Deep divefrom Episode 5Batteries8 October 202621 min

Bake It Twice — Why Sodium Ceramics Need a Head Start

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: Non-reactive sintering enhances density and ionic conductivity of NASICON solid electrolytes (arXiv)

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. One paper today: a ceramic that carries sodium ions, and the question of whether it matters how much of it you make before you bake it.

Host 2The paper is called Non-reactive sintering enhances density and ionic conductivity of NASICON solid electrolytes. It is by Andrea Cornelio, Gustav Graeber and seven colleagues at the Federal Institute for Materials Research and Testing, known as BAM, and Humboldt University, both in Berlin. It is posted on the arXiv, so it has not been through peer review yet.

Host 1And the claim, in one sentence?

Host 2If you let the ceramic fully form during the first heating step, before you press and sinter it, you get a denser electrolyte that conducts sodium ions better. They also found that the material loses about a fifth of its phosphorus during processing, and that adding extra phosphorus at the start fixed a stability problem against sodium metal.

Host 1That sounds like a recipe paper.

Host 2It is a recipe paper, and that is its value. The result is solid within one lab and one set of raw materials, and the effects are moderate rather than dramatic. We will get to exactly how big.

Why this field exists

Host 1Start with the basics. Why would anyone want a solid electrolyte at all?

Host 2This is established background. Every battery has an electrolyte between the two electrodes, a material that lets ions through but blocks electrons. In today's lithium-ion and sodium-ion cells that is a flammable liquid, and replacing it with a solid is one route to safer cells.

Host 1And the solid version lets you do something else too.

Host 2It may let you use pure metal as the negative electrode, which the paper points out raises energy density. A metal electrode stores far more charge per gram than the graphite or hard carbon used today. The catch is that metal tends to grow filaments through electrolytes and short the cell.

Host 1Why sodium rather than lithium?

Host 2The paper's argument is the usual one: sodium is abundant, cheap and spread widely around the Earth's crust. It also has a very negative reduction potential, minus two point seven one volts, so cells can still reach useful voltages. The trade-off, as background, is that sodium ions are heavier and larger than lithium ions.

Host 1And NASICON is the name of the ceramic.

Host 2NASICON stands for sodium super ionic conductor. The first and best-studied composition was introduced by Henry Hong and John Goodenough in nineteen seventy-six, and the paper calls this family one of the most promising sodium solid electrolytes. It conducts well, nearly all of its current is carried by sodium ions, and it is thermally and chemically stable.

Host 1What does it actually look like, chemically?

Host 2It is a framework of zirconium, silicon, phosphorus and oxygen with sodium ions sitting in channels, and those ions hop through the open framework. You can swap some phosphorus for silicon and add sodium to keep the charge balanced. The paper studies two versions, which I will call sodium three and sodium three point four, after how many sodium atoms each formula unit carries.

Host 1Why those two?

Host 2Sodium three is the classic Hong and Goodenough composition. Sodium three point four, with more silicon and less phosphorus, has become the newer benchmark, because that balance of sodium ions and vacancies gives faster transport. The paper cites room-temperature conductivities of up to five millisiemens per centimetre reported for it.

Host 1Put five millisiemens in context for me.

Host 2As rough background, liquid electrolytes in commercial cells sit around ten millisiemens per centimetre. So a ceramic at a few millisiemens is in a usable range for a thin separator, which is why the number gets attention. It is a ballpark, not a figure from this paper.

What was known before

Host 1So the material is known. What was the problem before this paper?

Host 2Reproducibility. The paper notes that samples with nominally identical compositions report widely different conductivity, activation energy, critical current density and cycling stability across the literature. The usual explanation is processing: how dense the ceramic gets, how the grains grow, what sits at the grain boundaries.

Host 1Walk me through how this ceramic is normally made.

Host 2The common route is called solid-state reaction. You mix cheap powders, here sodium carbonate, zirconium oxide, silicon dioxide and ammonium dihydrogen phosphate, and you heat them. There are two hot steps: calcination, which drives off gases and starts the reactions, and then sintering, where you press the powder into a pellet and heat it until the grains fuse into a dense solid.

Host 1And people have tuned sintering a lot.

Host 2A great deal, the paper says, while calcination has had comparatively little attention. When the authors compiled published recipes for sodium three made from these same raw materials, calcination temperatures ranged from roughly nine hundred to twelve hundred degrees Celsius, with no consensus. So some groups sinter powder that is already NASICON, and others sinter a mixture that is still reacting.

