Molten Salt Reactors: Why Thorium-Based Nuclear Has Been Promised for 60 Years and Still Is Not Here

## The 1965 Proof of Concept Nobody Followed Up On

Between 1965 and 1969, the Oak Ridge National Laboratory in Tennessee operated the Molten Salt Reactor Experiment (MSRE) — a 7.4 megawatt-thermal test reactor that ran for four years using liquid fluoride salt as both fuel and coolant. It worked. The reactor operated reliably, demonstrated continuous fuel processing, and showed that liquid-fueled fission was controllable and safe in practical operation.

Then the program was cancelled.

The cancellation was not primarily technical. The Nixon administration's Atomic Energy Commission, under Glenn Seaborg's successor, prioritized the Liquid Metal Fast Breeder Reactor (LMFBR) — a sodium-cooled design that would breed plutonium from uranium, supporting the existing uranium-based nuclear weapons complex. MSR research did not align with weapons material production goals, and in a resource-constrained environment, it lost the political competition. Alvin Weinberg, Oak Ridge's director and the MSRE's champion, was removed from his position in 1972.

The MSRE's success — a demonstrated, operating molten salt reactor — has been cited by MSR advocates for six decades. And six decades later, the technology still has no commercial deployment. Understanding why requires separating the genuine engineering obstacles from the institutional and economic barriers.

## Fluoride Salt vs Chloride Salt Chemistry

Modern MSR designs split into two main branches based on their choice of salt carrier.

**Fluoride salt reactors** use mixtures like FLiBe (lithium fluoride and beryllium fluoride) or FLiNaK (lithium, sodium, and potassium fluorides) as the solvent in which fuel — uranium or thorium fluorides — is dissolved. FLiBe has favorable neutron moderation properties (the beryllium acts as a moderator), making it suitable for thermal spectrum reactors that can run on thorium. It is also relatively stable and has low vapor pressure at operating temperatures. The MSRE used a FLiBe-based system.

The problems with FLiBe are practical and severe. Beryllium is toxic, making handling and processing hazardous. The lithium must be enriched to lithium-7 (naturally occurring lithium-6 absorbs neutrons problematically and produces tritium), and Li-7 enrichment is expensive and generates waste. FLiBe at operating temperature (650–700°C) is highly corrosive to structural alloys.

**Chloride salt reactors** use magnesium chloride or sodium chloride-based mixtures. Chloride systems operate at higher temperatures (700–800°C), allowing higher thermal efficiency. They support fast neutron spectra, enabling more efficient actinide transmutation. TerraPower's Molten Chloride Fast Reactor (MCFR) is the leading chloride-based commercial program.

Chloride's engineering challenges are different: chlorine has higher neutron absorption cross-sections that affect fuel cycle economics, and the thermodynamics require attention to chloride volatility at high temperatures. Chloride systems do not naturally support the thorium fuel cycle as efficiently as fluoride systems.

## The Thorium Fuel Cycle: Proliferation Resistance and Waste Reduction

Thorium-232 is not itself fissile. It is fertile — it absorbs a neutron to become thorium-233, which beta-decays to protactinium-233, which beta-decays again to uranium-233, which is fissile. The thorium fuel cycle therefore requires a neutron source (usually a small amount of fissile material like U-235 or Pu-239) to initiate, but once running, it breeds its own fuel from abundant thorium.

The proliferation resistance argument is legitimate but not absolute. Unlike conventional reactor fuel cycles, the thorium cycle does not produce significant quantities of plutonium-239, the material used in most nuclear weapons. U-233 produced in a thorium cycle is contaminated with U-232, which has a decay chain producing hard gamma radiation that makes weapon construction difficult without sophisticated shielding and remote handling.

However, U-233 itself is a weapons-usable material, and pure U-233 (if the U-232 contamination is removed) can be used in nuclear weapons. The proliferation resistance is therefore a function of implementation — it is better than the standard uranium-plutonium cycle but not categorically proliferation-proof.

The waste reduction claim is more straightforwardly accurate. The thorium cycle produces fewer long-lived transuranic actinides than the uranium-plutonium cycle. MSRs operating with online fuel reprocessing can transmute the actinides they do produce into shorter-lived fission products. The result is waste with a radiotoxicity timeline of hundreds rather than tens of thousands of years — a genuine and significant improvement in long-term storage burden.

## The Corrosion Problem: 60 Years and Still Unsolved

This is the fundamental engineering challenge that separates MSR from commercial deployment.

Fluoride salt at 650–700°C is an extraordinarily aggressive chemical environment. It attacks virtually every conventional structural alloy — stainless steels, nickel-based superalloys — through selective oxidation of chromium. The MSRE used Hastelloy-N, a nickel-molybdenum alloy developed at Oak Ridge, which showed reasonable corrosion resistance over the four-year experiment but not at the decades-long scale a commercial reactor requires.

