 This is the storage X symposium. I'm E-Tray again. This is our fourth event. From the previous three events, we have been seeing great attendance, great success, great talks, and great questions from the audience. So this week, right now, we are having another two outstanding and leading experts in the world to join us to give Storage X symposium. I would like to welcome Professor Claire Gray from University of Cambridge and Professor Goethe Seder from UC Berkeley to join us today. There are two very exciting talks waiting for us. Let me give a very brief introduction about these two speakers. Professor Claire Gray is a Professor at University of Cambridge, a member of a fellow or lawyer society. Claire's career on the energy storage is very interesting. She actually started from MMR, a very powerful technique, but many, many years ago, evolved into using MMR, starting welfare problems, and expand her program broadly. That's a very interesting trajectory. Our second speaker, Professor Goethe Seder, at UC Berkeley and also Lawrence Berkeley Lab, also has a very interesting trajectory as well. He's a well-known theorist doing computational study of materials. Also, many years ago, evolved into doing battery-related work. Not only that, he also evolved from theories to becoming a combination of theory and computation and experiment together. It's a very interesting trajectory for both. So I think today's symposium, having both of them to look at the careful problem, is very complementary and exciting. I really look forward to the two talks, as well as the question section, at the end also the panel, with that, I will hand this over to Professor Claire Gray first to start today's program. Claire, please. Well, I'm delighted to be here, and thank you very much, and thank you to the NCE and Will for the invitation. I'll just wait for my slides to come up, and then I'll start. So what I wanted to do today was to focus on just two aspects of my work on NCA materials in 811, looking at the structure, the dynamics, and the degradation of these materials. And you're going to have to work with me, because there was a bit of a lag. I wanted to really look at the issues associated with moving towards increasingly nickel-rich materials, whether they're 811 or NCA, and the increased degradation that you see, particularly as you go to higher voltages, and also to reflect on why there's some initial capacity losses. And this work is very much a combination of work that I started when I was still at Stonybrook in the US almost 10 years ago now, with NEXIS, the Department of Energy Center, and then moving forward on with the Faraday in the UK. So I want to start by reviewing the work we've done on NMC 811 on the structure and the dynamics. Then I want to talk about the first capacity loss, which is a work done in collaboration of Karina Chapman's group, Karina now being in Stonybrook, and then move on to looking at the structural fatigue and long-term degradation in NMC materials and end with some of our current studies in future directions. So as I said before, the NMC 811 material is done within the context of the Faraday institution and the battery degradation project that I lead, but we're really trying to use a whole variety of different methods to interrogate the mechanisms for degradation of 811. But I'll just focus on some of the initial studies, largely done by my postdocs in Cambridge, particularly Karina Marker and Chow Chu. And so in collaboration, again, with the Argonne lab and Camilla and Karina, we've used the Ampex cell that they developed a number of years ago as part of the EFLC project to look at cycling and doing in situ NMR. And so this is a classical in situ NMR, sorry, in situ diffraction study of NMC 811, showing the classic changes. This is the 003 peak. You see it shifting, first of all, to lower 2 theta and then up to two higher, back to higher 2 theta, which you can then from a refinement see the classic pattern of a layered structure where the layers expand and then they compress above about 0.7, 0.7 removal of the material, or about 200 milliampers per gram. So looking at this, then, comparing the C parameter drop and the lithium layer collapse, what's interesting is that you see this change in the C parameter at about 0.65 and this dramatic decrease in the C parameter. But if you unpick it and you separate the changes in the layer spacings, be it the lithium layer or the transition metal layer, the key point I want to show you here is that the lithium layer itself doesn't collapse until about X equals 0.75. So 75% of the lithiums have been removed from the structure. So the question we asked then was what was the impact of this on the lithium mobility? So I just want to remind people of some seminal work done many years ago by our next speaker, Gerd Seder and his former PhD student at the time, Kisa Kang, who together with my group looked at the role of the lithium spacing in determining mobility in layered materials. And so what Gerd's group did by DFT was to determine the activation barriers as a function of the slab layer. And a lithium mobility in the lithium layer involves a hop from an octahedral site through a tetrahedral site to another octahedral site. And the tetrahedral site position acts as the transition stage or the intermediate in the hop from the octahedral site to the octahedral site. And this distance between the tetrahedral site and the metal in the layer below is a direct control of the activation for that mobility of that hopping process. And so the smaller the spacing in the lithium layers, the closer this contact, the higher the activation barrier. The other trend is as you go from nickel two plus to three plus to four plus, as the material becomes increasingly oxidized, the activation barrier increases. Again, because of this increased propulsion. So just look at these numbers. There are about 300 to 400 milliolectron volts for that activation barrier. So what we then did was to examine the materials by NMR. So what we're doing now is to use X-situ NMR. And so what I mean by that is we make lots of different batteries. We stop them at different states of charge. We pull them apart and we do NMR spectroscopy. We're doing lithium seven NMR and when we start, the initial material contains a very broad signal. And it's broad because the lithium sees a whole variety of different transitional metal ions in the layers above and below. And for a number of years, we've basically developed an understanding of how to interpret these NMR spectra in terms of the numbers and types of paramagnetic ions. And so if you have a nickel two plus ion with two EG electrons, you can have different shifts depending on whether the nickel points directly at the lithium via a so-called 180 degree interaction or 90 degree interaction. And this is nicely described in Dongli Zhang's work a number of years ago. But basically, if you're a lithium and you see nickel two plus and nickel three plus and nickel four plus randomly distributed, you get this very broad pattern that we can model using close to random distributions. But interestingly, when you start to pull the lithium's out, you can see this noticeable narrowing of the spectrum. And then at the top of charge, you actually see much more clearly resolved peaks. And we can assign this particular peak here to what lithium surrounded by one manganese and this one to lithium surrounded by two manganese because each lithium that sees one manganese in a 90 degree interaction. So this lithium oxygen manganese 90 interaction gives you about a 250 ppm shift. So that's all explainable. At this point, all the nickel is four plus so it's diamagnetic. The question is, what's going on there? So that mystery was explained by Katarina's studies where she simply took the materials and she heated them up. And when you look at a material that's barely dilithiated, you see very little change in temperature. It's a very little change in spectrum as you change the temperature. But in contrast, if you moved to 0.25 and you heat the material up, you see this dramatic narrowing of the lines and that's indicative of increased mobility. So the lithium ions are hopping between the different sites that give you this broad distribution and so that you only see the average shift coming from all the average environments. Now this is a well-renowned phenomenon in NMR called two-site or multiple-site exchange. And the bottom line is we know how to model it. And so these are simulations done by Katarina where she assumes, first of all, a random distribution of lithiums being surrounded by a random distribution of transition metal ions. And then she can then put in hopping between these sites and then model the spectrum as you start removing lithium from the material. And you can see that the spectrum are well-modeled by a model where you've got some rigid ions remaining but most of them by the time you've come to X equals 0.5 are mobile. So what does that mean? It means we can extract the hopping rates as a function of different states of charge. And what you can see is that when you remove more than 10% of the lithiums, you get a dramatic increase in the lithium mobility, reaching a maximum at some point between 0.4 and 0.6, while as at 0.75, it's dropped off dramatically. We can extract activation barriers, the errors in the activation barriers coming from the assumptions in our modeling. But the bottom line is the activation at barriers for lithium transporter with the order of 350 millivolts, which is very close to that predicted earlier on by the simulations or the DFT calculations of Gauss-Sader in this group. And the point also is that the mobility we get agrees very well with that obtained by GITT, where about above 0.75, you get a rapid decrease in the lithium mobility. So let's go back and look at the diffraction data and try and pull it all together. So remember I said that the C parameter collapses at 0.65, so that's not where we see the decrease in the lithium mobility. We see the decrease in the lithium mobility rather at 0.75, but if you unpick the different layer spacings of the transition metal layer versus the lithium layers, you see that's where the lithium layer starts to collapse and increases the activation barrier for transport and the lithium mobility decreases. And so the implication, of course, is if you want to have access high rates in NMC materials, you should cycle within the regimes where the lithium mines are moving. So let's unpick this a little bit more and think about starting to cycle the materials. So moving on to the next topic, I want to look at this issue of the first irreversible capacity loss on the first cycle. And so this irreversible capacity loss seems to be quite consistent amongst a whole variety of different nickel rich materials, whether it's 811 or NCA's. And work by Karina's group in Stony Brook and also Stan Whittingham's has shown that this is largely kinetic in origin. So if you, for example, go from room temperature to higher temperature, and this is Stan's work, you can actually reduce the first capacity loss. And so we wanted to look at what the implications of lithium mobility were on this process. So this is describing now work that was published recently in JAX, which was a collaboration between myself and Karina's group and built very much on earlier work in NEXIS, looking at the role of the impact and growth of surface layers in collaboration with predominantly Glen Amitucci's group. So I want to point one thing out that this first