Deep divefrom Episode 10Batteries7 November 202619 min

Why Heavy Metals Keep Battery Oxygen in Line

A deep dive into one paper: what was known before, what the researchers did, what the numbers say and what they don't.

The paper: Transition-Metal-Oxygen Covalency Controls Ligand-Hole Stability in Li-Rich Cathodes (Hsu, Lomeli et al., arXiv preprint)

MP3

Chapters

Transcript

The script the two AI voices read from. The times are worked out from the script, not heard, so a line can start a few seconds early or late.

Read the whole episode

Cold open

Host 1You heard the trailer — this is the whole story. Welcome to the Hastrup Review of Science — where artificial intelligence reads real science, so you don't have to pretend you understood the paper. Today, why some battery materials can give up a lot of their lithium and survive, while a close cousin falls apart and starts breathing out oxygen.

Host 2The paper is Transition-Metal-Oxygen Covalency Controls Ligand-Hole Stability in Lithium-Rich Cathodes, by Kuan Hsu, Eder Lomeli and colleagues, with Wanli Yang and Thomas Devereaux as senior authors, from Stanford, the SLAC laboratory, Berkeley's Advanced Light Source and POSTECH in Korea. It is a preprint, not yet peer reviewed. It is mostly a theory paper: they simulate X-ray spectra and compare them with measurements, most of which were published before.

Why cathodes matter

Host 1Before we get into ligands and holes, can you remind me what a cathode actually does inside the battery in my phone?

Host 2The cathode is the side of the battery that stores lithium when the battery is empty. When you charge, lithium ions leave the cathode and travel to the other side, and when you use the phone they come back. How much lithium a cathode can give up and take back, without damage, sets how much energy the battery holds.

Host 1And where does the metal come in, the cobalt and nickel people argue about?

Host 2In the wider picture of the field, cathodes since the late nineteen seventies and eighties have mostly been layered oxides: sheets of a transition metal like cobalt or nickel bound to oxygen, with lithium in between. Every lithium that leaves takes an electron with it, and something in the material has to give up that electron. Classically, the textbook answer is the metal, which shifts to a higher charge state.

Host 1So if I have this right, the metal is like a bank account that pays out one electron for every lithium that leaves?

Host 2That is the classic picture, yes, and it has a limit. Once the metal has given up what it comfortably can, the account is empty. That is where this paper's materials come in.

What was known before

Host 1What makes a cathode lithium-rich, and why would anyone want one?

Host 2A lithium-rich cathode simply has more lithium than metal in each formula unit. The three compounds here have two lithiums, one metal atom and three oxygens, compared with one lithium per metal in the conventional ones. More lithium per unit means a higher theoretical capacity, so more stored energy per gram.

Host 1If the metal account runs dry, who pays out the electrons for all that extra lithium?

Host 2Oxygen, at least partly, and that is the heart of a long debate. In the manganese version, the manganese already starts at a formal charge of plus four, which leaves it very little room to give more. So the extra capacity has been linked to the oxygen itself giving up electrons.

Host 1Is it bad when the oxygen gives up electrons?

Host 2It can be. The paper recounts that the manganese compound shows big voltage hysteresis, structural damage and oxygen loss on the first charge, and that oxidised oxygen can pair up into oxygen-oxygen dimers, peroxide-like species, and eventually molecular oxygen gas. Gas inside a battery is exactly what you do not want.

Host 1You said hysteresis; what does that mean in plain words?

Host 2It means the voltage when you charge and the voltage when you discharge do not match, so you lose energy each round trip, a bit like a gearbox that slips. It is one of the visible signs that something irreversible is happening inside.

Host 1Is the manganese one the only material in the family, or are there better-behaved relatives?

Host 2There are, and they are the reason for the paper. The ruthenium version reversibly gives up about one lithium per formula unit with low hysteresis. The iridium version goes further, about one and a half lithiums per iridium, reaching a formal iridium charge of plus five and a half with comparatively little structural damage.

The open question

Host 1So why does the heavier metal behave better, if experiments already show oxygen taking part in all three?

