Home About My Books Résumé Chronological World of Wonders

Why Not Nuclear

Created/Modified: 2019-09-24/2019-10-02

Nuclear power is back in the news after a hiatus of decades, and I'm hearing a lot of arguments "against nuclear power" that come straight out of the 1970s.

I thought it would be useful to give my view of where we stand and why, and what the best way forward is, even though I know that no one is listening to this. I am a nuclear physicist and mechanical engineer and engineering physicist who has worked in technology development for decades, which means that even if I'm not precisely an expert, I'm still a member of the professional elite that modern populists will reject out of hand in favour of whatever prophetic guru or raving pundit happens to be in accord with their pre-existing conditions.

But as someone who has been arguing since the '80s for advanced nuclear of one kind or another as an effective way to lower our impact on the planet, it seems like I should at least make a token attempt to sway public opinion.

On the other hand, all I've got to offer are data and knowledge, which many people don't find persuasive.

This is long, so here's a handly Table of Contents. If you already have a good grasp of how data-driven policy-making works, the evidence that nuclear reduces greenhouse gas emissions, the way technology progresses, the role of neutrons, and conventional nuclear engineering, feel free to skip ahead to Modern Nuclear. I wanted a relatively comprehensive capture of my current thinking in one place, so people who are less familiar with that long list of introductory topics will be able to easily answer some of the questions that might arise when they read the core of the argument in the later sections.

  1. Being Data Driven: a simple test to see if this is worth your time
  2. What's Driving the Discussion: climate change and evidence that nuclear reduces CO2 emissions
  3. Technological Progress is a Fact: talking about aviation the way anti-nuclear politicians talk about nuclear would leave us grounded
  4. A Brief History of Nuclear Power: more than you probably want to know about neutrons
  5. Nuclear Engineering: putting nuclear physics in to practice
  6. Modern Nuclear: these aren't your grandparents nuclear reactors
  7. So Why Not Nuclear?: some dated and embarrassingly bad "arguments" put forward by people whose politics is more important to them than science
  8. But REALLY, Why Not Nuclear?: a few legitimate concerns
  9. But ...: are your politics more important than the future of the planet?

Being Data Driven

I don't want to waste anyone's time, so here's a useful exercise to decide if you should stop reading now.

It's very simple: write down three things you have changed your mind about as an adult, ideally in your late 20s or beyond, based on new information. Maybe first-hand experience, maybe second or third hand information, but data: numbers, natural histories, systematic observation, controlled or natural experiments, Bayesian inference, what-have-you.

Instances where you changed your mind, as an adult, based on data.

For me, the list would include these three, and many others:

If the reader can't write down any similar list of ideas where they have changed their mind as an adult, there is likely no point in reading any further, because the odds are overwhelming that they are unable to change their mind in the face of new data.

No one comes to adulthood with ideas and beliefs that just happen to be consistent with everything they will encounter in life, so if someone has never changed their beliefs as an adult based on data, they are almost certainly data-resistant, not hyper-intelligent, uncannily prescient, or intuitively tuned in to the deep truths of the universe.

The other possibility, I guess, is that the fact I've changed my mind based on data many times throughout my life means I'm not very smart.

What is Driving the Discussion

Before getting into the details of modern nuclear, it's worth considering what has brought nuclear back into public consciousness thirty years late.

Here is what global CO2 emissions looked like over the past half century or so with the number of nuclear plants being built overlaying it:

global CO2 emissions

See how the growth curve in CO2 emissions gets bumpy in the late '70s and early '80s? That's new nuclear capacity coming on line, helped a bit by the recession of the early '80s.

If we had continued to build nuclear capacity instead of coal capacity in the '80s and '90s, our climate problem today would be smaller. Our nuclear waste problem would be bigger, but we know how to solve that, as I'll explain below.

To get a sense of the impact of nuclear, consider Europe.

Nuclear fraction and carbon emissions in Europe

There are roughly three groups of countries: high-nuclear (> 50%), mid-nuclear (25-50%), and low nuclear (< 25%).

The high-nuclear countries--France, Slovakia and Hungary--all have carbon emissions of less than 7 tonnes per person per year.

Mid-nuclear has a bit of a bimodal distribution. Bulgaria and Sweden have less than 6 t per person per year, while Finland, Belgium, and the Czech Republic have over 10.

Low nuclear countries also exhibit wide variations, some of which may be explained by geography or GDP per captia. Luxembourg, weirdly, as an extremely high carbon footprint (21 t per person per year!) as does Estonia (14), while neighbouring Latvia and Lithuania have values around 4.

So there is more to carbon emissions than nuclear (duh) and I've not even bothered to break out transportation vs power generation, because I believe in giving the weakest argument possible for my case.

Weak arguments are far more robust than strong arguments. Weak arguments already have all the attack vectors built in. If anyone wants to look at these numbers and say, "BUT YOU HAVEN'T CONSIDERED..." well, yeah, I actually have.