cycle irreversible capacity loss is much more pronounced in NCA than in lithium cobalt oxide. And I'll come back to that later on. So if you do a constant current charge and then discharge, you can see that you come back to about 0.9 lithium back into the material, but if you actually hold it at 2.7 volts for, in this case, 24 hours, we can really almost get back to 97% of the capacity. And this is the work done by Antonin Grenier. And we can then look at this process by NMR and show the same thing. And so in work that Phil did, he showed that if you hold the material, again, at a constant voltage for 2.7 volts, you get almost all the lithium back in again. And you can see this via the lithium NMR. And you can also see this by the aluminum or aluminum NMR depending on where you are in the world. And the aluminum NMR is very sensitive to both the surface impurities, but it's also sensitive to the aluminum aluminum in the bulk. And you can see that only on holding at 2.7 volts, do you get all of the aluminum back in again. So careful work by Antonin, looking at the diffraction pattern, using a single phase model to fit the patterns on charge and then discharge, raised red flags, particularly when looking at the R factors or the profile fits at the beginning of charge with single phase. And so what he found was that he really needed to model this with two phases. I just want to make one point before I move on and show you this analysis and to say that what he's seeing is very distinct from the two phases that you see. If you take, for example, an NCA or an NMC811 and you leave them in the air and they build up a lithium carbonate or hydroxycarbonate coating. And there you really will see this two phase material because of the impedance associated with getting the lithium through the lithium carbonate. And this two phase process disappears if you look after your materials well or after the first and second and third cycles whether you remove the carbonate electrochemically. So what Antonin did was then to move beyond the single phase analysis to fit the first process with the two phase reaction. And so what you're seeing here is the C parameters and the A parameters changes with the two phases growing in. So you have a second phase with a larger C parameter which gradually grows in percentage. And by the time that you're at 0.75 that corresponds to essentially the full phase fraction while the initial material dies. So is there a role for mobility in this? There must be otherwise I wouldn't be linking the two stories together. So what Phil did was to go back and do the NMR again of the pristine material and then looking at the partially dilithiated materials. And again at about 0.86 lithium he sees a noticeable change in the NMR patterns as a function of temperature. This is variable temperature NMR. By the time you're at 0.71 you get again this dramatic line narrowing indicative of again lithium mobility. And a more sensitive way of looking at this is to look at the so-called spin-spin relaxation or T2 time. And so this is a work of Katerina's where by quantifying how fast the lithium ions move in the time scale of this measurement which is about a hundred microseconds you can again look at the onset of lithium mobility that's happening at about 0.86 or just before 0.86. So the point is you start to get rapid lithium mobility as you start to pull lithium out of the structure. So let's try and pull all this together and go back to the idea of how lithium ions move in these layered materials. So remember we're getting hops from octahedral sites across through this intermediate tetrahedral site and the hop becomes allowed and this is a work nicely summarized in a review article by Anton van der Ven when you have a die vacancy mechanism nearby. So the hop will go to a tetrahedral site and when there's two vacancies here the repulsion for the next part of the hop process is removed. So you imagine you start off on your NCA material you've got almost a full lattice and there are very few vacancies so the lithium mobility is very poor. So you then can imagine or do a thought experiment where you pull a little bit of the lithium out of the surface of the material and this is to form phase B. So you have composition A that's almost completely stuffed of lithium, composition B that's partially delitiated and then you can ask the question where does the next lithium come from? Are you going to pull it out of phase A which would be the thermodynamically phase or because of the difficulties of the kinetics of pulling out phase A where you basically have very reduced lithium mobility what the diffraction suggests is that there you start to go via a kinetic phase or pathway where it's preferable to pull the lithium out of phase B because it's just easier because you have much more rapid lithium mobility in this phase. And so what you see then is at least in the first part of the the lithiation is this removal of the lithium by this two phase behavior when you get to a regime where at 0.76 where the lithium mobility is really fast there are no more kinetic limitations when you go back to a single phase behavior. So just to be very clear this is not two phase behavior based on a thermodynamic reason that's for a kinetic reason. And it has interesting implications because on the way back you don't see this phase segregation. And I'm going to try and explain this as clearly as I can with these little cartoons. So on the way back you've got a phase B and remember this is the phase that has rapid lithium mobility. And as we're starting to put the lithium back so that the lithium is now almost completely full you can imagine again this surface state that's almost lithium 100% but because the phase B has got rapid lithium mobility you can actually get much more rapid equilibrium. And so you can continue to add lithium into this material until you finally hit this state where the lithium mobility is too low. And so kinetically you can't get more than about 90% of lithium into the material and only by holding this low voltage can you force the lithiums into this fully stuffed phase. And so using these arguments it's very straightforward to rationalize why on charge you get two phase on discharge you get these single phase behaviors. So then I wanted to return very briefly back to why the kinetics are or why there's a difference in lithium cobalt oxide. And I wanted to remind people of the work that Michel Menetray did many years ago where he made these really stoichiometric lithium cobalt oxides and showed that if you hit the stoichiometry spot on you could really get essentially zero coulombic or 100% coulombic efficiency in the first charged discharge cycle. And what we believe is at the origin of this is the distinction between the metal insulated transitions and the lithium cobalt oxide insulated to a metal. And in this metallic phase essentially the energy landscape is very flat for lithium because all of the cobalts or the transition metal ions have the same oxidation state. And so there's none of this pinning that goes on in these semiconductor systems such as the NMCs and the NCAs. And so once you start introducing dopants into the lithium cobalt oxides whether they're nickel free passes or even lithium excess you go back to this process where you have some first loss in coulombic inefficiency in the first cycle. But if you have a really stoichiometric material you get very little pinning of this interface. And so you can get a very rapid movement of the interface through the phase and you don't see this behavior. And so experiments are in progress to quantify this phenomenon particularly by looking at the lithium mobility in these materials. So I want to now move on to the last topic and this is now moving back to 811 and again coming back to the work of the Faraday institution. And I want to look at particularly the role of what happens when you move now in full cells. So these are NMC811 graphite full cells in LP57 where we're looking at the role of different degradation processes in capacity loss. And so these are materials coming from Targrey and Argonne National Lab in the camp facility where you can see that when you cycle at low voltages so for four volts and 4.1 volts we're getting very respectable capacity retention. But once you go above 4.2 and 4.3 volts so that you get some good initial capacity it drops off very rapidly. And these are cycling data from West Dose in Cambridge. So just to go back then to remind you about the structural evolutions on charging. So this was the pattern I showed you initially where you take an 811 and you charge it and there's the expansion the C and then the collapse. What we're now then doing is what happened we're looking at what happens to these diffraction patterns as you cycle the NMC811 for many time cycles. So this is an 811 that's seen 348 cycles. So we're charging at C over two and then we hold it at 4.2 volts in a constant voltage. And then when you look at these 003 you can see that there's no longer a single phase at these high voltages. Instead you have what we call the active phase and then you have a component that's responding much less rapidly. So what Char-2 did was then to take the material out of the full cell, charge it against lithium so that he could control the voltages very well. And so this is now the diffraction pattern of the 811 Char-2 4.3 volts. And if you blow up the 003 reflection which is telling you something about the layer spacing was directly related to the layer spacing you can see that in order to fit this we need to invoke at least three different rhombohedral patterns. So there's no evidence for any 01 phase everything is the normal rhombohedral structure but it's clear that there are multiple C parameters. And so we just going to distinguish between them as three different components that being the sort of minimum needed to fit the patterns that the most fatigue phase, the intermediate phase and the so-called active phase that looks very much like it was that it does the same thing as the material in the first cycle. I just want to point out that gas tigers, groupers and others have also seen similar sorts of phenomena in NCA materials on multiple cycles. So what we've done then is to move to the light source and used laser thinned coin cells so that we can use a standard coin cell now to study these long cycle experiments without any worry about moisture and contamination. And so this is a comparison between our in-house coin cells versus these long duration experiments and we're getting very good electrochemistry or at least very similar electrochemistry. And then this allows us to interrogate what's going on after multiple cycles. And so what we do is we come back sporadically once a month and we collect the diffraction pattern on charge and discharge to follow what's going on. And what you can see is as you progress from cycle each sort of a hundred of cycles we're getting increasingly more fatigued. And by the time you say a cycle 915 there's no active phase left. The whole material is essentially all fatigue. So there's none of this collapse of the C parameter of the chopper charge. So let's think about this a little bit more. So this is just a summarizing what we're seeing. So the fatigue phase is essentially getting to 4.2 volts and it doesn't continue any further along this higher voltage process which is then associated with the collapse of the C parameter. So we then started to think about what was going on in the system. So one of the questions