The Case for Separating Fission from Electricity Production

2023 April 6 Twitter Substack See all posts

Nuclear power could level up by producing high value products. Thanks to Casey Handmer for sparking some of these topics.

The Vision

The current paradigm for nuclear energy is to harvest low-quality heat to produce steam. Neutrons, other radiation particles, and radioactive isotopes are byproducts that make the heat more expensive to collect.

But what if improving technology like photovoltaic panels or AI-powered robots flipped the paradigm? Could it make sense to produce energy-releasing isotopes that generate electricity when paired with semiconductors while wasting low-quality heat? Could these devices power things like flying cars? What if all the nuclear reactors could be in designated places like Idaho National Lab or former fusion bomb test sites? Do the potential costs make any sense? Can we produce enough isotopes?

Utilization of Radioisotopes

We use radioisotopes for PET scans, cancer treatments, sterilizing equipment and food, inspecting welds, evaluating oil and gas reservoirs, household smoke detectors, etc. They also have promise for increasing the power density of nuclear devices.

Radioisotope Basics

Radioisotopes are unstable isotopes of an element that undergo radioactive decay, releasing ionizing radiation until they reach stable isotopes. They emit energy in three ways:

  1. Alpha Particles:

    Alpha Particles are composed of two protons and two neutrons. They are relatively large and heavy and have a low penetration power. A sheet of paper or even the outer layer of human skin can stop them. However, they can cause significant damage if ingested or inhaled.

  2. Beta Particles:

    Beta Particles are high-energy electrons or positrons emitted during the radioactive decay of some radioisotopes. Beta particles are lighter and more penetrating than alpha particles, but thin layers of plastic, glass, or aluminum can stop them. You don't want to ingest or inhale beta particle-producing materials.

  3. Gamma Rays:

    Gamma Rays are high-energy photons that have no mass or charge. They are the most penetrating form of ionizing radiation and can pass through most materials, requiring thick shields made of lead or concrete to stop them and protect biological life.

Radioactive isotopes can emit more than one type of radiation, usually because they decay into other unstable isotopes that release different particles.

Desirable Traits of Radioisotopes for Electricity Generators

Isotopes need to check many boxes to be useful for power generation, especially in transportation applications. A few are:

  1. Power Density:

    Many radioisotopes store tons of energy, but only a few release it fast enough to power a car or aircraft.

  2. Half-life:

    The energy from radioisotopes decays exponentially. Relatively few radioisotopes have half-lives of more than a year. There is usually a tradeoff between half-life and power density. Longer-lived isotopes tend to have lower power density.

  3. Radiation characteristics:

    Alpha or beta emissions usually require less shielding than gamma rays, but the energy of the particles matters because higher-energy particles need more shielding. Strontium-90's decay chain emits a very high-energy beta particle that collides with other atoms and releases high-energy photons known as Bremsstrahlung radiation. As a result, strontium-90 requires more shielding than many isotopes that produce gamma rays.

  4. Availability and Production:

    The best candidates have high yields as fission products, or manufacturers can produce them by bombarding a parent isotope with neutrons with few side reactions.

There aren't many candidates and no perfect isotopes.

Historical Radioisotope Generators

Radioisotope generators have powered space probes, small devices, and remote facilities like lighthouses, where conventional power sources were not feasible.

In space exploration, radioisotope thermoelectric generators (RTGs) convert the heat generated by radioactive decay into electricity, providing a reliable power source for long-duration missions. Examples include Voyager 1 and 2, Apollo, the Mars rover Opportunity, and the New Horizons mission to Pluto. The primary radioisotope used in RTGs has been plutonium-238, known for its long half-life and acceptable energy density.

Radioisotope thermoelectric generators also provide a stable power source in inaccessible locations, eliminating the need for frequent maintenance and fuel replenishment. Strontium-90 has been the primary isotope for these applications because of its acceptable power output and lower cost than plutonium-238. It has extensive radiation shielding requirements that make it less useful for usage on space probes. A hundred-watt generator might weigh hundreds of kilograms. Solar panels and batteries have absorbed many terrestrial uses for RTGs.

