 Alright, good morning from Stanford University. My name is Will True. I'm the co-director of StorageX Initiative. It's a great pleasure to welcome everyone back after a bit of a break from our seminar series. And today I'm delighted to be hosting two of my academic colleagues to talk about the fundamentals of ionic transport. So let me just give a brief introduction to the topic and then I will introduce our speakers. So ionic transport really underpins almost all of electrochemical technologies. And over the past 30 years we have really seen a huge avan in the understanding in materials for lithium-ion batteries. However, as we think about the post lithium-ion battery world, many new materials need to be discovered and new mechanisms need to be understood. For example, many of you have heard talks from our industry partners on solid state batteries. And you've also heard talk about transport of ions beyond lithium from other academic colleagues. So that is the theme of our seminar today is to really explore how we can better understand and therefore control ionic transport in these materials. So we have two speakers. We have Professor Wolfgang Zeyer from the University of Minster and also Professor Kimberly C. from Caltech. So Wolfgang is someone who has performed one of those magic tricks of transforming himself from one field to another and to another, which is really quite external at the pleasure of knowing him. For about 20 years actually, where he was initially studying charge transport in thermoelectric materials. And I think as history would have it, he got excited about ionic transport and mixed conduction in thermoelectrics while he was a postdoc at Northwestern University. And then he made this really big leap of faith to take some of the same concepts for understanding heat transport and charge transport in thermoelectrics and transformed it to understand solid electrolyte as their use for solid state batteries is it's quite rare to see people reinvent themselves so rigorously from really really disconnected fields. So that's why it is being so exciting to see the latest progress from Wolfgang's group and Minster where is now a chair professor. So Wolfgang we're really excited to hear from you today and the latest with phonons and ions and other things in solids. All yours. Thank you for the kind introduction and the first invitation so we've known each other for 12 years, not 20. I don't think we're that old. But I think. Yeah, so I was, I was really happy about the invitation because usually I get asked to talk about solid state batteries and not too much about ionics and I really wanted to put some of the thoughts that we've had over the past years and recently on ionic transport with the solid state chemical concepts there. But at the end of the side I'll show a little bit of like transport limitations in solid state batteries as well because I think we do need to make that jump from a local jumping ion to how it is moving in composites. Now why do we do all this. Well, the, I think the motivation really is if we if we go away from a liquid electrolyte to a device that has a solid electrolyte. Ideally we can use a little metal anode. It may make them smaller, we won't have polarization issues, and they might be safer and so currently this is pushed towards solidification. And what we really need that is, when you're the liquid cell you have a liquid nicely percolating all active particles and wetting them nice. So ion transport is quite good. And once we're in solids we have solid solid contacts also interfaces and things become a little bit more difficult we need a good percolation of the ions through an electrode materials that are processable. We have decent electrolyte stability and we need a higher on the conductivity and for the longest time I thought well, higher on the conductivity just means that we don't want a large IR drop between our electrodes. It's still the case but we may need to think about which ionic conductivity are we talking about that needs to be high. And so what my group mostly does I would say is trying to understand ion transport and materials and trying to push them and here's a here's an overview of different materials classes on the left conductivities on the right activation and you can see in, for instance, the so called a girl of class and I'll get to those in a minute. You have a large spread of ionic conductivities of orders of magnitude within the same structural time. And this tells us that there is a lot that we need to understand of how compositions and structure effect ion transport and how we may be able to push those. Now, what is ion transport I think ion transport is a very simple conceptual process where an iron sits on a ladder side jumps to an empty adjacent ladder side by that and it's displayed flat is a little bit all this costs energy and this gives you a delta G. And this delta G is we sort of measure this as an enthalpy and activation there and controls the overall ionic conductivity. The structure itself determines the energy landscape. So let's say our list here is happy and a treat a trifle coronation but it's unhappy and a full full coronation. The final state will be higher in energy, which means, even if we overcome the activation barrier for the lithium and sits on the saddle points, it's most likely to jump backwards then forwards. And this combination really matters. And if you go to the textbooks you see a list of things that seem to matter that just number of available sites or defects if there's no, if there's no space to jump in here then the IMA is most likely not going to jump in some sort of dimensionality three dimensional conductors are better than one dimensional conductors. We think of that polyhedral of our mobile iron are well connected and we can see that here schematically let's say in HCP ladders and irons it's on a tetrahedral side and wants to jump to its adjacent tetrahedral sites the space sharing this doesn't cost a lot of energy point to EV. Why, well if you think of this as a chemical reaction, we don't have a change of reaction coordinate but we can think of it like this, we change our coordination from four to three to four this doesn't cost a lot of energy which is leaving one bond and forming a new bond. This is a good jump path. In transport we would have to go by an octahedron to the next tetrahedron would change in coordination from four to three to six. Let's say might be unhappy. So in a higher intermediate state activation barriers higher and this is one of my favorite examples to show how the structural polyhedral arrangement alone of the mobile iron affects your energy landscape. Now we do have a lot of influences or ideas how we can influence iron transport. One of them and I think this is a typical one is that we just try and synthetically blow up the unit cell by substituting the material open up the fusion pathways. This displacement is not needed as much for the iron to jump and we always believe that then the activation barrier decreases in the conductivity. And today I'm going to show something that challenges that this believe maybe a bit too simplistic let's put it this way. And there's also concepts of softer more polarizable lattice I quickly touched is I'm not going to talk about phonons too much today. And in general, the strength of bonding interaction matters and these are things I'll slightly touch on. I split the talk in three parts. One is some of the work that we've been doing on understanding the grid lights and this is, I apologize. Some people have seen parts of these slides before but this is our workhorse were over the last I would say seven years we've been trying to refine our understanding of how transport occurs. Then I'll talk about the general usefulness of structural descriptors. And I'll show a bit on transport limitation in solstice. Now, why, why a gear that's well the gear lights are a fun materials class to work with and lucky for us, I would say they're also mostly used in solstice batteries these days because they're just really good ionic conductors. And they're also very easy to make enough in principle simple crystals. So we have a, these are guides are biofossilates appears for base materials and the general units, the formula is li six ps five x and x can be a highlight is a highlight that sits on forms in FCC center FCC lattice of the halide ions, and you can have these orthothiophosphate tetrahedral on the octahedral sites, and you can have half of the tetrahedral sites occupied by free sulfur s to minus and this gives us then our formula and the lithium somewhere in that unit. The somewhere in the unit cell was for a long time believe to just be clusters of lithium that sort of form around the free sulfur and iron, and you have fast rotation between these ions between the sites to listen sites, and that the long range jump needs to sort of go from one cage to another. For a long time we've wondered why this is the case if you do some sort of topological analysis of cages, they're very isolated the distance between the cages is quite high. And you can, you can imagine that the longer distance is the more difficult