We can however make a fairly robust statement about an interesting question from these numbers. Consider the distribution of per capita annual carbon emissions from non-nuclear countries:

The interesting question is: what are the odds of three randomly selected countries from this distribution just happening to all have per capita emissions of less than 7 t/year? That threshold is how I've characterized the high-nuclear countries above, and is based on casual observation.

If we randomly select three countries from the 14 non-nuclear countries in Europe, we get three below 7 t/year just 10% of the time. So it's far from impossible that what we're seeing in the data is just random chance, but 90% of time it isn't what we are going to see.

And remember, this is restricted solely to Europe (including, for the moment, the UK) so all these countries are at least somewhat developed and in the same general part of the world. I'm not comparing apples to orangutans.

I could push this analysis a lot harder, but the point should be clear: it is at least reasonable to think that maybe having a high fraction of your power generated from nuclear relates to having lower-than-average per capita carbon emissions.

That's why some people who think that climate change may be an existential threat are once again interested in nuclear. Me, I just like radiation, in the hope that a freak accident will give me super-powers, or at least better hearing.

Technological Progress is a Fact

Commercial nuclear started in 1954 and new development stopped, post-Chernobyl, in the late '80s, not even 40 years later. For comparison, commercial air travel got off the ground--as it were--in 1913, with KLM (the longest-lived commercial air transport company still operating under the same name) founded in 1919. So put the date for commercial air travel at 1914, and ask what air travel would look like today if technology development had stopped in 1954, two years after the introduction of the first commercial jet aircraft, and we had built no new designs since then.

Air travel would be incredibly unsafe and inefficient compared to what we actually have, after another half-century of research and development.

1952 was a pretty typical year for air travel at the time (I did not cherry pick this, I goofed in my search and got 1952 instead of 1954, which I haven't even looked at... but if you want to fact-check me and look at the years around 1952, please do so! I'm reasonably confident of what you'll find.):

So there you have it. Air travel is clearly inherently unsafe, vulnerable to terrorist attack, and inextricably linked to military uses. It not only kills passengers, but people on the ground. One crash narrowly missed an orphanage.

We can extrapolate from these incidents (and I've just listed one major incident for each month of 1952... there are many more) that as air travel grows over the next several decades the death rate--both among passengers and completely innocent people on the ground--will grow exponentially with it. Failures in air traffic control, already evident in these data, will escalate as human controllers are overwhelmed by the impossible task of guiding these flying death-machines through the sky.

Obviously we must ban air travel because none of these problems can ever be fixed.

Thankfully, we didn't ban anything, and the past few years have been the safest ever seen in absolute terms. Fewer people died in aviation accidents in 2018 than did in 1954, although air traffic has increased by nearly a factor of a hundred since then.

Technological progress is a fact.

To claim that any particular technical problem is inherently unsolvable has always been a bad bet. There may be economic or other practical limitations that prevent a solution from being adopted at scale, but it has always been a good bet that if we want something enough, we will figure out a way to do it.

This is not a moral judgement. One can easily argue that we ought not to want certain things. Fine. Make that moral argument. It hasn't worked against non-procreative sex, being gay, being trans, using mind-altering substances for fun, taking the Lord's name in vain, working on Sundays, or dancing in the streets. But maybe the desire for cheap, clean, abundant, and safe power is more malleable than at least some of those things. I personally doubt it, but people are welcome to try.

Personally, I think we ought to fly less. I've been fighting the good fight against air travel--which is dreadful for the environment, no matter how safe it has become--by limiting my own flying as radically as I can for over a decade, with an average of less than one long-haul flight per year in that time, and often going for years without setting foot on an plane, all while running a successful international consulting business. Interestingly, there is currently a huge push to make aircraft more environmentally friendly, with a local short-haul outfit (Harbour Air) planning to go all-electric in the next few years, and save money in the process.

Electric and hybrid aircraft are being developed because a lot of people are catching on to the idea that air travel is bad for the environment, but still want to keep on doing it. Telling them they ought not to want that doesn't work for sex, drugs, or hip-hop, and there's no particular reason it will work for anything else. If people want something badly enough, we almost always figure out a way to do it.

A Brief History of Nuclear Power

Nuclear physics is complicated. Sub-atomic particles behave in ways that are hard for most people to imagine, and it's difficult to come up with analogies that explain them, mostly because of quantum.

That said, I'll try to give a basic explanation of what happens in a conventional nuclear reactor and why.

The atomic nucleus consists of two kinds of particle: neutrons and protons. I envision them as kind of base-ball like solids, all clumped together under the influence of the strong nuclear force, with tiny little electrons whizzing around them out in the distance someplace, a good hundred metres away from my baseball-scale nucleus-thing.

All of our normal experience of atoms is mediated through the electrons, which are responsible for most of the chemical properties of elements. Atomic physics is mostly about electrons, which are are stable particles that interact with a long-range force (electro-magnetism). Nuclear physics is mostly the physics of neutrons, which are unstable particles that interact with a short-range force (the strong nuclear force). The detailed nature of the nucleus doesn't matter much in every-day life, so most people don't think much about neutron physics. But it matters a lot in a reactor.

