Where Does the Radiation Go?

A Rule Nature Doesn’t Bend

Start with something that isn’t up for debate: energy and mass are conserved. In a nuclear reaction, when two atomic nuclei fuse together, the tiny bit of mass that vanishes has to reappear as energy, carried away by specific particles in specific, well-measured proportions. Physicists have measured these proportions so many times, in so many experiments, that they function like a recipe nature is required to follow.

For two deuterons (that’s the nucleus of deuterium, a heavier version of hydrogen with an extra neutron) fusing together, the recipe looks like this: about half the time you get tritium (another hydrogen isotope) plus a proton; about half the time you get helium-3 plus a neutron. Only rarely — about once in every ten million fusion events — do you get ordinary helium-4, and when you do, that event releases its energy as a single, powerful burst of gamma radiation.

This matters because gamma rays and neutrons are both forms of radiation you can detect from across a room. If you warm up a cup of water using deuterium fusion, ordinary physics says your lab should also be getting flooded with neutrons. That’s the deal. You don’t get to keep the heat and skip the radiation.

The Strange Claim

A handful of experiments in a field called “solid-state fusion” (you may have also heard the older name, “cold fusion”) report something that seems to break this deal. They report excess heat coming out of a deuterium-loaded metal electrode, helium-4 building up in amounts that roughly track that heat, occasional traces of tritium above what you’d expect from background sources — and hardly any neutrons at all. Far, far fewer than ordinary fusion physics would require.

Think of it like a friend telling you they’ve been running five miles a day for a year but somehow haven’t burned a single calorie. Maybe they’re lying about the miles. Maybe there’s a totally different explanation for why they look fit. Or maybe — and this is the interesting possibility — there’s something genuinely new going on that doesn’t follow the usual rules of “exercise burns calories.” You don’t get to decide which of these is true by arguing about it. You get to decide by measuring things very, very carefully.

An Open Question Worth Sitting With

Nobody has identified a nuclear process that would let two deuterons fuse into helium-4 without producing neutrons, tritium, or a gamma ray. If solid-state fusion is real, it would mean nature has a rule scientists haven’t found yet — a genuinely novel nuclear channel. That’s a huge claim, which is exactly why it needs huge evidence. The way to settle it isn’t debate; it’s building instruments sensitive enough to catch whatever radiation is (or isn’t) actually there.

How Do You Audit Something Invisible?

This is where a whole different group of scientists gets interesting: the people who spend their careers doing environmental radiation monitoring — checking whether a nuclear power plant, a hospital, or a contaminated site is leaking radioactive material into the air, water, or soil. Their entire job is finding a small, real signal buried inside a messy, noisy natural background. That is precisely the skill this puzzle needs.

Here’s the toolkit they’d bring to a solid-state fusion cell, and what each tool would be listening for.

Helium detection — the hardest one. Helium-4 is already floating all around us — it makes up about 5 parts per million of ordinary air. Any experimental cell that leaks even slightly (and most lab equipment leaks a little) will pull in outside helium, which can easily be mistaken for helium made inside the cell. That ambiguity is a big reason a major U.S. Department of Energy review in 2004 came away split on whether the reported helium was real. The fix isn’t more arguing — it’s leak-tight equipment and mass spectrometers precise enough to tell “helium that leaked in from the room” apart from “helium made by a nuclear reaction inside the cell.”

Gamma-ray and neutron detectors. Specialized detectors can either spot the specific gamma-ray signature that helium-4 fusion should produce, or — just as usefully — rule it out. A well-designed instrument that finds nothing is still telling you something important: it sets a hard ceiling on how much of this reaction could possibly be happening. In fact, the field’s own literature backs up how well this works. A careful 2016 attempt to reproduce a solid-state fusion heat claim found essentially nothing (a signal of 6 milliwatts, indistinguishable from zero, across 231 tries) — and traced the tiny amount of noise that did show up to ordinary hardware quirks like calibration drift, not to anything nuclear. Careful measurement caught the artifact.