we had was associated with into cracking between the secondary particles by the primary particles and the large secondary particles of the more traditional 811s. And so we started to investigate this via single crystal 811. And you can see a material now that's also been cycled again. This is aged in a full cell for 500 cycles pulled apart against lithium metal. And again, you have the same phenomena of fatigue phase where the C parameter doesn't collapse and an active phase that is behaving normally. So it doesn't seem to be associated with the particular primary secondary configuration that you get in the traditional 811s. So we then asked the question is there something about say the anti-site mixing or the disorder that's changing the lithium mobility? And so you've seen these lithium mobility experiments before for the first cycle where we could model the spectrum as a function of temperature. So remember we see as you go up in temperature the lines collapse which is indicative of lithium mobility. And we can compare the first cycle with the sample after a thousand cycles. And you can see while there is some increase of disorder the lithium ions are still mobile at 4.2 volts. So it's not that the lithium ions are trapped because of the mobility per se. So then let's examine this whole structural question. And so when we compared all of the data from the synchrotron diffraction from the in situ the ex situ the lab diffractometers we saw that this fatigue phase always ended up at the same state of charge. So basically there was always about 25% of the lithium left in the structure. And this corresponded to the point of the C over A ratio just before the collapse of the C parameter. So this seems to be structural in origin and the state of charge is independent of the method is the fatigue phase is always the same. So then the question is what was the relationship with surface reconstruction? And so of course the surface reconstruction of these layered materials to form rock salts on the surface. So this is showing now work of Johan and Layla from Liverpool looking at the electron microscopy the stem images of the same materials where you see the classic layered materials moving into the bulk but then a rock salt cubic structure on the surface. And so what we found was extremely interesting is when you analyze particles in more detail so these are materials that have been cycled for about 300 times and comparing a pristine one with an age one you can still see open channels. And so this is material that's been charged to 4.2 volts and then held at 4.2 volts and then examined by TM. And you can look at age materials that have very similar spacings as the pristine materials. So here's the pristine spacing and here's the open channel in an age material where they're very similar but then when you have a rock salt structure you see that there's a rock salt spacing which is larger and that is seeming to pin the transition metal layers moving into the bulk of this material. So we're going to at least hypothesize that the origin at least of some of this fatigue is related to the surface reconstruction. So we know that on discharge we can get the lithiums back in again. So it's not per se the issue of the lithium not being able to move through the rock salt layer. So there's definitely going to be an impedance effect but it's not enough to force the lithiums back in again. It just seems to be sufficient to stop the lithiums coming out at this last point at the last between above 4.2 volts. And what's extremely interesting is that the relationship between the spacings of the rock salt for example in a nickel oxide based one exactly corresponds to the same spacings of the NMC at this 75% top of charge. And so we can calculate the strain in the C direction and the A direction. So in this direction versus the layer spacing. And you can see that the fatigue phase basically stops when this strain has been minimized in the C prunter. So we're going to at least hypothesize that the rock salts are helping to pin these layers apart and that's making it very difficult to pull the residual lithiums out. And of course this now requires that we do more mechanical modeling to understand this, to do DFT calculations to understand the energetics of this process but also to explore the role that doping and coatings may play in changing the structures of some of these rock salts and mitigate or changing these processes. So I just wanted to end on discussing a little bit about where we're going next. And so in work of Wes Doze and Michael DeVolder we've been trying to really pull apart where the degradation processes occurring and using the sort of approaches of slippage pioneered by the argon group. What we're doing is taking a full cell data decomposing it based on half cell voltage profiles obtained for the anode and the cathode and what that allows you to do instead of looking at the DVDQ, you look at, instead of looking at the QDV, you look at the DV, sorry, the QDV, instead of looking at the DVDQ you look at the DQDV which allows you to add in a linear fashion the components that come from the anode and the voltage. And when you do that you can see that the capacity retent loss is largely associated with a slippage on the anode. You can more, in addition, you can pull out the loss of capacity from both the anode side and the cathode side. And you can see that the anode maintains its capacity in the sense that you don't have to compress the amount of the fraction that you use from the anode. And the same thing from the NMC at 4.1, again, the scaling factor remains the same. So it per se is not degrading substantially if you only go to 4.1 volts. What's happening is, is you're tying up the lithium in the electrolyte so that you no longer access as much of the cathode as part of this process. And this is what's shown in this so-called slippage curve where you're plotting the capacity loss versus the capacities of slippage on the anode. And so this is all good to 4.1, but when you now start to move above to 4.2 and 4.3 and you start to see the processes that I was describing with these fatigue phases, then you start to see additional processes coming in from the cathode starting to degrade. And so then just sort of think about why the anode is, why despite the lack of unit 4.1 volts, the lack of capacity loss per se of the anode and the cathode, just lithium inventory loss due to the SEI formation, it's clear that a lot of this is coming from the electrolyte oxidation of the cathode. And of course, we're not the first to say this, this is some beautiful work by Hubert Gasthaig and his group. And so what we've done, and this is now work of D.D. Wrinkle, is to really start to unpick a lot of these reactions starting off with lithium-cold oxides and of course moving on to the NMCs to try and unpick what are the reactions due to oxygen loss? What are which of them are simply chemical and which of them are faradaic? And that's a talk in its own right. And just to end then to say, this all wouldn't have been possible with great teams both in Cambridge in my own research group and I mentioned most of them moving on, a strong collaboration of Kate Dukakis group in Cambridge on the TM, Michael DeVolda on the slippage collaborations with Sarah and Chu in the diamond light source, Johanna and Leila on the TM and the whole degradation project. And of course last but not least, the NMCs team for all of its collaboration, wonderful collaborations over the years and particularly in the studies I talked about recently, Karina's group, Howe and Camilla. And thank you very much for listening. I think that was almost perfect timing. Well, thank you Claire for the very exciting talks. You discuss very deeply into this high nickel content, how it works, what's the degradation mechanism and so on. From the audience, there are a number of questions already. Let me give you the first one. It looks like somebody really knowing you, say Claire, great talks as always. The question is in the pristine MMCA11, what are the precise nickel manganese cobalt oxidation state? So manganese is simple, it's always four plus. The nickel is gradually increasing up to four plus. The cobalt, I know that traditionally people think that the cobalt is really only being oxidized at the top of charge. I think our initial study suggests that the cobalt is almost starting to participate, not right from the word go, but it's gradually kicking in as you start putting up. So I don't think it's as straightforward as saying, it's all nickel and then it's all cobalt. Yeah, the same person also asked, did you see any evidence of nickel three plus? The disproportionation reaction to nickel two plus and four plus. So I'm just to, one thing I didn't point out was when we look at the NMR, in order to model the NMR spectrum, what you have to do is the nickel, it's obviously starts off with the mixed nickel two, nickel three, but you have to model it as a quick hopping between nickel two plus and nickel three plus. So on the NMR timescale, it's always mixed oxidation. So the timescale of electron hopping is faster than the NMR timescale. And so as you gradually increase the, the you proceed on charging, that you get a gradual increase in the oxidation state up to four plus. And so in our hands, it doesn't really make sense to think about the yarn teller distortions. We don't see an evident, we never see any evidence of a static yarn teller coming from the three, the nickel three plus. That's not quite the case when you start looking at the aluminum doped NCA systems. And this is work of Yain and Seymour and others and Nicole Treesfoo and Nexis. There was some evidence of the aluminum pinning the nickel EG electron. So there was some evidence of a static yarn teller in that case, but otherwise we don't see it. And then in terms of the rock salt, well, I mean, I believe that that rock salt is intrinsically associated with nickel being reduced down to low oxidation states of two and maybe some three, but that's when we haven't looked at directly at this point. Mm-hmm, okay. So next questions. Some years ago, nickel lithium nickel oxide was not considered as a promising candidate. Now we have 81, you know, 80% of nickel. Now even more going up and EV. So how far are we from just lithium nickel oxide and EVs? Yeah, I guess this is probably much broader question beyond what you just talked about. Yeah, and I mean, I'm cognizant of the fact that there are plenty more experts on this subject listening, possibly listening to this call than I am. But what I would like to say is I think the issue is in the electrolyte and solving that surface, that couple surface reconstruction with oxidation of the electrolyte. And I think, you know, there'll be many people who've made progress on additives and dopants. And to me, that's the issue is you have to solve that electrolyte oxidation issue to fully enable a very high nickel rich material that will cycle for many cycles. I think the NMCs themselves are fit for purpose. You know, the bulk structures, they largely cycle. Yeah, I agree with you actually, Claire. You know, this question coming from me. Do you have this powerful MMR technique? Once the surface reaction taking place with electrolyte, there's a transition matter's potential going to the electrolyte. So with your MMR technique, because the co-ordination environment change, would you be able to also quantify the solution? Yeah, to be honest, that's another study that is particularly work of Jennifer Allen, who's