Whereas in chemistry and everyday life we are mostly concerned with elements, like hydrogen, oxygen, iron, and uranium, in nuclear physics we are mostly concerned with isotopes. "Element" and "isotope" are just concepts of convenience that name things that are important to us in different contexts.

An element has a specified number of protons (and therefore specific chemical properties) but an unspecified number of neutrons (which don't affect chemical properties significantly), while an isotope has a specified number of protons and neutrons (which are both important to nuclear properties). For example, C-12 (pronounced 'carbon-12') is the most common isotope of carbon, with six protons and six neutrons, but C-14 also occurs, with six protons (which is what makes it "carbon") and eight neutrons, which don't change its chemical properties to speak of, but do make it slightly radioactive.

Protons have positive electric charge, which is balanced by the negative charge of the electrons once you're far enough away. In the nucleus itself, though, those positive charges are repelling each other: opposite charges attract, like charges repel. This makes it harder for nuclei to hold on to protons.

The strong nuclear force is basically a contact force. Think about it like the neutrons and protons are dipped in molasses. When they touch, they stick.

The strong force holds the nucleus together against the electro-static repulsion of the protons, but the heavier a nucleus is, the more neutrons it needs in the mix to remain stable. Too many neutrons, though, and they fall off.

As we add neutrons and protons the nucleus becomes more stable until we get to iron. After that, each additional neutron or proton is a bit less weakly bound, until we get to the actinides, like uranium and thorium. These are so weakly bound that some isotopes can spontaneously fission, falling apart into two daughter nuclei with about half the number of protons and neutrons, and a few neutrons left over, because the extra neutrons aren't needed any more and are in excess of what the smaller nuclei can hang onto.

What happens to those extra neutrons? There are two or three of them left over in most cases.

They go bouncing away through whatever material they find themselves in, like the ball in a pinball machine. Because the mid-range nuclei that result from fission have higher binding energy than heavy nuclei like uranium and thorium, there is quite a bit of energy left over from the fission process--some of the mass of the original nucleus has been converted into energy via Einstein's famous E = mc2 relation--and most of it ends up in the neutrons. So they come away pretty energetically.

As they bounce around they lose energy by banging off of the other nuclei in the vicinity. They can travel a few metres in most materials while doing this. If they happen to encounter a nucleus that can absorb another neutron, they may get gobbled up and become part of another (possibly now-radioactive) nucleus. If they live for long enough (about fifteen minutes, which never happens unless they are in space) they have a good chance of decaying into a proton and an electron: free neutrons are unstable, thankfully, because if they weren't the universe would be flooded with them and life would be unlikely at best, even weirder than it actually is at worst. Or maybe the other way around.

Eventually, the neutrons slow down to the point where they are in thermal equilibrium with the atoms around them. When this happens they have the same chance of gaining and losing energy on average in any collision with an atom. They're moving at about 2200 m/s (600 km/hr) at room temperature, faster at higher temperatures.

Despite their speed, most of them don't get very far. Thermal neutrons are easily absorbed by a very wide range of other nuclei. The hydrogen in ordinary water will happily absorb one to make deuterium. Heavy water, which has deuterium in place of ordinary hydrogen, is very unwilling to absorb a neutron, which is one of the many tricks that went in to the CANDU reactor design in Canada, which has provided the province of Ontario with 60% of its power, safely, cleanly, and often quite cheaply, for the past 40 years.

But most other materials are pretty accommodating where an extra neutron is concerned, and will happily take one on, absorbing a neutron and changing into another isotope in the process.

One of the things that can happen when a heavy nucleus like certain uranium and thorium isotopes absorbs a neutron is that the nucleus undergoes fission. It splits apart, creating more extra neutrons in the process, and releasing energy while doing so.

This is the key to nuclear power: if we can create a situation where for every fission, exactly one neutron goes on to create another fission, we have a stable chain reaction. Too few, and the reaction dies out. Too many, and CNN gets to provide 'round-the-clock live coverage.

Fortunately, the odds of a neutron causing fission increases as they run down to lower temperatures. That means that if the reactor starts to heat up, the reaction will slow down. This "negative temperature coefficient of reactivity" is one of the things that helps keeps conventional reactors stable.

"Delayed neutrons" are another important source of stablity, which are also important to reactors that depend on faster neutrons. These are emitted by radioactive fission fragments that first undergo a beta-decay and then emit a neutron. Because beta-decay lifetimes are long, this process can take minutes. This allows external control of the reaction to keep it balanced at a point where the reactor is sub-critical in terms of the prompt neutrons from fissions, but super-critical as a result of delayed neutrons.

"External control" in this case means control rods made of cadmium or similar material that has a very high neutron absorption probability, but doesn't undergo fission so doesn't produce any extra neutrons. Materials like this act as neutron "sinks" that mop up any free neutrons that run into them. More on these materials below.

All of this should hopefully reinforce the idea that nuclear physics is complicated and nuclear power is not as simple as it might first look. Reactors are expensive for a lot of reasons, some of which are the result of burdensome and needless regulation, but many are due to the real complexities of building a reactor that is safe and stable.