it is for a jump to occur. And so, why are these so good ionic conductors, what we actually found is that they're, they're actually more sites in there, for instance there's one side that really connects these cages. So this is the, for one we need to understand the substructure to the substructure more, and we need to use this as, as, as, as tuning knobs for transport. Something that we found very early on is that there is a site disorder in these materials so I said the sulfur sits here and listen for from these cages around it and here's a highlight position. And in reality when you make these materials that is order a chloride anion has the same anionic radius in the sulfur anion. So just synthetically you have some self as I'm pouring sitting here some sulfur sitting here. And when you change the composition from chlorine to bromine to iodine your halide becomes larger this this order decreases. And it seems to see back in 2017 that there's a strong dependency of the conductivity on the site disorder with arbitrarily fit optimum in here. But clearly this disorder seems to have an influence on the conductivity and we can calculate the so called bond valence some maps fancy word for this thing how much space is for the lifting to diffuse and if there's no disorder you see that they pretty much just rotate around the cages as soon as you turn on this order, you connect these cages. And we, we just wanted to understand why this is the case. So, a little bit more structural work. We set down and decide okay let's really look at the list of structure of our of materials that we can perfectly control in terms of synthesis and storage dimension. So let's look at the iodide based agiridide, and take temperature painted neutral diffraction and just sort of show a radial histogram of the lesson distance never close together and we only see two positions. As soon as we go to the bromide, there's more disorder, the cages expand. So that this order gives us a third position. And once we're in the chloride where we have roughly 50 50% disorder between the sulfur and the chlorine, the cage has expanded much more. This is remarkable because you can imagine your ID is larger which means units are larger and then you go to bromine the unit cell strings for the chlorine the unit cell shrinks further, but the cages the lithium ions expand. And that is to say well let's let's do some sort of metric that describes the spreading out of this cage. And this isn't an average distance and average radio distance away from the center and I am, which we call our mean to describe this, this lifting structure spreading out. And if you plot this are mean versus the disorder, it increased, you can easily envision this in terms of electrostatic interactions, if we only have sulfur sitting here itself or two minus more electrostatic interactions over cages smaller, you start putting a one minus an iron on it and the cages start to expand. And this is now when we actually try to use to tailor the listen substructure but before I get into that let's talk about what this disorder actually is. Is it thermodynamics is it kinetics. So, you can make these materials this is a bromide agiridide easier to distinguish bromide with extra diffraction. And we can make the material by a standard route of two weeks just annealing shutting off the furnace and you get roughly 20 22% site. You can also take the material from 550 degrees Celsius and just quickly quench it. You get a much higher this war. So it seems that at higher temperatures we have hired his order and by quenching we can medically sort of trap the site to sort of thermodynamically they want to be disordered higher. If we are correct then we should be able to do a slow cooling approach to lower that is order and if we slow cool over five days and more my student didn't want to slowly adjust the furnace in the lab. We get to 13% this order and the ionic conductivity scales with this disorder. I plucked this back into this initial graph really see well if we play with this order. We can tailor the conductivity. Let's push this understanding further, instead of just quenching from 550 we now decide okay let's just quench it from different temperatures. Because if it's really thermodynamically and tropical driven that it wants to disordered higher temperatures, the temperature should matter. And what we do see is that this order increases with increasing annealing temperature. So if we look at nutrient refraction data, yes for mental data here we see that this radio distribution of the lithium really expands with increasing this order and here down here plotted as an average charge on that structural site and have an issue molecular dynamic simulation sort of corroborate that. So we can pretty much see that in so called maximum entropy maps from a nutrient refraction that the more disorder you have the more these your core density of lithium is really spread out. AMD shows this I'll quickly show this because there's a little bit more visual, where if you have 0% this order to listen really just rotating these cages and as soon as you turn this order to fly half through the unit so you get a very high ionic sensitivity. And so we can either plot the ionic conductivity versus the site this order we can plot the ionic conductivity versus this this radial distribution this this average distance of lifting away from the center. And so, clearly disorder and the lifting distributions are affected by affecting each other, maybe they're affecting each other but at this point I would say that this order affects looking distribution. And this is something that we can tailor synthetic. And we can use this descriptor to further by compositional changes, increase the ionic conductivity because in the end all you need to do is take out that additional sulfur and plug an anion in there that is more that is less charged. So, a highlight in the end. So, let us back to something that we found 2018 knowing now that this order really affects illicit distribution. But in 2018 we we start substituting the, the, the iodide agiridide with Germania. And what we saw is that there's the activation barrier at some point really drops the more Germany put in the more lithium you put in. It goes by on some sort of inflection point we hadn't seen that before and this seemed to correlate the sort of onset of this ordering. And back then this was, I think, the fastest, fastest lifting conductor, and luckily, there's there's more now out there. And it was such a good semiconductor we could actually make well loaded solid state batteries with it, but we've never really understood why this actually became such a good ionic conductor. And so, this is something that we looked at in the collaboration with Martin Buchan in a teograd via NMR. So what we're able to see is that when we increase the germane content we start to turn on sulfur iodine site disorder. If this comes from the additional listening is something we can we can argue, but we do see that at some point that this order increases. We see that suddenly we start to populate more crystallographic lifting sites. So there's one more than there's another one coming in and so you really see that from this topologically isolated clusters we start connecting them we start connecting them for the substitution turns on some sort of an eye in this order, and by that, our listening spreads out, and we can map this out again as this average Radial distance, and this increases quite a lot, even more so than just the halides and the last six PS five series. And I doing that, we pretty much move from these isolated clusters to very well connected listening clusters lowering the sort of flattening the lens. And I think now we're at a point in the agiridides where I believe that we understand that we need an eye disorder that creates charge immunogeneity in the system that leads to a very strong listening distribution and the better the listening is distributed. The faster the ionic transport, or the other way around I guess the faster I don't transport the more distributed is when we do instructions. So I think at this point, I hope that we understand agiridides well. That being said, I think two years ago I would have told you that we know everything about this class of materials now I'm telling you that I think we know everything about it, and I hope in two years. We can make amends to that. Okay, and this was some of the overview of the agiridides that we've been playing. I want to come back to my introduction to structural static structural destruction, so to say, so I mentioned that an ion that wants to jump from one side to another needs to overcome an activation. And you can really think of this as well the I needs to display flat is a little bit and there's evidence of showing that. The typical approach is to increase the so called structural bottle widening the units are lowers the activation barrier and gives you a higher higher ionic transport. And I wonder if this is true. Well, let's see. We decided to use a model system to study this sodium ion conductor. It's in principle in a three PS four so a sodium phyro phosphate for phosphorus we