Many pacemakers had betavoltaic cells for power. Betavoltaic cells are similar to solar panels and directly capture a beta particle instead of using its heat to run a thermoelectric generator. These nuclear-powered pacemakers fell out of favor once electrochemical batteries improved.

The low efficiency and high cost of conversion devices, the need for heavy shielding, and astronomically expensive isotopes made historical radioisotope generators absurdly niche products.

Modern Improvements Needed

The candidate isotopes that could power vehicles and have a reasonable near-term chance of usefulness are cobalt-60, strontium-90, and cesium-137 because they can be made easily in existing reactors. Of the three, only cobalt-60 will have the power density to power aircraft anytime soon. Prometheum-147, uranium-232, thorium-228, tritium, and curium-244 are more fringe but have longer-term potential. The deal killers are:

  1. Prohibitively Heavy Shielding:

    Virtually all of these isotopes or their contaminants produce high-energy photons. There are several possible strategies to reduce shielding.

    Shielding works at the atomic level by robbing energy from high-energy particles when they collide with parts of an atom. Denser elements usually reduce shielding thickness (higher chance of collisions) but also weigh a lot. There are modern materials, like plastic doped with heavier elements, that can be more mass-efficient.

    Radiation increases linearly with added nuclear material, while shielding decreases radiation exponentially. So shielding mass fraction decreases for larger power modules.

    Isotopes that emit particles with lower energy require less shielding. Energy-robbing interactions like the photoelectric effect become more probable when particle energy is lower.

    The shielding can also be the electricity generator, saving weight.

  2. Poor quality energy converters:

    Traditional RTGs or betavoltaic cells are only 3% to 7% efficient, adding significant mass and cost. But some off-the-shelf photovoltaic cells might work as a betavoltaic material because beta particles can create lower-energy secondary electrons when colliding with the PV cell. Efficiency can exceed 10%.

    Scintillators are materials that emit visible light when hit with high-energy photons. Scintillating materials run the gamut from high-performing materials like strontium iodide to the plastic used in dinner plates. These systems can reach nearly 20% efficiency and serve double duty as shielding. There are always integration challenges, but photovoltaics and scintillation are mature technologies.

    Thermionic cells are a solution with high-performance potential but lower technology readiness. They could reduce shielding while reaching 30%-40% efficient conversion of gamma rays.

    Direct conversion of alpha particles to electricity is still in the early stages. Alpha emitters tend to be more challenging to manufacture, limiting usage to niche applications.

The upside is that better solar cells and materials knowledge allow the construction of higher-performance radioisotope generators.

Beachhead Markets

Possible markets are space, defense, and terrestrial aerospace.


Traditional space will likely remain a small market. Satellite companies are comfortable with solar panels and batteries. NASA only builds a few deep-space probes and has different requirements than terrestrial users.

Defense - Aerospace

Defense is the market crying out for a solution like lightweight radioisotope generators. Small drones have captured mindshare in the latest Ukrainian war but can only fly for <20 minutes. That may work in a trench warfare slugfest where the enemy is only a thousand meters away. It is less useful in a scenario like the US fighting China, where drones must travel hundreds or thousands of kilometers.

A DJI Phantom 3 drone uses around 20 watts and weighs ~1.2 kg, with a quarter of that weight being the battery. Roughly six grams of cobalt-60 could provide the same power in a conversion device using PV cells and scintillators. The trouble is that the required material/shielding weighs several kilograms.

A better option might be using medical isotopes. They rarely get love in the radioisotope generator space because they have half-lives measured in days or hours, inadequate for space or remote power. But small drones won't need to fly for years. The gross watts per gram of the isotopes can be ~10x-20x that of cobalt-60, allowing them to sustain adequate energy levels for several months. Most importantly, their gamma rays are often much lower energy, decreasing shielding and dramatically improving total power density. Powering something like the DJI Phantom 3 would be a breeze with an isotope like iodine-131. The cost per gram will start at astronomical levels, but if the alternative for a long-range surveillance mission is using a 1000 kg conventionally-powered drone, then you can afford to pay it.