been with Chris, has been looking very nicely at that. It turns out we can look at it in principle. You can do it by ESR spectroscopy, but you can also do it by NMR. And it's because the ions that go into solution, are paramagnetic and they have very definable changes. And so you can look at the liquid electrolyte and you can actually see indirectly what metals are there. And so this is something we're actively looking at. So at this moment, I'm not sure that I know the answers to what's more important. Is it the metals in there, or is it the electrolyte oxidation? But I do think they're all intrinsically coupled. And that once you particularly go down to lower oxidation states, then you get some more metal dissolution as well. So, and what I didn't talk about, or I moved maybe too fast, the work we were trying to do on looking at electrolyte degradation is to put things like OHARA glass in the middle so that you can decouple, process happened the cathode from the anode and then you compare the two. So you can look at the role of the crosstalk between the two and unpick the two. And we have this view of SCI formation as being very basic chemistry. It's all close to lithium metal or lithium hydroxides. But what's going on on the anode is much more acidic. And so if you put a blocking electrode in or something that stops that crosstalk, you get some very different chemistry and you can really pull apart some of these phenomena. That's a very good experiment to do. Yeah, that's a really good experiment. That's where we are in that. I think that's all of the exciting stuff coming out of that. Okay. Our next question, Claire. What's the fundamental reason the lattice C increase first, then follow sudden drop during charge? I'm very cognizant of the fact that my next speaker could probably give a whole tone on that or something. Okay. No, no, I can answer it. You don't need a theorist to explain this one. Let's pull the lithiums out of the structure. Then the metal oxidation increases. And so then you get a repulsion and that comes up. And then basically when the lithiums come out, the transition metal, the metal oxygen bond becomes more covalent using a chemist term that reduces the partial charge on the oxygen. And so you get a collapse. And so you've got these two competing phenomena. It's a repulsion between the increased repulsion between the layers of the metal part. But then you've also got changes in the covalency of the oxygen coupled with just simply having the lithiums propping the layers apart. But I think it's actually a much more subtle thing than that. There's also roles of you end up polarization of the oxygen charge clouds and dipoles. And I think what we need to do is look very critically at why you get these rapid collapses. And what is the role of essentially staging or pinning effects of metals in those layers? Why is it subtly different from 8-1-1 and NCA? And I'm sure that the next speaker can comment later on and give his views on the subject. Yeah. Well, this question actually also be answered. This is a great example in teaching a solid-state chemistry. I think this could be used right in the future. Next question, do you think the pinning of the layers by rock salt applies deep into the core of a particle? Yeah. So that's the multi, I mean, in so far it's a multi-million. But it's the big question that we're wrestling with. So what we see is that if you analyze the diffraction patterns of the single, not the single crystal, the large normal ones with small secondary particles. So the primary, sorry, the large secondary particles, small primary particles. The primary particles are about 200 to 500 nanometers. If you look at the refinements of those, there's not a lot of broadening. So what that suggests is that there's not a large distribution in cell parameters. And this effect is coherent across about 200 nanometers. So then you ask the question, how can you have a surface thing that pins in those sort of distances? And I kind of think what is what we're wrestling with is is that really a big enough effect? And I know that there's a mechanical people out there who's sort of maybe pulling their hair at this point. But I think what's sort of going on is it's coupled with the point where the voltage is fairly flat. You're at this point where if you were looking at a pure lithium nickel compound, it wants to form an 01 phase. But it doesn't form an 01 phase. And so I think it's the sort of stuff we really don't understand, why a material moving with just a little bit of cobalt, a little bit of manganese doesn't form an 01 phase. You know, that's the 01 phase is very clear and diffraction. We're not seeing it. But yet it really does want to at that point. The voltage is quite flat. And you actually see a spike in the DGDV plot, which most people would then say that was the 01 transition, but we're not seeing it. And yet it's wanting to have this massive collapse that rock salt seems to be pinning it at least to a certain extent. And I just think that's something that really needs to be unpicked. And to see what the energetics of that process are and why is it just the sort of, it's not an increase in anti-site mixing. We're not seeing any change or noticeable change that would cause this effect. And again, just to reiterate, the lithium goes back in again. So you can force it back when you go all the way down to four volts and below. You can get the lithium back in again. So there's no inherent kinetic limitation of getting through that rock salt layer. There may be a kinetic limitation of pulling it out because of where you're sitting, the lithium mobility is dropping. So there's a number of related phenomena going on. Yeah, well, Claire, let me report back to you. People are really excited about your talk. There's so many questions flowing in. I think probably we can take just two more. So one question now goes broader about the MMR technique. What's your comment on using this technique for other benefit systems? Sodium, potassium, multivalent. So yeah. Well, I mean, as you know that if it has an NMR, yes, my group has largely looked at it. Sodium is easy. Sodium is very similar in some ways to lithium in terms of its properties. So that's straightforward. I showed you an example with the aluminum or the aluminum, phosphorus, fluorine. Potassium is more difficult. It's a more difficult nucleus. So you have to sort of ask the question, is it the best technique for that? Are there easier ones that potassium is in NMR nucleus? It's just a much harder one. So the problem with NMR is not like diffraction where you can just always do the same diffraction experiment essentially. And you kind of know how to analyze the data. Here you really, it's much more bespoke. You've got to think about what are the interactions, whether it's the paramagnetic ions, whether it's the quadruples, which is the distortion of the charge cloud, sorry, the nuclear charge received then interacts with electric field gradients. So in principle, it's a very powerful method, but you've got to choose your experiments. OK. I think one last question. We're giving the problems showing up with high nickel content. This audience asks, but it's better to use high nickel content in solid-state electrolyte instead of liquid. In solid-state, this could be a whole another issue coming up. I mean, there are others who are experts, but just look at that C-clamp collapse and that expansion and collapse. You've got to design a solid-state system that will deal with those expansion and contraction. And that's a lot of stress and cracking and fraction fatigue you're going to have to do. So I don't personally, I wouldn't necessarily want to never say no, but I don't see it as sort of instantaneous, easier system. I think if you're looking at composite systems where you've got polymers and you build in some flexibility, maybe, but it's not a simple solution to this problem. And you have to mention, of course, what if we're going up to 4.3, 4.4 volts, you need oxides that are stable at those sorts of voltages. And the reality is that having a material that's stable at those high-oxidation voltages is also challenging from a solid-state electrolyte perspective. And so oxygen loss is also an issue with those systems. And again, this is a lot of theory work on. Thank you so much, Claire. I know we cannot, we don't have enough time to answer all the other questions. I think this is just great. Thank you. Now let's move on to Geh Seder. Geh, you are ready, right? So let's. Ready. Well, thank you for setting up this symposium. And thank you for all those of you who are staying on, especially our friends in Asia who, for whom I know, it's really late. So Claire gave a really beautiful talk on some of the newest NMC materials. So my talk will be a continuation of that in the sense that we're sort of trying to understand what may come after the NMC series. So I'm going to talk a bit about the science of disordered rock salts, the role of short-range cation order, and how it can be induced to make high-rate capable DRX materials, the important role of fluorination, which gives stability to the materials. And then I'll get into some very recent work on partially-disordered spinels. And in the end, I'll talk a bit about the importance of electrolyte issues. Again, already stressed by Claire on the NMC materials. So you've heard a lot about the NMC materials, and they are great materials, right? There is nothing wrong with them. They are clearly going to be the workhorse of the industry for a very, very long time. They are highly optimized. But there are limits to them. And there are, first of all, chemical limits to them, that there's a reason they're called NMC. They contain nickel, manganese, and cobalt. And that's pretty much the only elements you'll ever see in these layered materials. And that comes from their electron physics, that if you make a layered material, if you pull the lithium out, then, in many cases, transition metals want to migrate into the lithium layer. And the only elements that do not do this are elements with very strong octahedral ligand field stabilization. So these are essentially elements that have the T2G electron shell heavily occupied. Things like cobalt 3 plus, manganese 4, nickel 4, nickel 3, and to a minor extent, nickel 2. And that's why you see these elements so prevalent in the NMC series. They make the material layered, and they also hold it layered. And so it's not like we're going to make good layered materials with all the periodic table. There's probably also limited improvement remaining in NMCs as we hit 700, 800 watt-hours per kilogram. And then there is, of course, a potential resource issue. Everybody knows about the cobalt issue. But of course, the nickel issue is just a severe. If you want to make a terawatt hour per year of NMC, then you need a million tonne of combined cobalt and nickel. And with NMC811, that's now mostly nickel. And to put that in perspective, that's today 40% of the nickels of the world production of nickel. I used to think that one terawatt hour was an aggressive estimate. A study in Korea just pronounced that by 2030, we will be making three terawatt hour. So at three terawatt hour, you are consuming all of today's nickel production. And batteries are already the largest nickel consumer after the solid state, after the stainless steel field. So if you want to think of novel cattled materials, we've done this for almost 20 years now. We've gone back and forth between different materials classes. And