can substitute antimony and for sulfur we can substitute selenium. We can make instead of a single series of substitutions, we can make a full, why call it the box or box approach, where we go from the sulfide to the selenide and certain steps we go from the selenium phosphate to the selenium and timonade. We can go from the fire phosphate to the fire and timonade and so on. What we do know is that the two structural polymorph that are so similar that for now let's assume that they're the same structure, but we do see a slight change of structure along a series, but the jump paths are quite similar so I think I'm the analysis that we're going to do is quite here. So what do we see. And this is a lot of work of my student of a lot of years and frustration I guess for him to figure out what are the structural descriptors and now I'm telling you. So what do we see, we see a typical behavior that if you substitute something the unit cell changes. So, on the left cationic substitution. We're going from phosphorus to antimony in the sulfide in the selenides the selenides have a larger unit cell and going from phosphorus to antimony expands the unit cell, 89%. And the anionic substitution going from sulfur to selenium your unit cell expands more than in a cationic substitution. This is already cool to see what we do see is that the unit cell volume correlates with an average ionic radius. This is actually just an arithmetic ionic, ionic radius of all of the components that we have in the system, remarkably well correlating something that I wouldn't have expected but we're happy with that. What else do we see. Well usually we try to look at polyhedral volumes harder changing because that's our structure battling right. So we can look at these bio phosphates, or these these polyhedral poly anionic groups, and we see that in the cationic substitution sulfide selenides by putting antimony on the phosphorus side, these for leader expand quite a lot 40 to 50% of their volume. This is an increase. If we do an anionic substitution so replacing sulfur with selenium it's like half of that. So the anion doesn't expand as much as a cationic substitution does. So, locally, the type of substitution glaze seems to matter. Interestingly enough, when we do a cationic substitution, the sodium. When we do a cationic substitution, polyhedron, it's the eight full quarry coordination, barely increases if we can't I only substitute. If we can I only substitute increases by 10%. My gut feeling would tell me, hey, then the anionic substitution is the one that we need to use to increase the ionic transport. The unit cell, the MS for polyhedron changed a lot more here, even though this this pool here doesn't change why we think that the reason is that it we're just distorting more. So we know that if we go from phosphorus to antimony this tetrahedron is expanding a lot. We know that the this polyhedron does barely expand, but it seems to distort there's an angle that we can describe. And despite expansion it just extorts more in the angle changes a lot but not the volume so from the PS4 to the SPS for you see a large change in the angle and barely change the ball. So what we see here is that our typical belief of what's just thrown element in there and which is expanding the unit cell we're expanding all polyhedron is not necessarily correct. The type of substitution clearly matters locally and this is something we should look at. We would have never caught that had we not done this box approach, because what we would have seen is that irrespective of the substitution a bottleneck changes, our political volume increased, but not as much as really depending on the type. Let's find a descriptor for this bottleneck. There's two that we can use and I'll show to one is this sodium polyhedron and one is a distance and I and I and distance that the sodium is to jump through. Let's use this here for now. There's also barely changes in the cationic substitution it changes more the anionic substitution. Let's use this as a structural bottleneck for the iron jump from one point even to the next one. Okay, so we can measure ionic transport of all these materials and the error bars to some extent our trip tickets, not in the full series. Some would have my head off but to make sure that these trends are correct. We did that we can unfold this box and you see a really cool trend of decreasing activation barriers depending on where you start. But this only looks this nice because we're plotting it this way. Let's plot the activation barrier versus the anion and I and distance that we have. And this is a beautiful linear trend, and this was a paper that we published in 2018 saying, oh, if you blow up the unit cell, you get better transport looks really great. So we're looking together, not so much anymore. What do we see here, we see that for a given bottleneck distance, we get to activation barriers, so the structural bottleneck cannot be a good descriptor for an activation barrier. We can also plot this for the sodium, political volume and we get a similar. And I think this shows us that the bottleneck itself. So the main contributor to transport may not hold up in all materials. Had we only done a single substitution like we once did. We would have said that this is a dominant factor but doing this two dimensional substitution shows us that it's not. And something I will mention that we're trying to work on is figure out that dynamic influences on iron transport. So if you think of the iron sitting in a in a in a mnemonic oscillator, it needs to jump over an activation barrier and if we soften the lattice for increase this this this in a mnemonic oscillator, we usually get higher on a transport lower activation barriers and this is something you can find in literature, and pretty much just going from oxide to sulfides going from chloride to iodide going from sulfides to the entire anionic polarizability soft the lattice as we increase the conductive. But it's becoming challenging to really measure this and in the in the past we measure this wire. Speed of sound measurements and measuring the bi frequencies or in elastic neutral scouring to get a real look at how the phones are moving. And here I'm proposing two different descriptors that are on the lab scale easier to access. If we talk in these parabola, we can think of thermal expansion. And what we can do is we can use these materials, measure thermal expansion, and from the thermal expansion coefficient and their heat capacity we can define an anionic bulk modulus. So in the same approach, we can just look at the melting point. I said that the bonding situation the overall bonding influence may matter and so if we have a stiff rigid lattice, we should have small displacement of the irons and high melting temperatures, and the soft lattice should have lower melting temperatures of course we can measure the melting temperatures of the materials. So we can then look at how melting temperature and anionic bulk module I actually correlate and they correlate quite well so to lower your bulk modulus to lower your melting temperature, but this sort of chemically make sense. Let's see if they can be used as descriptors, we can plot the activation barrier via versus the melting temperature and we can plot the activation barrier by versus the bulk modulus. So this is a much better average descriptor in this class of materials to for the ionic transport. And so I think I'm, what I want to say is that clearly ion transport is more complicated and we currently think it's not just structural there's just, there can be a lot more factors at play, and we would have only found that out. Had we not done a two dimensional substitution but I can also tell you that we've done this on other compounds where you see a beautiful trend of an activation barrier with a bottleneck. And so I think this is something that we need to do is as annoying as it may be from doctoral student to map all of these out but this is something we need to do to really understand transport in our materials. All right, in my, my final few minutes that I have will and I'll try to hurry up. I want to talk about transport limitations and saws that bad. And in this case I'll talk about lifting sulfur solid batteries. And what we always see is that these lifting sulfur the source that batteries give us similar capacities as conventional themselves so with a liquid electrolyte, but usually a higher over potential. And so this just made us think about how transport actually happens in this composite. How does this work. Well, the transfer that I've shown you so far is consolidated electrolytes and let's just move from one side to another and these are easy transport. In a solid said battery composite. Things become tricky, they become more tortures. So let's say we put sulfur in here and suddenly the transfer path is no longer direct torture. So we can then ascribe an effective iron transport that is a function of the volume fraction of our solar electrolyte how torturous the pathway is, and the only conductivity that we went in with with the electrolyte. And in truth in these lifts himself ourselves with carbon in