Host 1That second one has a name.

Host 2Reactive sintering: the crystal phase forms and the pellet densifies at the same time. Non-reactive sintering means the phase has already formed before densification starts. The paper cites one earlier study that skipped calcination entirely, a fully reactive one-step route, which reached about one point four eight millisiemens per centimetre but only about ninety percent relative density.

Host 1And why would reacting during sintering hurt?

Host 2As background, chemical reactions in a pellet can change volume, release or consume phases and compete with the processes that close pores. The paper frames it as an open expectation that the extent of reaction before sintering should affect densification and microstructure. Nobody had tested it systematically.

The open question

Host 1So the open question is simple.

Host 2Under identical sintering, does it matter how far the reaction has gone in the calcined powder? And if it does, does that show up in density, in conductivity, and in how the electrolyte survives against sodium metal? They set out to isolate that one variable.

The experiment: making the powders

Host 1Let us build it. How did they isolate calcination?

Host 2They made one large batch of mixed raw powder per composition and split it into identical sub-batches, so particle sizes would match. Each sub-batch was calcined at a different temperature, nine hundred, one thousand or eleven hundred degrees, plus twelve hundred for sodium three point four, always for twelve hours. Then every sample went through the same milling, pressing and sintering.

Host 1How did they choose those temperatures?

Host 2From thermal analysis of the raw mixture: they heated it while recording weight loss, heat flow and the gases coming off. Below about eight hundred degrees, the ammonium phosphate and the sodium carbonate decompose and release ammonia, water and carbon dioxide. Above that, nothing more leaves, and intermediate compounds form.

Host 1And then the NASICON itself appears.

Host 2At a third stage there is a heat signal, an absorption immediately followed by a release, which they read as NASICON crystallising out of a short-lived liquid. That happens between one thousand and eleven hundred degrees for sodium three, and between eleven hundred and twelve hundred for sodium three point four. So the newer composition forms about a hundred degrees later.

Host 1Did the weight loss match what the chemistry predicts?

Host 2No, and they say so. The expected total loss was about seventeen percent for sodium three and sixteen percent for sodium three point four, but they measured about seven and ten and a half percent. Their explanation is that some gas had already left before the furnace, and they noticed ammonia escaping during milling.

Host 1Then they checked what actually formed at each temperature.

Host 2With X-ray diffraction and a fitting method called Rietveld refinement, which estimates the weight fraction of each crystalline phase. At the lower temperatures the powder was mostly an intermediate sodium zirconium silicate, with leftover zirconium oxide. At the top temperature, NASICON dominated.

Host 1Numbers?

Host 2Sodium three calcined at eleven hundred degrees was ninety point two percent NASICON by weight, with about five percent silicate and five percent zirconia. Sodium three point four calcined at twelve hundred reached eighty-four point two percent NASICON, with the silicate still the main phase at eleven hundred. One caveat in the method: amorphous material is invisible to this quantification, so these are fractions of the crystalline part only.

The experiment: sintering

Host 1Then the baking itself.

Host 2Pellets were pressed at one hundred and fifty megapascals and sintered for twelve hours, at twelve twenty-five degrees for sodium three and thirteen hundred for sodium three point four. They were buried in their own calcined powder to limit sodium and phosphorus evaporation. Afterwards they were ground to about one millimetre, polished and annealed at nine hundred degrees to clean the surface.

Host 1Why different sintering temperatures for the two?

Host 2They watched pellets shrink in a heating microscope. Both started shrinking around eleven hundred degrees, but sodium three point four densified more slowly and needed higher temperatures, which matches earlier reports that more phosphorus helps densification through low-melting glassy phases. The chosen temperatures aim for similar final densities while staying below melting.

Host 1And within a composition, sintering is identical.

Host 2Yes, that is the core of the design. Only the calcination history differs, so any difference in the final pellet can be pinned on it. The trade-off is that one fixed sintering schedule is tested, which matters later.

Result one: density

Host 1Results. First, density.

Host 2After sintering, density rose steadily with calcination temperature in both compositions, and electron microscope images showed less residual porosity. The densest pellets were sodium three calcined at eleven hundred and sodium three point four calcined at eleven hundred and twelve hundred. Three discs were measured per condition.

Host 1How much denser?

