261023 – Mercury into Gold via Fusion Reactor Neutrons

by | Oct 23, 2026 | Physics | 0 comments

Fusion Gold: The Physics Behind the Mercury-to-Gold Claim

· @Thomas Lee Abshier, ND

On 21 October 2026, Marathon Fusion, a San Francisco fusion-fuel startup, announced that in September it had turned mercury into real, stable gold at ten times the rate CERN recorded in 2025. The announcement drew millions of views and an immediate question from the markets: could laboratory gold do to bullion what laboratory diamonds did to the diamond trade?

My position is this. The physics is sound, and it is not new; nuclear physicists have known how to turn mercury into gold since 1941. The claim itself is plausible but, as of this writing, unmeasured: the company has published no mass, purity, or cost. And the real significance of the announcement is not gold at all. It is what it reveals about the economics of a fusion neutron, and on that measure gold is one of the least valuable things a neutron can make.

This essay walks through why chemistry could never make gold, how a nucleus can, why fusion is suited to the job, what the “ten times CERN” claim actually measures, and why gold is unlikely to follow diamonds.

Why chemistry could never make gold

An element is defined by one number: the count of protons in its nucleus. Gold has 79. Mercury, its neighbor on the periodic table, has 80. Lead has 82.

Chemistry rearranges electrons. A chemical bond involves energies of a few electron volts. Holding a nucleus together involves energies of millions of electron volts per particle, roughly a million times more. No flame, acid, or reagent reaches into the nucleus, so no chemical process can add or remove a proton.

That is why the alchemists failed. Their goal was not absurd; their tools were operating on the wrong layer of matter. Changing one element into another is a nuclear process, and it became possible only once physicists could fire particles at nuclei.

The nuclear route: mercury-198 to gold-197

Gold in nature has exactly one stable isotope, gold-197: 79 protons and 118 neutrons. Any process that ends there produces gold that is chemically and physically identical to mined gold. Marathon’s route gets there in two steps.

  1. Knock out a neutron. A fast neutron strikes mercury-198 (80 protons, 118 neutrons). If it carries enough energy, it ejects one extra neutron, leaving mercury-197. Physicists call this an (n, 2n) reaction: one neutron in, two out.
  2. Let the nucleus decay. Mercury-197 is unstable, with a half-life of about 64 hours. It decays by electron capture: the nucleus absorbs one of its own inner electrons, and a proton becomes a neutron. The proton count drops from 80 to 79, and the nucleus is now gold-197, which is stable.
^{198}\mathrm{Hg} + n \;\rightarrow\; ^{197}\mathrm{Hg} + 2n, \qquad ^{197}\mathrm{Hg} + e^{-} \;\rightarrow\; ^{197}\mathrm{Au} + \nu_e

None of this is speculative nuclear physics. In 1941, Sherr, Bainbridge, and Anderson at Harvard bombarded mercury with fast neutrons and produced gold isotopes. What Marathon proposes is not a new reaction but a new place to run it: inside the blanket of a fusion power plant, at industrial scale.

Why fusion neutrons rather than fission neutrons

The (n, 2n) reaction has a threshold. The incoming neutron must supply the energy that binds a neutron inside mercury-198, a little over 6 million electron volts (MeV). Below that, the reaction simply does not happen.

A fission reactor is a poor source of such neutrons. Fission neutrons are born with an average energy near 2 MeV, and only a small tail exceeds 6 MeV; the reactor then deliberately slows them further. A deuterium-tritium fusion reaction, by contrast, releases every neutron at 14.1 MeV, more than double the threshold.

There is a second, subtler advantage. A fusion plant must breed its own tritium by letting neutrons strike lithium in a blanket surrounding the plasma, and it needs extra neutrons to do so. Designers already plan a “neutron multiplier” layer, usually lead or beryllium, that turns one fast neutron into two. Mercury-198 can serve that same role: each (n, 2n) event makes a gold atom and also adds a neutron to the breeding economy. That is the basis of Marathon’s claim that gold comes without sacrificing electricity or tritium.

What “ten times CERN” does and does not mean

The CERN benchmark is real but was never an attempt to make gold. In May 2025, the ALICE collaboration reported that lead nuclei passing close to each other in the Large Hadron Collider sometimes shed three protons through their intense electromagnetic fields, briefly becoming gold. The peak rate was about 89,000 gold nuclei per second. Over the 2015–2018 run, about 86 billion gold nuclei formed, a total of 29 picograms, and each one shattered against the accelerator’s hardware within a fraction of a second.

Marathon’s wording, “high-rate alchemy,” suggests the comparison is a rate. Ten times 89,000 per second is roughly 900,000 gold atoms per second. Sustained for a full day, that is about 25 picograms, a quantity far below anything a jeweler’s scale could detect. My own rough estimate: a compact laboratory deuterium-tritium neutron generator aimed at enriched mercury-198 could plausibly reach such a rate, so the benchmark is a modest one.

The claim still marks one genuine difference. CERN’s gold was a fleeting, radioactive isotope that existed for an instant. Marathon says its gold is stable gold-197 that stays in the sample. That is the meaningful part of the announcement, and it is exactly what an independent laboratory should verify by mass spectrometry. As of 23 October 2026, the company has released no gold mass, isotopic purity, recovery efficiency, or cost per gram.

The catches: enrichment, radioactive gold, cooldown

Three practical problems stand between the reaction and a gold bar.