Tritium: The Tracer That Can’t Be Faked

If helium is the hardest signal to pin down, tritium is the cleanest. It’s a radioactive isotope of hydrogen that no ordinary chemical reaction can produce — only a nuclear one can. It decays slowly (with a half-life of about 12 years) by releasing a weak particle that standard lab equipment can detect down to very small quantities. If a deuterium cell is genuinely making tritium above the natural background level, and that amount rises and falls with how the experiment is run, that’s about as close to a nuclear smoking gun as this field is going to get.

There’s also a second reason to take tritium seriously: it’s already a regulated environmental hazard, with a legal limit in U.S. drinking water. That means the exact careful measurement techniques needed to test the physics claim — leak-tight sampling, precise counting, background subtraction — are the very same techniques required to keep workers safe. Testing the science and protecting the people doing the testing turn out to be the same task.

Keeping the Books Honest

One more thing environmental science brings to the table: habits for making sure a measurement can be trusted by someone who wasn’t in the room when it was taken. Every raw reading, calibration, and step of the analysis gets logged in a way that lets an outside reviewer retrace exactly how a final number was produced. And in some fields, when a scientist analyzes their own data, they don’t get to see the true value until after they’ve locked in how they’re going to analyze it — a practice called blind analysis. It’s a safeguard against the very human tendency to see what you’re hoping to see.

None of this proves a heat signal is real or nuclear. What it does is remove an entire category of doubt — the suspicion that a result only looks impressive because of how it was measured or analyzed. That’s valuable whether you’re trying to prove the effect is real or trying to show it isn’t.

So Why Bother?

It’s worth being honest about where the odds currently sit. In 2019, a team backed by Google re-examined these claims using modern equipment across roughly 420 samples and did not find excess heat — and they published that negative result rather than burying it. That’s the realistic starting point. At the same time, that same project pointed to specific gaps in older experiments, especially around the materials science of loading enough hydrogen into a metal to even give an effect a chance to appear. In 2023, the U.S. Advanced Research Projects Agency–Energy (ARPA-E) committed roughly $10 million across eight research teams, including groups at MIT, Stanford, and Lawrence Berkeley, to re-test these claims with modern instruments. The current stance isn’t belief or dismissal — it’s measurement.

And that’s really the whole point of this story. These are radiological claims, which means they get settled by the people who are experts at counting radioactive decays and characterizing background noise for a living. Bring the full toolkit — sensitive helium detection, gamma and neutron spectrometry, tritium counting, and an honest chain of custody for the data — to a well-built experiment, and one of two things happens. Either you find a radioactive inventory that ordinary chemistry can’t explain, which would be a genuinely major discovery. Or the careful audit closes the case in a way even a skeptic has to accept, and a stubborn, poorly understood corner of physics finally gets put to rest.

Both outcomes count as a win for science. Neither one is available to anyone who isn’t willing to actually do the counting.

Why This Is a Good Habit of Mind, Not Just a Physics Story

This whole piece is really a case study in how scientists handle an extraordinary claim: not by shouting it down or cheering it on, but by asking what specific, checkable prediction it makes, and then going and checking. “Where did the radiation go?” is a question anyone can ask. Answering it takes patience, the right instruments, and a willingness to publish the answer even if it’s boring.

That combination — curiosity plus rigor — is the actual engine of scientific progress, in this field and in every other one.

Editorial note: This primer presents an introductory synthesis of solid-state fusion's relationship to environmental science and radiation safety, adapted from a technical primer for a general audience. The underlying nuclear claims of SSF/LENR remain scientifically contested. Readers are directed to primary experimental literature, and to the companion expert-level primer, for a more detailed and fully sourced treatment.

Introduction: The Ledger With No Overdraft

Start with the version of the puzzle that has nothing to do with belief. A nuclear reaction is a bookkeeping system with no overdraft facility. Fuse two deuterons and the mass that disappears has to reappear as kinetic energy carried by specific particles, in ratios set by nuclear structure. Roughly half the time you get tritium and a proton; roughly half the time helium-3 and a neutron; only about once in ten million events do you get helium-4, which sheds its 23.8 MeV as a single hard gamma.¹ Warm a calorimeter with deuteron–deuteron fusion and, on the standard branching, you must also flood the room with neutrons.