Host 2The text gives the trend and a figure rather than a table of values, so I will not put a number on it. What it does state plainly is the direction: more NASICON in the powder before sintering, denser electrolyte after. That is the paper's central finding.

Host 1Did the final crystal phases differ?

Host 2No, and that is important. After sintering, every sample was NASICON with monoclinic zirconia as the main impurity, a common one for this route. So the calcination history did not change what the pellet ends up being, only how well it packed together.

Result two: the missing phosphorus

Host 1Then the chemistry. Did anything get lost?

Host 2They dissolved sintered samples and measured the elements with a plasma emission technique. Sodium, zirconium and silicon were similar across calcination histories, so calcination did not change element retention. But every sample was short of phosphorus by roughly twenty percent.

Host 1Every sample.

Host 2Every sample, both compositions. They attribute it to phosphorus oxides evaporating during calcination, before the phosphorus is locked into the NASICON structure. That is an attribution; they did not measure the evaporation directly at that step.

Result three: conductivity

Host 1Now conductivity. How do they measure it?

Host 2They put gold on both faces of a disc and measure impedance, the response to a small alternating voltage over a wide range of frequencies. Fitting that spectrum separates two contributions: bulk, the resistance inside the crystal grains, and grain boundary, the resistance at the junctions between grains. Total conductivity combines both, and bulk is mostly set by composition, while grain boundaries reflect microstructure.

Host 1So which one moved?

Host 2Both, but grain boundaries much more. For sodium three, going from nine hundred to eleven hundred degree calcination raised bulk conductivity by eleven percent and grain-boundary conductivity by seventy-eight percent. For sodium three point four, from nine hundred to twelve hundred, bulk rose forty-nine percent and grain boundary one hundred and thirty percent.

Host 1And the total, the number that matters?

Host 2For sodium three point four, total conductivity went from one point six zero to two point nine five millisiemens per centimetre, close to a doubling. For sodium three it went from zero point seven three to zero point nine eight. Every electrolyte had a sodium transference number above zero point nine nine, meaning essentially all the current is carried by sodium ions.

Host 1Is two point nine five good?

Host 2It is good, not a record. The paper itself cites up to five millisiemens per centimetre for this composition elsewhere. The point here is the comparison within one controlled series, not the absolute value.

Host 1Why would bulk change at all, if bulk is about the crystal?

Host 2The authors suggest fewer leftover secondary phases or better crystallinity from higher calcination. They phrase it as a possible explanation, not a demonstrated one. The stronger claim is about grain boundaries: denser pellets, better grain contact, fewer resistive regions.

Result four: against sodium metal

Host 1Conductivity is not the whole story for a battery, though.

Host 2Right, the electrolyte also has to survive sodium metal on both sides. They build symmetric cells, sodium, electrolyte, sodium, and push current back and forth. One key number is the critical current density, the current at which the cell starts to fail.

Host 1How did sodium three do?

Host 2The sodium three pellet calcined at eleven hundred reached a critical current density of one point seven five milliamps per square centimetre at room temperature, under about ten megapascals of stack pressure. At a gentle zero point one milliamps per square centimetre it plated and stripped sodium stably, in one-hour steps. That is one cell type, one composition.

Host 1And the better conductor, sodium three point four?

Host 2That is where the twist is. The best sodium three point four pellet started out well, but after about one hundred and seventy hours of mild cycling its resistance began dropping and the voltage became erratic. A resistance that falls over time sounds good, but here it is a warning sign.

Host 1Explain why falling resistance is bad.

Host 2If sodium metal starts forming thin deposits inside pores and along grain boundaries, the electrolyte begins conducting electrons as well as ions. That is called a soft short circuit: not one clean metal bridge, but a gradual leak. Both bulk and grain-boundary resistance fell together, which the authors say fits a soft short better than lost contact or a single dendrite.

Result five: adding phosphorus back

Host 1And the phosphorus comes back in.

Host 2Their hypothesis: sodium three point four starts with less phosphorus than sodium three, so losing the same twenty percent could leave more phosphorus-poor regions and secondary phases at the grain boundaries. Sodium three lost phosphorus too but stayed stable. So they made sodium three point four with twenty percent extra phosphorus in the starting mix, calcined at twelve hundred degrees.

Host 1Did it work?