there. So we have three phases. And if we're honest and solid said batteries we have a fourth phase that is porosity. So the effective ionic conductivity will matter. And what we see is when we measure these is that your ionic conductivity of your pure solar electrolyte. Once you decrease the volume fraction of the solar electrolyte so you put more active material in there, in this case carbon and sulfur. The only conductivity decreases in the effective ionic conductivity crease over orders of magnitude. We're familiar with transport in liquid cells. This would be the program model, and we're dropping a lot more. So we're losing a lot of conductivity just by making our composites, and I believe that this drop in resistance, this increase in resistance gives us this large polarization in our cell large potential. We're schematically saying that the more torture it is to slower your iron transport. And I think there's two ways that we can optimize iron transport, we need to think no longer just about ionic conductivity. Yes, one approach is to say, okay, we're currently at five to 10 milli Siemens per centimeter of my on a conductor. We dropped by two orders of magnitude you probably need 100 milli Siemens per centimeter for my on a conductor to be good enough in thick or high loading solid batteries. Challenging work for us to do that that's that work on these materials, but maybe we need to think about structuring thinking about the composite microstructure and optimizing the microstructure to really get fast. So final comment. Something that I want to show another issue that we have with low effective ionic conductivity. So we can look at these batteries by a mutual tomography and mutual radiography to really look at reaction fronts. And what we see is the following so we have a cell in a beam that we're discharging so we're converting sulfur to lithium sulfide and whatever you see now here in bright white is sort of the lithium sulfur forming so that's the lithium that you see moving first off. And we do that during the first discharge as a function of depth of discharge. And what we see is here down the separator appears to current collector that look at this video, but there's a reaction from moving from a separator to current collector. You can see that here visually as as a generation that this is really growing from the separator up to the current collector. So again in the video. And plot this as a median rate off of change of the attenuation you really see that at 0% of discharge is moving towards the current. This is a problem. This is a problem because that means that we have in homogeneous reaction fronts in the solid state battery it also means that when you run your cell and you think that all the charges that you're the electrons that you're getting out is coming from the composite it may just be at the separator so you may have local overcharging effects that kill your electrolyte for instance at the separator. But why is that the case. Well, it's the effect of transport that is limiting. So in this case, or in general, if we had a composite were electronic and ionic transport was infinitely fast. At every reaction point we would have plenty of lithium and plenty of electrons, we would have a beautiful uniform current distribution reaction. Now in this case we have slow effective transport so the electronic transfer is fast we have a lot of carbon in there. So the ions are supplied by the separator. There's enough ions close to the separator so we have more of a reaction from here and slowly percolating upwards. If it was flipped and electronic transport was limiting, then your reaction front will start at the current collector but you will still have that. So it's not just that we need faster ionic conductors to get rid of over potentials and and have thicker electrodes, but we also need fast ionic conductors to make sure that we don't see reaction forms and reactions in all solid batteries, and we need to balance our ionic and electronic transport because it's important that electrons and ions come to the reaction site at the same time. Otherwise, you'll always have these in in homogeneity. And yeah, with that I thank you for your attention I hope I was able to convince you that well for one listen here that's a fun to work with and that there's a lot of underlying structural questions still. I hope I was able to show that structural descriptors may or may not be ideal and I think that really depends on the system that that one is looking at and there's no one. One idea fits it all when we love fits it all to ionic really to look at that's independent materials. And I hope I was able to convince you that solid batteries might be more tricky than we think in terms of transport in a composite and that's not just high on the conductivities that we need but we need them to be fast in a solid battery composite. And with that, of course, I need to thank funding the collaborators in my group, and I'm more than happy to answer the questions may have. Thank you. Wolfgang thank you for that very exciting talk connecting so many different aspects together. So maybe I thought we can talk a little bit about. Oh, let's see about the solid state chemistry of transport so you're talking about this bottleneck right in terms of a bond length and at least sort of in freshman chemistry, we always think about the bond length also to be quite related to electronegativity in an ionic system. So, are you also saying that maybe there's also intrinsic dig coupling also with electronegativity or think that's another degree of freedom that is separate from the bond length and the softness of the last. I don't know if it's an independent degree of freedom, I mean they're they're intrinsically related right if you have a lower charge or a lower electronegativity things are not as strongly bound so we know the reason why sodium ion conduct sodium and the solid state are faster than lithium is because lithium is just way too sticky electrostatically and we can then talk about electronegativity. There's just moving away from the sulfides to the highlights, because lithium is less stuck to a chloride and I and into a highlight that to a sulfide. They're all interrelated, I would say, and it's, in my opinion, something that we need to look at in each individual system, what is the main contributor. Right, and there was also a related question from the audience on bond valence model so have you guys look at that as a potential descriptor or it's using very different contexts elsewhere. So bond valence is in different contexts probably mean like that you can sort of figure out oxidation states of ions and solids. Well, you can map out these geometric pathways, calculating this via bond valence sums. It's a good, I would say for us this is always a good start, it gives nice pictures of okay what could be potential pathways, but it's all static. Right, so in MD at least your ions moves there's, there's more and more evidence that in the materials when the ions are moving that the anions are truly moving out of the way this is something that you're not capturing in BBS so I think MD is much stronger there. But whenever you look at look the structure itself and we're listening maybe sitting a calculated landscape by BBS may give an indication if well if that is a high energy site. It may not be populated and this may be wrong so I think checking BBS against structural data is always a good start if your structural data makes sense. Everything in terms of structure should be taken with a grain of solid as well. Because in the end in these files and kind of we arbitrarily chop up our unit cell and say oh this could be a position right by these fast conductors if the density is in many ways to spread out. Well, the next question is for me, this is kind of a high level question. You know, in the catalysis field, you know, we have seen the success of the descriptors work really well, especially when consistent computations are done to obtain those descriptors. But of course in many systems, the descriptor is not single dimensional it's sometimes multi dimensional, which it makes it quite hard to understand. So what can I'm curious of your philosophy here is do you think there is a sort of a single dimensional descriptor that could explain everything, or do you think the reality is more complex and sort of as you have shown there. You know something, you know more than one descriptors needed to represent the system. I think more than one descriptor is needed, but I think all of these descriptors are interleaded some extent, right. Like we said bond length, polarizability, electronegativity, the structure clearly plays a role. Then the disorder, right and this is, if we have. If you just use a bond length descriptor, then you start putting a disorder in and you have charges all over the place, the way the defects, this ordering of materials. There's a reason why if you ball mill things, they become better, because you just introduce a lot of ball mills and in these harsh sentences. So, a lot of things play a role, I would say. It would be nice. And it would be nice to do it like the catalysis field but