There was no missing North Korea’s latest nuclear explosion. It went off on September 3, 2017, beneath a mountain at the country’s nuclear test site. It shook the ground so powerfully that it registered on distant seismometers as a magnitude 6.3 earthquake.

But what if there were ways to detect a nuclear weapons program early on, long before it succeeded in setting off an explosion? That’s the goal of researchers working on remote sensing techniques, such as satellite instruments to spot uranium mining or chemical detectors to sniff for byproducts of uranium processing. By looking for signs of weapons being developed, experts can intervene and try to stop further, more dangerous progress.

R. Scott Kemp heads the Laboratory for Nuclear Security and Policy at the Massachusetts Institute of Technology in Cambridge, Massachusetts, which aims for “peace and security through creative science.” Describing himself as a “policy physicist,” he is a rare bridge between the technical world of nuclear engineers and the political world of policymakers. He has studied the history of the gas centrifuge, used to enrich uranium for nuclear weapons; devised theoretical methods for telling the difference between a real and a fake nuclear warhead; and advised the US State Department as it was laying the groundwork to negotiate with Iran over its nuclear weapons program.

Kemp explored the promises and limits of detecting hidden nuclear weapons programs in a recent article in the Annual Review of Earth and Planetary Sciences.

This conversation has been edited for length and clarity.

Why is it important to try to monitor nuclear weapons programs earlier in development?

The goal is to try to prevent proliferation prior to it becoming a fait accompli. It’s very hard to get people to give up weapons that they already have. It happened when the South African apartheid government collapsed. They abolished their nuclear program prior to handing over control to the new government.

The international community has focused on detecting the production of fissile materials, the nuclear explosive fuels that are needed to make nuclear weapons. Typically you have to synthesize these through some kind of process that is very expensive, can be protracted and may have signatures that could be detected at a great distance. Look at the facilities that were built to support the Manhattan Project just before 1945. One of those, the uranium enrichment plant called K-25, produced material for the bomb in Oak Ridge, Tennessee. At its peak it consumed more electricity than the entire city of Detroit. That’s a detectable facility.

The problem is that the technology has changed. It has become more efficient and compact, and these sorts of facilities are easier to hide. It’s an arms race of hiding versus finding.

What sort of techniques could be used to detect these types of smaller facilities?

The unfortunate conclusion of the article is that there are very few hopes for finding these facilities. There might be advances in areas like hyperspectral satellite imaging, to more accurately characterize soil and rocks and try to discern whether these rocks that you’re seeing are related to uranium mining.

A couple techniques that might be explored are related to ultrasensitive detection of chemical effluents. Some of these processes might release trace amounts of unusual chemicals — for example, there’s a solvent that is used for extracting plutonium. When it’s exposed to radiation, the radiation can break some of those molecular bonds, creating strange and esoteric chemicals that might be released into the environment. And so advances in ultrasensitive detection of certain chemicals might enable detection at a distance.

We are looking at using carbon nanotubes in this way. These tiny tubes can have a chemical group affixed to them that makes them nonconductive. But if one of the strange molecules created by exposure to radiation comes floating through the air and attaches to this chemical group, it activates the carbon nanotube to become conductive.

The constraint is not merely a technical one. It’s not sufficient to show that you can see a given activity through some technical means. Whatever those technical means are, they have to be small enough, cheap enough and non-intrusive enough to be deployed around the planet to find these facilities without creating political problems. The question becomes, can you find something that is practical?

Here’s an example of what won’t work. People have talked for a long time about using antineutrinos, exotic particles produced through the process of beta decay. Antineutrinos are wonderful in that they can travel through the entire Earth without ever being absorbed, which makes them very appealing as a potential signature of, for example, nuclear reactors. But they are also very difficult to detect. The antineutrino detectors we would need to find those reactors would have to be so big that they would be a cubic kilometer in size. The cost of instrumenting them is beyond reason.

How easy is it to circumvent detection?

