Cold Fusion & LENR: From Metal Lattices to Mini Reactors?

October 19, 2026

What if fusion’s future isn’t one enormous power plant, but many smaller reactors serving homes, vehicles, and neighborhoods? That’s the possibility physicist and electrical engineer Lynn Bowen sees in solid-state fusion: deuterium fuel held inside a metal lattice, where she proposes that energetic nuclear reaction products could help sustain further reactions. Her approach puts materials science at the center of a problem usually associated with extreme plasma temperatures and magnetic confinement. But the distance between a promising reaction model and a practical generator isn’t measured in atoms alone. It runs through fuel loading, particle energy losses, radiation detection, heat extraction, and the controls needed to turn nuclear activity into dependable power.

An Engineer’s Approach to a Nuclear Problem

Bowen’s route into fusion began with an interest in nuclear theory, followed by a career built around materials and electronics. She earned a physics degree from the University of Missouri in 1977 and a master’s degree in electrical engineering from MIT in 1980. Her subsequent work included the semiconductor industry and teaching physics at Colorado Mountain College. Alongside that career, she pursued independent research into the strong nuclear force—the interaction responsible for binding protons and neutrons together. When Martin Fleischmann and Stanley Pons announced their cold-fusion claims in 1989, she was already thinking about how nuclear behavior might fit into a different theoretical framework.

The controversy that followed became part of the field’s identity, but it isn’t where Bowen wants the discussion to stop. In our conversation, I described what seemed to be a gradual broadening of interest in fusion approaches and the behavior of atomic lattices. Her response was hopeful rather than triumphant: “I hope so, yes. And I hope it continues.” Her position is that the materials knowledge available today deserves attention on its own merits. She argues that an explanation should account for the microscopic environment inside the metal, rather than treating the solid host as merely a passive container for hydrogen.

Her published work includes A Lattice-Assisted Conventional Explanation for LENR, Part I, which appeared in the Journal of Condensed Matter Nuclear Science in 2025, spanning pages 36–83. LENR stands for low-energy nuclear reactions, an umbrella term used in the research community surrounding cold fusion. Bowen describes her own approach as an attempt to explain the reported phenomena using conventional nuclear processes. “I’m sticking with conventional physics,” she said. The proposed departure lies in how a highly loaded lattice changes the conditions experienced by reacting particles—not in declaring that energy conservation or ordinary nuclear accounting no longer applies.

That distinction matters for the technology she ultimately imagines. Bowen isn’t proposing that developers simply shrink a conventional plasma reactor and install it in a garage. She’s describing a different reaction environment that, if it can support controlled, useful output, might allow a different system architecture. Her preferred future is “a reactor in your car or a reactor in your house or a reactor for a neighborhood.” Those are prospective applications, not specifications for an existing product. They establish the engineering destination; the rest of her argument concerns the physical pathway that might make it reachable.

The Lattice Is More Than a Container

In Bowen’s usage, solid-state fusion means proposed nuclear reactions involving fuel held within a solid host material. That host could be palladium, or potentially another metal or alloy, although her discussion centers on palladium loaded with deuterium. Deuterium is hydrogen-2: its nucleus contains one proton and one neutron, rather than the single proton of ordinary hydrogen-1. A neutral deuterium atom also has an electron; its nucleus alone is called a deuteron. That terminology matters because fuel preparation concerns hydrogen within a material, while the proposed nuclear reactions involve the nuclei themselves.

Fuel loading describes how much deuterium the metal contains relative to its palladium population. Bowen explains an 80% loading as approximately 80 deuterium atoms for every 100 palladium atoms. At 100% loading, the ratio becomes approximately one deuterium per palladium atom. In chemical shorthand, those ratios can be represented as PdD₀.₈ and PdD₁.₀. They’re atomic ratios, not weight percentages, conversion efficiencies, or statements that 80% of the fuel has reacted. For a developer, that distinction separates the condition of the starting material from the eventual performance of a reactor.