Nuclear Engineering

That's it for the physics. Now for the engineering, which is the business of ignoring most of the physics and using a bunch of practical guidelines and workable approximations to create useful machines.

It is relatively easy to build a working nuclear reactor as device that will maintain a stable chain reaction for a few months or a year. That's a physics experiment. It's relatively hard to build a working nuclear power plant that will generate electricity safely and cheaply for decades. That's engineering.

Conventional nuclear reactors consist of several main components:

In many designs the moderator and coolant are the same: water or heavy water. Regular (light) water absorbs neutrons quite readily, so light-water moderated reactors have to burn enriched uranium. Heavy water reactors, like the CANDU, can burn natural uranium. I'll talk more about enrichment below.

The problem for a nuclear power plant--as opposed to just a reactor--is that we want to make steam from the reactor, so we can drive turbines to generate power. But we need liquid water in the reactor core to moderate the neutrons and keep the core cool. So we need water to be both a liquid and a gas under the same circumstances and at the same time, which my friends in the philosophy department call a "contradiction".

There are two tricks engineers use to get off the horns of this dilemma.

One trick is to change the circumstances by keeping the water in the core under pressure, and running a lower-pressure secondary loop that creates steam. This has the advantage of keeping the primary coolant--which could get contaminated with radioactives--out of the steam turbines. These are called Pressurized Water Reactors (PWRs).

But it also means the core, where all the exciting highly radioactive stuff is happening, is full of water under high pressure and high temperature, which is basically a steam explosion waiting to happen, and a single steam explosion in your reactor core can ruin your whole day. Fortunately they very rarely happen.

An alternative trick to to run the core at lower pressure and allow the coolant to boil within it and drive the turbines directly, then run through a heat exchanger to cool it, and back into the reactor. These Boiling Water Reactors (BWRs) are a surprisingly popular design.

Either way, nuclear reactor cores are difficult engineering environments. Neutrons do all kinds of damage to most metals, which builds up over time, causing cracking, corrosion, and other things starting with "c". Since cracks in the core can result in steam explosions in PWRs, this makes reactors a maintenance nightmare, especially since all the work has to be done via remotes due to residual radioactivity even when the reactor is powered down.

When stable isotopes absorb neutrons they often become unstable isotopes, which can have long half-lives. Furthermore, when an unstable isotope does decay, they change their chemical nature, so an iron crystal in a metal pipe now has, for example, a cobalt atom in the middle of it where an iron atom should be. This kind of thing builds up stress in the lattice and creates opportunities for things to go wrong.

Still... a typical fossil power plant burns a railcar full of coal every fifteen minutes, releasing poisons like lead and arsenic--whose half-life is infinite--and dumping enough carbon dioxide and ash into the atmosphere to have significant long-term effects on the Earth's climate. Since there is increasing evidence that those effects are going to be large enough to end global civilization and kill hundreds of millions or billions of people, investigating alternatives seems like a good idea, even when those alternatives are politically unpopular due to an effective smear campaign run by non-experts for a couple of decades many years ago.

Modern Nuclear

Like commercial passenger aircraft in the early 1950s, conventional reactors have some issues. They're complicated and awkward in ways that may one day be put to shame by the streamlined and elegant designs of the future.

The single biggest problem with conventional reactors is their use of water as a coolant. This creates the worst possible environment inside the core for things to go catastrophically wrong. Even though loss of life in nuclear accidents like Fukushima is tiny compared to the numbers coal plants kill day in an day out all over the world, the potential for damage to the core, which effectively writes off the reactor as a useful machine, is significant.

The modern alternative to water is molten salt, usually lithium fluoride, but sometimes other salts as well. This immediately fixes two major problems with conventional reactors: the core in molten salt reactors runs at low pressure and high temperature--typically around 700 C--which means making steam in a secondary coolant loop is easy, and the possibility of a steam explosion in the core is zero because there is no water present.

Furthermore, in the most advanced current designs, the fuel can be mixed directly into the molten salt, so the fuel and the coolant--rather than the coolant and the moderator--are coupled.

Engineering has two very basic tricks: one is taking two functions that used to be separate and putting them together. This increases efficiency.

The other is taking two basic functions that used to be together and taking them apart. This increases robustness.

For example, the Wright Brothers' first aircraft were steered via "wing warping": the whole wing structure twisted to turn the plane, so their wings had two functions, lift and control. This weakened the wing structure but at the time produced superior control to ailerons, which had been invented by English scientist Matthew Piers Watt Boulton decades previously. Ailerons are separate components, mounted on the wing, that were refined in the years after the Wright's first flight, and became the universal means of flight control within a decade.

This is the way invention actually works: outside of straight-up violations of physical law, no one is able to predict what will be successful, or possible, until it is actually tried. People who confidently proclaim that nuclear will "never work" should all be incredibly rich, if they are really capable of accurately intuiting which technologies will be ultimately successful and which will not. So far as I know, this is not the case. A few rapacious anti-nuclear shills have been able to make millions on their institutes and books and pricey speaking engagements, pandering to the prejudices of their fellow-travellers, but true to their anti-science roots, they never bother to check their predictions against reality. If they did, they'd find their data-free prognostications were rarely close to the mark.