Host 2That is exactly the open question. Measurements on oxygen in the ruthenium and iridium compounds show clear oxygen participation, so the simple story that heavy metals just do all the work does not hold. The paper argues that because heavy metals bond so strongly with oxygen, it has been hard to say who is doing what, and there is no consensus.

Host 1I keep hearing ligand hole in the title. What is a ligand, and what is a hole?

Host 2The ligand here is just the oxygen surrounding the metal. A hole is a missing electron, so a ligand hole means the electron that was removed came out of the oxygen rather than the metal. In chemists' shorthand, the metal looks unchanged but its oxygen neighbours are short one electron.

Host 1And covalency, the other word in the title, is that just how strongly the metal and oxygen share their electrons?

Host 2Yes, exactly. High covalency means the metal and oxygen orbitals are strongly mixed, so the electrons belong partly to both. The heavier metals, ruthenium and iridium, have larger, more spread-out outer orbitals than manganese, so they mix with oxygen more.

Host 1Then there's the phrase negative charge transfer that I've seen in the abstract. Can you put that in everyday words?

Host 2It describes which way the cheapest electron move goes. In a normal, positive charge transfer material, it costs energy to move an electron from oxygen onto the metal. In a negative charge transfer material, that move actually lowers the energy, so the oxygen spontaneously hands electrons to the metal and holes pile up on the oxygen even before you charge.

Host 1So both strong sharing and negative charge transfer put holes on the oxygen. Is the paper's point that those are two different routes to the same-looking result?

Host 2That is precisely the point. You can have lots of oxygen holes because the material is deep in the negative charge transfer regime, or simply because the metal and oxygen share strongly. The authors argue these two are often confused, and that only the first is the dangerous one.

What they did

Host 1How do you actually see where the electrons are inside a solid? You can't exactly look.

Host 2With X-rays from a synchrotron, tuned to the energy where a specific element absorbs. X-ray absorption spectroscopy tells you which empty states exist on, say, oxygen or the metal. Resonant inelastic X-ray scattering goes further: it measures the energy of the X-rays that come back out, which reveals the excitations inside the material.

Host 1And what did these authors add on top of the measurements, since you said it is mostly a theory paper?

Host 2They built models and calculated what those spectra should look like, then checked them against experiments. If a model reproduces the measured spectra at both the oxygen edge and the metal edge, you can read off what the electrons are doing inside it. They did this for each compound before charging and at one or more charged states.

Host 1What kind of models are we talking about, roughly?

Host 2Two kinds, used side by side. One is a charge transfer multiplet model, a small cluster of one metal and its oxygens, whose output includes how the ground state splits between configurations like metal-with-one-oxygen-hole and metal-with-two-oxygen-holes. The other is a first-principles approach: standard density functional theory, plus a code called OCEAN that solves the Bethe-Salpeter equation to compute spectra for the whole crystal.

Host 1Why bother with both? Wouldn't the bigger, whole-crystal one simply be better?

Host 2They capture different things. The small cluster handles the strong interactions between electrons on one metal well, which dominates manganese, but it is localised by design. The crystal-wide code captures electrons spreading out, and the paper notes it shows a smooth continuum of excitations where the cluster gives discrete ones, so agreement between the two adds confidence.

Host 1Whose experimental data are they comparing with? Did they measure it all themselves?

Host 2Mostly not. The manganese and ruthenium spectra are reproduced from earlier publications, and the iridium metal-edge data too. The new measurements are oxygen-edge spectra of the iridium compound at different charge states, taken at two beamlines of the Advanced Light Source, with samples prepared at POSTECH.

A word from our sponsor

Host 2This episode is brought to you by Spreadwell, the sock for people who get holes.

Host 1I used to get one big hole at the big toe. With Spreadwell, the hole is shared evenly across the whole sock, so you never see it.

Host 2Spreadwell socks are now ninety percent hole by weight, and warmer than ever.

Host 1Last winter my hole delocalized over the entire street. The neighbours have never been so cosy.