Electrolysis is one way to produce the loaded material, not an alternative to using a solid lattice. Applying an electrical bias to a palladium electrode can drive deuterium loading from an electrolyte. The apparatus may therefore contain a liquid while the proposed primary fusion environment remains inside the metal. Bowen emphasized that other loading methods are possible; the defining feature is the fuel-bearing lattice. This also creates an important accounting boundary for a future device: the electricity used to prepare and maintain its material state must be distinguished from any nuclear heat or electrical output it subsequently produces.

Her interpretation differs from Edmund Storms’s emphasis on specialized nuclear-active environments associated with cracks or gaps. As Bowen puts it, “To me, the lattice is the environment. To Edmund, it’s the cracks.” Both positions assign importance to material conditions; they differ over the environment required for the proposed nuclear process. Identifying the lattice as essential doesn’t mean Bowen has established that reactions occur uniformly throughout every loaded sample. It means that, in her explanation, palladium’s role extends beyond holding fuel near a special defect: the loaded material itself helps create the conditions for continued reactions.

A Dense Fuel Supply Isn’t the Same as a Powerful Reactor

Bowen’s density comparison is one of the most striking numbers in the interview. She cites approximately 6 × 10²⁸ deuterium atoms per cubic meter in a loaded palladium lattice, compared with a plasma example of 1 × 10¹⁹ particles per cubic meter. Dividing those figures produces her quoted factor of six billion. The comparison arose during our discussion of tokamak plasmas, and that context matters: it shouldn’t be read as a universal comparison with every form of plasma or every fusion experiment. Her argument concerns the concentration of stored fuel within a solid host.

Expressed in smaller volumes, her figures correspond to approximately 6 × 10²² deuterium atoms per cubic centimeter in the loaded metal, versus 1 × 10¹³ per cubic centimeter in the plasma example. Those are unit conversions of her stated numbers, not additional experimental measurements. The contrast helps explain why a solid host is interesting: a compact volume can contain a substantial population of potential reaction partners. But it doesn’t follow that the material produces six billion times more power. Stored fuel density says nothing by itself about how many nuclei have suitable energies, how often they fuse, or how much energy is required to maintain the conditions.

Bowen reduces the collision problem to four factors: the population of targets, the population of projectiles, the reaction cross section, and the distance those projectiles can travel. Her analogy is a room filled with balloons. The number of balloons struck depends on how many balloons are present, how many projectiles enter, how large the targets effectively appear, and how far the projectiles travel. Nuclear cross sections add another complication: they depend on the reaction and the particles’ energies. More densely packed fuel can offer more opportunities for encounters, but an encounter still has to occur under conditions that permit fusion.

A separate NASA research program offers useful context without validating Bowen’s specific mechanism. NASA’s lattice-confinement fusion work has used deuterated metals, including erbium, to hold fuel far more densely than in magnetic-confinement plasmas. In the reported experiments, an external photon beam above approximately 2.9 MeV initiated processes that produced energetic particles and subsequent reactions. NASA explicitly distinguishes an initially cool bulk lattice from energetic individual particles within it, and identifies increased reaction rates as necessary for appreciable power. The comparison supports the importance of the material environment while also illustrating the central challenge: dense fuel and observed reactions aren’t yet the same thing as a self-powered generator.

Where the Particle Energy Goes

The heart of Bowen’s proposed mechanism is an energy-transfer problem. An energetic charged particle moving through matter loses energy through interactions with the material. Electronic stopping transfers energy into electronic excitation and ionization. Nuclear stopping transfers energy through collisions that produce recoiling atoms and changes in particle direction. Both slow the incoming particle; nuclear stopping doesn’t preserve all its energy or automatically constitute fusion. Bowen’s interest is in where the transferred energy goes. Energy given to another deuterium nucleus might, in her model, help prepare another reaction opportunity instead of being dispersed primarily into the electronic system.