In the case of modern nuclear, mixing the coolant and fuel has a lot of advantages in terms of both efficiency and robustness, which wins both ways because we get increased robustness by splitting the functions of the moderator and coolant, and the increased efficiency of mixing the coolant and fuel.

To start with, it isn't possible for the reactor to leak coolant without leaking fuel, and modern designs have a catchment area beneath the core that will collect any leakage into dispersed channels to prevent any further chain reaction from occurring. Inside the core, a coolant leak is also a fuel leak, so if the low-pressure coolant does leak out, the reactor shuts down. Loss of fuel will do that.

There are many additional advantages.

Liquid salts are less corrosive than water, because almost everything is less corrosive than water. Water is a horrible problem from an engineering point of view. It is very nearly a universal solvent: almost everything dissolves in it, especially at high temperatures. Liquid salt coolant solves many corrosion and materials issues.

Even better, if the fuel is mixed with the coolant, a number of issues with fuel integrity also go away. In conventional reactors uranium fuel is typically in small cylindrical pellets or slugs that are slid into channels in the core. These get damaged by radiation and "poisoned" by fission decay products.

Conversely, as anyone who has tried to punch the ocean knows, liquids are very hard to physically damage. No matter how strong a solid is, it has a crystal or rigid amorphous structure of some kind that neutron radiation can disrupt in multiple ways. Liquid salts just go on being liquid.

Poisoning is a dominant practical issue in reactor operations--to the extent that its neglect was instrumental in creating the Chernobyl disaster--and is the reason why conventional reactors are designed to run continuously. When a uranium nucleus undergoes fission it typically ends up with two fragments, each around half the mass of the original. So U-235 produces two daughter nuclei with masses around 235/2 ~ 118, which is the atomic mass of the most common isotope of tin.

The fission process is quantum, so two identical nuclei will quite happily split into a wide range of daughter isotopes, with everything from cadmium to barium being fairly copiously produced. Momentum and energy considerations favour roughly-equal masses for the daughters, but it's a matter of probabilities, with some isotopes being preferentially produced due to angular momentum effects in the phase space of the decay.

Iodine-135 is copiously produced as a daughter product, and decays to Xe-135 with a six or seven hour half-life. Xe-135 itself decays with a 9.2 hour half-life, but while it exists it has a stupidly high neutron absorption probability, something like ten thousand times higher than U-235 itself, which is already 500 times more likely to absorb a neutron than average nuclei.

This means Xe-135 eats neutrons like crazy, and reactors need a little extra "push" to maintain a stable reaction. By pulling out the control rods a bit further than would otherwise be required, extra neutrons are created that are absorbed by Xe-135, effectively "burning it up" in the process. This maintains the reactor in equilibrium... while it is running.

When a reactor is shut down, the accumulated I-135 that has been built up will continue to decay with a 6.7 hour half-life. The Xe-135 that is produced by this process decays with a 9.2 hour half-life. So for the first 12 or 18 hours, Xe-135 builds up. Forty-five minutes to an hour after shutdown, for most reactor designs, there is so much Xe-135 in the fuel that it is impossible to restart the reactor until most of it has decayed away, a day or so later.

This is a huge issue for practical nuclear, which can only be taken out of operation for 45 minutes, or more than 24 hours. It's like god is teasing us, or nature is being deliberately unkind, by making an otherwise wonderful power source as awkward and difficult to use as possible.

As often happens, though, if we are sufficiently clever, we can beat the unkindness of nature.

Molten salt reactors with liquid fuel are sufficiently clever. Xenon is a noble gas, and one of the features of noble gases is to be effectively insoluble in pretty much anything. Noble gases come out of solution. That means in a liquid-fuel molten salt reactor, Xe can be purged from the system continuously. It never builds up so long as there is enough heat to keep the salt liquid.

This means molten salt reactors can be run up and down in power very quickly, letting them be used to load-follow much the way gas turbines are, and this means liquid fuel molten salt reactors could fill in the gaps left by renewables until we figure out the storage problem. Since natural gas only looks good compared to coal--which is a bit like saying a murderer looks good compared to Jack the Ripper--this is worth keeping in mind.

Furthermore, it isn't just xenon that can be removed from the fuel: all fission products can, as part an external circulation loop. This allows a much higher fraction of the fuel to be burned. Typically only about 3% of the U-235 in a reactor undergoes fission before the fuel has to be removed and reprocessed, due to a variety of issues, mostly related to buildup of fission products. Liquid fuel reactors can achieve over ten times this, and get something like a billion kilowatt hours out of less than 100 kg of uranium.

This also means most of the transuranics (plutonium and the like) created in the reactor get burned up in the process, absorbing neutrons and decaying into something stable, or undergoing fission themselves. This greatly reduces the long-lived nuclear waste, although since nuclear waste disposal is a solved problem this is not as big a deal as it was in the 1970s. The separated fission products, being lighter, tend to have relatively short lifetimes.