A subset of solid-state fusion (SSF) experiments report the opposite ledger: measurable excess heat, helium-4 accumulating in rough proportion to that heat, tritium above background, and a neutron flux orders of magnitude too small to balance the books.² For an environmental scientist this should read less like heresy and more like a familiar kind of case. Either something is producing energy without the expected radiological signature, or the measurements are catching an artifact. Both possibilities are questions about detection, background, and mass balance. They are your questions.

So here is the one worth putting at the center. If the heat is nuclear, where is the radiation that must accompany it, and can we account for it atom by atom and becquerel by becquerel? The value of that question is that it cannot be answered by argument. It is answered by counting, and by the discipline of knowing what your instrument would have seen if the effect were real.

Section I — The Audit Any Heat Source Has to Pass

Fusion power sets a hard floor on the reaction rate, and the floor is where the anomaly bites. Take the reported pattern at face value: helium-4 accumulating in step with the heat, at roughly 23.8 MeV per atom. Known nuclear structure makes helium-4 from two deuterons only through the radiative branch, about one event in ten million; the other ten-million-minus-one throw off a neutron or a triton. To supply helium-4 at the rate the heat implies through conventional fusion, the main channels would have to run some ten million times faster still, spraying neutrons at a rate of order 10¹⁸ per second. Nothing of that kind is seen. That is the real size of the deficit, and it is why the positive claim is not slow ordinary fusion but a claimed novel channel, one that yields helium-4 without the neutrons, tritons, and 23.8 MeV gamma that known physics requires. A channel like that would contradict some of the best-tested nuclear physics there is, which is exactly why the prior against it is low and the evidentiary bar high.

The weaker version of the same accounting is still worth stating, because an instrument checks it directly. One watt of deuterium fusing to helium-4 corresponds to about 2.6×10¹¹ events per second.³ Had that watt instead come from ordinary deuteron–deuteron fusion, the neutron branch alone would light up any detector in the building. Either way the room should not be quiet, and in the anomalous cells it largely is.

The environmental discipline already owns the correct framing for this. Radiation protection is, at bottom, a mass-and-activity balance: what went in, what is present now, what is leaving, and whether every pathway is accounted for within stated uncertainty. Applied to an SSF cell, that same balance becomes the sharpest possible test of the nuclear-origin claim. If the mechanism were conventional fusion, the product ratios would be fixed in advance and the audit would be simple bookkeeping. The anomalous claim is precisely that the ratios are not the conventional ones, that helium-4 arrives stripped of its usual escort of neutrons and gammas, and that unconventional inventory is itself the extraordinary part. Either way the claim is falsifiable in the strong sense: it predicts specific activities and specific isotopic shifts, and those either appear at the required level or they do not.

Keep two statements rigorously apart, because most of the field’s confusion lives in the gap between them. “Excess heat was observed” is a calorimetric claim. “The heat is nuclear” is a claim about products and rates. The first can be true while the second is unproven, and an environmental audit is precisely the instrument that separates them. You are not being asked to referee a theory of the lattice. You are being asked whether the radiological inventory adds up, which is a measurement.

Section II — What Environmental Science Can Actually Measure

The most useful thing the field brings is a mature craft of finding small signals inside large, structured backgrounds, and of proving the signal is not the background wearing a disguise. Most of the tools the SSF question needs already exist in routine regulatory use, and the one that does not is a reach into an adjacent specialty rather than an invention.

Take helium first, because it carries the heaviest interpretive load and the worst background problem. Helium-4 is a permanent 5.24 parts-per-million component of air, and its isotopic cousin sits at an atmospheric ratio near 1.4×10⁻⁶.⁴ Any cell that leaks, and most apparatus leaks a little, breathes in a helium reservoir that dwarfs the claimed nuclear yield. That is exactly why a 2004 Department of Energy review split on the helium evidence: several reviewers judged the reported levels close enough to background to admit contamination.⁵ The honest reading of that split is not “case closed.” It is a specification. It says the measurement must be done leak-tight, with static-vacuum or high-resolution noble-gas mass spectrometry able to resolve the claimed excess against air, using the helium-3-to-helium-4 ratio as a fingerprint of origin. This is the one part of the audit that reaches past routine regulatory counting: separating helium-4 from the deuterium molecular ion at mass-to-charge 4, where the two species differ by only about 0.026 u, demands resolving power well beyond a standard quadrupole and belongs to noble-gas geochemistry rather than to a typical radiochemistry bench. That is a reason to recruit the specialty, not a barrier. The correlation between excess power and helium that the field leans on hardest was reported three decades ago, and has been challenged since on statistical and calorimetric grounds rather than simply left alone;⁶ what it has never received is a modern, blinded, contamination-controlled repeat by a laboratory that measures helium for a living.

