On the feasibility of fissile material production with inertial electrostatic confinement fusion devices

1 Introduction

Inertial Electrostatic Confinement (IEC) is a method to contain a plasma by means of electric fields instead of the magnetic fields used in wider known electromagnetic confinement systems such as stellarators or tokamaks. When a light element gas mixture such as a combination of hydrogen and/or helium isotopes is selected for ionisation into a plasma, and sufficiently high energies are reached, both methods can be used to achieve nuclear fusion in the plasma (Moynihan et al., 2023). Depending on which isotopes are fused together, different fusion products are produced, which in turn can fuel further fusion reactions. Most thermonuclear fusion reactions, such as the fusion of ²H with ³H, produce neutrons, whereas the fusion of ²H with ³He produces protons instead of neutrons which is referred to as “aneutronic” fusion. The fusion reaction of two ²H atoms is of particular interest to the scientific community, because it has a 50% chance to produce ³He and a neutron, and 50% chance to produce ³H and a proton, in addition to ⁴He.

A major difference between electromagnetic and electrostatic confined fusion systems from an engineering point of view is the simplicity of construction of IEC fusion systems compared to their electromagnetic counterparts. An IEC fusion system in its simplest form consists of a vacuum chamber with two concentric electrodes, to which a voltage of several kV is applied. When a suitable gas mixture is fed into the chamber at low pressure and then ionised, the positive ions are attracted to the cathode from all directions and accelerated towards it by the electric field. If sufficiently high energies are achieved, there is a chance that ions accelerated to the cathode from opposing directions collide and fuse. This elegant and simple concept was demonstrated in the 1960s by Farnsworth and Hirsch, who built the first electrostatically confined fusion system and patented it (Philo, 2025; Hirsch, 1967). Although the majority of fusion research has since focused on electromagnetic confinement with the hope to develop a sustained fusion reaction that can produce net positive electrical power, IEC fusion has received considerable interest from the scientific community as a means to produce high energy neutrons and protons rather than for its potential application in electricity generation, including for applications such has neutron activation analysis (NAA) (Eftekhari et al., 2016; Bhattacharjee et al., 2021; Miley et al., 2011) and neutron imaging (Takakura et al., 2022; Bakr et al., 2021; Kerr et al., 2022). The simplicity of an IEC fusor system and low construction cost make it attractive as a particle source. A community of “fusioneers” exists on the internet (Pasmann et al., 2020), and dozens of individuals have reported peer-reviewed home-built IEC fusion devices (Paul, 2009), including one as young as 12 (Lazendic-Galloway, 2022; Guinness World Records, 2020; Busby, 2019). The low cost also makes IEC fusion systems an attractive instrument for plasma and high energy physics research, with many schools, colleges, and universities having built comparable IEC fusion systems for educational and/or academic purposes (Maximilian Claus Constantin Messmer, 2019; Rasmussen et al., 2020; Raymond, 2020; Fettahoğlu, 2021; Alexandru, 2019; Jacob van de Lindt, 2020; Hermens et al., 2019; Bowden-Reid, 2019a).