you need ideally high throughput studies right where you have a lot of data, but making these materials consolidating them sponsoring measuring and being. Then you have porosity. Are you sure that it's not also grain boundaries contributing. Everything that we do might be faulted because we don't know anything about the grain boundaries of sulfides because they're so soft that you can't see them and even at low temperature sometimes you can. I was going to ask both you and Kim that question later but we'll save that question for that that's a much longer discussion. I couldn't agree with you more often that the descriptors could be quite complex. But I think what gives me a little bit of hope is that because everything is interrelated there should be a higher level descriptor that in principle can tie things together but maybe we don't fully understand it quite yet. There are two more questions from the audience and let me just ask them really quickly. So, in terms of your anion substitution. Have you also looked at the change in the electronic conductivity the partial electronic conductivity in the anion substitution case. In the from sulfur to selenide or in the chloride bromide itines. Yeah, if you put a cell in there, the color changes, and I think it becomes more gray I think this is enough. So, I would assume the electronic conductivity increases, we haven't, we haven't looked at it for a simple reason. And this is also part of a much longer discussion, it's very difficult to measure electronic transport properties of these systems. What one currently does is to blocking electrodes and polarizing, but this is not a. This is not how you should do it right you need to have back in the experiment you need non blocking electrode and blocking electrode and then do your polarization, but none of these materials are stable against lithium or sodium. We're running these experiments without fixing the chemical potential of lithium or sodium. So, whatever you get out as electronic conductivity is bound to be wrong. Yeah. Last question from the audience. So related to this question of interfaces and stability so how does the substitutions that you talked about affect the electrochemical redox window I presume based on what you said this may be a challenging task as well. Yeah. So, I think the thermodynamic window is fully determined by your decomposition pathways. It's just gets tells us wherever or has long whatever decompose into we can just calculate that into a thermodynamic window, which means as long as there's a sulfur in there. The decomposition will create a sulfide, which means the electrochemical window in let's go from chloride to bromide to either it will stay the same. The selenide will change it. So, as long as we're within the sulfur in the in the sulfide, I don't think we can increase the thermodynamic stability of these materials, we don't know anything about kinetic stability. I mean, we can, we see that kinetics like the practically the stability changes a little bit on depending on how you change the your electrolyte. When you think about when when things just decompose at an active material, then now whatever form that interface determines the pathways of decomposition. Is it electronic conducting is the island conducting what are the faces like we're now in a multi phase, fixed space, and even theoretically, there's barely anyone mapping this out how sort of kinetic decomposition windows show up once you start decomposing things. So very challenging question and I have a lot of thoughts on it, but I don't have any idea how to properly solve this right now. As you say, more experiment for your doctoral students to do. Oh, well, thank you very much and please sit tight and then now let me ask you to introduce Kim our second speaker and we'll come back for a panel discussion at the end of the session. Well, thank you well and thank you Wolfgang for the very nice talk. Let me introduce our second speaker. Professor Kim C from Caltech. Kim received his PhD from you Santa Barbara after that. She went on to do a postdoc and University of Illinois, Obama champagne. I think about 2017 she joined in Caltech faculty. She has been doing very interesting research on and I am redox. I believe that's the topic she's going to talk about today. Well over the last few years. Kim's research has been recognized with multiple very prestigious award, including backman young investigator award. The Volkswagen BSF science award and electrochemistry. The pack our fellowship for science engineering and most recently, all face on naval research young investigator wall. We start Kim, I'll let you take over. Thank you so much for that kind introduction and thanks for the invitation to talk here today. It's my pleasure to tell you about the work we've been doing in and I on redox and alcohol alky rich metal sulfides. Before I jump into the main topic, I just want to sort of broadly introduce what my group is interested in. So to do that I'm just going to throw up a schematic of a conventional lithium ion battery that contains lithium and circulated graphite anode lithium metal oxide cathode that usually contains some amount of cobalt and a solvent based electrolyte like lithium hexafluoride phosphate and carbonate solvents. And so the goal of my group is really to get away from from this type of chemistry, because of limitations associated with lithium ion batteries. And so those limitations I consider there to be three that we can try to address with the chemistry. So the first is the cost. So of an EV battery most of the cost is in the materials and of the materials cost most of that cost is in the active cathode material. So if we can actively change the chemistry in the cell we can change the cost of the battery. The second is resources. The main components in the lithium ion battery aren't the most sustainable things we could think of to put in a widely spread technology. So lithium, for example, is geographically isolated in Argentina, Bolivia and Chile. At least the most accessible reserves are there. It's also mined in Australia. And then cobalt, which I mentioned is one of the main metals in the cathode is found primarily in the Democratic Republic of the Congo it's mined in a copper vein in the south part of that country. And there are some humanitarian issues associated with cobalt mines we'd like to get away from cobalt. And then finally the last issue is capacity. So I would argue that lithium ion interpolation chemistry is reaching its theoretical limitations. So we have a timeline of some of the canonical interpolation materials that have been developed since 1976 and stands winning handed with the MTS to. And now we have different kinds of materials that are similar to lithium cobalt oxide which of course john good enough and then to the 1980 and was commercialized in 1991, where we can substitute different metals under the cobalt site like nickel and manganese to the so-called nmc type materials. And these nmc materials are our state of the art cathode electrodes. And the benefit of the nmc materials is that we can reach near theoretical capacity so the orange bars here are showing the experimental capacity that that's reached in the lab and then the open bar shows that theoretical assuming we can remove and reintegrate all the available lithium in the structure. And so what you can see with nmc is that it's reaching that theoretical limit. And in general there's an interpolation limit to this sort of paradigm of one electron per transition metal that we do when we have interpolation chemistry. So to go beyond these types of issues with lithium ion batteries we try to think of new chemistry is we can put in the cell that that try to address these issues. And so our battery looks a little bit different. And we have things like we're very interested in metal electrodes were different interested in different kinds of electrolytes, and we're also interested in different types of cathodes. And so some of the projects in my group include working on different working ions compared to lithium so we're very interested in things like magnesium zinc and calcium. And these divalent cations are difficult to move in the solid state so we have projects trying to understand fundamentally how do we understand the ionics of divalent ion conductors. And if you're interested in that topic because maybe you were listening to Wolfgang's talk earlier gave great talk on ionics and you can look at some of these papers from our group looking at divalent ionics and trying to understand how to move these divalent cations through solid state materials. We also have projects where we're very interested in metal interfaces. So if we move to these next generation cations they're only energetically competitive if we use metal electrodes as the anodes. And so we have to understand how to do metal deposition stripping and we have to understand how to stabilize the interface of that metal electrode against electrolytes and a lot of these metals are very reactive and form very stable oxides. And then lastly which is what I'm going to focus on today, we have projects that where we're interested in