Why borrow a chip-making process for the Moon?
The Moon is not short of oxygen. It is short of cheap ways to break silicon–oxygen and metal–oxygen bonds without building a high-temperature foundry in vacuum.
As examined in our baseline lunar ISRU mass analysis, oxygen accounts for roughly 40–45% of lunar regolith by weight, bound inside silicates and oxides such as anorthite (CaAl₂Si₂O₈), pyroxene, olivine and ilmenite (FeTiO₃). Conventional dry-regolith architectures rely on extreme heat to overcome those bond energies: hydrogen reduction operates around 900°C–1,050°C but works mainly on iron-bearing ilmenite; molten salt electrolysis runs near 900°C–950°C in liquid calcium chloride; and carbothermal or direct molten regolith electrolysis requires 1,600°C–1,850°C.
On 20 November 2026, Rajani K. Vijayaraghavan (Dublin City University) and Aidan Cowley (European Space Agency) outlined in The Conversation a different industrial parallel. In semiconductor fabrication, engineers do not melt silicon wafers to remove silicon dioxide (SiO₂) insulating layers. Instead, they place the wafer in a low-pressure radio-frequency (RF) chamber filled with fluorinated gas. Energetic plasma electrons dissociate the gas into highly reactive fluorine radicals that break Si–O bonds while the bulk wafer remains near room temperature or a few hundred degrees Celsius.
Working across Dublin City University (Rajani K. Vijayaraghavan, Aneeta Jaggernauth, Seán Kelly, Niall McGearalt, Patrick McNally and Robert O’Connor), Durham University (Brian K. Tanner) and the European Space Agency (Aidan Cowley and Beth Lomax), the team tested whether standard cleanroom etch gases—sulfur hexafluoride (SF₆) and octafluorocyclobutane (C₄F₈)—can strip oxygen directly from lunar regolith simulants at temperatures below 400°C.
The engineering question
Does lowering the reactor temperature from ~1,600°C to <400°C via fluorine plasma reduce total landed system mass, or does the mass of fluorine makeup and regeneration equipment cancel the thermal savings?
How fluorinated plasma breaks silicate bonds below 400°C
In thermal reduction, the entire mineral mass—and the refractory vessel holding it—must be heated until chemical equilibrium shifts toward oxygen release. In a non-thermal or low-temperature RF plasma, energy is delivered selectively to free electrons rather than the bulk rock lattice.
When an RF electric field energises a low-pressure fluorinated gas such as SF₆ or C₄F₈, electron-impact dissociation splits the stable parent molecules into atomic fluorine radicals (F•),SFₓ fragments, or fluorocarbon radicals (CF₂•, CF₃•) alongside energetic ions. Fluorine is the most electronegative element on the periodic table; a free fluorine radical has a stronger affinity for silicon and metallic cations than oxygen does.
Idealised silica etching reaction: SiO₂(s) + 4F• → SiF₄(g) + O₂(g)
In semiconductor reactive-ion etching (RIE), two mechanisms work together: chemical attack by neutral fluorine radicals and physical bombardment by ions accelerated across the plasma sheath. When applied to silicate minerals in lunar soil:
- Silicon-oxygen cleavage: Atomic fluorine attacks Si–O bonds at the grain surface, converting solid silica units into volatile silicon tetrafluoride gas (SiF₄, sublimation point −95.5°C at 1 atm) and liberating oxygen.
- Carbon-assisted reduction in C₄F₈: When fluorocarbon plasmas (C₄F₈) are used, carbon-fluorine radicals (CFₓ) also scavenge oxygen atoms from the mineral lattice to evolve gaseous carbon monoxide (CO) and carbon dioxide (CO₂) alongside O₂.
- Moderate thermal assistance: While the plasma itself is non-equilibrium, raising the substrate temperature from 200°C to 400°C accelerates surface desorption and diffusion, increasing oxygen evolution rates without approaching the 1,100°C+ sintering and melting threshold of lunar basalt.