The dominant metric of interest for fusors operated as a high energy particle source is the neutron production rate (NPR) or proton production rate (PPR), respectively. In a conventional IEC fusor, the NPR is proportional to the fusion reaction rate. Ions must collide with a sufficiently high kinetic energy, so there is a trade-off between collision probability and the mean free path of ions. If the gas pressure is increased, the probability of collisions also goes up because there are more ions in a given volume, however, ions are also more likely to collide before reaching the required energy threshold to fuse. This conundrum has historically capped the NPR of Farnsworth-Hirsch IEC fusors to 1 0 5 − 1 0 7 neutrons per second (Rasmussen et al., 2020; Masuda et al., 2001; Gu et al., 1995; Wolf, 2015), which is sufficient for most fundamental experiments involving neutrons or protons, including transmutation experiments and producing measurable quantities of radioactive materials, but challenging for applications such as the in-situ production of medical radioisotopes. One of the methods explored to break through this status-quo is by combining IEC fusors with solid state fusion concepts. Solid state fusion gets around the gas pressure limitation by immobilising fusion fuel in a solid state target such as the metal electrode itself. In this configuration, the electric field in an IEC fusor accelerates plasma ions towards a stationary target with a fuel density that is drastically higher than would be possible in the conventional approach, and the probability of fusion events occurring increases accordingly. This approach is referred to as “embedded fusion” or “solid state fusion”, occurring at or just below the surface of the target electrode. This concept was pioneered by Cipiti and Kulinski in 2003 for the ³He(d,p)⁴He reaction in an IEC fusor (Cipiti and Kulcinski, 2003), building on the work of Thorson et al. at the University of Wisconsin-Madison in 1998 (Thorson et al., 1998). Solid state fusion with specifically the ²H(d,n)⁴He reaction in an IEC fusor was first published by Nobirio et al. at Kyoto University in 2007 (Noborio et al., 2007), work continued by Bakr Arby et al., in 2018 (Bakr et al., 2019). It has received wide interest from the scientific community including NASA which has conducted experiments with deuterated erbium targets (Steinetz et al., 2020; Pines et al., 2020). Researchers at the Eindhoven University of Technology (Huisman, 2017), University of Sydney (Bowden-Reid, 2019b), and Portland State University (Jeffrey, 2022) also report increased neutron flux with IEC fusor designs using deuterated electrodes. Research has been done on the implantation of deuterium in zirconium metal with the specific purpose to increase the fuel density in IEC cathodes (Bakr et al., 2025; Verbelen et al., 2022; Verbelen et al., 2021). An increased particle production rate means that such fusors could potentially be used to transmute larger material quantities, potentially turning them into viable alternatives to research reactors for the production of medical radioisotopes (Verbelen and Scott, 2025) and radioactive sources for industrial applications. Production of medical isotopes with fusion devices is already being explored commercially by multiple technologies, including IEC fusors (Wallace-Smith, 2023; Heltemes et al., 2020).

Some of the excitement growing around IEC research progress, however, is countered by security concerns arising from low cost neutron generators capable of producing fast neutrons at high flux. Such fusors could theoretically be used to produce a variety of radioisotopes that pose a general nuclear security hazard, with fissile isotopes such as ²³⁹Pu, being well-known components of nuclear weapons, chief amongst them. A concern associated with high neutron flux sources that can be built subtly and at relatively low cost, could result in the production of sizeable quantities of fissile material outside regulatory control, turning such high flux fusors into a nuclear non-proliferation risk. Where in the past, the main barriers to proliferation were a combination of knowledge, materials and facilities/technology. In the current landscape, due to the internet and associated AI tools, the ‘knowledge’ barrier to proliferation has been significantly reduced, meaning that limiting access to materials and closely monitoring the technology development landscape is more important than ever for guarding against proliferation. The question is them raised whether IEC technology upsets the contemporary nuclear proliferation context; could it facilitate a nuclear weapons program by reducing the capital investments, resources, and/or manpower required, or by reducing the time necessary to produce sufficient quantities of fissile material? The technology to build reactors for plutonium production, plants for uranium enrichment, and spent fuel reprocessing has been in the public domain for nearly 70 years, however, in comparison to such elaborate industrial facilities, an IEC device is easy to build with widely available components, and can be constructed at marginal relative cost. A member of the general public with access to knowledge in the public domain can build a neutron producing IEC fusor with small footprint that effectively exists outside regulatory control (Moynihan et al., 2023; Lazendic-Galloway, 2022; Guinness World Records, 2020; Busby, 2019). Fettahoğlu (Fettahoğlu, 2021) demonstrated the feasibility of a home-made aluminium vacuum chamber to build an IEC fusor, further eliminating the need for specialised ultra high vacuum (UHV) parts.