here's really where I have come out. You need obviously high energy density, or you're not a player. Every other condition is secondary. You really need high energy density. And hopefully, we can also make chemical diversity. Because again, if we stick with one element, essentially less abundant element, there's no way we can scale to multiple terawatt. And then on the anion side, it's becoming pretty clear that we need oxides. If we do polyanions, unless we can do multi-redox, the energy density will not be high enough because we just get too much space taken up by the polyanion. And even more with the oxides, you need dense packing of them, which is typically represented in FCC type packing. So this path leads you fairly naturally to the disordered rock salts. So first of all, the chemical diversity, if you don't care about having a specific ordering of the cations, then you can use all kinds of chemistries. Because some of them will disorder, some of them will give you other ordering. So if we can make a disordered rock salt work, we can actually use a much more diverse set of elements. Rock salts, of course, have the FCC packing of oxygen. And that will give you the high crystal density. And therefore, if you have high specific capacity, also the high specific energy. So the big issue with disordered cation materials is whether you can make lithium diffuse in them. So we have all worked extensively on layered materials where there are very well two-dimensional pathways for lithium spinels, where there are three-dimensional tunnels penetrating lithium iron phosphate, where there are one-dimensional tunnels. So the question is, how do you diffuse when you have a cation disordered systems where there are no specific sites for either the lithium or the transition metal cations? And Claire gave a wonderful introduction already to how diffusion occurs in disordered rock salts or in any rock salt geometry in general. Again, rock salts are FCC anion packed. The cations are usually tetrahedral. So the way you diffuse from one octane to the other is through the phase-sharing tetrahedral site. And the energy of that site is set by the size of it, as well as by the metals that are nearby. You always need a die vacancy. So one of these, what we call gate ions is vacant. So the question is what the other one is. In a layered material, the fourth ion is a transition metal. And that's what you see here. And this works really well when this tetrahedron is large. And that's what Claire talked about. So if the lithium slab spacing contracts, then the activation energy goes up. Because when lithium diffuses in here, we have more of a repulsion with the transition metal and the kinetics degrades. You don't have this if you could make this channel. So if you could make a tetrahedral activated state where there's only lithium, then this site and this site would be vacant in a diffusion process. So when the lithium goes in, it only phase shares with another lithium. And in layered materials, you don't have these environments. But when you make lithium-access materials or where you disorder, you will statistically get all kinds of environments. So some of these will appear. And indeed, they have lower activation energy as a function of tetrahedron height. The black line is this 0TM channel, as we call it. The red and the blue lines is this 1 transition metal channel. And an important thing is that this 0TM channel, the activation energy through it is much less sensitive to the size of the tetrahedron. So that means you can make more compact structures and still have good diffusion. So of course, migration through that channel by itself is not diffusion. To have diffusion, you need percolation of these channels. And what we found is if we study random cation distributions in a rock salt, you get percolation when you have about 10% lithium-access. So this is the basic principle. In disordered rock salts, diffusion occurs through these channels. And they statistically percolate through the structure. So you don't have a well-defined path like in a layered material. But you have these kind of like tortuous pathways through the material. And this actually works. Here's a material we made early on, a lithium molychromium oxide. Here's a stem picture. It's fully disordered. It's really a rock salt. But you see you can cycle one lithium per four million. So 280 milliamp hours per gram. Here's another material. This related class been studied a lot. 20% lithium-access manganese niobium material fully disordered. And I see, again, well over 200 milliamp hours per gram. So the model of diffusing through percolating zero TM pathways really works. There's been lots of material studied now. And again, our first aim is kind of achieved here. There is sort of very diverse chemistries in play. And actually, we did a recent study of the papers published in this field, which is still very young. And the colored metals are all involved in these DRX materials. And what's kind of amazing when I looked at this, almost every transition metal that actually can be used, that's practical, has actually been used in the DRX materials now. Almost all the 3D materials, all of the 4Ds that are reasonable. Of course, we're not going to use technetium or rhodium or palladium or silver and some of the 5Ds. So clearly, when you go away from the cation order, you can use lots of metals. You can, in general, get quite high capacity. A lot of these materials hit 7, 800, even approaching 1,000 watt-hours per kilogram. And I'll go a little more in depth on some of these materials. A lot of them can be manganese redox-based. Some of them are nickel redox-based. But we have extensively investigated the manganese redox materials. Since the original invention of these materials, there's a few things we have found out. So in the original work, we assumed the cations were randomly distributed because in X-ray diffraction, you see no superstructure peaks whatsoever. It's just the rock salt. If you do careful electron diffraction, you actually see what's called short-range order phenomenon. So short-range order is a deviation from statistical randomness. It means that some ions are slightly more correlated to be neighbor to another one. It's not domains. It's not bad long range, or it's purely a statistical phenomena. And if you look at electron diffraction, so these dark spots are the black peaks, and then these sort of fuzzy circles are the short-range order. And I'm comparing here two materials that we thought would behave very similar. This is a 20% lithium excess. Manganese is the redox couple, and a titanium 4-plus stabilizer. And all we do here is we change the inactive element to titanium to zirconium. And we actually thought that would make a better diffuser because zirconium is bigger. But what you actually see is the opposite. If you look at the performance, the zirconate is actually very poor performance. The lithium manganese titanate, you can cycle about 80% of lithium performing unit, again up to 200 million parts per gram. And what we now know is that this is actually due to the short-range order. If you do a simulation and you plot the zero-tm channels, this is the titanate, this is the zirconate, because of the typical near-neighbor correlations, the short-range order correlations in the ion, the titanate has very good percolation, as is shown here by the green tetrahedron. These are the zero-tm channels. As you can see, they connect really well. Whereas in the zirconate, you have exactly the same composition, exactly the same heat treatment, but you actually have very poor percolation. And this is consistent with the observed short-range order and performance. So this is important because it means that you can use short-range order to tune the rate capability. And what we have done here, and unfortunately, we're not ready to disclose this composition, through compositional modification and annealing and heat treatments, we can tune the short-range order to get better percolation and, therefore, get better rate capability. And here's a DRX material that at 20 milliamp per gram gives almost 1,000 watt-hours per kilogram, but holds really good rate capability prior to some of the, in contrast to some of the prior DRX materials. For example, 50 milliamps per gram, even at 500 milliamps per gram, so almost 2C, we get very good rate capability, very good capacity. And we've even discharged this material at two amperes per gram, so relatively speaking, an 8 to 10 C rate. And this is purely done by optimizing the very local short-range order. Okay. The second part that's important about DRX materials is that unlike layered materials, you can substitute part of the oxygen by fluorine. And this, I have to admit, was a surprise to us. This was not a design. This was sort of a fortunate accident, I like to say. If you look at a layered material, if you look at the anion environment, there are three metals. This is the metal layer and three lithiums. This is the lithium layer. And this anion site has too many metal neighbors to be substituted by fluorine. The energy is way too high, and that's why people have tried very hard, it's very hard to bring fluorine into a layered material. But if you make a DRX oxides, because you have, first of all, lithium excess, and then you have statistical disorders, you create all kinds of environments, there are certain environments where the anion is coordinated by a lot of lithium, maybe four, maybe five, maybe then six lithiums, and their fluorine can substitute. We now know from an MR that it is exactly these environments in which lithium substitutes. This is work done by Raphael Clémat at UC Santa Barbara, who's really been a tremendous asset in this research on DRX materials, because it's very hard to see fluorine in a material because you don't really see it in the fraction, so you really have to use an MR. And it shows very clearly that the fluorine is near paramagnetic metals and is therefore incorporated in the bulk. With simple salt-state synthesis using simply lithium fluoride, nothing sophisticated as a precursor, we can get between five and 20% of fluorine incorporated depending on the chemistry and the lithium excess. If we go to metastable synthesis, mechanochemical ball milling, we can get up to 33% and maybe even higher. Fluorine incorporation does remarkable things for you. I would say the first one is important. It reduces the average anion valence. And of course that means because of charge balance, it lowers the average cation valence in the discharge state. And so that's good because that means if you charge the same number of electrons, you end up in the charge state in a lower valence. And lower valence is better. It's more stable against oxygen release. So either you can go to a higher capacity or if you do the same capacity, you end up in lower valence and therefore you will have better safety. So we were even able to lower the valence in the synthesizing till so much that we could use manganese too. And this is important because in nature there are not many double electron redox couples, right? There is of course nickel, which is the most important ingredient of NMCs. And then there's vanadium, which we can't really use in layered materials, but others have shown you can use in polyanone materials. But very few people have used double redox on manganese because most of the time we synthesize materials as manganese three. So we were able to synthesize materials as manganese two by putting in a lot of fluorine. This is 