Cobalt-60 makes more sense for larger aircraft, especially high-altitude, long-endurance surveillance drones. The mass ratio of the energy converter and shielding is dramatically better at higher power levels. Its energy and power density can shrink the size of the drones and increase the loitering time. An example is the MQ-9 Reaper, which weighs 2200 kg empty and carries 1800 kg of fuel. A power pack using 50% purity cobalt-60 might weigh 611 kg, allowing the airframe and powertrain to decrease in size, letting the power pack shrink further. Loiter time would increase to years, making maintenance the limitation on flight time instead of fuel.

Savannah River Site's cost to produce a few kilograms of cobalt-60 was ~$140/gram (inflation-adjusted) in the 1960s, equivalent to ~$12 per gallon of diesel before considering savings from downsizing the aircraft, reducing maintenance, or simplifying logistics. The downsides are you must buy years of fuel upfront, and the fuel gets burnt whether you use it or not. It might be prudent to switch power packs between aircraft to increase utilization.


At some point, other industries might want small, long-endurance drones that reduce drone wrangling labor and improve practicality without requiring much larger, more expensive aircraft.

Cobalt-60 radioisotope generators might be a good fit for vertical take-off and landing aircraft like the Lillium. It'd make the most sense to size a small battery pack for take-off and landing while the generator handles cruise speed and recharging the batteries. It could fly across oceans instead of for an hour.

Defense - Naval

Underwater drones are another area with massive potential. The power density requirements are much lower than aircraft and could use strontium-90 or cesium-137 sources that are present in spent fuel. The material would go the furthest in a small intelligence-gathering craft like an underwater glider.

The Bottom Line

Electric power train, materials, and energy conversion technology have been plowing ahead, making radioisotope-powered vehicles more practical. Small aircraft and underwater drones benefit the most because internal combustion engines are too large or impractical, and batteries lack adequate energy density.

Long-Term Markets

Production costs of isotopes could decline at scale, making more uses practical. Batteries will still be cheaper for most applications, but some require more energy density. Larger aircraft, more military vehicles, and places with poor sunlight could be candidates. Eventually, engineers will take advantage of radioisotope generators to design new, optimized vehicles.

Scaling Radioisotope Production

How can we increase production from kilograms to tonnes to megatonnes?

Using Existing Reactors and Mining Nuclear Waste

Nuclear power plants can produce cobalt-60 by exposing cobalt-59 to neutrons inside the reactors. Most production is between 5% and 50% purity with high neutron efficiency. CANDU heavy water reactors make most cobalt-60, but increasing demand from the medical industry has pushed suppliers to increase production at existing sites and develop techniques to produce cobalt-60 in pressurized water reactors. US reactors could produce ~35 tonnes of pure cobalt-60 per year with a perfect neutron economy. Realistic capacity is a tiny fraction of that amount because existing reactors don't preserve neutrons. One larger drone might use 50 kg every few years, and a Navy destroyer would require 10-20 tonnes! Supply is inadequate if demand spikes because of improvements in radioisotope generators.

Strontium-90 and cesium-137 are fission products present in spent nuclear fuel. Fissionable isotopes are 3%-5% of commercial fuel. Strontium-90 and cesium-137 are around 12% of that mass. The US produces 10-15 tons per year of these radioisotopes. They have a fraction of the power density of cobalt-60 and can only provide a few megawatts worth of power devices per year. Their half-lives are much longer, allowing some mining of previously spent fuel. Spent fuel has hundreds of isotopes, making their recovery messy. The final power densities of strontium-90 and cesium-137 are only a quarter of their potential because of contamination from different isotopes of the same element and mixers to make them insoluble in water.

Cost is an issue for mining spent fuel because it requires complicated processing for small volumes. Strontium-90 was ~100x more expensive than cobalt-60 per watt, but processing more than a few kilograms would drop the cost. Part of the high cost comes from reducing worker exposure to radiation using shielding or substituting them with robots. Better robots could make manufacturing simpler.

Any new radioisotope demand will quickly outrun the supply from existing reactors and spent fuel.

Purpose Designed Reactors

Producing electricity from fission reactors creates many design challenges. Water-moderated and cooled reactors require high-pressure and high-temperature coolant, adding cost and complexity. Steam turbines and their associated equipment are a significant portion of nuclear power plant CAPEX. Nuclear power plants usually have complex systems to manage the concentration of neutron poisons that help control the reactors, adding cost to waste neutrons! Power plants must be relatively close to cities because power lines are inefficient and expensive.