Even ‘high flux’ fusors do have much lower NPR than, for instance, a fission reactor, so it is reasonable to expect them to not be especially effective at fissile material production. What is not obvious, however, is whether material for a nuclear weapon may be produced by many fusors operated in tandem over an extended timeline. The present work makes a technical assessment of the proliferation risk posed by enhanced IEC devices for conceptual production of ²³⁹Pu from ²³⁸U:

U 238 + n 0 → U 239 → 23.45 min β − N p 239 → 2.356 d β − P u 239 (1)

Or from natural thorium, T h 232, which can also be transmuted into fissile U 233:

T h 232 + n 0 → T h 233 → 21.83 min β − P a 233 → 26.98 d β − U 233 (2)

A simple calculation of absolute neutron requirements is given in the next section, followed by a review of how appropriate fusion-energy neutrons are for this purpose. Finally, the is a broader analysis of how IEC systems scale in comparison to conventional proliferation technologies (nuclear reactors/piles or uranium enrichment). Henceforth, ‘fertile isotopes’ and ‘feed’ will refer to ²³²Th and ²³⁸U, and ‘fissile isotopes’ and ‘product’ will refer to ²³³U and ²³⁹Pu. The international atomic energy agency (IAEA) deems 8 k g of either ²³³U or plutonium ¹ a “significant quantity” (SQ) of material, that a new proliferator could use to build a first weapon, accounting for manufacture losses and the learning required to achieve smaller-core weapons. The following analyses will assess the feasibility of producing this ‘target’ quantity of either isotope.

2 Modelling fissile material production

First, this work takes first principles approach to assess the viability for a single IEC device or array of multiple devices to produce the SQ of fissile material needed for a warhead. It is assumed that the IEC devices may be operated in either D-D or D-T fuel modes, producing 2.45 M e V and 14.1 M e V neutrons respectively. As a conservative model for order-of-magnitude estimations, it is also assumed that a feed of pure fertile precursor (100% ²³⁸U/²³²Th) is used; all produced neutrons are of appropriate energy to cause a transmutation by Equations 1, 2; and a neutron efficiency of 100% ² is achieved: each given neutron will produce one fissile atom, and there will be no further absorptions, e.g., from ²³⁹Pu to ²⁴⁰Pu. So, for a neutron source intensity of J (i.e., neutrons per unit time), the mass production rate, d m d t, for a target isotope, is given by:

d m d t = J A N A (3)

Where N A is Avogadro’s number, and A is the molar mass of the target isotope.

There is no clear highest sustained neutron production rate from D-T IEC devices, but the record D-D IEC fusor is 3.8 × 1 0 8 s − 1, which may achieve around 9.5 × 1 0 10 s − 1 with D-T (Wulfkühler et al., 2024; Ramzanpour and Pahlavani, 2018). Using this latter intensity would yield 3.8 × 1 0 − 11/3.7 × 1 0 − 11 g s − 1 of ²³⁹Pu/²³³U respectively, which to produce the 8 k g of fissile isotope needed for a warhead, would take a time of Equations 4, 5:

t P u = 8000 g 3.8 × 1 0 − 11 g s − 1 = 2.1 × 1 0 14 s ≈ 6.7 × 1 0 6 years (4)

t U = 8000 g 3.7 × 1 0 − 11 g s − 1 = 2.2 × 1 0 14 s ≈ 6.9 × 1 0 6 years (5)

Indeed, this is vastly longer than the half-lives of each isotope (24 k a for ²³⁹Pu, 160 k a for ²³³U), which means that with this setup of a single fusor, it would be physically impossible to generate 1 SQ of material; correcting for the decay of fissile material as it is made, with a half-life of λ, and generalising for a neutron efficiency ϵ, the mass m accumulated after any given time is expressed by Equation 6:

m = J ϵ A λ N A 1 − e − λ t (6)

Or, re-arranging for the time to achieve a given mass (Equation 7):

t = − 1 λ ln 1 − m λ N A J ϵ A (7)

It would be possible to use many fusors in tandem, however, in order to achieve a sufficient NPR to counter radioactivity. The record neutron production rate for any electrically driven fusion - reported as 4.6 × 1 0 13 s − 1 ³ by SHINE Technologies (Heltemes et al., 2020) - provides an example of the neutron production rates that are feasible today (which would be the equivalent to an array of ≈ 500 IEC fusors). Inputting this intensity to Equation 7 gives an operating duration of 17700 years for ²³⁹Pu, and 14700 years for ²³³U. The times to produce sufficient ²³⁹Pu and ²³³U for other J are shown in Figures 1, 2, for a given ϵ = 1.

