While Oak Ridge gets all the molten-salt glory, Los Alamos quietly operated the world's first liquid-metal-fueled reactor — the Los Alamos Molten Plutonium Reactor Experiment (LAMPRE-I) — from 1961 to 1963. It ran for over 2,500 hours at full power, using molten plutonium-iron eutectic fuel at 650°C, and was then shut down, dismantled, and effectively forgotten. The follow-on LAMPRE-II was canceled before construction. Sixty-three years later, every "advanced reactor" startup pitching liquid-fueled fast reactors is rediscovering ground that Los Alamos already walked.

What it was. LAMPRE-I was a 1 MW thermal fast-spectrum reactor designed by David Lillie's team at LASL. The fuel was not solid pellets but a liquid plutonium-iron alloy (Pu-2.5 wt% Fe) contained in tantalum capsules, cooled by flowing sodium. The vision was breathtaking: a reactor where fission products could be continuously removed from circulating fuel, where xenon poisoning was a non-issue, where you could refuel without shutting down, and where the fast neutron spectrum bred more fuel than it burned. A 500 MW commercial version was sketched in LA-3540 (1966).

Why it died. Several reasons converged:

• Tantalum embrittlement. The capsules degraded faster than expected from plutonium attack at 650°C.
• The AEC's bet on solid fuel. Argonne's EBR-II and the eventual Clinch River Breeder Reactor sucked up all the fast-reactor funding. Solid oxide fuel was "good enough" and politically safer.
• Plutonium proliferation panic. By the late 1960s, anything involving liquid plutonium and online reprocessing was politically radioactive.
• No champion. Lillie moved on. Los Alamos pivoted to weapons work. The institutional memory evaporated with the staff.

Why it deserves a second look in 2026. Three things changed:

• Refractory alloys. We now have TZM (titanium-zirconium-molybdenum), ODS steels, and tantalum-tungsten-hafnium alloys that survive molten actinides for decades, not months. Idaho National Lab's recent work on Mo-Re cladding shows < 5 µm/year corrosion rates in molten Pu-Fe.
• Online reprocessing is now real. Electrorefining and fluoride volatility processes developed for IFR (Argonne, 1984-1994) and recent pyroprocessing demonstrations at INL mean you can actually skim fission products from a flowing fuel stream.
• The waste problem. A LAMPRE-style reactor can burn its own actinides plus LWR spent fuel. The U.S. has 90,000 tons of spent fuel sitting in dry casks with nowhere to go. A fleet of fast-spectrum liquid-fueled burners could reduce that to fission products with ~300-year storage requirements instead of ~300,000 years.

The technical record is shockingly complete. LA-3073, LA-3540, and LA-2833 contain the full operational data — neutronics, corrosion measurements, thermal hydraulics. Yet when TerraPower, Oklo, and Southern Company pitch their fast reactors, LAMPRE is almost never cited. It's as if the 2,500 hours never happened.

The Chinese, of course, noticed. CIAE in Beijing referenced LAMPRE explicitly in their 2019 roadmap for molten-fuel fast reactors. The Russians built BR-5/BOR-60 with liquid sodium-bonded fuel, drawing the same lineage. America invented this technology, ran it successfully, wrote it up, and walked away.