33% fluorine, add a high valent ion to further lower the valence of the transition metal. And so these are manganese two plus compounds either with niobium or with titanium as a high valent charge compensator. And so these now start as manganese two. You can charge up to get very high capacity. So we're over 300 milliamp hours per gram here. Again, over 3000 watt hours per liter even approaching 4000 watt hours per liter. But what's really important here, the valence state of manganese is still only four plus in the charge state. And that's a quite stable valence state for manganese unlike say nickel four plus, which leads to a lot of oxygen release. So this is an important role of fluorine. The other thing it does is it reduces oxygen loss. As you know, almost all cathode materials at the top of charge in the first few cycles lose some oxygen. That's true for NMCs, that's true for some of the RX materials. And if I compare here, this is done with Dems with Brian McCloskey. This, the red curve is the oxygen loss. When you charge, this is a DRX oxide. This is a nickel compound. When you charge it to 4.3 volt, you start right away, start seeing oxygen loss come off, okay? If you fluorinate that material even lightly, in this case, seven and a half percent, what you see is that the oxygen content is significantly reduced already. And I'm not an expert in this, but according to Brian McCloskey, the oxygen content coming off here is already much lower than what comes off NMC materials. I think what's even more remarkable is that when you go to high fluorination content, so these are materials that are 33% fluorine substituted, we charge to five volts and essentially no oxygen comes off whatsoever. Basically there is almost undetectable oxygen loss. So clearly fluorine stabilizes these materials and we are not 100% sure exactly what causes this, but there is really nice modeling work going on. This is in Christine Persom's group and surface characterization by Wang Li-Yang at LDNL, which are showing that there's clearly a small amount of fluorine segregation to the surface. And again, fluorine as an anion is really, cannot really be oxidized unless you go to six-fold. So you may have some built-in surface protection here in these materials, because some of them still have bulk oxygen redox, but not surface oxygen redox. So these DRX materials have outstanding energy numbers. The in red here is the oxides on a specific capacity versus average voltage plot. The green are the oxyfluorides. And again, from the abundance perspective, right, you see a lot of materials here with manganese, which would be a much more earth-abundant material to make cathodes with than some of the things we use now. So in the last part of my talk, I wanna show you that you can extend the partial disorder concept to not just to rock salts. And this is really kind of one of the design experiences that I'm probably most proud of because this is one we actually did think through, particularly hard to see how you would make a good material. And it sort of worked out. So I wanna remind you, we like these zero TM channels. They are very good for mobility. So in a DRX material, we rely on them statistically appearing because we just randomize the cations. So we started thinking, what's actually, could you push the long range order or the short range order in a direction where you make these channels? So not just randomly. And actually, the answer stared us in the eyes, right? This is actually a spinel. If you look at a spinel, and I've here drawn an unusual setting of spinel, if you look at the tetrahedral environment in a spinel, they are perfectly segregated between pure lithium, so zero TM and four TM. And that's perfect. In a layered material, they are one TM and three TM segregated, which is also quite good. And so in a spinel, you already have these environments pre-made and they actually form these connecting pathways which are the tunnels in spinel. And so spinel indeed is a very good lithium diffuser, but there are problems with spinel, right? In the nineties, we all tried to use spinel for batteries and that field is sort of dead. And the reason is the following that one can easily cycle spinel between LIM and 204 and MN204. This is mostly a solid solution. There is some minor phase transitions there, but this only gives you 140 milliamp hours per gram. What you really wanna do is cycle it all the way from MN204 to Li2MN204, so you would have close to 300 milliamp hours per gram. But of course, as we all know, that doesn't work really well because this part is a first order phase transition. And you can see that from the flat voltage profile and a first order phase transition is bad because it occurs by definition inhomogeneously in a material and therefore you get cracking, you get all kinds of negative effects on cycling. The origin of this first order transition is the fact that in a spinel, you have the manganese octahedral and the lithium tetrahedral. And when you try to add an ion to that, like say by intercalation, you necessarily need to phase share with some other ion and cations don't like to phase share. And so what the system does instead, it pushes all the ions octahedral. And this is an important thing to know if you wanna make a structure in an FCC anion framework that shares both tetrahedral and octahedral ions, then spinel is the highest you can, the best you can do. It is the highest amount of cations you can put in a structure. As soon as you wanna put in more, you have to phase share ions. And so what is our strategy to get rid of the two phase region? Well, we decided we're gonna partially disorder the spinel. If you partially disorder the spinel, for example, you distribute some of the manganese between 16C and 16D sites. The 16D sites are normally vacant. It's well known from physics that if you create intrinsic disorder, you can kill first order phase transitions. Sometimes they become second order and sometimes they simply go away. And this collective effect of growing from LI-M and 204 to LI-2M and 204 will go away. So that's strategy one for the design. Strategy two was much more harder. And this comes to the issue that what I really would like to make is a compound in between these two end members. And again, the problem is in reality, they phase separate. So what does phase separation mean that you cannot make something there because it will phase separate into spinel and into the litigated spinel. Now, why do you wanna be, so you wanna be cation excess. You wanna have more than three cations per four anions. And the reason is that we now understand from work on anodes, cathodes, and conductors that if you have phase sharing cations, you get high mobility. This was shown in this paper that we did with Feng Wang that this is the reason that lithium titanate is a very high rate spinel because when you add a little bit of lithium, it creates these metastable phase sharing contacts where a tetrahedral and an octahedral ion phase share. And you could think of it as a high energy defect. And that high energy defect now moves very fast. So what we wanna make is a spinel with these high energy defects so that we have high rate capability. So we wanna make the material cation excess and we also wanna make it lithium excess because if we make it lithium excess then we will have a lot of lithium to cycle back and forth and we will have lower cation repulsion and we can fluorinate it if we have a lithium excess. Okay, so here's the strategy. We wanna disorder partially to kill the two phase reaction. We don't wanna fully disorder it because then we're back at rock salt. We wanna get cation excess so more than three cations per four anions to get high mobility. We wanna get lithium excess for good percolation and we wanna get fluorination for stability. So here's two compounds we made and that we published on recently. So as you can see, this is four anions. It's a 0.3 fluorine, but if you add up the lithium and the manganese, it is almost 3.3. So we have about 10% cation excess over the spinel and it is also heavily lithium excess. The material that I'll refer to as FO3 is the fluorine O3 and the material I'll refer to as FO6 is just with fluorine O6. You can make these, they do look like spinel with the neutron refinement on it. And as you can see, this is a partially disordered spinel. The tetrahedral site is partially occupied about half, but now you see that this, while there is full occupancy on the 16D site, the 16C site is also occupied somewhat. So you do have some cation disorder in this material. And they work really well. Here's the cycling curves of these materials. We get, again, over 300 milliamp hours per gram approaching 1,000 watt-hours per kilogram. Even when we do a cutoff at only at two volt, we still get over 300 milliamp hours per gram and 1,000 watt-hours per kilogram. So these are becoming very high energy density materials. Capacity retention is good, but degrades when you go to a higher voltage cutoff and all come to that issue. It turns out that this is mostly an electrolyte issue, not a materials issue. And capacity retention also improves with higher fluorination. But here was the sort of shocker. This is the rate capability. First, in a more normal electrode formulation, the first curve is 100 milliamp-hours per gram. But what you see, we can go all the way up to 2,000 milliamps per gram and still get over 200 milliamp-hours per gram capacity. So although this is a partial disorder material, again, very high rate capability, if we dilute the electrode to have less active mass and more carbon, we can even get to 20 amp-hours per gram. Clearly showing that these materials are not intrinsically rate limiting. And so while early DRX materials have low rate capability, some of these newer materials clearly can show very high rate. These materials do not undergo a lot of change during cycling. Here's the site refinement on the different positions. At the end of discharge, there's a little bit of change, but we're actually not convinced yet that that's real, but that that may be part of the refinement. So the materials seem to intrinsically, relatively stable against cation migrations. So I wanna end with some of the challenges for these materials, that because these materials have more of a voltage slope than layered materials, you'd like to go to high voltage to charge them. And that definitely degrades cyclability. So here's a manganese titanium system. And what you see, for example, if we charge to 4.4 volts, you know, capacity increase a bit in the beginning, and that is fairly flat. When you increase the charging voltage to 4.6 or even 4.8, you get higher capacity, but you also see that the capacity fade is a little higher. We have done work in a larger direct team with Gu Yongchen at LBNL and Robert Costecchi, and there is strong evidence that this is actually a breakdown of the LIPF6 salt on the surface of these materials. We use a classic electrolyte for this today, LIPF6 in carbonate solvents. And it is not clear how much of this is due to the material because there is significant electrolyte breakdown. Also in DEMS, you see a CO2 evolution that is clearly arising from solvent breakdown. We tried a little bit already with different electrolytes. We are by no means electrolyte experts, but you right away see some improvement if you take the FSI salt in DMC, which is the red curve. You already see better cycling behavior. So with that, let me wrap up. I think DRX materials, I think I'm really convinced are