She describes this competition using a “crossover energy,” where one stopping mechanism becomes more important than the other. In her interview explanation, she places an illustrative crossover near 0.1 MeV in an amorphous material such as water, contrasts that with 1–5 MeV in metals, and discusses 10–20 MeV in connection with the behavior she attributes to superconducting materials. She then places the reactions of interest around 0.01–1 MeV. These are Bowen’s proposed quantitative comparisons, not universal stopping-power specifications. Their applicability to particular projectile species, loading conditions, and palladium samples is a central part of what her mechanism needs to establish.

The superconductivity discussion also requires a distinction between a material and its operating state. Bowen says palladium deuteride begins superconducting around 10 kelvin, roughly −263°C, and acknowledges that the apparatus being discussed isn’t necessarily operating in that superconducting state. Calling a material a superconductor therefore doesn’t establish that its electrons exhibit superconducting behavior at the temperature of a proposed LENR reactor. The engineering question is more specific: what are the actual stopping characteristics of the loaded material under the conditions in which the reactor would run?

In Bowen’s proposed cascade, nuclear reactions release products carrying substantial kinetic energy. During the interview she mentions product energies on the order of 2, 3, and 14 MeV, depending on the product. Those particles can transfer energy to deuterons through collisions, potentially enabling subsequent reactions. For the process to sustain itself, enough useful energy must reach enough suitable reaction partners before particles slow down or escape. That’s the proposed connection between microscopic collisions and continuing heat production. The interview explains how she thinks an operating cascade could continue, but it doesn’t supply a complete startup procedure or a measured balance of all the competing losses.

What the Neutron Measurements Would Have to Show

Bowen’s model gives neutrons a prominent role, making radiation measurements part of the reactor-development problem rather than an optional confirmation. She emphasizes energetic neutrons around 14 MeV and contrasts their detection with that of thermal neutrons around 0.026 electron volts. Her concern is that the surrounding apparatus and a detector’s energy-dependent response can separate the number of neutrons generated from the number actually recorded. In her explanation, the metal, electrolyte, water bath, detector position, and detection mechanism must all be considered before interpreting a low count rate.

Her geometry example makes that argument tangible. A detector presenting a circular area 2 centimeters in diameter, positioned 30 centimeters from an isotropic point source, would intercept approximately one part in 3,600 of the emitted directions before accounting for intervening materials or detector response. That follows from comparing the detector’s projected area with the surface area of the surrounding sphere. Real apparatus can complicate the calculation through scattering and an extended source geometry, but the example illustrates a genuine design issue: source emission, particles reaching an instrument, and recorded events are different quantities.

Bowen goes further in the interview, describing high-energy neutron detection as essentially impossible. That broader characterization shouldn’t be adopted as a general engineering fact. As an independent reference, NIST researchers have reported a segmented scintillator and helium-3 spectrometer designed for 1–200 MeV neutrons and characterized in 2.5 MeV and 14 MeV fields. The relevant development question is therefore not whether energetic neutrons can be detected, but whether a particular measurement system has the required sensitivity and calibration. Low counts alone can’t establish a large hidden source; a source estimate must account for the instrument’s measured response.

Bowen also points to CR-39, a plastic detector in which charged particles leave damage trails that chemical etching makes visible. She interprets certain triple-track observations as evidence of energetic neutron interactions with carbon. The corresponding published interpretation involves ¹²C(n,n′)3α: carbon-12 breaks into three helium-4 nuclei, with an outgoing neutron also present. Each alpha particle contains two protons and two neutrons. That proposed interpretation concerns a nuclear signature, not a direct measurement of useful reactor power. Connecting the tracks to a particular source, reaction rate, and heat output requires additional experimental evidence.

A Water Bath and a 2.22-MeV Checkpoint

Bowen’s proposed gamma-ray experiment attempts to turn the surrounding water into part of the measurement system. An energetic neutron leaving the core can undergo collisions that reduce its energy. If it is subsequently captured by a hydrogen-1 nucleus, the reaction produces deuterium and a gamma ray: n + ¹H → ²H + γ. For capture of a slow neutron, the characteristic photon energy is approximately 2.22 MeV. The hydrogen-capture line is established nuclear physics; what remains to be tested is how effectively Bowen’s proposed arrangement could use it to investigate the activity she attributes to a palladium–deuterium system.