Host 2Spreadwell. Do not wear above four and a half volts.

What they found

Host 1Back to the manganese compound. What did the model say about its starting state, before any charging?

Host 2That it is already strongly negative charge transfer before you charge it at all. The ground state is an even split between a state with one oxygen hole and one with two, so the oxygen already carries a lot of holes. The simulated spectra matched the experiments at both the oxygen and manganese edges.

Host 1How can a spectrum tell you that? What is the actual signature they point to?

Host 2At the oxygen edge, there is a near-elastic feature at an incident energy of about five hundred twenty nine electron volts, which they tie to the excess oxygen holes. They contrast it with the known fingerprint of trapped oxygen gas, which sits around five hundred thirty one. At the manganese edge, features between about six hundred forty three and six hundred forty five electron volts point to strong mixing.

Host 1If the manganese one starts out like that, is it already unstable as soon as it's made?

Host 2No, and the paper is careful here. Clear excitations within the manganese itself, at about three electron volts of energy loss, show the manganese electrons are still fairly localised, which helps keep the uncharged compound stable. The trouble comes with charging, which pushes it deeper into the negative charge transfer regime and overloads the oxygen with holes.

Host 1So in plain words, manganese starts the race already near the cliff edge, and charging pushes it over?

Host 2That is a fair picture. The paper presents that deepening as a precursor to the oxygen dimers and instability seen experimentally, rather than the manganese going to plus seven, or so-called non-bonding oxygen states, which earlier theory work had proposed.

Host 1Is there any manganese material that gets away with it?

Host 2Yes, a sodium manganese oxide with three manganese to seven oxygens, which cycles reversibly beyond plus four. Their calculations for its charged state show compensation landing only on the oxygen, consistent with earlier work crediting stabilised oxygen hole polarons. It operates below four and a half volts, whereas the lithium manganese compound quickly climbs above four and a half, so a strongly negative state can be lived with when other factors cooperate.

Host 1What about ruthenium? Does it start in the same place as manganese?

Host 2No, and that is the key contrast. Uncharged lithium ruthenium oxide also has plenty of oxygen-hole weight, spread across states with zero, one and two oxygen holes. But the model says that weight comes mainly from strong ruthenium-oxygen sharing, not from a strongly negative charge transfer state.

Host 1What changes in the spectra when they pull out one lithium from the ruthenium compound?

Host 2That charged state was reached experimentally by cycling to four point one volts. Both edges show the two main absorption features moving apart, with the lower one, around five hundred twenty nine electron volts at oxygen, growing stronger. The authors read that as electrons leaving the bonding orbitals shared between ruthenium and oxygen, and the model's ground state shifts to an even spread across one, two and three oxygen holes.

Host 1So the extra holes get spread out over the whole metal-oxygen network instead of piling up on one oxygen?

Host 2Yes, that is the mechanism they propose. The added charge is delocalised across ruthenium and oxygen together, which offsets the drift toward a strongly negative state. But it has a finite range: beyond about one lithium per ruthenium, earlier work shows oxygen dimers and oxygen vacancies appear.

Host 1And iridium, the heaviest one, is that where the trend holds most strongly?

Host 2The model puts uncharged lithium iridium oxide as a small positive charge-transfer system with large covalency, its ground state spread across one, two and three oxygen holes. On charging to one lithium, it becomes only slightly negative, with an even spread across one, two and three holes. They followed it down to half a lithium per formula unit, formal iridium plus five and a half, and their new oxygen-edge measurements agree with the simulations along the way.

Host 1Did anything in the iridium spectra actually change clearly, or is it subtle?

Host 2Subtle at the iridium absorption edge, which only shows one broad feature because the excited state is so short-lived. The clearer change is predicted in the scattering spectra at the iridium edge, where intensity shifts toward charge transfer excitations, roughly five to ten electron volts of energy loss in the crystal-wide code. Nobody has measured those scattering spectra yet.

Host 1Does iridium also break down eventually, or does it just keep going?