a novel path for cathodes with high energy content, with broad chemistry appeal, and therefore hopefully with earth-abundant materials. There's good and bad news. The optimization space of these materials is very large, which means we probably only have cracks to surface, no pun intended, but there's many cation chemistries you can use. You can do different levels of fluorination. You can change the short range order. As we've shown a high rate is possible. We have several more high rate materials on which we will publish soon. Another benefit is they have high stability against oxygen loss. So that allows you to go to high voltage, but at high voltage, you clearly have electrolyte issues. At this point, the evidence of how much materials break down, the risk is not clear. There does not seem to be much. If any at all. And another aspect that is probably more important for solid state batteries that DRX materials have, because they are cubic, they have by definition isotropic volume change, not like layered materials with our anisotropic, and they are often small. They are not always small, but some materials, we have one material that has zero volume expansion, for example. You can go to high capacity and you can extend the partial disorder concept. With that, I wanna thank my group, I really have a fabulous group of people. Some of the experimenters, Zhen Yang Leng and Hui Wenji, who did the spinels and my largely all women theory team, in addition to Bin Wang, who works on the theory of these materials. And I also wanna acknowledge all my collaborators in the DRX program of the US government and the participating national lab. So with that, I thank you and I'll be happy to take questions. Well, thank you, Gert. This is a very nice deep dive into this all the rock salt structure. Just report back to you and also Claire. In the last four times, multiple times of our symposium, our Nobel friends, Stan Wittingham has been always in the audience and watching all the talks and listening. The first question actually come from Stan. Gert, for you. So you actually touched upon this a little bit, is the wattage curve of this all the rock salt, spread very widely. A Stan's question is, in terms of commercialization, the practical ability of that, do you have comment on that because the spreading of the wattage is too wide? Yeah, well, thank you for that question and congratulations, Stan. I've never gotten to congratulate you in person and it's probably gonna take a while. You see it before we meet in person. So first of all, if you look carefully at some of the newer materials, it's not as sloped as some of the highly disordered materials. So it's definitely a little more flat, but having said that, it will always be more sloped than say an NMC, just because they're disordered and the NMCs are ordered. I'll give you my opinion. I'll give you the opinion of, well, I'll first start with the opinion of companies I've talked to. Their opinion is that if it has high enough energy density, we don't care. That's their opinion. But it would not be used, say, for electronics, right? In single cell devices, there's too much complication with dealing with highly variable voltage, but in multi-cell devices like EVs, the answer is that if it really is worth it, we don't care. And what I've done now is, so when we publish, we always cycle between high voltage regimes, but even when we clip for some of these, the discharge voltage for some of these materials to two volt or 2.5 volt, they still have energy density, specific energy density in the range of 800 to 1,000. So the partially disordered spinel, for example, if you clip at two volt, it's still over 1,000. If you clip at 2.5 volt, I think it's like 920, 930 or so. So the lower voltage you could just clip, charging to one, discharging to one volt just makes better papers. But the high voltage is, of course, important, right? And there, I think we really do worry about electrolyte issues. So the next question, how to quantify the extent of the cation disordering and the given materials? You know, that's a good question, right? Because, you know, every time the battery field moves into new directions and new materials, we need new techniques, and I think this may be one of these areas as well, right? So we're gonna have to learn new techniques. And obviously, so we do it with a variety of techniques. We use electron diffraction, which is not fully quantitative, which gives you the type of short range order. We use neutron PDF, so neutron pair distribution function analysis. And then, of course, there is an MR, right, which you've seen before. And what we have found is that if you put these three together, you can definitely get a quite good quantification of the short range order. You know, in the old days, people did this with X-ray diffraction on single crystals, but I don't think, you know, that's something that's gonna happen anytime soon, right? But, you know, I have faith in my characterization, friends, that there's some amazing characterization and development happening, so. Yeah, Claire is one of them, right? So right here. So next question, how to analyze the stability of a given material when you have increased or decreased of the cataract disordering? Yeah, one question from the audience. That's a good question. You know, I haven't really thought much about that. So I'm not sure we have seen significant changes of stability with varying the short range order. You know, there's always a chance that when you... So some of these materials are fairly thermodynamically stable. They're synthesized at high temperature and disordered. But for example, the high rate spinel today is made by mechanochemical ball milling. So there's always a risk that you drift from the metastable state into something more stable. We have not seen that. So in very high manganese content materials, we have seen bulk changes, quite reminiscent of the old work of when people made lithium manganese oxides and they changed all into spinel after a few cycles. But besides the very high manganese contents, we haven't seen a lot of bulk change. And again, surface seems to be quite stable as well. Though the low fluorination materials undergo some oxygen loss and densification just like NMCs. It's just hard to see because they're already rock salt. Yeah, yeah. So next question, good. Is there direct data to confirm lithium transfer and the layer structure is faster than in the rock salt phase and spinel structure? Oh, that's a, you know, I think that depends a bit. You know, if you paid attention in clear stock, right? The lithium diffusivity depends quite a bit on the state of charge. So it's not the same for any state of charge. You know, I would probably, if you compare them, DRX materials are all over the place. Some are definitely slower than layered materials. But some of the ones that I showed here to DR actually remarkably fast and probably just as fast or faster than layered materials. So I don't know if I can give a conclusive answer on that. I think as I tried to show today, tailoring the short range order is critical for getting high rate capability in DRX materials. Yeah. So next question is from a student, he or she said, this is probably broader, these questions about where it's possible to design a solid iron conductor with n-iron disordering, which may overall n-iron framework soft. So to facilitate the lithium ion transport. And the question is with n-iron disordering, is that the question? That's right, the n-iron disordering. Yeah, to make it overall, you know, the framework soft. Yeah. You know, people have tried, right? So unfortunately many n-iron, many anions are not very admissible with each other, right? So obviously you can make oxyfluorides, but with limited chemistries, oxysulfides can be made, but it's quite limited how well you can mix anions. So I think that can definitely be done. There are definitely crystal structures like that, but maybe not enough to play with. Yeah. So next question, how does the foreign ions change the lithium-hopping energy? Ah, good question. And the antimethyl-texture hydrocytes. Yes. We studied that extensively, and there's actually a paper that just came out from the group in Europe from Alex's growth group that showed consistently what we find that it does not change the local barrier much at all, which is a bit surprising to us, but it's interesting, our results are not published, but independently these two groups came to the same conclusion. What it does change though is the percolation path. So the effect of fluorine is much stronger on changing the statistical percolation, and the reason is that fluorine grabs the lithium. Fluorine really wants to grab the lithium, and so at small amounts of fluorination, the fluorine grabs the lithium and lowers the percolation ability because it's sort of like attracts the lithium. When you go to higher fluorination, that's not an issue anymore, but the barriers themselves are not dramatically effective. Yeah, very interesting. So next one, does the difference between zirconium and titanium and DRX also relate to the differences in electronic conductivity? Which one might be worse with zirconium? You know, it's possible that I don't think so because you don't expect the titanium four plus and the zirconium floor plus to contribute much to the electronic conductivity. The electronic conductivity is probably mostly from the manganese and the oxygen. It probably is more due to the percolation. Like with all materials, some of these materials are okay, electron conductors, some are not. And so with DRX materials, there is some care that has to be taken in electrode making just like with our old good friend, Luthymar and Phosphate. I think what the battlefield has learned that if you work with good electronic conductors, you can be really sloppy with your electrode making and it always works. If you work with bad electron conductors, you have to be very careful with your electrode making and how you bring carbon in because, so electron connectivity is rarely limiting for transport into the particle. Once the electron reaches the particle surface, you can easily do that back of the envelope calculation. But if you have a bad conductor, then you have very bad percolation of the electrons through the composite cathode. And again, that's what we all learned from Luthymar and Phosphate, right? But that problem can clearly be solved as that material has shown. Yeah, yeah, for sure. Next one is a little bit broader about this disorder, rock salt. So there's a person asking, is the cation disorder also promising for sodium ion batteries as sodium has larger radius compared to lithium which makes it harder to mix with transition metal iron? That's a question, Mark. Yeah, it's a great question. I used to think that it was impossible to disorder, sorting with metal and I've been proven wrong. There are actually a few papers out in the literature where sodium is disordered with transition metals and we actually repeated some of that work and it's indeed true. You can actually disorder sodium with some transition metals, which I think is remarkable, like you say, because sodium's so big. What we found is though that the concept is not nearly as useful for sodium because the migration mechanism of sodium in the cation disorder structure seems to be different and so it may not be as useful. But somebody can prove me wrong, maybe somebody will make it useful but you can't disorder them, which is remarkable. That's cool, yeah. So next