The corrosion mechanisms are multiple and interactive: oxidative dissolution of chromium from the alloy grain boundaries, embrittlement by tritium and helium produced in the lithium salt, and tellurium embrittlement from fission products dissolved in the salt that attack alloy boundaries. Each of these mechanisms can be addressed individually, but addressing all simultaneously in a single alloy system with commercial-scale fabrication and decades of service life has not been demonstrated.

Current approaches include modified Hastelloy-N formulations, silicon carbide composite cladding, carbon-carbon composites, and molybdenum-based alloys. All show promise in laboratory experiments; none has been qualified for commercial reactor service. The NRC's licensing framework, which requires detailed material degradation data over projected service life, effectively demands this qualification data before a commercial license can be issued — creating a chicken-and-egg problem: the data comes from operating a reactor, but the reactor requires the data to get a license.

## Current Programs: TerraPower, Terrestrial Energy, and China's TMSR

Three programs represent the current serious contenders for MSR commercialization.

**TerraPower's Molten Chloride Fast Reactor (MCFR)**: Backed by Bill Gates and partnered with Southern Company, TerraPower is developing a chloride fast reactor that would use spent nuclear fuel as its initial fuel source, transmuting it while generating power. The MCFR completed a pre-application review with the NRC. The demonstration reactor (MCRE) is targeting initial operation in the late 2020s at the Idaho National Laboratory site.

**Terrestrial Energy's IMSR (Integral Molten Salt Reactor)**: A Canadian company targeting a 400 MWt (185 MWe) reactor designed as a sealed, integral unit with a projected seven-year design life before the core is replaced as a unit. The IMSR uses LiF-based fluoride salt with low-enriched uranium. It has completed Phase 1 of pre-licensing review with the Canadian Nuclear Safety Commission. Targeted deployment: early 2030s.

**China's TMSR-LF1**: The Thorium Molten Salt Reactor Liquid Fuel 1, operated by the Chinese Academy of Sciences at Wuwei in the Gobi Desert, became the world's first operating molten salt reactor since the MSRE when it reached first criticality in 2023. It is a 2 MWt experimental reactor using liquid thorium-uranium fluoride fuel, designed to generate operational data for a planned 373 MWt demonstration reactor. China is the most serious state-level investor in MSR technology in the world, with a decades-long program and institutional continuity that Western programs have lacked.

## Regulatory Pathway: NRC vs CNSC

The US Nuclear Regulatory Commission was built around solid-fuel, water-cooled reactor designs. Its regulations, standard review plans, and material qualification databases are largely calibrated to light water reactor technology. Licensing an MSR requires either adapting existing frameworks or using the NRC's 10 CFR Part 53 pathway for advanced non-LWR reactors.

The Part 53 rulemaking, finalized in 2024, provides a performance-based, technology-inclusive framework that is more accommodating for MSR designs. But even under Part 53, the evidentiary requirements for material qualification, safety analysis methodology validation, and accident scenario demonstration remain formidable. A realistic first-of-a-kind MSR license in the US will likely require 10–15 years of regulatory engagement from serious application submission to fuel loading.

Canada's CNSC has shown more flexibility, reflected in Terrestrial Energy's relatively smooth pre-licensing progress. Several Canadian provinces have expressed interest in IMSR deployment, and Canada's regulatory culture around advanced reactors has been somewhat more pragmatic than the NRC's.

## Private Fusion vs MSR: The "Obsolete Before Completion" Argument

MSR advocates argue that the technology offers near-term (2030s–2040s) deployment on proven fission physics, unlike fusion which remains perpetually decades away. Fusion advocates counter that MSR faces its own unresolved engineering challenges and that the long-term physics of fusion (no long-lived radioactive waste, theoretically unlimited fuel from seawater deuterium) make it the right bet for the 21st century.

The realistic assessment is that both timelines are optimistic. MSR proponents who claim 2030s commercial deployment are banking on corrosion material solutions that have not yet been demonstrated at scale, regulatory processes that historically take longer than planned, and construction cost control that the nuclear industry has systematically failed to achieve for 40 years.

The honest answer to "why has thorium MSR been promised for 60 years and is still not here" involves three factors: the MSRE was cancelled before it could generate the operational data needed to build the next step; the remaining engineering challenges (primarily corrosion and materials) are genuinely hard; and nuclear regulation is structurally conservative in ways that make first-of-a-kind deployment slow and expensive regardless of how good the underlying physics are.

What is different in 2026 compared to 2006 is that the physics is better understood, computational materials science has accelerated alloy development, regulatory frameworks are being modernized, and China's TMSR-LF1 is generating real operational data. The technology is advancing. Whether it advances fast enough to matter before other low-carbon alternatives saturate the market is the question that will determine whether MSR becomes a footnote or a foundation.

Molten Salt Reactors: Why Thorium-Based Nuclear Has Been Promised for 60 Years and Still Is Not Here

// COMMENTS

ON THIS PAGE