Host 2It breaks down too. Past plus five and a half, earlier experiments saw peroxide-like oxygen species. And a differently structured iridium compound reportedly goes deeper with less hysteresis, which tells the authors that crystal structure matters jointly with the electronics.

Why it matters

Host 1So stepping back, what is the single principle they want us to take away?

Host 2That the number of holes on oxygen is not what decides stability. What matters is how the holes got there and where they live: strong metal-oxygen sharing spreads them over a network, while an already strongly negative charge transfer state makes them pile up. In the authors' words, covalency, rather than the absence of oxygen holes, is a key factor.

Host 1Does that settle the old argument about whether the metal or the oxygen does the work?

Host 2They suggest the question itself is framed wrongly. In their view, the various kinds of oxygen redox people have described can all be seen as rearrangements of ligand-hole density in a negative charge transfer state. So rather than metal versus oxygen, it is about how the shared metal-oxygen system handles the holes.

Host 1What would a battery designer actually do differently with this?

Host 2The paper's suggestion is to pick the metal carefully. Doping heavy metals into manganese-type cathodes could stabilise the extra holes and allow more lithium out per formula unit, while doping with metals that take no part in the reactions, such as tin, costs energy density. They also argue it may be possible to stay below about four and a half volts and avoid the oxygen gas regime altogether.

Host 1Ruthenium and iridium sound expensive. Are we going to see iridium in my next phone battery?

Host 2Almost certainly not, and the paper does not claim so. In the wider picture, iridium is one of the rarest metals mined, so these compounds are model systems for understanding. The hope is that the principle carries over to cheaper materials, perhaps as small additions.

Honest caveats

Host 1What can we not conclude from this paper?

Host 2We cannot conclude that covalency has been proven to cause stability. This is three compounds, interpreted through models fitted to spectra, and three data points make a trend, not a law. The paper also presents the quantitative ground-state weights in supplementary tables that were not in the text I read, so I can only give the qualitative picture.

Host 1You mentioned many of the experiments were borrowed. Is that a weakness?

Host 2It is a limitation worth naming. Reusing published data is normal, but the ruthenium metal-edge scattering and the iridium metal-edge scattering simulated here have no experimental counterpart yet, and the authors say so themselves. They also note that oxygen deficiency in real ruthenium samples may explain some mismatch between experiment and simulation.

Host 1Are there people in the field who'd push back on the main conclusion?

Host 2There are. One study they cite is titled, plainly, covalency does not suppress oxygen gas formation in the ruthenium and iridium cathodes. This paper's answer is that covalency does not prevent it, only delays it, which is consistent with both ruthenium and iridium failing eventually.

Host 1And the small cluster model, does it have blind spots of its own?

Host 2Yes. By design it treats one metal site and its neighbours, so it cannot describe holes spreading over many sites, the very delocalisation the argument leans on. That is why the crystal-wide calculations matter, but those handle strong electron interactions less well, so neither model alone carries the conclusion.

What to watch next

Host 1What experiment would most convincingly test this next?

Host 2The authors point to one directly: scattering at a different iridium absorption edge, at around two and a half thousand electron volts, a softer and more accessible X-ray range, which should show the predicted change in the ground state as the iridium compound is charged. Measuring the scattering at the ruthenium edge, in the tender X-ray range, would fill the other gap.

Host 1And beyond those measurements, what would tell us the design rule really works?

Host 2Mixed materials. If adding a little ruthenium or iridium to a manganese cathode extends the stable charging window the way the model implies, that would be practical evidence. A cheaper metal with similar metal-oxygen sharing doing the same would matter even more.

Sign-off

Host 1So if I had to tell a friend in one sentence, it would be: it's not how many holes the oxygen has, it's whether the metal helps share them?

Host 2That is a good sentence, with one addition: sharing buys a wider safe window, not unlimited safety. Every material here still fails if you pull out enough lithium.

Host 1That's the deep dive for today. Every weekend, this same feed carries the roundup: the week's new papers, in one short episode. This has been the Hastrup Review of Science.