Enrichment. Natural mercury is a mixture of seven stable isotopes, and mercury-198 is only about 10% of it. Fast neutrons striking the other isotopes produce unwanted products, so the feedstock must be enriched. Marathon’s 2025 reference plant calls for 622 tonnes of mercury-198. That is why the company’s second announcement, a plasma centrifuge it says broke a 40-year-old Soviet record for isotope separation, matters as much as the gold.

Radioactive gold. Fast neutrons do not only knock out neutrons; sometimes they knock out a proton. That (n, p) reaction turns mercury directly into radioactive gold isotopes such as gold-198 and gold-199, which decay within days. Other pathways yield longer-lived gold isotopes, such as gold-195 with a half-life of about six months. These are chemically inseparable from gold-197, so the only remedy is time.

Cooldown. Marathon’s own estimates put the storage time at roughly 7 to 18 years before the gold is clean enough to sell, depending on how pure the feedstock is. The company frames this as a trade-off: pay more to purify the mercury, or wait longer to sell the gold. Either way, fusion gold is a slow-maturing asset, not a printing press.

The economics of a single neutron

The clearest way to judge fusion gold is to ask what one fusion neutron is worth when spent on different products. The answer shows that gold roughly matches electricity, and that both are dwarfed by medical isotopes.

The arithmetic is short. Each deuterium-tritium fusion releases 17.6 MeV and one neutron, so one gigawatt of fusion heat means about 3.6 × 10^20 neutrons per second, or about 1.1 × 10^28 per year. Marathon’s 2025 paper models about 2 tonnes of gold per year per gigawatt of heat, which is about 6 × 10^27 gold atoms. That implies roughly one gold atom for every two fusion neutrons, an ambitious figure that depends on the (n, 2n) multiplication and on nearly every neutron being put to use.

At $4,062 per troy ounce, one gold atom is worth about 4 × 10^-20 dollars. The same neutron’s share of electricity, at 40% conversion efficiency and 5 cents per kilowatt-hour, is worth about 2 × 10^-20 dollars. These are my own estimates, and they agree in order of magnitude with Marathon’s February 2026 paper.

Use of one fusion neutron Approximate value per neutron (US$)
Actinium-225 for targeted cancer therapy (Marathon, Feb 2026) 10^-10
Gold-197, at ~0.5 atom per neutron 2 × 10^-20
Electricity, at 40% efficiency and $0.05/kWh 2 × 10^-20

Scaled up, a one-gigawatt-thermal plant would produce about $260 million of gold a year at today’s price, against about $175 million of electricity. That is the basis of the claim that gold could double a fusion plant’s revenue. But the table’s first row is the real lesson: a neutron spent on a scarce medical isotope is worth about ten billion times more than one spent on gold. The first fusion neutrons sold will go to medicine, and the company’s own announcement thread said more about medical isotopes than about gold.

Will gold go the way of diamonds?

I do not think so, and the reasons are physical before they are financial.

Diamonds were cheap to make; fusion gold is not. A lab diamond needs a press or a vapor-deposition chamber, equipment a mid-sized factory can buy and multiply. Production expanded quickly, and wholesale prices for lab-grown stones fell 90 to 95% from 2015 levels. Fusion gold needs a gigawatt-class fusion power plant. No fusion plant yet sells electricity, and Marathon lists its present device at about 1 kilowatt of fusion power, a millionth of a gigawatt.

Gold is a stock, not a flow. Marathon’s study suggests about 5 tonnes of gold a year per gigawatt of electricity. One hundred such plants would add about 500 tonnes a year, about 14% of 2025 mine output of 3,672 tonnes. But roughly 222,600 tonnes have been mined in all of history, and most of it still exists. Five hundred tonnes a year would add about 0.2% to the world’s gold per year, a pace that markets have absorbed from mining for generations.

Fusion gold limits itself. Every neutron spent on gold is a neutron not spent on breeding tritium, on medical isotopes, or on other products. If gold prices fell sharply, plant operators would simply stop loading mercury, because the neutron would be worth more elsewhere. A source that switches off when prices drop cannot drive prices to the floor the way cheap diamond factories did.

There is one real point of difference from diamonds, and it cuts the other way. Lab diamonds can be identified by specialists; fusion gold, once its radioactivity fades, cannot be told apart from mined gold. Gold’s value would therefore rest more than ever on its total supply rather than on its origin. That supply, for the reasons above, will grow slowly. It is a curious detail that Anglo American, owner of De Beers, is among Marathon’s backers: the company that lived through the diamond shock appears to be betting on the new alchemy rather than against it.

Conclusion

The alchemists’ dream was never forbidden by nature, only by chemistry. A 14 MeV neutron can do in an instant what no furnace could do in a thousand years: remove one neutron from mercury-198 and let the nucleus settle into gold.

Marathon Fusion’s announcement is best read as a credible but unverified demonstration of an old reaction in a new setting. Until the company publishes the mass, isotopic purity, and cost of what it made, “ten times CERN” remains a headline rather than a measurement. And even if every number holds, the deeper story is the value of a fusion neutron. Gold is roughly an even trade with electricity, while medical isotopes are worth ten billion times more. Fusion, when it arrives, will make gold, but it will make it last and slowly, and the price of gold will be set, as it always has been, by the vast stock already above ground.

References

The per-neutron values, the daily gold mass at ten times CERN’s rate, and the plant revenue figures are the author’s own order-of-magnitude estimates from the sourced inputs.

Submit a Comment