From these basic calculations it seems that, at present, IEC fusors (and indeed other contemporary fusion neutron sources) fall extremely short of exhibiting neutron production capabilities able to generate nuclear weapons isotopes in useful quantities and timescales. As a rough limit for feasibility, a neutron source, or array of sources, would require an intensity of J ≈ 6.4 × 1 0 16 s − 1 to make 8 k g of fissile material in 10 years. In reality, the neutron efficiency, ϵ, will be lower than 1, which will drive up the source requirements as not every neutron will cause an absorption and transmutation. Depending on the exact geometries of a system, many neutrons will be lost by simply leaving the region of fertile material, or never entering in the first place - unless the source is surrounded by an infinite sphere of fertile material, which is not physically feasible. Any moderation or structural materials will cause some parasitic absorptions, and the interactions with fertile nuclei are by no means guaranteed to produce the desired product every time. A more detailed assessment of some of these limitations are examined in the following section.

3 Neutron energy analysis

The ideal case where each neutron emitted by the source corresponds to one fissile nucleus being produced, is unrealistic and incomplete. This section will review the effects of the neutron energy landscape to better assess the nefarious use of IEC fusors. As the neutrons are formed by fusion reaction, their energies leaving a source will be approximately 2.45 M e V (for D-D fusion), or 14.1 M e V (for D-T fusion), which both have a much greater chance of fissioning an irradiated fertile nucleus, than being captured for transmutation. The relationship between incident neutron energy and capture/fission cross section ratio is given in Figure 3, and shows that good conversion rates (with minimal fission) occur up to 100 k e V, which shall be referred to as the ‘intermediate’ energy spectrum. This is also a fairly typical fast breeder fission reactor neutron energy. This quality comparing fission and transmutation rates can also be expressed by the capture efficiency (Equation 8):

ϵ c = σ c σ c + σ f (8)

The next neutron energy consideration, is the interaction with the product. As levels of the products build up, there is potential for losses, either by fission or capture to non-fissile isotopes. The products are prone to fission easily, as that is their value. The proportion of product will approach a limit, depending on relevant neutron cross sections. Clearly, a practical setup would need to account for some regular reprocessing to extract the product before it is lost to fissioning. There is also the consideration of the rare instance of an intermediate nucleus (e.g., 239 U, 239 N p) undergoing another neutron interaction before decaying, changing its final state. This effect will be ignored, however, because of to the short half lives (minutes to days), compared to the decade of more of operations required for sufficient neutron fluences.

Finally, there is the matter of physical dimensions. For minimal neutron losses in the ideal case, to maximise production rate, bulk solid of uranium or thorium metal would be needed, isotropically around a source or sources, with a thickness on the order of the neutron mean free path, l. Thermal neutrons (25 m e V) have the largest cross sections for absorption, thus very small l, which would be useful for practical situations where minimal material requirements are important.

Of course, a design based on intermediate or thermal neutrons would require moderation of the neutrons. Increasing moderation would lead to some losses by absorption in the moderator (‘over-moderation’), and amplifies any geometric problems (e.g., neutron leakages, volumes of fertile material needed), but leads to a different neutron environment, which may overall be more efficient. However, fusion neutron cross sections are small enough that most neutrons will fail to scatter at all through a moderator that does not over-moderate (Picton et al., 1982; Gad, 1978). A layering system (moderator-feed-moderator-feed) may be optimal, but this adds to the complexity.

Numerical data relevant to each of these points are expressed in Tables 1, 2, for neutrons at each fusion and moderated energy. As is evident, the advantages and disadvantages for each are broadly similar for both fertile feed isotopes.

TABLE 1. ²³⁸U cross sections, capture efficiencies ( ϵ c ), and neutron mean free paths ( l ), for different incident neutron energies (Brown et al., 2018). ²³⁹Pu239Pu limit represents the theoretical equilibrium concentration of the product, from σ c feed / ( σ c feed + σ absorption product ).

TABLE 2. ²³²Th cross sections, capture efficiencies ( ϵ c ), and neutron mean free paths ( l ), for different incident neutron energies (Brown et al., 2018). ²³³U limit represents the theoretical equilibrium concentration of the product, from σ c feed / ( σ c feed + σ absorption product ).