Gamma spectrometry is the second pillar. A high-purity germanium detector resolves gamma lines finely enough to identify specific nuclides and, just as importantly, to set quantitative upper limits when nothing is there. A credible null is a scientific product in its own right, provided the sensitivity is stated: an upper bound on activation or on a 23.8 MeV signature, at a known confidence level, constrains the nuclear-origin claim whether or not it finds anything. Neutron spectroscopy, using energy-resolved time-of-flight detection, does the same work on the branch that defines the anomaly, converting “no neutrons seen” into a numerical ceiling on the deuteron–deuteron rate.

The field’s own literature already shows this discipline catching artifacts, which is the strongest argument that the tools are up to the job. A careful 2016 replication of a laser-stimulation heat claim returned a clean null (6.1 ± 21.6 mW over 231 trials) and, more valuable than the null itself, traced the limiting noise to hardware: thermocouple behaviour, calibration drift, and end-cap instability of a few hundred milliwatts.⁷ That is a low-level measurement done right. It is also a warning that the artifact floor is often physical rather than statistical, and that no amount of reanalysis lowers a hardware floor.

Section III — Tritium: A Test and a Safety Question at Once

Tritium is where the empirical question and the health-physics question become the same question, which makes it the natural entry point for this field. It is a genuine nuclear tracer that no chemical process can counterfeit. It has a 12.32-year half-life and decays by a soft 18.6 keV beta to helium-3,⁸ a signature that liquid scintillation counting reads routinely, with electrolytic pre-enrichment pushing detection limits into the single becquerels-per-litre range.⁹ If tritium appears above background in a deuterium cell and tracks operating conditions, that is a nuclear event demanding an explanation. Reports of exactly this exist in the SSF literature, and they are best treated as reported rather than established: sporadic, unevenly replicated, and in need of the same leak-tight, blinded, background-resolved treatment as the helium.

The measurement discipline here is unforgiving in a productive way, because tritium is also a regulated environmental hazard and the analytical chain is already codified. Tritium occurs naturally from cosmic-ray spallation in the upper atmosphere, so a background always exists and must be characterized before any excess is claimed. Its regulatory footprint is concrete: the U.S. drinking-water limit is 20,000 picocuries per litre, set against a 4 mrem-per-year dose basis.¹⁰

For SSF this cuts two ways at once. A rigorous tritium balance is simultaneously the cleanest available test of a nuclear channel and a straightforward radiation-safety obligation, since tritiated water is mobile, is absorbed readily, and demands containment and monitoring wherever it is produced at all. The same becquerel you count to test the physics is the becquerel you count to protect a worker.

This is the substantive bridge, and it runs in the direction environmental scientists will find most persuasive. You do not have to accept any claim about the lattice to see that a deuterium-loaded electrolytic system is a radiological environment that deserves proper characterization. Doing that characterization well happens to be the same act as adjudicating the science.

Section IV — Monitoring, Provenance, and the Credibility of a Number

A reported activity is only as good as the record behind it, and this is the second place the field’s methods matter. Regulated environmental measurement long ago formalized what it means for a result to be trustworthy. The ALCOA+ criteria (attributable, legible, contemporaneous, original, accurate, and then complete, consistent, enduring, available) amount to a demand that the final number be re-derivable from its raw record by someone who was not in the room.¹¹ SSF’s reproducibility troubles are, in large part, provenance troubles: raw streams that were never preserved, calibrations that cannot be reconstructed, analysis choices made after seeing the data. A field that already lives under chain-of-custody rules can import that scaffolding wholesale.