essentially removing cobalt from the cathode and generating electrodes with high energy densities. And to do that we're looking at reversible multi electron redox. And the point of this project is to try and bypass this intercalation limit. And that's set by one electron per transition metal. So the work I'll tell you about today is done by Andy Martin, which was my first post stack. Steve Kim, Josh stack and Charlie handsome. And we've recently had more team members join this effort, Ishaan, Michelle, Xiaotang and Colin. And so really everything I'm talking about today is is due to their, their heroic efforts and setting these multi electron processes. So just to put multi electron and perspective, I'll go back to this lithium cobalt oxide material that we all know and love in conventional cathode materials so lithium cobalt oxide is a metal oxide of layered metal oxide material. And you can reversibly remove one or half a lithium per cobalt because there's an irreversible phase transition that happens if you go past that. But theoretically you could get all the lithium out of this material and if you did that you would pull one lithium out per cobalt, and that would give you one electron oxidation on the cobalt. And if you were to interpolate one lithium per cobalt you would have one electron reduction per cobalt. This is sort of paradigm of one electron per transition metal allows us to operate in this regime where we're doing intercalation chemistry and by that I mean that as we intercalate the lithium the structure doesn't really respond in any meaningful way there's there's small contractions and expansions but the crystallographic symmetry is highly highly related to the pre-lithiated structure. So I'm beyond this paradigm, and we now start to put more lithium in the material than we have transition metal, and then that moves us into the so called lithium rich regime, where we can have higher capacities. So over here on the left I have my lithium metal oxide sort of paradigm, and now I'm comparing that to these lithium rich structures where lithium is substituted for some of the transition metal in the metal layer. So in this case one third of the metals have been substituted with lithium, and these phases are of the formula Li2M03 and have been known for quite some time. And these lithium rich metal oxide materials you can now assume that I can deintercalate more than one mole of lithium per metal. And if I do that, then the question becomes where's the charge compensation so do I have multi electron oxidation on my metal, or can I can poke an ion redox in these materials for the charge compensation mechanism. And this is really important understanding where those electrons are coming from because it's a it's a way to start to target issues of the multi electron redox like reversibility and history assist. And the benefit of this multi electron redox is that we can reach much higher capacities. So in lithium metal oxides, like lithium cobalt oxide, we have 3D metals which are technologically more exciting than 40 metals. But we're limited by one electron per metal so our capacity is hovering around 300 million bars per gram. But with these multi electron materials, even though with the metal oxides we have to go to 40 metals like ruthenium. We can still get higher capacities theoretically because of the two electron oxidation. And so that's why these types of of chemistry is are really exciting. So a lot of the field is focused on understanding multi electron redox or anion redox and metal oxide electrodes. And what we decided to do with switch to sulfites. The reason for that is that when you try to do anion redox and oxide materials, the oxide oxidation potential is very high in voltage or very low in energy. And so in many times, many, many times you get electrolyte decomposition in conjunction with anion redox if you're lucky or in lieu of anion redox if you're unlucky. So this electrolyte decomposition really convolutes the electrochemistry so you don't really know where your electrons are coming from so it's really hard to calculate capacities. And it also convolutes the spectroscopic tools that we use to understand anion redox so you can have oxygenated products as a function of electrolyte decomposition so you don't really know what's happening with the oxide redox at all. And so it moves to sulfides however the sulfide P states are much higher in energy than the oxygen P states and so we can get sulfide redox at much lower potentials compared to oxide redox. And so these sulfide redox potentials lie well within the electrolyte stability window. We know that very well from different types of conversion chemistries. And so sulfide redox then offers us a way to understand Anna and redox from a fundamental point of view. It's not necessarily interesting but it's also technologically interesting and that's because when we go to lithium rich metal sulfides we can now use 3D metals again. So for example here's a phase li2 MS2 so in this case the metal is iron. And so li2 FES2 is a lithium rich metal sulfide material that contains highly abundant and easily obtainable iron. And so that makes it very interesting. And the other reasons sulfides are compelling choice to study for Anna and redox is because of the mechanism and the reversibility compared to the oxides. So here I'm comparing li2 FES2 versus li2 Ru03. And so this is data out of our lab but these materials have been studied for quite some time so li2 Ru03 was first studied of course by John Goodmuff in the 80s. And then li2 FES2 has been studied by people like Jeff Donn, Ruchel, Emma Kendrick for example, also since the 80s. And comparing these two apples to apples has brings up some striking differences between the mechanism. And so the one that I'll point out initially is this difference between the cycle one and the cycle two charge curves. So both of these materials are getting over one electron out of the formula unit meaning that we're deintercalating more than one lithium out of these materials and in fact we're deintercalating a lot more than one electron it's 1.75 electrons. So we're getting 1.75 moles of lithium out of the material. That's a lot of material to deintercalate. And if you look at this and compare li2 Ru03 to li2 FES2, what you'll see is that the charge curve for li2 FES2 for cycle one and cycle two are nearly identical, there's a slight capacity phase. But the mechanism for the charge compensation mechanism is largely maintained, even though we're pulling 1.75 electrons out of this material and we can get about 1.75 back in. With li2 Ru03, however, if you look between cycle one and cycle two, there's drastic changes in the shape of the curve and the voltage of the curve. So that's telling us that whatever processes are happening on cycle one or causing irreversible changes to this material, and that's likely associated with these degradation mechanisms. So this reversibility of this sulfide material is also another compelling reason that we should study these sulfides for amine and redox. So what we wanted to do was understand the charge compensation mechanism of li2 FES2. So there are some characteristic regions in the charge curve, there's a sloping profile up about to a half an electron. And there's a plateau region where we get another electron out of the material in the charge curve. And then on the discharge we kind of have another plateau and another sloping region. So we wanted to understand where these electrons are coming from and if we indeed have multi electron oxidation that is due to charge compensation by the anion. So we turned to X-ray absorption spectroscopy to study the electronic structure of the material. So this is the iron and the sulfur K edge of li2 FES2. And iron in this material is tetrahedral. So it's a split site lithium iron 50% occupied site. And it's edge sharing within this metal layer. And then you have this octahedral lithium layer, which is also edge sharing that separates the metal layers. So in the iron K edge you can see this pre-edge feature here. This pre-edge is associated with a slight distortion in the tetrahedral iron so that you can see this iron is displaced off of center from the tetrahedron. And that gives rise to this pre-edge feature. And then we have this, the rising iron K edge here that tells us about the oxidation state of the metal. And we can look at the first derivative of the rising K edge to give us a better idea of where the oxidation state or where that rising edge sits. And then we have the sulfur K edge spectrum, which is shown down here. The sulfur K edge spectrum has this really beautiful pre-edge feature. This pre-edge feature is the sulfur 1S to iron 3D transition that arises due to the covalent iron sulfur linkages within this material. And then we have the sulfur edge here. So now that you're sort of benchmarked into the X-ray absorption, we can now oxidize this material and see what happens to the iron and the sulfur K edges. So as I oxidized the material, what I see in the iron edge is that this pre-edge feature actually increases in intensity, which is telling us that we have a more of a tetrahedral or more of a