Thermal neutrons are very well captured by fertile isotopes, and don’t require lots of material to interact, due to the low l, but the fissile products also capture and fission very well, meaning regular reprocessing is required with very low concentrations of product, such that it is extracted before it is lost. This is less of an issue for ²³³U production, however, due to the significantly higher ²³²Th capture cross section and lower ²³³U cross sections. Further demand for regular reprocessing will occur from any material that does fission - fission products can dramatically reduce the neutron population available for production (consider that X 135 e has a 2.65 M b thermal absorption cross section). Inevitably there will be some neutron losses due to parasitic absorptions by the moderator, and the increased distance from the source may induce geometric losses and increased feed material quantities. Although, these will be highly dependent on the exact moderator material and design.

Fusion-energy neutrons have problems with capture efficiency, namely, they very readily fission the fertile isotopes, especially with D-T neutrons, where the capture cross sections are so small that the fissile product will almost immediately be fissioned before building up. D-D neutrons can accumulate relatively high concentrations of product, meaning it is likely the best choice of unmoderated fusion energies, regardless of the higher neutron intensities that can be achieved with D-T systems.

There is some advantage to such high proportions of fission, as the population of neutrons will be multiplied; along with (n, xn) reactions, neutron multiplication in 238 U has been measured to be at most 3.3 from D-T neutrons. Though notable, this not the orders of magnitude required to bridge the feasibility gap. Thorium is expected to be similar due to broadly comparable cross sections. D-D will be significantly lower, as there are no (n, xn) reactions (which account for about half of the multiplication in D-T) and significantly less fission, also with fewer neutrons per fission (Brown et al., 2018; White, 1962; Mirnov et al., 2020; Ethvignot et al., 2005; Howe, 1984; Dubey and Kumar, 2023). The use of fission adds fission products to the equation for a would-be proliferator; the increase in fissile material breeding comes with an associated increase in reprocessing difficulty. Consider that a nuclear pile reactor may be a significantly simpler and more effective solution if significant reprocessing capabilities were available. Unlike with slower energies, the build-up of fission products does not have a significant effect on the neutron economy (e.g., 135 X e only has a capture cross section of 6.5 and 1 m b) for D-D and D-T neutrons respectively (Brown et al., 2018; Mirnov et al., 2020).

Intermediate spectra offer the best distribution of cross sections, giving close to no fissions of the fertile and fissile isotopes; the product builds up quickly and without need for rapid reprocessing. The need to remove neutron poisons is also lower, due to less fission and less significant capture cross sections. The same issues of moderator-induced neutron loses that exist for thermal neutrons are also present for the intermediate energies, but to a lesser extent, as less moderation is required. The problem now is that the cross sections in general are quite small; great volumes of fertile material are needed to avoid terrible geometric efficiencies (i.e., meters thick blocks of U or Th).

In all, no neutron energy is ideal so some compromise would have to be made by a prospective proliferator, but none of the options presented here are particularly desirable, and a real neutron spectrum will be some combination of each, so all effects must be accounted for.

4 Engineering practicalities

Finally, this section will review how a proliferating IEC system would stand from a practical standpoint and compare it to existing proliferation technology. The key drawback of IEC fusors is the very principle of their operation: they need a high power electrical supply. One of the best reported IEC fusors runs at 1.5 k W, for J = 3 × 1 0 9 s − 1 (Wulfkühler et al., 2025). Figure 4 shows the range of the most power efficient IEC devices (i.e., neutron production per kilowatt). It is difficult to predict improvements in energy efficiency, but a simple, conservative, linear power-intensity scaling to a 3 × 1 0 16 s − 1 NPR, for instance by a vast array, demands a continuous 15 G W from a would-be proliferator, which is likely an excessive undertaking for even the largest economies. Additionally, all of this power will result in heat release from the facility, which would be challenging to hide from remote observers. For IEC fusors, the highest NPRs can only be achieved for a short time using pulsed systems. An instantaneous neutron intensity of 4.70 × 1 0 9 s − 1 has been achieved in this way but required 576 k W (Wulfkühler et al., 2025). The best option to provide a dedicated gigawatt electrical supply would likely by an on-site fission reactor, in which case it would be simpler to just use such a reactor to produce fissile materials directly. An improvement in electrical efficiency of several orders of magnitude would be required for an IEC route to become desirable.

