exhaust contains deuterium, tritium, helium, protium, seeded radiators, water, hydrocarbons, nitrogen, oxygen, dust, and corrosion or activation products. No one separation step handles the mixture. The vacuum train delivers a bounded stream to isotope separation, detritiation, impurity removal, waste, and reinjection systems.

Leakage must be found, not merely detected

Pressure rise confirms that a vacuum region is gaining gas but may not locate the source. Helium mass-spectrometer tests, residual-gas analysis, sectional isolation, acoustic sensing, tracer methods, temperature response, and local pressure instruments provide different evidence.

After activation, inspection access may be robotic and time-limited. The plant therefore includes permanent segmentation. Each sector can be isolated and tested. Double barriers with interspaces can reveal which side is leaking. Cooling circuits near vacuum carry chemistry and pressure monitoring that help distinguish water ingress from air or fuel release.

A leak budget allocates allowable throughput among welds, seals, penetrations, valves, windows, and diagnostics. The budget includes degradation and measurement uncertainty. A component that passes its factory limit may still leave no system margin when thousands are assembled.

ContainSeparate pressure domains

Define barriers, differential pressures, and relief direction for each region.

MovePreserve conductance to pumps

Reserve exhaust geometry before blankets, ports, and shielding consume it.

KnowLocalize gas and inventory

Combine sector isolation, composition sensing, and material accountancy.

RecoverRegenerate and restore remotely

Process captured gas and replace failed train components without exposure.

Clean assembly is part of operation

Vacuum quality begins during manufacturing. Machining fluids, marker ink, fingerprints, trapped cleaning agents, porous deposits, and inaccessible fasteners can become long-lived sources. Materials and processes need vacuum compatibility, radiological compatibility, and a qualified cleaning route.

Subassemblies receive leak tests before later layers hide their joints. Assembly records preserve test gas, sensitivity, temperature, acceptance limit, and exact configuration. Temporary closures and purge states are controlled like flight hardware.

Recovery time belongs in the plant schedule

A plant does not earn availability merely by avoiding catastrophic leaks. It must pump down after planned maintenance, condition surfaces, regenerate pumps, verify purity, and return to plasma operation within the outage window.

CloseVerify boundaries and configuration EvacuateRough, bake, and condition surfaces OperateFuel, exhaust, and monitor inventory RegenerateRecover gas and restore capacity

The vacuum system establishes the clean, low-pressure canvas on which fueling acts. The next design problem is to place deuterium and tritium where the plasma needs them without surrendering inventory control or relying on edge injection that never reaches the core.