How data resistant people deal with these facts is a curious and unpleasant question, and just thinking about it makes me sad. The world is full of new information, and "updating our beliefs in the face of new information" is the most fundamental duty of a Bayesian adult. Sure we all have different priors and biases, but nothing absolves us of the responsibility for shifting how plausible we think a thing is when faced with new data that's relevant to it. Nothing excuses faith.

Molten salt designs span the range from fast reactors--which are especially good at burning up their own waste--to epithermal or thermal reactors. Thermal and epithermal designs are more like conventional reactors in their underlying physics, as they slow neutrons down using a moderator. Moderation is usually done with graphite, which has a bad history because conventional reactors run at low temperatures and this allows radiation damage to build up in the graphite. This can cause the core to catch fire, which is bad. Fortunately, because molten salt runs at much higher temperature, the damage to the graphite gets continuously annealed out, and graphite moderation becomes much less of a problem (I never thought I'd say that, as I don't much like graphite as a moderator, but hey... that's what the data suggests.)

Molten salt designs have both negative temperature coefficient of reactivity and negative void coefficient of reactivity, which means if you pull the control rods out the reactor gives a little bump in power, heats up, and stays stable, and if you lose the coolant, it just shuts down, as described above. Molten salt reactors are inherently stable.

The most common salt used is lithium fluoride, which has a melting point of around 400 C. It has low toxicity (143 mg/kg is the LD50 rats, which happens to be precisely equal to the LDLO--the lowest observed lethal dose--of acetaminophen in humans).

Uranium and/or thorium fluoride salts are mixed in to the coolant, and the whole mass is reasonably well-behaved as industrial chemicals go.

The one wrinkle (there is always a wrinkle) is that the lithium typically has to be isotopically enriched. Naturally occurring lithium has both mass six (three protons, three neutrons) and mass seven (three protons, four neutrons), and 6Li is much more likely to absorb a neutron (twice as likely as uranium) which we don't want happening in a reactor core. Fortunately, we already know how to do this, as 6Li is highly sought after for use in hydrogen bombs, so for many years most commercial lithium was already somewhat enhanced in 7Li, and making it moreso is not that difficult. Natural lithium is already over 92% 7Li.

That's where the nuclear industry is going, if it is allowed to go there. But there are a lot of people who don't want it to go there.

So Why Not Nuclear?

Sometimes I see people giving "reasons" they don't think nuclear should be one part of our approach to climate change, which may be a civilization ending event. They go like this:

Let me stop there, because it's getting silly.

"What about the waste?" is the "What about mental illness as a cause for mass shootings?" of the nuclear "debate". It isn't so much a question as a declaration that the 1970s was a great decade, the best decade, with the best arguments. So many great arguments. I mean... so much winning. Really. Waste. What a winning decade. Argument.

In fact, we've known what to do about nuclear waste for decades, and if you click on the link you'll see it isn't some pie-in-the-sky proposal, but an actual engineering report on an actual facility including detailed system drawings and performance measurements.

Vitrification and burial works.

Nuclear waste is a solved problem.

At least, it would be, if anyone would let us implement the solutions at scale.

People who ask about nuclear waste as if it wasn't a solved problem look to me like prohibition proponents look at public health conferences dealing with drug abuse. Maybe they're well-meaning, but when people ask them, "Are you in the loop about the last few decades of research and development on this topic?" they may as well reply, "Wait, what??? There's a LOOP???!!"

If these are the best "arguments" against nuclear power, one can only assume that the people who make them never fly, because flying forty years after commercial aviation started was still really dangerous and barely beginning to develop ways of coping with simple things like "flying over water" and "managing air traffic around busy airports" and "stopping people with guns from taking over planes", which in fact still isn't really a solved problem.

If the problems that nuclear power has today are inherently unsolvable, those problems with flying must have been inherently unsolvable as well, so flying today cannot be any safer than it was back then, because the arguments are premised on the idea that technological progress does not happen, which is false.

But REALLY, Why Not Nuclear?

There are some reasons. But, you know, actual reasons, not recycled bullshit from the 1970s that was only marginally relevant then and completely outdated today.

The big one is: it might not be needed.

Renewables are growing faster than anyone but me predicted ten years ago, to the extent that renewable capacity in Europe is growing faster than demand and in parts of the southern US renewables are putting the financial viability of utilities at risk as people drop off the grid.

Solar and wind are ramping up like crazy. Storage solutions are still a long way behind, but for rich people in sunny climates getting off the grid is a practical reality today, and who cares about people who aren't rich and living in the sun?


The real question isn't "nuclear or not?" but "how much nuclear?" Some of us think "quite a bit, and a lot of money for research on new designs" is a good answer. In particular existing nuclear should be cherished and maintained. Shutting down nuclear power stations in the age of climate change is an extremely high risk endeavour.

Consider this: would you rather shut down or keep running a power station that has operated safely for decades when shutting it down will increase the risk that a civilization ending event might occur?

Is shutting off a machine that has operated safely for decades really such a huge risk reduction compared to staving off global catastrophe for long enough to get other solutions in place?