Meanwhile, research reactors like the TRIGA design provide neutrons for research instead of producing power. The original design case was to be so safe that undergraduate students could run them. And they don't need containment structures. A combination of fuel design, size, and operating at atmospheric pressure without exotic coolants gives them that capability.

A reactor optimized to produce only isotopes could be a fraction of the cost of traditional nuclear power plants. And they could be in a few clusters away from cities.

Each kilogram of U-235 can produce approximately 0.38 kg of cobalt-60, 0.06 kg of strontium-90, and 0.06 kg of cesium-137. A $1000/kg price for 5% uranium-235 fuel would equate to ~$50/gram of raw material costs for cobalt-60. Most of the revenue would cover the CAPEX of the facility because a plant with the same thermal capacity as a 1 GWe nuclear power plant would be producing ~450 kg per year of product. ~99% of the energy would be waste heat, but it is not unprecedented. The oil industry flares massive amounts of natural gas because it is too expensive to bring to market. If cobalt-60 sells for $140/g, the reactor cost would need to be <$1000 kWe to pencil out (I'm using $/kWe to compare to nuclear power plant costs even though it's nonsensical). Some early customers might be willing to pay much more.

There is a tradeoff between capital costs and fuel enrichment level. The reactor can be smaller and have higher neutron intensity if more uranium-235 is in the fuel. The ideal enrichment amount might fall if reactor CAPEX decreases. Eventually, the industry might breed fuel in fast reactors, increasing the available fuel supply. The raw material cost could fall by as much as 95%.

The Department of Defense or the Department of Energy could regulate these facilities if they are the early customers rather than the Nuclear Regulatory Commission.

The solution might seem ugly because of the waste heat, but power and energy-dense fuels for transportation are valuable. Costs for the uranium burners could fall as we build thousands of copies on the same sites and iterate the design to drive further simplification. Both CAPEX and raw material costs will need to decrease to reach prices competitive with diesel/jet fuel on an energy basis.

Raw Material Availability

Cobalt, conversion materials, and uranium are possible bottlenecks.

Cobalt is plentiful compared to current uranium-235 reserves. It makes up something like 0.001% of the Earth's mass, so there is ample opportunity to find new supplies.

Relatively rare iodine is in the best scintillation materials. The supply of iodine can increase, but it would still take a lot of effort and expense to find new iodine-rich brine reserves and exploit them. Quantum dots embedded in plastic are one possible long-term solution.

Thermionic cells for gamma rays use materials like rhenium and gold. There may be opportunities to substitute lower performance but more available materials like lanthanum.

Land-based uranium-235 reserves seem pretty meager compared to what demand could be. There are certainly more reserves to find, though. The ocean has ~4 billion tons of uranium, and the expectation is that more uranium will leach from the ocean floor into the water as we remove it, maintaining a balance. But the uranium price required to recover these reserves might be too high to support more than niche uses of cobalt-60. If cheaper ocean extraction does not emerge, we might breed fuel to make uranium-238 useful.

Towards a Radioisotope Future?

It seems likely that improved radioisotope generator designs would kick off a gold rush for radioisotopes. Existing nuclear reactors are our source in the short term, but they are struggling to meet medical industry demand. Even minor usage from the military would overwhelm them.

Uranium-burning reactors could be a solution to increase supply. The supply of fissionable mass will need to increase from new reserves or through fuel breeding.

Power supplies for micro drones using medical isotopes seem like low-hanging fruit that would work even at very low conversion efficiencies, allowing iteration and improvement of generator designs. The most important uses of radioisotopes will be for aviation, especially once engineers create new aircraft to take advantage of the technology. It will take time to mature new reactors and expand production to handle larger aircraft, let alone ships or cars.


This report has extensive information on many isotopes.

Some quick data:

Cobalt-60 releases 17.4 watts/gram with a 5.3-year half-life and emits beta and gamma radiation.

Strontium-90 releases 0.95 watts/gram with a 28-year half-life and emits beta particles.

Cesium-137 releases 0.42 watts/gram with a 30-year half-life and emits beta and gamma radiation.