Her interest in the method grew from Monte Carlo simulations conducted with a co-author. Such simulations can follow particle interactions through specified materials and geometries, helping predict where radiation travels and what a detector might observe. They’re useful for designing a measurement before equipment is assembled. But a transport calculation that begins with an assumed neutron source doesn’t independently demonstrate that the core generates that source. The source mechanism and the detector response are connected problems, and each needs its own evidence.

Bowen identifies five practical conditions she wants combined: a palladium–deuterium reactor, a surrounding water bath, measurements taken while the reactor operates, a gamma detector positioned close to the bath, and enough energy resolution to distinguish the relevant spectral features. “I haven’t seen anybody doing a replication as yet,” she said. More specifically, she says she hasn’t found a published experiment satisfying her full set of conditions in the literature she reviewed. That is a statement about her search and proposed test—not proof that every related measurement has been absent from the field.

From an engineering standpoint, the value of such a test would depend on its controls as much as its detector placement. Background measurements, energy calibration, reactor-on and reactor-off comparisons, and synchronized records of electrical input, temperature, and radiation would make the result more interpretable. These are validation requirements implied by the proposed measurement, not a completed protocol supplied in the interview. A convincing outcome would connect a reproducible spectral change to the operating apparatus and then determine what that signal says about the source. That would be a development milestone, rather than a shortcut around the remaining power measurements.

Turning Nuclear Heat Into Useful Electricity

Even a reliable nuclear heat source would still need a practical material system. Bowen identifies palladium cost as an obstacle and argues for finding a less expensive host that retains the useful properties she attributes to the loaded lattice. She mentions nickel and alloys as possibilities in the broader discussion, but the interview doesn’t establish an interchangeable substitute. A replacement would have to be evaluated for fuel loading, reaction behavior, and operation over time—not merely selected because its purchase price is lower. The expensive component isn’t necessarily the biggest cost once processing, lifetime, and replacement are included.

For electricity generation, Bowen proposes borrowing from pressurized-water reactor engineering. It’s important to distinguish that general proposal from a detailed plant design. In a conventional pressurized-water system, as described by the Nuclear Regulatory Commission, the primary coolant carries heat to a steam generator; water in a separate secondary loop becomes steam that drives a turbine-generator. Pressure, temperature, heat exchangers, pumps, and the condenser are all part of the system. Applying that architecture to a different heat source wouldn’t eliminate those components or establish that the source can operate at a suitable temperature.

A simple engineering illustration shows why the surrounding machinery matters. Assume—not as a prediction for Bowen’s system—that a device converts thermal energy into electricity at 20% efficiency. Producing 5 kilowatts of gross electrical output would require 25 kilowatts of thermal input, leaving approximately 20 kilowatts to reject as heat. Pumps, controls, and other auxiliaries would further reduce the electricity available to an external load. The numbers are hypothetical, but the accounting is unavoidable: a compact reaction region doesn’t automatically produce a compact electrical generator, especially when cooling equipment must accompany it.

The interview doesn’t provide the operating temperature, pressure, measured thermal output, conversion efficiency, or auxiliary power consumption needed to specify such a machine. That limits what can responsibly be inferred about its size or performance. It also identifies the next engineering questions. A development program would need to distinguish useful heat from gross electrical output and net electrical output, then relate those quantities to sustained operation. Producing an interesting thermal anomaly, maintaining a nuclear reaction, and powering an external load are different milestones, even when the long-term ambition connects all three.