Blind analysis is the companion practice worth transplanting. Where a small signal rides on a large, uncertain background, a secret offset frozen before unblinding removes the analyst’s expectation from the answer, and it is standard in nuclear and particle measurement for exactly the situations SSF finds itself in.¹² None of this makes a heat signal real or nuclear. That is the important limit. A validated, blinded, provenance-complete pipeline yields a trustworthy result, which may be a clean null. What it removes is the whole category of objection that has dogged the field for a generation, the suspicion that the number is an artifact of how it was produced. Removing that suspicion is worth as much to a skeptic as to an advocate.

Section V — Why Take It On

The base rate deserves to be stated plainly, because pretending otherwise is how a field loses its audience. When a Google-convened team re-examined these claims with modern methods across roughly 420 samples, they did not reproduce excess heat, and they published the negative result.¹³ That is the honest prior. The same paper, though, mapped specific corners of the parameter space, especially the materials science of reaching and holding extreme hydrogen loading, where the older work was simply not good enough to conclude anything. In 2023 ARPA-E committed about $10 million across eight teams, including groups at MIT, Stanford, and Lawrence Berkeley, to settle the low-energy-nuclear-reaction question with modern instrumentation.¹⁴ The institutional posture is no longer belief or dismissal. It is measurement.

That posture is an opening for environmental science specifically. The claimed phenomena are radiological claims, and radiological claims are adjudicated by people who count decays, resolve isotopes, and characterize backgrounds for a living. Bring low-level counting, noble-gas mass spectrometry, gamma and neutron spectroscopy, and a regulated chain of custody to a well-characterized cell, and one of two things happens. Either an inventory appears that conventional chemistry cannot explain, which would be a discovery of the first rank and would need every safeguard your field knows how to build. Or the audit closes the question on terms a skeptic can accept, with sensitivity attached, and a poorly mapped corner of the parameter space is retired. Both outcomes are wins for the discipline, and neither is available to anyone unwilling to run the count. The radiation, if it is there, will not tally itself.

Editorial note: This primer presents a scholarly synthesis of solid-state fusion's relationship to environmental science and radiation safety. The underlying nuclear claims of SSF/LENR remain scientifically contested. Evidence tiers (A = replicated/consensus, B = single peer-reviewed source, C = preliminary/contested) are noted inline. Readers are directed to primary experimental literature for empirical evaluation.