What the DCU, Durham and ESA experiments actually observed
A wafer in a semiconductor fab is a flat, ultra-pure single crystal of silicon or a uniform film of SiO₂. Lunar regolith is a jagged, multi-phase powder containing plagioclase feldspar (anorthosite), pyroxenes, olivine, ilmenite and impact glass. The peer-reviewed diagnostic studies behind the November 2026 article—published in Scientific Reports (Jaggernauth et al., 2026) and Planetary and Space Science (Cowley et al., 2026)—used X-ray energy dispersive spectroscopy (EDS), X-ray diffraction (XRD) and X-ray photoelectron spectroscopy (XPS) on Lunar Mare Simulant (LMS) to see what actually happens at the grain scale.
| Plasma gas | Gases detected | Mineral phase selectivity (EDS / XRD) | Grain-surface consequence |
|---|---|---|---|
| SF₆ (sulfur hexafluoride) | Measurable molecular O₂ | Reacts almost exclusively with the anorthosite (calcium-aluminium silicate) component of the simulant | Localised loss of Si Kα and O Kα EDS signal; forms a surface fluoride layer causing XRD strain-broadening |
| C₄F₈ (octafluorocyclobutane) | O₂ alongside CO and CO₂ | Broader multi-phase reaction; does not show exclusive anorthosite isolation | Fluorocarbon radical etching without the same localised anorthosite XRD peak strain-broadening |
Why the surface fluoride shell is the critical finding
When pure SiO₂ is etched on a silicon wafer, the reaction product (SiF₄) is a gas that pumps away, continuously exposing fresh SiO₂ underneath. Anorthosite in lunar regolith (CaAl₂Si₂O₈), however, contains calcium and aluminium alongside silicon.
While silicon leaves as volatile SiF₄ and oxygen leaves as O₂, calcium and aluminium fluorides (CaF₂ and AlF₃) are non-volatile solids at 400°C. As Jaggernauth et al. documented via XRD peak strain-broadening and XPS, a solid fluoride layer forms on the exterior of the mineral grains. Once that fluoride crust thickens, it acts as a diffusion barrier that slows further radical penetration into the core of the grain—explaining why static powder yields plateau and why different minerals react at markedly different rates.
How plasma processing compares with thermal and molten-salt routes
Every lunar oxygen architecture chooses where to pay its penalty: in temperature, in feedstock selectivity, or in chemical reagent recovery. Placing the DCU/ESA fluorinated plasma route alongside the four established dry-regolith benchmarks makes the trade-off explicit.
| Process route | Operating temp | Feedstock & working reagent | Core mechanism | Primary systems bottleneck |
|---|---|---|---|---|
| Fluorinated plasma etching (SF₆ / C₄F₈) | Below 400°C (bulk substrate) | Bulk lunar regolith + fluorine-bearing gas (SF₆ or C₄F₈) | RF plasma generates fluorine radicals that break Si–O and metal-oxide bonds at grain surfaces | Surface fluoride layer passivates grains; fluorine must be recycled >90% from SiF₄ and metal fluorides |
| Hydrogen reduction of ilmenite | 800°C – 1,050°C | Ilmenite-rich (FeTiO₃) feedstock + recycled H₂ gas | Hydrogen reduces iron oxide in ilmenite to metallic iron and water vapour, followed by electrolysis | Low total oxygen yield (typically 1–4% of bulk regolith mass); restricted to titanium-rich mare sites |
| Carbothermal reduction | 1,600°C – 1,850°C | Unbeneficiated silicate regolith + recycled methane / carbon catalyst | Carbon reduces molten silicates to CO / CO₂, which is converted to water and electrolysed | Extreme refractory thermal losses, slag handling, and carbon loss locked in metallic carbides |
| Molten salt electrolysis (FFC Cambridge) | 900°C – 950°C | Solid sintered regolith cathode in molten CaCl₂ electrolyte | Electrochemical de-oxidation in molten salt bath; oxygen evolves at an inert anode | Anode erosion, molten salt volatilisation and freeze/thaw recovery, batch salt makeup mass |
| Molten regolith electrolysis (MRE) | 1,600°C+ | Direct molten regolith (no added chemical flux required) | Direct electrolysis of molten silicate melt at extreme temperature | Containment material survival, high radiative power draw, and inert anode lifetime in molten slag |
The comparison shows why plasma etching is attractive for early or low-mass payloads: operating below 400°C eliminates refractory ceramic linings, liquid-metal corrosion, and the radiator or insulation penalties of holding a pressure vessel at 1,600°C–1,800°C during lunar operations. Furthermore, as a dry gas-solid process, it avoids handling molten CaCl₂ salt baths in one-sixth gravity.
The halogen mass trap: why fluorine recycling decides viability
Lunar regolith contains very little native fluorine. Any fluorine consumed by the reactor—or locked permanently into discarded CaF₂ and AlF₃ tailings—must be launched from Earth.
We can quantify the recycling threshold directly from charge balance. In every lunar oxide—whether silica (SiO₂), calcium oxide, alumina, or anorthite (CaAl₂Si₂O₈)—displacing one mole of dioxygen (32.00 g O₂, carrying four units of negative charge as two O²⁻ ions) requires exactly four moles of atomic fluorine (76.00 g F, as four F⁻ ions).