Control, Tritium, and the Possibility of Hybrid Systems

Bowen is direct about an unresolved requirement: control. She says researchers still need better ways to increase a reaction, decrease it, or shut it down quickly. That’s particularly important for a proposed cascade that uses products from previous reactions to enable subsequent ones. A developer would need to measure how the process responds to changing inputs and disturbances, rather than assume that controlling the loading current necessarily controls the nuclear output. The interview doesn’t establish response times, shutdown margins, or a validated control strategy. Those omissions don’t settle whether the mechanism can work, but they do define important barriers between an experiment and a product.

The fuel cycle presents another distinction. Bowen emphasizes deuterium as her starting fuel and contrasts it with the tritium requirements of deuterium–tritium fusion. Tritium contains one proton and two neutrons and is radioactive; the NRC gives its half-life as approximately 12.3 years. As broader fuel-supply context, the Department of Energy identifies tritium availability and breeding as significant challenges for prospective D–T power plants. Bowen says LENR can produce tritium, but the interview doesn’t supply a verified production rate, recovery efficiency, or cost. Nor would starting with nonradioactive deuterium make a neutron- and tritium-producing system automatically free of radiological requirements.

That possibility prompted a complementary application during our conversation: could a solid-state system help supply fuel to a larger fusion reactor? I raised the idea of using tritium production alongside a tokamak, and Bowen described it as a possible hybrid arrangement. The appeal is that the technologies wouldn’t necessarily have to compete for the same function. One might contribute to another’s fuel cycle or energy system. But useful integration would require more than detecting tritium: developers would have to establish sustained production, recovery, containment, and the energy balance of the combined process. None of those quantities was specified in the interview.

Mobile applications raise the same questions under tighter constraints. Bowen discusses cars, and our closing exchange extends the possibility to maritime systems such as submarines and aircraft carriers. A small core could be attractive, but a vehicle must carry the complete system: heat removal, conversion equipment, controls, containment, and whatever radiation protection its measured emissions require. These are engineering implications of the proposed applications, not features of a demonstrated Bowen reactor. A neighborhood installation and a passenger vehicle could therefore face very different paths to practicality, even if both began with the same underlying reaction technology.

From a Research Community to a Development Program

Bowen’s work also belongs to a community whose knowledge extends beyond any one proposed mechanism. She names Edmund Storms as a major influence despite their different explanations, recommending his book The Science of Low Energy Nuclear Reactions as an introduction to the field. She also remembers Norman Cook and Alan Smith as scientific influences and friends. In our discussion, I added Jed Rothwell’s LENR-CANR archive as a resource for preserving the literature. That history matters because future experiments don’t start from an empty page: they inherit measurements, methods, disagreements, and unresolved questions.

Her confidence is stronger than a cautious technology forecast. During the interview, she says the field has already produced self-sustaining reactions and usable output energy. However, that statement isn’t accompanied in the conversation by an identified device’s complete power balance, operating duration, independent replication record, or engineering specification. Elsewhere, she says she wants to perform further experiments herself and hopes others will attempt the proposed gamma-ray test. Those statements should remain distinct: confidence in reported LENR results, confidence in her explanation, and demonstration of a controllable commercial reactor aren’t interchangeable achievements.

The development sequence implied by the discussion is demanding but concrete. First, establish reproducible operating conditions and measurements that distinguish the proposed signals from backgrounds and artifacts. Then relate those signals to a defensible energy balance, sustained output, and controlled changes in operation. After that come material lifetime, useful operating temperature, heat extraction, and a complete system that delivers energy to a load. Bowen doesn’t give a funded development budget or a commercial delivery date in the interview. Assigning either would require information the conversation doesn’t provide.

The possibility remains substantial: a material-engineered route to fusion that could eventually support energy systems closer to the places they’re used. But the most important word in “mini reactor” isn’t “mini.” It’s “reactor”—a device whose behavior can be measured, reproduced, controlled, and integrated into a useful machine. Bowen’s work connects a proposed microscopic mechanism to that larger ambition. The path from metal lattices to mini reactors would be built through experiments and engineering results, with each measured step making the possibilities more concrete.