less distorted tetrahedral iron center. And what we also see is that the rising edge shifts to higher energy. So this is suggesting that iron is oxidized in this initial region where we are doing the sloping region here from one to about half an electron. So initially we have iron oxidation of iron two to mixed iron two, three, so we're only getting half an electron now. If we look at the sulfur K edge, the biggest difference here is a rise or an increase in intensity of this pre-edge feature. And so an increase in intensity of the pre-edge feature is tightly correlated to the covalency of the material. And this is work that has been done by Ed Solomon, for example, looking at iron based proteins. We've done a lot of work proving that the pre-edge feature here is associated with the covalency. And so this makes a lot of sense because as we oxidize iron two to iron two, three, we get an increase in covalency of the iron sulfur bond. So that makes a lot of sense. So this initial oxidation is really localized on the iron. So what happens then if we go to the plateau region and we take another full electron out of this material. So now I'm showing that in red. And so we see that the iron K edge, we still have tetrahedral iron as indicated by this pre-edge feature. And what's very interesting is that even though we're taking a full electron out of this material, the iron is not oxidizing. So the position of the iron rising age stays essentially the same as it did at this 2.5 volt point. However, if we look at sulfur K edge, we see huge differences in the spectroscopy. So we see the evolution of this new pre-edge feature. We also see a big shift in the sulfur K edge. And so this is suggesting that sulfur is indeed oxidizing. And we can look to some standards to understand why there's a new pre-edge feature that grows in as the material is charged. And of course the material that we want to look at in great detail is FBS2, which has sulfur sulfur dumbbells. So here I'm showing Pyrite FBS2 that structure. And you can see that all of the sulfurs in this material have sulfur sulfur dumbbells are persulphides. And if you look at the pre-edge feature in FBS2, you can see it lines up nearly identically to what we measure in the oxidized LA2 FBS2. And so this is suggesting that we have oxidation of sulfide to persulphide. But I'll note that this initial pre-edge feature is maintained. So we still have sulfide character in this material. And so we have a mixed sulfide persulphide material after this oxidation has completed. And that makes a lot of sense because initially we take half an electron out, so we get mixed iron 2, iron 2, 3, but this full electron that's essentially coming from sulfur, we believe, is really localized on the S2 minus. If we were able to reduce all of the sulfur in this material, then we could get more than one electron. So it makes sense that some of our sulfurs maintain sulfide character as we do the oxidation. So we can then discharge the material to see if these oxidations are reversible. And indeed what we see is that the iron edge shifts back to its original position. And the sulfur K edge also shifts back to its original position. And I'll note there are some wiggles out here that have changed. And this is due to some irreversible structure changes that occur in the material. But electronically, the oxidation and the reduction are very reversible. So we can think about how the structure is responding to this oxidation and you might expect that it's responding quite significantly because we're pulling a lot of electrons out of the material and we're pulling a lot of lithium out of the material. And so we did operando x-ray diffraction to look at the structure response. And here I'm just going to focus on the 001 reflection of the LI2FDS2. And what we see in the sloping region is that we get a solid solution like behavior where the peak is simply shifting as a function of oxidation. But in the plateau region, we see the evolution of a new peak in the x-ray diffraction suggesting a two phase type mechanism, which makes a lot of sense considering we have this really flat plateau in the charge curve. What you'll also notice is that the intensity of the reflection is decreasing. And that's because we're losing a lot of the long range order in this material. And one of the only reflections that we maintain significant intensity for is the 001. So we maintain some registry within the C-axis. But other than that, we've broken a lot of the symmetry within this material, which has caused all of our reflections to decrease in intensity. When we reduce the material, you can see that the lattice contracts back or expands back to its original position. But we do not get back the crystallinity that we had before. So the intensities are still low. So even though we're able to do that reduction and actually the form of efficiencies are quite good, the structure is not able to return back to its original phase exactly. And if we look at the electrochemical impedance spectroscopy, here I'm showing some representative nitrous plots along the charge and discharge curve. And then here I'm plotting the charge transfer resistance as a function of oxidation and reduction. And what we see is that as the material is charged, the impedance increases as a function of charge, and then it starts to decrease as a function of discharge, but it doesn't quite get back to its original position. And that's because we don't have exactly the same material that we started with. And so this can lead to some capacity fade in the material. But now that we understand something about the structure as a function of oxidation, we can at least say in this sloping region where we have the solid solution like behavior, that the structure, the pristine structure and the structure at two and a half volts are pretty similar. And so that allows us to do some density of states calculations. And so we collaborated with Anton Vanderbent and his student for us to do partial density of states calculations on both li2 fs2, and then the partially oxidized li1.5 fs2. So if we look at the pristine phase over here on the left, we see that they're near the Fermi level. There is a majority of iron D states. And so this makes a lot of sense and this correlates really well with our X-ray absorption that we've initially seen iron, we initially see iron oxidation when we charge this material. After you dilute lithium from this material, you see a re-hybridization of the iron gain of sulfur p states to give us a more covalent iron sulfur bond. And you can actually see that in the electron density map shown here. And what this does is it essentially pushes up the sulfur p states closer to the Fermi level and this is now what allows us to access the sulfur p states and start doing an an redox in this material in lieu of iron redox. So if we're doing an an an redox and we're going from sulfide to persulfide, and we see structural changes in the X-ray diffraction, where exactly are those persulfide bonds? And this is actually still an outstanding question, but we can start to look at this by understanding how the iron changes as a function of oxidation. So we have a good local structural probe for the iron local structure with the iron XSAPS data. And so here I'm just showing the first gel correlation. So the first gel remember is just this tetrahedral iron where we have iron sulfur bonds. And we're looking at the bond length for those iron sulfur bonds and the coordination number as a function of oxidation and reduction. So we fit the XSAPS and we can then pull out these two parameters in the first coordination shell. And what we can see is that here's the charge curve. Here's the discharge curve and I've plotted the bond length and the coordination number on top of the charge and discharge curves. So initially for this charge we see a decrease in the iron sulfur bond length because we have this more covalent iron sulfur bond we've oxidized iron that makes a lot of sense. But what's really surprising is that when we do this sulfide to persulfide oxidation and the plateau, we don't really see any change in the iron local structure. So both the bond length and the coordination number are relatively similar. Even though we have this huge structural transformation, we have a two phase region in our diffraction and we start to lose registry in the material. We really don't see a change in the iron local structure. And when we discharge the material, you can see that the bond length and the coordination number is somewhat recovered to their original positions. And so we have a hypothesis for how we might be able to form sulfur sulfur bonds and not change the iron local structure very much. So initially we believe that lithium is deintercalated from the tetrahedral sites and this is from calculations from Anton's group that has shown that the tetrahedral sites are more favorable. Once you form tetrahedral vacancies, you've now gone from edge sharing tetrahedra to corner sharing tetrahedra. And we think that these corner sharing tetrahedra can now tilt towards each other to form a sulfur sulfur bond at the apex. And so forming sulfur