Host 2In these tests, yes. The phosphorus-compensated electrolyte showed a normal critical current test, with voltage rising steadily until failure began at five point zero milliamps per square centimetre. It reached a total room-temperature conductivity of three point eight two millisiemens per centimetre and cycled stably for over four hundred hours, with the electrolyte's own impedance signal staying constant.

Host 1So the extra phosphorus raised conductivity too.

Host 2It did, from two point nine five to three point eight two. That means two things changed at once, stability and conductivity, and the main text does not say how much phosphorus actually ended up in the final pellet. The stabilising mechanism, phosphorus-poor grain boundaries, is a hypothesis the experiment supports but does not directly image.

A word from our sponsor

Host 1Do you find yourself lying awake, worrying about unwanted hair? Then you are in luck.

Host 1The Hastrup Review of Science is brought to you by epilators. A rotating head of small tweezers removes hair at the root, so the result lasts for weeks rather than days.

Host 2I use mine regularly. It is reassuring to remove something from the root instead of only the surface.

Host 1They come corded or cordless, with several speed settings and attachments for different areas, and they are widely available right now. Some models can be used in the shower.

Host 2Thorough, repeatable and not especially quiet. A very dependable product.

Host 1Epilators. Our referral link is in the show notes.

Why it matters

Host 1Back to the paper. Step back: what is the principle here?

Host 2Separate phase formation from densification. If the crystal has already formed before you sinter, sintering can spend its effort closing pores and joining grains, and you get a denser, better-conducting ceramic. The authors argue the reaction state of the powder is a key processing parameter that people have been setting more or less arbitrarily.

Host 1Who does this change things for?

Host 2Mainly labs and pilot lines making NASICON by this route. It gives them a concrete step to report and control, calcination temperature and the resulting phase content, and a reason why two groups making the same nominal compound get different numbers. The second lesson, check your final phosphorus content, is cheap to act on.

Host 1And what does it not mean?

Host 2It does not mean sodium solid-state batteries are ready; there is no full cell with a cathode in this paper. It does not show the best possible NASICON, and it does not prove calcination is the only source of the scatter in the literature. It shows that, in one well-controlled setup, calcination history moves density and conductivity in a consistent direction.

Honest caveats

Host 1Give me the caveats in one place.

Host 2First, a single sintering schedule per composition, twelve hours at one fixed temperature. A longer or hotter sintering might let the reactive powders catch up, so the advantage is shown for this schedule, not in general. Second, one set of raw materials and one synthesis route, which the authors themselves acknowledge.

Host 1Keep going.

Host 2Third, the electrochemical tests are few: symmetric cells only, and the text does not report how many cells were tested per condition. The stack pressure of about ten megapascals is high, and room temperature drifted between twenty and twenty-five degrees with no active control. Fourth, the best sodium three point four results bundle two changes, higher calcination and extra phosphorus.

Host 1And the phosphorus story itself?

Host 2The twenty percent loss is measured, but where it goes and why it hurts sodium three point four specifically are hypotheses. The sodium three series also topped out at eleven hundred degrees calcination, so the two compositions are not tested on an identical grid. And one practical point that is my observation, not the paper's: a longer, hotter calcination costs energy and time, a trade-off the paper does not weigh.

Host 1Anything reassuring?

Host 2The design is clean, with one powder batch split into identical sub-batches, and the trends point the same way across density, microscopy and impedance. The authors also deposited their data openly on Zenodo. And it is a preprint, so peer review may still sharpen it.

What to watch next

Host 1What should we watch for next?

Host 2Whether the same rule holds for other routes, such as sol-gel or solution-assisted synthesis, which the authors flag themselves. Whether other groups reproduce the density and conductivity gains with their own raw materials. And whether a pellet that is fully reacted before sintering can match these results at lower sintering temperatures or shorter times.

Host 1And for the phosphorus?

Host 2Direct measurements of where the phosphorus goes and what the grain boundaries look like after cycling would test the hypothesis. The real test, though, is full cells with a cathode, at lower stack pressure, over many hundreds of cycles. That has not been done here.

Sign-off

Host 1That is the deep dive: Non-reactive sintering enhances density and ionic conductivity of NASICON solid electrolytes, by Cornelio, Graeber and colleagues, a preprint on the arXiv. The link is in the show notes.

Host 2Every Friday this same feed carries the roundup episode: five new papers as trailers, and you pick which become deep dives. This has been the Hastrup Review of Science.