One of the things the Annual Reviews piece looked at was, could you detect uranium particulates that come out of a uranium enrichment plant or chemical conversion plant used to produce uranium-based weapons? You might be able to detect these on the order of 100 kilometers away. The problem is, if they put a HEPA filter on the air outlets of the plant, they can filter out 99.9 percent of the particulates and immediately destroy the signature for which you built the detector.

That has to factor into your decision about how much money to spend on this new detector scheme, when it can all be unraveled with a $20 air filter.

Then how can we defuse the threat of nuclear proliferation in the modern age?

Perhaps instead of trying to catch people and police the world, we need to change or remove the core motivations for acquiring nuclear weapons.

We’ve had 50 years where very few countries had the ability to make nuclear weapons, and those that did could be readily monitored. But now we are in a situation where just about every country can probably make nuclear weapons, and just about every country can probably hide it from our technical detection. We can hope that human intelligence and other modes of spying will save us, but the case history there is not great. It took 16 years to discover that Libya had a nuclear weapons program. Fortunately, in those 16 years Libya made almost no progress. That might not be the case with the next country that attempts to acquire weapons. And so we are in a sense living on borrowed time.

“Now we are in a situation where just about every country can probably make nuclear weapons, and just about every country can probably hide it from our technical detection.”

Scott Kemp

How did you get into this field?

I did an undergraduate degree in physics, but after graduation my interest began to wander towards the international stage. I decided to investigate the intersection of the two, and by some Brownian motion I stumbled upon and worked for Richard L. Garwin, who is perhaps the paragon for a national security physicist. I soon learned about the significance of nuclear issues, mainly from Frank von Hippel, and the general lack of attention they were receiving as the nuclear threat of the Cold War was supplanted by newer, peacetime concerns such as climate change. Nevertheless, as nuclear war was still one of the most credible paths to ending civilization, I felt I should take up the cause. And so I began to get educated, first through a PhD in public policy and then through a more practical education at the State Department, before I finally found my home in academia doing hybrid technical-policy research.

What keeps you up at night?

North Korea. Or to be fair, I would say the interaction between the US program and the North Korean program. The North Korean program would be less concerning to me if we had a more pro-engagement stance in the United States. It’s not just their program alone — it’s the politics between the two.

You were involved in helping lay the framework for the Iran nuclear deal, in which six countries agreed to lift sanctions on the country in exchange for its slowing its nuclear program. What happens if President Trump pulls out of the deal, as he has threatened?

Iran can restore its ability to produce nuclear weapons. Within about a year they can be at a state where they can make a weapon in a couple of weeks, maybe less. That is the price. If you throw out the deal, you give them a weapons program. More to the point, you give them a motivation to pursue weapons. All sorts of hellish scenarios can be imagined.

What are you working on now?

A way of doing nuclear archaeology. You can imagine many scenarios where you begin to have diplomatic engagement that rests on trusting the other side’s claims. Let’s say we start talking to the North Koreans, and they agree to turn over the highly enriched uranium they could use to produce nuclear weapons. We want to know that they’re really turning all of it over. Right now there’s no way to do that verification. I’m working on a way to reconstruct the historical operation of nuclear facilities so that we can.

How would that work?

All these nuclear processes involve radioactive materials, and every radioactive material produces a unique kind of radiation. Those particles damage the crystal structure of surrounding objects. For example, you might imagine a uranium warhead component damaging the metal shelving it sits on. Roughly 1 percent of that damage stays permanently frozen into the metal. It’s a really tiny quantity.

The question is, can we forensically analyze how much damage is frozen in, and use that to determine how much uranium has been there, and that it was indeed uranium. If we can, perhaps we can verify the claim there was a nuclear weapon sitting on that shelf for 25 years. The techniques for doing this have never been developed. We are starting from scratch to try to understand more about how radiation interacts with matter, and how the damage occurs and changes over time. If it works, it could be very powerful for many issues in arms control and nonproliferation.