That sets a hard stoichiometric floor of 2.37 kg of active fluorine cycled for every 1.0 kg of oxygen liberated. For our standard 1,000 kg oxygen reference batch, the reactor must cycle at least 2,375 kg of atomic fluorine—equivalent to 3,043 kg of SF₆ parent gas before counting pressure vessels or unreacted gas losses. Calcium and aluminium oxides do not raise that stoichiometric ratio, but they make closing the loop much harder: their fluoride products (CaF₂ and AlF₃) remain locked as non-volatile solids on the spent grains instead of pumping out as gas like SiF₄.
| Fluorine loop closure | Recovery rate | Earth F makeup for 1,000 kg O₂ | O₂ produced per kg of lost F | Mass-payback verdict |
|---|---|---|---|---|
| Open loop (no recycling) | 0% | 2,375 kg F | 0.4 kg O₂ / kg F | Net mass loss: launching 1,000 kg O₂ directly from Earth requires less mass than launching 2,375 kg of fluorine. |
| Early bench loop | 80% | 475 kg F | 2.1 kg O₂ / kg F | Marginal: reagent loss consumes nearly half of the oxygen mass savings before counting hardware. |
| Industrial loop (90%) | 90% | 237 kg F | 4.2 kg O₂ / kg F | Viable pilot regime: reagent makeup drops to 119–238 kg per tonne of O₂. |
| Closed loop target (95%) | 95% | 119 kg F | 8.4 kg O₂ / kg F | Viable pilot regime: reagent makeup drops to 119–238 kg per tonne of O₂. |
| High-efficiency loop (99%) | 99% | 24 kg F | 42.1 kg O₂ / kg F | Target production regime: reagent makeup is minor compared with hardware and spares mass. |
The table exposes the first law of halogen ISRU: an open-loop fluorine plasma etcher cannot save launch mass. Without recycling, liberating 1,000 kg of lunar oxygen consumes at least 2,375 kg of fluorine reagent—more than double the mass of simply bringing liquid oxygen from Earth. Even at 80% recycling, 475 kg of fluorine is lost per 1,000 kg of oxygen produced.
To outperform Earth resupply once reactor mass, power systems and spares are included, a lunar fluorine-plasma plant must recover and recycle at least 90% to 99% of its fluorine. That requires two downstream chemical stages:
- Gas-phase separation and SiF₄ reduction: Separating O₂ (and any CO / CO₂) from volatile SiF₄ and unreacted SF₆ / C₄F₈, then cracking SiF₄ via secondary plasma or electrolysis to deposit elemental silicon and return fluorine to the plasma feed.
- Solid-tailings defluorination: Stripping fluorine back out of the solid CaF₂ and AlF₃ crust left on the spent anorthosite grains before dumping the tailings back onto the lunar surface.
Three engineering gates between a wafer chamber and a lunar mine
The DCU, Durham and ESA researchers explicitly note that their experiments establish proof of concept and that yields must rise substantially before any lunar deployment. For an engineering team evaluating whether to fund or build a plasma ISRU prototype, three hardware gates separate the current bench test from a flight payload:
1. Continuous grain agitation and shell removal
A plasma sheath penetrates only the exposed surface of a powder bed, and the resulting CaF₂ / AlF₃ reaction layer passivates each anorthosite grain within nanometres to micrometres. A scalable reactor cannot treat a static tray of dust. It will need a rotary drum, fluidised bed, or acoustic/mechanical tumbler operating inside the low-pressure RF cavity that continuously turns the powder and abrades the fluoride crust off the grains without blinding the vacuum pumps with abrasive lunar fines.
2. Specific energy (kWh per kg of stored O₂) vs thermal routes
Running at a bulk temperature below 400°C lowers conductive and radiative heat loss, which is a genuine advantage over a 1,800°C carbothermal vessel. However, RF plasma generation is electrically intensive: much of the input RF power goes into gas ionisation, wall recombination, and vacuum pumping rather than productive bond cleavage. Until a bench test reports kilowatt-hours of total electrical input per kilogram of separated, purified oxygen, low bulk temperature should not be equated with low total power demand.
3. Vacuum lock seals and halogen corrosion
Moving dry, electrostatic lunar regolith from the exterior vacuum into a pressurised (millitorr to Torr) plasma chamber and discharging spent tailings requires valves that seal repeatedly against abrasive silicate dust. When the working gas contains atomic fluorine radicals, elastomer seals degrade rapidly unless shielded or replaced with metal-to-metal or perfluoroelastomer stages. Valve lifetime and fluorine leak rate per batch cycle should be measured early.
