Inertial Electrostatic Confinement Fusor
A glowing star-in-a-jar built by hobbyists — real fusion, but with a fundamental physics ceiling on how far it can scale.
Confines a fusion plasma electrostatically between two concentric, transparent metal grids inside a spherical vacuum chamber. A negative voltage of 30–80 kV on the inner grid accelerates deuterium (or D-D / D-T) ions inward; ions converge on a virtual cathode at the centre, oscillate radially, and fuse during head-on collisions. Simple variants are buildable by undergraduates and high-school students for under $5,000 and produce 10⁵–10⁷ neutrons per second — genuine, repeatable, easily verified fusion.
Deuterium gas is admitted to a vacuum chamber at 1–10 mTorr. The inner grid (typically tungsten or stainless steel) is biased to −30 to −80 kV with respect to the grounded outer chamber. Background D₂ molecules are ionised by the high-voltage glow discharge; D⁺ ions accelerate radially inward through the transparent inner grid, converge near the geometric centre, and form a 'virtual cathode' poissoir region. Ions that miss in their first pass oscillate back and forth across the centre, occasionally colliding head-on at relative energies of ~30–80 keV — well within the deuterium-deuterium fusion cross-section peak (~70 keV). Each fusion releases either ³He + n (2.45 MeV) or ³H + p (3.0 MeV) with roughly equal branching. Total neutron yields of 10⁵–10⁷ /s are routine; record laboratory fusors have reached ~3 × 10⁹ neutrons/s. The fundamental obstacle, identified rigorously by Rider (PhD thesis MIT, 1995): non-equilibrium IEC plasmas thermalise via Coulomb collisions much faster than they fuse, so any non-thermal ion-energy distribution that boosts fusion rate decays into a thermal distribution that radiates and conducts energy out faster than fusion replaces it. Rider's analysis showed that under broadly applicable assumptions, no non-equilibrium fusion plasma — IEC, polywell, or otherwise — can be net energy positive.
In plain terms
Picture a metal cage shaped like a tiny basketball with thousands of volts pulling positive ions toward its centre. The ions fall in, miss each other most of the time, race past the centre, fly back out, and fall in again. A handful collide head-on at the middle and fuse. You can see the glow from across the room and even count the neutrons — but the same physics that lets it work also says it can never break even on energy.
Also called: IEC fusor, Farnsworth–Hirsch fusor, demo fusor.
Why it matters
- Cheapest demonstration of nuclear fusion accessible to students and amateurs.
- Used commercially as compact 2.45 MeV (D-D) and 14 MeV (D-T) neutron sources for activation analysis, well-logging, and explosives detection.
- Drives education in plasma physics, vacuum technology, and high-voltage engineering.
- Provides a clean test bed for the theoretical limits of non-equilibrium fusion plasmas — Rider's 1995 thesis remains the standard reference.
Where it lives
- Phoenix Nuclear Labs / SHINE neutron sources (commercial IEC)
- Adelphi Technology IEC neutron generators
- University of Wisconsin–Madison fusion lab IEC research
- Hundreds of amateur fusors documented at fusor.net
Key numbers
- Cathode voltage: 30–80 kV
- D-D fusion peak σ energy: ~70 keV
- Hobbyist neutron yield: 10⁵–10⁷ n/s
- Lab record yield: ~3 × 10⁹ n/s (University-scale devices)
- Cost (entry hobby build): ≈ US$2k–5k
Surprising facts
- Philo T. Farnsworth — best known for inventing electronic television — also patented the IEC fusor in 1968; his prototype produced measurable neutrons.
- Taylor Wilson built a working fusor at age 14 in his parents' Arkansas garage in 2008, becoming the youngest person to achieve nuclear fusion.
- A typical hobby fusor running on D-D releases 10⁶ neutrons per second — enough to be detected by a He-3 counter across the room, but invisible to the eye except for the violet plasma core.
- Robert Bussard's polywell variant (a magnetically shielded IEC) showed promising scaling in early-2000s tests, but his EMC2 group never demonstrated net energy and the company closed in 2011 without resolving Rider's objection.
Open questions
- Anomalous polywell scaling claims: EMC2's 2005–2010 polywell tests reported scaling of fusion rate with B-field that, at face value, looked more favourable than Rider's analysis allowed. Independent reanalysis suggests the apparent advantage was an artefact of measurement geometry rather than new physics. We label any 'energy-positive IEC' claim as speculative.
- Cathode-grid sputtering plateau: Operational fusors lose performance when the high-voltage grid sputters and contaminates the chamber. The sputtering yield depends on grid material, voltage, and pressure in ways that no published model fully captures, making device-to-device performance hard to predict.
A short history
Philo T. Farnsworth filed key fusor patents in 1966 and was granted the principal IEC patent in 1968 (US 3,386,883). Robert Hirsch refined the device in 1967 while working under Farnsworth, demonstrating neutron yields up to 10⁹ /s in pulsed mode. Robert Bussard proposed the magnetically shielded 'polywell' IEC variant in 1985. Todd Rider's 1995 MIT PhD thesis (supervised by Lawrence Lidsky) provided the rigorous proof that non-equilibrium fusion plasmas — including all IEC variants — cannot reach net energy under broadly applicable assumptions; that result has stood for thirty years. Commercial IEC neutron generators (Phoenix Nuclear Labs, Adelphi Technology) emerged in the 2010s for industrial neutron radiography and activation analysis.
Sources
- Farnsworth, US Patent 3,386,883 (1968) (https://patents.google.com/patent/US3386883A)
- Hirsch, J. Appl. Phys. (1967) — IEC neutron yields (https://aip.scitation.org/doi/10.1063/1.1709771)
- Rider, T. H., MIT PhD thesis (1995) — fundamental limits on plasma fusion systems not in thermodynamic equilibrium (https://dspace.mit.edu/handle/1721.1/11412)
- fusor.net — amateur IEC fusion community archive (https://www.fusor.net)