Is there anyone outside of the German or Japanese parliaments who really think it's a good idea to decommission reactors we already have running, that pose less risk to the public than flying to an international climate conference does, and that will definitely help us bridge the gap between where we are today and where we want to be in terms of green house gas emissions?

There must be, I guess, as those politicians are pandering to voters who they think will find those policies popular, but it seems to me an incredibly cynical and cavalier decision, that puts the planet at higher risk for no good reason.

So even if it turns out renewables really can ramp up fast enough, do we feel OK about putting an option that experts think is a pretty good one on the shelf because a bunch of activists convinced people to be afraid of it when it was 1950s-aviation-unsafe?

I don't think so.


Isn't it proliferating? Don't reactors lead to bombs? (Thought I was going to try to skate by on that, didn't you?)


There have been times in my career when I've been a lot more tepid about nuclear than I am now because I was sceptical about climate change and I was painfully aware, as a Canadian, that everyone we sold reactors to turned around and made a bomb, or tried to make a bomb.


That was then. Here's an unfortunate reality, made possible by the big strides in development of gas centrifuge technology in the 1980s: you don't need a reactor any more to make a bomb. All you need is uranium ore and some completely conventional engineering.

There are several types of nuclear weapon. Uranium bombs, plutonium bombs, and hydrogen bombs are the major technologies. Uranium and plutonium are both fission bombs, working on the same basic principles as nuclear reactors. When people worry about nuclear proliferation, they are mostly worried about fission bombs, both because they form the core of fusion (hydrogen) bombs, but also because even though they are "small" compared to hydrogen bombs they really can still ruin your whole day.

Uranium is really, really easy to build a bomb with. Terrifyingly easy. But only uranium that is highly enriched in U-235. Natural uranium, which contains just 0.7% U-235, can't explode. It's even hard to use as reactor fuel.

Plutonium, on the other hand, is fairly hard to make explode, requiring an implosion-type weapon with carefully designed and manufactured explosive lenses, precise machining, and extremely precise timing in detonation to ensure a symmetrical implosion. Plutonium weapons can be thought of as an orange that has been sliced into pieces and then explosively reassembled.

Unlike uranium, it is relatively easy to produce plutonium... if you have a reactor, which is effectively a copious source of neutrons. Some of the U-238 nuclei in reactor fuel will capture a neutron, converting them to U-239 which beta-decays to Pu-239, a highly fissile material suitable for bomb-making. Significantly, plutonium can be chemically separated from the fuel, which is much easier than the kind of isotopic enrichment required for uranium bombs.

The only issue is that if left too long in the reactor, the Pu-239 can absorb another neutron, becoming Pu-240, which has a high rate of spontaneous fission. This means that if there is too much Pu-240 in the mix, neutrons from spontaneous fission will set the bomb off before it gets properly assembled by the implosion, causing it to fizzle. Keeping the fuel cycle of a reactor short fixes this problem by not giving time for much Pu-240 to build up.

The unfortunate reason why "fission reactors are proliferating" is no longer a very good argument is that isotopic enrichment is now much, much easier than it was back in the day, thanks to gas centrifuge technology, which means it's much, much easier to jump from "enriching uranium sufficiently to build a reactor to produce plutonium to build a bomb" to "enriching uranium sufficiently to build a bomb." No reactor required. All you need is uranium ore. Or maybe sea water.

Isn't technological progress grand?

So while it used to be true that having a nuclear reactor made a big difference to weapons production, today it is considerably less of a barrier to not have one. As far back as the 1970s, Pakistan took a pure-uranium approach to nuclear weapons, and in fact when they detonated their first bomb in 1998 it was a uranium bomb. I didn't actually know that when I started writing this, but I did know that if I were to set out to build a bomb it would use U-235 and focus on enrichment, not Pu-239 and focus on a reactor.

This means the "incremental proliferation potential" of reactors has gone down, albeit not for a great reason.

To put it simply: advances in gas centrifuge technology since the 1980s, and more recent work on laser separation, mean that nuclear reactors no longer have any unique role to play in nuclear weapons production. If they didn't exist, states and state-like-actors could still produce copious numbers of uranium bombs.

So objecting to nuclear reactors on the basis of their proliferation potential is a little like banning cars on the basis of their use as getaway vehicles in bank robberies. Other options exist, and when that is the case, targeting one of them doesn't have as much effect.

Gas centrifuges don't depend on anything but conventional technology. They are a challenging but essentially straightforward mechanical engineering problem.

So eliminating civil nuclear programs would put a minor dent in state's ability to build nuclear weapons today. The effect wouldn't be zero, but it would be an increasingly long way from effectively preventing them from getting the bomb.

Pakistan didn't need a reactor to build a bomb twenty years ago. No one else needs one today.

So... do we really want to eliminate an effective tool against climate change to do nothing much about people's ability to build nuclear weapons?

And because molten salt reactors burn up much more of their fuel and produce much less waste as a consequence, they also produce less fissile material at the end of their fuel cycle. This is not precisely "non-proliferating", though, because one could easily inject U-238 into the coolant and extract the Pu-239 on a continuous basis. But the spent fuel itself would be essentially useless to terrorists wanting to make a bomb, so that's an improvement over conventional nuclear as well.