Sources and assumptions
This analysis was compiled and verified on 21 November 2026 following the 20 November 2026 publication by Rajani K. Vijayaraghavan and Aidan Cowley in The Conversation and its underlying 2026 peer-reviewed papers in Scientific Reports and Planetary and Space Science. The halogen mass-balance table is an original Lyon Industries stoichiometric calculation derived from oxide charge balance to establish minimum fluorine recycling thresholds.
Model inputs
| Input | Value | Basis and limit |
|---|---|---|
| Oxygen reference target | 1,000 kg O₂ | Matches the Lyon Industries lunar ISRU baseline for propellant and life-support comparison. |
| Reference stoichiometry | SiO₂ + 2F₂ → SiF₄ + O₂ | Charge-balance floor across lunar oxides (SiO₂, CaO, Al₂O₃, CaAl₂Si₂O₈): 4 moles of F⁻ (76.0 g) per mole of O₂ (32.0 g). |
| Gross fluorine throughput | 2.38 kg F per 1 kg O₂ | Calculated from molar masses (75.992 / 31.998 = 2.375); supplied as SF₆ parent gas (78.0% F by mass), this equals 3.04 kg SF₆ per kg O₂ before tankage. |
| Plasma bulk temperature | 200°C – 400°C | Reported low-temperature plasma processing regime from Jaggernauth et al. and Cowley et al. (2026), compared with 900°C–1,800°C thermal routes. |
| Reactive plasma species | SF₆ and C₄F₈ | Standard semiconductor dry-etch gases tested on Lunar Mare Simulant (LMS). |
| Mineral selectivity | Anorthosite preferential etching | EDS and XRD results (Jaggernauth et al., 2026) showing SF₆ reacts almost exclusively with anorthosite grains, forming a strained surface fluoride layer. |
Source trail
- [1] The Conversation: A new way to extract oxygen from lunar soil could support bases on the Moon (20 November 2026)
- Location
- Public research brief by Rajani K. Vijayaraghavan (Dublin City University) and Aidan Cowley (European Space Agency).
- Supports
- Adaptation of semiconductor plasma etching to lunar regolith simulants, lower operating temperatures, differential mineral reaction rates, and current yield limitations.
- Limit
- Science communication summary of laboratory simulant trials; does not specify a flight-sized closed-loop fluorine recovery plant.
- [2] Scientific Reports: Impact of oxygen extraction from lunar regolith simulant using fluorinated plasma processes (2026)
- Location
- Jaggernauth, A., Kelly, S., Tanner, B. K., O’Connor, R., Cowley, A., & Vijayaraghavan, R. K. (DOI: 10.1038/s41598-026-67891-2).
- Supports
- Extraction of measurable O₂, CO, and CO₂ from Lunar Mare Simulant using SF₆ and C₄F₈ plasmas; EDS, XRD, and XPS proof that SF₆ targets anorthosite and forms a surface fluoride layer with XRD strain-broadening.
- Limit
- Laboratory bench diagnostic study on simulant powders; does not demonstrate high-throughput continuous oxygen storage or fluorine regeneration.
- [3] Planetary and Space Science: Fluorine Plasma Processing of Lunar Regolith Simulant for Lunar in-situ Resource Production (August 2026)
- Location
- Cowley, A., Vijayaraghavan, R. K., McGearalt, N., Kelly, S., McNally, P., Lomax, B., & Tanner, B. K., Vol. 282, Article 106337.
- Supports
- Thermodynamic context, semiconductor SiO₂ etching parallels, temperature dependence (200°C to 400°C regime), and volatile fluoride / oxygen reaction pathways.
- Limit
- Pre-flight experimental and process-mechanism study; closed-loop halogen recycling and lunar dust feeding remain open engineering gates.
- [4] NASA NTRS: Automated Carbothermal Reactor Environmental Testing (2025)
- Location
- Sierra Space and NASA thermal-vacuum demonstration at ~1,800°C.
- Supports
- The high-temperature carbothermal benchmark (TRL 6 subsystem test) against which low-temperature plasma processing is compared.
- Limit
- Simulant chamber test; used here as the high-temperature silicate-reduction reference point.
- [5] Planetary and Space Science: Proving the viability of an electrochemical process for the simultaneous extraction of oxygen and production of metal alloys from lunar regolith (2020)
- Location
- Lomax, B. A., Conti, M., Khan, N., et al., Vol. 180, Article 104748.
- Supports
- The 900°C–950°C molten-salt FFC Cambridge electrolysis benchmark capable of extracting high oxygen fractions while leaving metallic alloys.
- Limit
- Requires molten CaCl₂ salt management and high-temperature electrochemical cells.