If you open a classic textbook on sandstone‑hosted uranium deposits, you’ll likely read a neat story: oxidised uranium‑bearing groundwater (U⁶⁺) meets a reducing agent – pyrite (FeS₂) or organic carbon – and U⁶⁺ is reduced to U⁴⁺, precipitating as pitchblende (UO₂). In that story, pyrite acts as a sacrificial fuel: it dissolves, disappears, and uranium takes its place.

Clean. Chemically intuitive.

But what we saw under the microscope never quite matched that picture.

A paradox: pyrite wasn’t dying – it was growing

In a large sandstone‑hosted uranium deposit in northern China, we observed abundant pyrite grains associated with uranium minerals. According to the classic model, those pyrite grains should have been corroded – pitted, etched, riddled with holes. Instead, they looked fresh. In fact, newly formed, arsenic‑rich pyrite (arsenian pyrite) was directly intergrown with pitchblende, or even wrapped around it.

In other words: uranium precipitated, but pyrite also precipitated. It wasn’t consumed – it grew.

That’s like throwing logs on a fire and watching the pile get bigger. Something was wrong with the textbook model.

Tracing the culprit: the hidden hand of arsenic

So we went in deep – from micrometre to atomic scale. Our team (Chengdu University of Technology, University of Regina, China University of Geosciences, Beijing) used EPMA, NanoSIMS, Raman spectroscopy, TEM and aberration‑corrected STEM.

One clue stood out: all the newly formed pyrite associated with uranium minerals was rich in arsenic (As), whereas the older pyrite was essentially arsenic‑free. More importantly, the arsenic‑rich pyrite showed a distinct lattice expansion – arsenic atoms are larger than sulphur, so when they substitute into the pyrite structure, they push the crystal lattice apart.

That lattice strain turned out to be the key to an unexpected electrochemical dance.

The eureka moment: a self‑driven galvanic couple

Anyone in materials science knows galvanic coupling: when two metals with different electrochemical potentials touch, the more active one becomes the anode and corrodes preferentially, while the less active one becomes the cathode and is protected.

In our system, the newly formed, strained, arsenic‑rich pyrite acts as the anode, and the pre‑existing arsenic‑poor pyrite acts as the cathode. The arsenic‑rich pyrite tends to lose electrons (oxidation), and those electrons are exactly what’s needed to reduce U⁶⁺ in solution.

So arsenic doping doesn’t kill pyrite’s reducing power – it *amplifies* it, by creating lattice strain and a galvanic couple. U⁶⁺ is reduced to U⁴⁺, forming nanoscale pitchblende particles on the pyrite surface, which then aggregate into larger uranium minerals.

This is a self‑organising process. Once a little arsenic enters the pyrite, the galvanic couple kicks in, and the reaction sustains itself – until the arsenic supply runs out or the pyrite is completely armoured.

We call this mechanism “arsenic‑induced lattice strain and galvanic coupling as a trigger for uranium mineralisation”.

Sleepless nights under the microscope

This wasn’t a sudden flash of insight. The hardest part was mapping arsenic distribution and uranium valence at the nanoscale – simultaneously. When the first NanoSIMS maps came out, showing arsenic and uranium almost perfectly overlapping on pyrite surfaces, the lab went quiet for a few seconds. We all knew: the paradox finally had an explanation.

Another challenge was ruling out interference from organic matter or other reductants. We spent months hand‑picking pyrite grains without organic inclusions, repeating experiments, checking and double‑checking. Only then did we trust that the electrochemical model stood on its own.

A word as a supervisor – and what my student taught me

Behind every paper there are not just data, but people – and a lot of perseverance. This one especially.

My student, Huijie Yu (first author), started his master’s work barely able to tell pyrite from magnetite under a reflected‑light microscope, let alone recognise different generations of arsenian pyrite. I remember a full month when he sat at the microscope every single day, taking photos, drawing sketches, mapping paragenetic sequences. I told him: “There’s no shortcut in mineralogical microscopy. If your eyes get tired, rest – then come back.” And he did exactly that.

Later came the NanoSIMS and TEM work. Huge datasets, often with poor signal‑to‑noise ratios. He would message me late at night: “Professor, is this arsenic peak real or just noise?” I taught him how to distinguish, how to run blanks, how to repeat measurements. He took notes carefully and never made the same mistake twice.

The hardest test was the submission process. We were rejected several times. Each time the reviews came back, Huijie was visibly crushed. Once he said to me, with red eyes: “Professor, what if we were wrong from the beginning?” I told him: “Rejection is normal – it’s not failure. A scientist isn’t someone who never gets rejected; a scientist is someone who, after rejection, calmly revises, does more experiments, and tells the story more rigorously.”

We printed every set of reviewer comments. He went through them line by line – even the harsh ones, thanking the reviewers for helping us see blind spots. Over several months, he revised the manuscript more than a dozen times and carried out three extra months of experiments.

When the acceptance letter finally came from Communications Earth & Environment, he sent me a message that I’ve kept. He wrote:

Thank you, Professor Song, for revising and polishing the manuscript with me over and over again, without ever losing patience. Even though we were rejected several times, you kept guiding me, encouraged me not to fear rejection, to be brave, to keep trying. That teaching is the biggest reason this paper was accepted – and it is a treasure for my entire future research career. Through all the trials, I’m lucky I didn’t let your hard work down.”

As his supervisor, reading that made every late night and every round of revisions worthwhile. But honestly, I owe him thanks too – his persistence and rigour are what made this counter‑intuitive model stand up. Research is never a solo fight. My whole research group – every discussion, every encouragement – played an indispensable role.

Why this matters

First, it revises a classic ore‑deposit model. Pyrite in sandstone‑hosted uranium deposits isn’t necessarily a consumed reductant – it can also be a “catalyst and a substrate for reduction”. This helps explain why some sandstone layers are rich in pyrite but poor in uranium: maybe they lack arsenic.

Second, it offers a new exploration indicator. If a sandstone contains both arsenic‑poor pyrite (old) and arsenic‑rich pyrite (new) coexisting, that horizon is likely a “hot spot for electrochemical uranium precipitation”.

Third, more broadly, this “dopant → lattice strain → galvanic coupling → enhanced reactivity” mechanism may operate in other metal systems (gold, copper, cobalt). It turns a mineral from a passive reductant into a “self‑regulating micro‑electrochemical system”.

What’s next

We are now testing whether this doping‑galvanic mechanism also drives the enrichment of other redox‑sensitive metals, and whether we can construct similar nano‑electrochemical interfaces in the lab for uranium‑contaminated water treatment.

If you work on pyrite, uranium deposits, or mineral–fluid redox reactions, let’s talk. The best part of science is when you think the story is finished – and the minerals themselves whisper: not yet.

And to everyone still fighting through revisions and rejections: keep going. Good results are on their way. Our team is growing stronger – and so will yours.

Article link: https://doi.org/10.1038/s43247-026-03511-1

This paper is published in Communications Earth & Environment. All authors thank the journal, the reviewers, and every member of our research group for their support.