Philosophers tell us that there's a thing called "deductive closure", which says that if you know something, you also know everything you could possibly deduce from that thing, like Sherlock Holmes on steroids. All with no effort or time taken.

Unsurprisingly, thinking doesn't work that way. Consistency is something we have to strive for with humility and great effort. It isn't easy. Me, I take pride in my humility, and work hard to follow the data where it leads me, and tease out the consequences of what I believe, slowly, awkwardly, stumbling along, but guided by what is in the world as much as possible, and not what is in me.

I've repeatedly said that if we want something badly enough its a good bet we'll get it.

So: what about solar with storage, which has the potential to Solve All Our Problems? Why not throw investment at that, despite relative lack of progress on storage for the past few decades?

I think we should, because it's the same argument in both cases: my argument for modern nuclear as one important approach is that we can't know it won't work and it quite plausibly will work. My argument against solar and wind with storage as our sole line of attack is that we can't know it will work, even though it's pretty plausible it will.

Because sometimes things fail.

No matter how hard we try.

No matter how much we want it.

Nuclear might be one of those things.

Renewables with storage might also be one of those things.

So it is consistent to take an approach that doesn't especially privilege renewables over nuclear when experts are telling us both have potential, and both have problems that we will have to work to overcome, and we may fail in either case.

Even though technological progress is a fact, and when we want something badly enough we find surprisingly innovative ways of making it work with remarkable frequency... sometimes we don't.

Overconfidence, one way or another, in technological prediction is something I see a lot of in people who have never put money or time into technology development. Lots of people are happy to say, "It'll never fly, Orville" right up to the point of takeoff. Lots of other people are happy to say, "We'll eventually work out the problems with exploding hydrogen and get those zeppelins back in the air! Nothing can stop us!"

Remember fractal image compression? Probably not, because it never worked.

I'm a student of failed technologies, and fractal image compression is one of my favourites. The idea was that simple mathematical formulae (fractals) could generate remarkably life-like images, so maybe we could take a real image and find a set of fractals that could generate it.

This was in the days when 300 baud dial-up was the standard, and getting pr0n off a BBS took forever, or so I'm told. Fractal image compression promised to speed that process up by orders of magnitude, and there was a point where it looked like it might be possible. But it wasn't.

Even though the technology "worked" in some limited sense, no one was ever able to figure out how to bring the speed up to the point where it was actually useful. Fifteen hours to encode a minute of video--even granted hardware was slower back then--was never going to become popular.

And as often happens, the "problem" got overtaken by other technological advancements: good compression is less important in an age of unlimited bandwidth.

It is often the case that competing technologies aimed at solving the same problem coexist for long periods of time. Sometimes the people working on one or the other stumble on a winning advantage, or the world changes in other ways to make the problem go away. No one argues about the edge versus the point in the age of firearms. Sometimes different people continue to use different solutions, because neither has any universally compelling advantage over the other.

What is always the case is that we are lousy at picking the winners and losers in advance.

So the flip side of "Technological progress is a fact" is "Technological progress can fail." Simply because we can and do move forward farther and faster than any reasonable person would believe possible, doesn't mean that the effort and cost to do so isn't vastly greater than any reasonable person would estimate, or that all developments will be successful. In fact, most of them will fail.

Think about technological obstacles like a mountain range, and research being exploration to find a path through it. Most of the routes we explore will go no-where. More often than not, eventually, we find a way through. But sometimes we don't, or find a route that will take a small determined party, but be useless for any kind of large-scale traffic.

My business for the past twenty years has been primarily focused on helping scientists, engineers, and businesspeople bring technology out of the lab and into the world. One of the biggest lessons I've learned in that time is that no one is any good at predicting what will succeed and what will fail. I'm certainly not, and anyone who is stands to make billions, so anyone who hasn't made billions is probably not really much of a technological prognosticator.

Renewables plus storage are absolutely worth investing in, but like nuclear they face significant technological challenges, particularly on the storage side. We have a good chance of solving those problems. But we might not.

If we don't, doesn't it make sense to have a second technology in development? One that experts say has great promise and is only opposed by non-experts like the people at the weirdly misnamed "Bulletin of the Atomic Scientists"?

Modern nuclear is at least as promising as renewables plus storage on a number of fronts, especially now that waste disposal is a solved problem.

Liquid fuel molten salt reactors solve a wide range of problems with conventional nuclear, including low waste production, zero meltdown potential, excellent burn-up characteristics, high temperature, low pressure operations, and inherent stability.

Waste disposal is a solved problem to anyone not still living in the 1970s, and gas centrifuge technology means having or not having a reactor no longer has much bearing on a state's ability to build nuclear weapons. Zero-entry mining eliminates risk to miners, and with molten salt designs costs and lead times will very likely come down, buying us time while we figure out what to do next.

So... why not nuclear?

Contact Home World of Wonders
Copyright (C) TJ Radcliffe