one, this sounds like from a student, Graywork, professor Cedar. The disorder rocks are showed very impressive energy density. So what about cycle life and what kind of failure would happen in the deletion state? And the suspended phase transition, this oxygen release or other factors? Well, you know, so it's early for these materials, right? It's not like we have super intensely tested them in full cells and in large cells. So we only know so much about failure mode. So maybe the easy one to answer is oxygen release does not seem to be a significant failure mode. I think that we've demonstrated multiple times. Clearly electrolyte breakdown, as I showed, is a failure mode on these materials. We know that at least 50% of the fade we see is due to electrolyte decomposition products buildup. We've worked on some tests. We don't see much materials degradation unless we go to very high manganese content. And then I wouldn't call it degradation, it's more change. So they seem to be fairly stable, these materials. But again, there's a lot still to investigate, right? We haven't done full cells with graphite annals and so where manganese could be an issue. But this is the pathway of all new materials, right? I like to remind people to read the first NMC papers from 1999 and 2000 and have a look at what they see or the first lithium-cold oxide paper. Or the first lithium-ion phosphate paper from John Gurunov. Yeah, we need a long pass to keep improving for sure, yeah. NMCs are 20 years old, I like to remind people so. Yeah, yeah. So I know we have a lot more questions right there. Maybe I'll ask the last one then I'll bring Claire also onto the stage so we can have more discussion. But this is for about, well, I think this is a really good one. The disorder rocks are compared to the layer. The electronic conductivity at the end rather you are going to be limited by something like Anderson localization induced by disorder that it ultimately affect the way performance of the batteries, yeah. I would say yes and no, right? So from what we've seen, the electron conductivity is definitely worse than many of the NMCs. But as I showed you, we've made, I've shown you two high rate materials, right? I've showed you one with the modified short range order and I've showed you the one with the partial disorder spinel and we have a third one on the way. So the electron conductivity does not seem to be a limitation for the rate capability if you make your electrodes well. Again, and this is the same as what I said before, right? You have to be cautious about how you engineered the transport paths. Yeah, I think there's always a way to solve this problem if it's a little bit poor and then you can do the coating and you have other methods to solve that. I mean, lithium iron phosphate was a wonderful teaching experience to all of us, right? Yeah, okay. Okay, good. Claire, let me also bring you up to stage again. We have about 10 minutes, right? Now I can ask some of the question I really like to ask. And first question from also related to one of the audience question actually for good, but I also thought this also could be for Claire as well. So looking at this, this person asked, I also asked, I actually kind of know the answer already. My own version of answer, but I want you to give your thoughts. This person asked, the top research group around the world, looks like you guys are all doing, avoiding polyane iron. I'm not sure that's completely true, first of all. What's the reason behind that? I think in good in your presentation, you can't have that a little bit, but I want both of you to express your thoughts. Polyane iron means there's a much big anti-multi-adult, right? So what's your thought? Looks like you didn't do much on the polyane iron. I think also Claire, you also didn't do much, right? You know, I think, I mean, I can answer first and Claire can answer if that's okay. So I know I've done almost 20 years of cathode research and design and most of it has been sponsored by companies and in the late 90s, early 2000, they came to us and they said, we want ultra-safe materials and they should be made out of polyane irons. And then we came up with polyane iron materials and they say, well, but they don't have high enough energy density. And I would say, well, what do you really care about, right? Oh yeah, it has to be safe and have high energy density, okay? And so I think we've gone around, right? We've gone full circle that we are, I think that the truth is that companies for most applications will not sacrifice energy density, period. And maybe for a grid, but even there, it's not clear that they actually will because energy density also relates to cost. That's just really hard to do with polyane irons, okay? You know, unless you do some of the work with multi-redox couples, like the work that Stan's been doing on Vanadium, right? If you can make that work, then maybe it's competitive, but most polyane iron systems are not competitive in energy density. And I have a graph, right? And I could put it on the web for people I can add to the presentation where you can show that the only anions that will give you a useful energy density are oxygen and fluorine. That's it. So otherwise there's no competition. So this is why we came back to the oxides and in our case, the disordered oxides. But you know, unless you're going into more esoteric reactions, like conversion reactions and so, if you need high density, you need an oxide. Yeah, Claire. I mean, I think Gert said most of it other than the other issue is because the conductivity is often lower, there's more work in terms of then processing the electrodes, which introduces a cost aspect. I mean, we have done work on phosphates and lithium ion phosphates and pyrophosphates. And of course, the borates give you a possibility of increasing the energy density, but so far they just haven't been competitive with oxide systems. So, and again, this is always the tension as an academic is that we are beholden to our funding agents and perhaps we should just have the confidence to do what we want to do, but then we may not get funded. So I've worked on lithium ion phosphate for a long time. It's got a nice and a more nucleus as well. Not the ball will land or but, you know, it helps. All right, so next one, you know, at the beginning of my introduction, I introduced both of you, you know, your staff from where your career is, you know, the skill set and then how you go, come in and tackle the energy storage problem. And I think this is mostly for students, young students. We have a lot of young students now online. Would you share your perspective? How you, you know, maybe that was a two decades ago, you know, roughly, right, about the time. It's a bit of a mandate. Yeah. Only two decades. I mentioned that, only two decades. How you, you know, really, how do you really get into the field? Why? And there's got to be a reason right there. Maybe Claire, you want to take this one first and we can give a good, a little bit of break. Well, I mean, just to, I mean, okay, I started off as an NMR spectroscopist, but I started off an NMR spectroscopist in a solid-state chemistry group. And I was at the University of Oxford, Tony Cheetham, and I was the only person doing NMR in, so, and I was in the, you know, I started off in the department that John Goodenough headed. So it wasn't as if it was a completely non-solid-state environment to grow up in. But then I started my career and I was looking at ionic transport. And we did some of the very earliest high-resolution measurements of ionic conduction, particularly fluorides. And so I was looking at these types of issues, but then I have to admit that I got into the battery field, partly from a conversation with Bill Bowden, the former jurisal chemist who sadly died, who was interested in lithium manganese oxides and he was interested in characterizing with spinels. And he sent me some samples and I said, well, I can do NMR on them. And it took me two years to figure out how to interpret the data. I know people have heard this story before, but I got back to him and he said, tough, Claire, we've canned the spinel problem. So that for me was the beginning. And I was like, oh, well, you've canned it, but I've got two PhDs on the project. So we're going to keep on going. And that was then how we started. Gerd and I, years ago, collaborating on the NMCs. And so that was mine, that was mine, but I was already interested in the sum of the questions and the structural origins behind this. And I've also been interested in just the whole energy agenda and the current zero carbon issue at the moment. So there was that trade-off. I was always working, I'd been a postdoc at DuPont looking at chlorofluorocarbons and looking at catalysis to make hydrofluorocarbons as part of the Montreal Protocol. So I'd always been in that environmental space. That's good. And Gerd. Oh, how did I get into this? Well, I had an experimental engineering degree and then a PhD to learn theory. And as a young faculty member, I tried to figure out something that wasn't being done by other people. So in those days, in the early 90s, everybody was doing theory on metals and semiconductors. So I go like, hey, nobody's doing oxides. So I developed theory for oxides. And then we sort of accidentally became aware of the lithium battery problem. This was the mid 90s, so it was still quite early. And I think that coming in with, like Claire, right? Claire came in with NMR skills to the field. And when there was very little of that, I came in with the sort of abinitial theory skills when there was very little of that. And I think to young people, I would say, it's not just about batteries, right? Try to learn some fundamental skills really well and bring them to the battery field, right? Because the battery field needs an enormous variety of diverse skills, right? Enormous characterization, theory, clever experiments, right? It's, I think the way for young people to get into it is sort of to develop an angle, right? Of something that they're good at and bring that to the field, sort of the way Claire and I brought either NMR or abinitial modeling to the field, so. Very good. I don't know whether good and Claire, I would tell you how I get into the batteries. Maybe I can share just one minute since I'm the moderator right here. So back quite a number of years ago, so I was certainly already started to get interested in energy. But very important one is when I visited MIT to give seminar, good, you still, I don't know whether you still remember that. I do. Talking to you and a few of your colleagues, right? They're learning about the solid-state chemistry, solid-state electrochemistry, the ion transfer, right there. That really gave me into the batteries. So that was my story. So I think with that, I think we are now really close to the end. I would like to thank both of you, good and Claire for the wonderful talks and also showing your insights by answering the questions. We will continue, of course, this exciting series. This is the fourth one. The fifth one will take place in two weeks. So next week, we will not have the symposium. We'll take a break for one week. But two weeks from now, we'll have Doron Orba and also Kang Shi, also another two experts to come back to give us the seminars. Thank you very much, good and Claire for joining us. And I would also like to thank all the audience. Some of you stick with us so late. I look forward to seeing you in two weeks. Bye now. Thank you. Bye. Thank you.