References & Footnotes

1. The three deuteron–deuteron channels and their approximate branching: D+D → T + p (Q ≈ 4.03 MeV, ~50%); D+D → ³He + n (Q ≈ 3.27 MeV, ~50%); D+D → ⁴He + γ (Q = 23.8 MeV, radiative branch of order 10⁻⁶). Standard evaluated nuclear data: D. A. Brown et al., “ENDF/B-VIII.0: The 8th Major Release of the Nuclear Reaction Data Library,” Nuclear Data Sheets 148 (2018): 1–142. ↩
2. The absence of a commensurate, energy-matched flux of nuclear products (neutrons, gammas, tritium, helium) accompanying the reported heat is a standing skeptical objection. U.S. Department of Energy, Report of the Review of Low Energy Nuclear Reactions (Washington, DC, December 2004). The general accounting argument is well supported (tier A); specific positive product reports are single-community and contested (tier C). ↩
3. One watt of D→⁴He fusion power corresponds to about 2.6×10¹¹ events per second (1 W ÷ 23.8 MeV, with 23.8 MeV = 3.8×10⁻¹² J). Two distinct arguments follow. Energy balance: had the same watt come from ordinary deuteron–deuteron fusion (Q ≈ 3–4 MeV), the roughly 50% neutron branch would produce a neutron field far above any tolerable occupational limit. Branching ratio, the stronger point: conventional fusion yields ⁴He only through the radiative branch, of order 10⁻⁶, so producing ⁴He commensurate with the heat by known channels would demand a main-channel rate about 10⁶ times higher, of order 10¹⁸ neutrons per second. The reported near-absence of neutrons therefore points to a novel aneutronic channel, if the effect is real, rather than to slow conventional fusion, and that is what conflicts with established nuclear physics. ↩
4. Helium-4 is a fixed component of air at a mixing ratio of about 5.24 ppmv, and the atmospheric ³He/⁴He ratio is roughly 1.4×10⁻⁶. M. Ozima and F. A. Podosek, Noble Gas Geochemistry, 2nd ed. (Cambridge: Cambridge University Press, 2002). This ubiquitous background is why helium claims live or die on leak-tight sampling and mass-spectrometric discrimination. ↩
5. U.S. Department of Energy, Report of the Review of Low Energy Nuclear Reactions (December 2004). Reviewers divided on the helium evidence, with several noting reported helium levels close enough to atmospheric background to admit contamination. The point is methodological, not dismissive: it is an argument for leak-tight, background-resolved measurement. ↩
6. M. H. Miles, B. F. Bush, et al., “Correlation of Excess Power and Helium Production during D₂O and H₂O Electrolysis Using Palladium Cathodes,” Journal of Electroanalytical Chemistry 346 (1993): 99–117. The energy-per-helium ratio is reported as roughly consistent (within about an order of magnitude) with 23.8 MeV per ⁴He. Tier C, contested: a single-group result that has been challenged directly, not merely left unrepeated, on calorimetric grounds (e.g., K. L. Shanahan, “A Systematic Error in Mass Flow Calorimetry Demonstrated,” Thermochimica Acta 387 (2002): 95–100) and in contemporaneous critiques of the helium measurement, as well as on the contamination grounds noted by the 2004 DOE review. ↩
7. M. J. Guffey, Y. Tang, and P. J. King, “Attempted Replication of Excess Heat in the Letts Dual-laser Experiment,” Journal of Condensed Matter Nuclear Science 20 (2016): 1–28. A published null (6.1 ± 21.6 mW over 231 trials at ~10 W input) in which the authors traced the limiting artifacts to hardware: thermocouple behaviour, calibration drift, and end-cap instability of order 130–460 mW. ↩
8. Tritium half-life 12.32 ± 0.02 yr; decays by beta emission to ³He with an endpoint energy of 18.6 keV and a mean beta energy near 5.7 keV. M. P. Unterweger and R. Colle, and see L. L. Lucas and M. P. Unterweger, “Comprehensive Review and Critical Evaluation of the Half-Life of Tritium,” Journal of Research of the National Institute of Standards and Technology 105 (2000): 541–549. ↩
9. ISO 9698:2019, Water Quality — Determination of Tritium Activity Concentration — Liquid Scintillation Counting Method (Geneva: International Organization for Standardization, 2019). Electrolytic pre-enrichment before counting pushes environmental detection limits into the single becquerels-per-litre range and below. ↩
10. Tritium occurs naturally, produced when cosmic rays interact with the upper atmosphere; the U.S. EPA drinking-water standard is a maximum contaminant level of 20,000 picocuries per liter, corresponding to a dose basis of 4 mrem per year. U.S. Nuclear Regulatory Commission, “Backgrounder on Tritium, Radiation Protection Limits, and Drinking Water Standards.” ↩
11. Medicines and Healthcare products Regulatory Agency (MHRA), ‘GXP’ Data Integrity Guidance and Definitions, Revision 1 (March 2018). The ALCOA+ criteria (attributable, legible, contemporaneous, original, accurate; plus complete, consistent, enduring, available) formalize what it means for a measurement to be re-derivable from its raw record. ↩
12. J. R. Klein and A. Roodman, “Blind Analysis in Nuclear and Particle Physics,” Annual Review of Nuclear and Particle Science 55 (2005): 141–163. A secret offset or masking scheme, frozen before unblinding, removes the analyst’s expectation from the result and is standard practice where small signals ride on large, uncertain backgrounds. ↩
13. C. P. Berlinguette et al., “Revisiting the Cold Case of Cold Fusion,” Nature 570 (2019): 45–51. A Google-convened, multi-institution effort that did not reproduce excess heat across roughly 420 samples but identified extreme-loading materials science as genuinely under-explored. ↩
14. Advanced Research Projects Agency–Energy (ARPA-E), “U.S. Department of Energy Announces $10 Million in Funding to Projects Studying Low-Energy Nuclear Reactions,” February 2023. Eight teams, including groups at MIT, Stanford, and Lawrence Berkeley National Laboratory. ↩