A valid historic comparator would be the two Windscale piles at Sellafield, each 180 M W, which were designed to produce a combined 90 k g of plutonium each year from natural metallic uranium fuel (Hill, 2013), and later repurposed to produce tritium. Pile technology is already well proven, does not require the complexities of fusor and power supply design, is a thousand-fold less conspicuous (going by thermal release), and does not need a large electrical supply. While single IEC devices are usually much smaller and thus easier to hide than a pile or enrichment facility, an IEC system array capable of large enough neutron outputs to fuel a weapons program will most likely be at least as large as the conventional facilities, at least because of the shielding requirements. Even in the unlikely case of minimal neutron leakage from the system, there will be a significant gamma flux to shield in addition. Of particular note would be sky-shine, needing shielding to stop detection from above; perhaps an underground facility would provide sufficient concealment, however this would be less convenient for a power supply and exacerbate any heat problems.

Further, D-T fusion (if used) would be reliant on the supply of tritium. For the neutron production asked for here, 18 P B q (≈ 50 g) of tritium is required over the duration of operations, which is currently not trivial to obtain (Pearson et al., 2018); the benefit of safeguards on tritium should be re-iterated.

5 Conclusion

The work presented herein has shown very clearly that even the highest intensity IEC fusion neutron sources are orders of magnitude too weak to produce kilogram quantities of fissile material for a weapon in any meaningful timescales. A neutron source in the 1 0 16 neutrons s − 1 range, as a bare theoretical minimum, is required for sufficient material to be made over 10 years. Granted, this could be achieved with an array of many smaller systems, but even 1,000 current IEC fusors in tandem would still take millennia.

A brief overview of the neutronics also suggests that there is no clear, efficient way to get high efficiency conversions of neutrons into fissile nuclei, meaning higher intensity sources will be needed. Additionally, regular chemical extractions of product from feedstock will be needed, possibly with extensive reprocessing infrastructure, due to the generation of fission products.

Indeed, a rogue entity intent on the production of a nuclear weapons, with the capability to acquire or fabricate the hundreds or thousands of IEC devices required to produce a weapon in good time, would likely find it much more attractive to simply build a uranium enrichment plant or basic nuclear pile/reactor to develop their weapons program. Even arguments for IEC devices being less conspicuous are wholly rebutted by their intense energy inefficiency.

Whilst further enhancements in IEC technology are likely over the coming years, only small incremental gains have been made since solid state fusion concepts were applied to IEC fusors for the first time in 2003 (Cipiti and Kulcinski, 2003), over a quarter of a century ago. An increase in > 3 − 5 orders of magnitude in NPR without a similar increase in power demand is unlikely. In the remarkable case of this occurring, then there would be a security motivation to regulate sources capable of producing in excess of perhaps 1 0 14 neutrons per second, accounting for the parallel operation of hundreds of devices, depending on their power requirements. This threshold, of course, would be highly context-dependent and is only an approximate guide; accessibility, ability to mass produce, capacity factor and many other practical factors should all be considered in a real scenario. The assertion is that at the present time, IEC fusor technology does not pose a proliferation risk as a low cost technical route to producing fissile materials. Therefore it is well placed as a candidate technology for sharing with non-NWS to support fundamental nuclear research, especially fusion, as well as a particle source for the production of certain medical radioisotopes. Finally, as tritium is key to the highest neutron production rates, proper safeguards on tritium access should be maintained.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Footnotes

Edited by: Arkady Serikov, Karlsruhe Institute of Technology (KIT), Germany

Reviewed by: Davide Pettinari, Polytechnic University of Turin, Italy

Yuanhao Gou, Tsinghua University, China

Footnotes

The IAEA makes no distinction in the isotopic composition of the plutonium, despite its great influence on suitability for a weapon. The only exception is that material with >80% 238Pu is exempt from all safeguards.

One could consider the addition of a neutron multiplying medium, e.g., Be, but here this will be treated as part of the neutron source intensity.

SHINE Technologies beam-gas target device, 2019.