sulfur bonds just requires a simple corrugation of the metal layer which will break a lot of the symmetry in the material but maintain some of the registry in the C axis, and also maintain the iron local structure as a function of oxidation. So just to summarize what happens with this material, initially you have one or half an electron oxidation, that half electron oxidation can really be thought of as a topotactic and deintercalation reaction where we have oxidation of iron two to mix iron two three. And that's just shown in this cartoon here where we're just simply removing lithium. And so oxidation occurs on the sulfide. So this next one electron is coming from sulfide, and it's a sulfide to per sulfide transition. And that sulfide to per sulfide transition causes a lot of structural distortions in the material and we think could corrugate the metal layers to form those sulfur sulfur bonds. And so this first part is kind of an interpolation type reaction and then this next part is starting to look like a conversion reactions where we have actually a lot of atoms moving and we have bond breaking and brown forming reactions occurring. And that second oxidation. So now that we have material that has that shows an iron redox we can use this material to start to understand on structure property relationships. And the first structure property relationship we wanted to get at was the role of covalency in the an iron redox. So covalency has been touted in the oxide and I'm redox literature as an enabling tool to enable an iron redox in the oxides and this is a list of some papers that talk about covalency in the metal oxide materials. As sort of a as, as, as the thing they need in order to get the oxide redox to occur. And this is non exhaustive these are just some that I pulled from the literature. And so we can use our metal sulfide materials to systematically change the covalency and determine how that might affect the an iron redox behavior. And so you can imagine doing that one of two ways you can either change the metal, or you can change the anion Wolfgang talked a little bit about this in the different context. And so what we ended up doing was changing the anion. So what we did was we took our li to fb s to n member, and we made li to fb se to, which is the selenide and member, and turns out these two materials are isostructural, which is not that surprising selenides and sulfides are very often isostructural. But what that allows us to do is go from sulfur piece p states and the material to selenide to selenium p states. And the selenium p states are much higher in energy than the sulfur p states. And so we should get greater orbital overlap of the iron gene the selenium p states and get a more covalent material as we go to the selenide. And so the, the other advantage here is that these two materials are iso structural and so we can actually make the solid solution in between so we can do a systematic substitution of selenium for sulfur. And just depending on our input stoichiometry have some statistical distribution of those two anions on the anion sublattice. And so that's what we did. So here I'm showing the x-ray diffraction for the solid solution, all the way from the sulfide and member to the selenide and member. And we can fit these x-ray diffraction patterns with a quantitative refilled refinement and plot the lattice parameters. And if we plot the lattice parameters as a function of stoichiometry we see that there's a linear correlation, which tells us that we have a statistical distribution of sulfur and selenium on the anion sites as we do the substitution so we have an anionic solid solution. So then when we do the electrochemistry we can see how this systematic change in the covalency going from sulfide to selenide affects the electrochemistry. And we see something that's very cool so as we go from sulfide, which is shown in red, to selenide, you see a systematic shift in the anion redox potential so the anion redox potential shifts to lower and lower voltages as you go to the selenide. And so we can actually plot this as a function of stoichiometry so the position of the anion redox plateau versus stoichiometry and we see a near linear correlation. And so what this is saying is that this material is very well electronically mixed. So as we oxidize we're pulling electrons out of bands that have selenium and sulfur character, and they have rehybridized together to form a single band so we're not just doing selenide oxidation followed by sulfide oxidation, it really is oxidation of both of those piece states due to this this rehybridization and the good mixing of the two anions. And you might expect that this shift might have something to do with kinetics, because, as we were sort of talking about earlier, the selenides of course will likely have a higher electronic conductivity than the sulfides which could lead to lower over potentials when we do the oxidation. So what we did was galvanosodic intermitent titration technique where we can measure the near equilibrium voltage of the oxidation for oxidation processes by letting the cell rest and intermittent times as we do the oxidation. And so that's what's shown in this solid line. And so what you can think of it for the bottom of these polarizations the bottom of these polarizations essentially tells us thermodynamic potential of the oxidation. So here I've drawn a dashed line that shows the average voltage for the anion redox oxidation voltage for the LI2FBS2. So as we start to put selenium in the material you can see that that thermodynamic oxidation potential starts to decrease systematically so as we put selenium in you can see it drops here by about 100 millivolts. And then as we go to the selenide M member we've now dropped that potential by over 250 millivolts, just by putting selenium in the face. And that's because of the change in the selenium P states compared to the sulfur P states. And we still have anion redox in this material. And one thing we wanted to understand was if we have formation of selenium selenium bonds just like we saw sulfur sulfur bonds in the LI2FBS2. One of the benefits of LI2FBS2 is that we can get better Raman scattering from LI2FBS2 than we could from the sulfide. And so that allows us to use Raman scattering to understand if we have selenium selenium bonds. So we can compare LI2FBS2 as we oxidize it to mark the site FES2, which has the selenium selenium bonds in the structure, and the selenium selenium stretches and FES2 are found around 217 wave numbers. So if we oxidize this material and do in situ Raman spectroscopy, what we see is a mode that grows in as we do the oxidation right around where we would expect the selenium selenium bond mode to be if we had selenium selenium bonds as a function of oxidation. And so we think that LI2FBS2 also forms selenium selenium bonds just like the LI2FBS2 forms sulfur sulfur bonds. So we can move forward and try to understand how the charge compensation mechanism has shifted as we as we change the cobalancy in this material. And so we turn again to X-ray absorption spectroscopy. But instead of showing you all of the X-ray absorption spectra, I have simplified them to just looking at a shift in the pre-edge as a function of oxidation and reduction. So to sort of orient you, we're first going to start with LI2FBS2 here on the top. And so here I'm just plotting the shift of the iron rising edge and the shift of the sulfur rising edge as a function of oxidation. So the electric chemistry is shown in black and reduction. And so what we see and what we already know is that as in the sloping region for LI2FBS2, we have first oxidation of iron as indicated by the shift in the iron edge. And currently the iron edge does not shift, but the sulfur edge shifts quite a bit. And so this is our sulfide to persulphide transition, and then both edges shift back as we do the reduction. So now we can compare what happens with LI2FBS2. So here is the selenide where now we've shifted the P states closer to these states and we have a more covalent metal anion bond. And what you can see is that throughout the charge per both the iron and the selenium are contributing to the oxidation. So if we have a more covalent material, we're actually going to push the anion redox away from the anion back onto the metal, because we have better mixing of that iron and the selenium electrons. And then when we do the reduction, you can see that they they shipped back to their original position. And what's very cool is that if we look at the mixed material where we have 50% sulfide and 50% selenide, you see that all three are contributing throughout the charge curve. So we see shifts in the sulfur edge throughout, we see shifts in the selenium edge throughout, and we see shifts in the iron edge throughout. So this increase in covalency has basically pulled sulfur back into the charge compensation mechanism at earlier states of charge. And it's pulled iron into the charge compensation mechanism at higher states of charge, because we have this more covalent material. And then we can do the reduction, you can see that they both they all shipped back to their original positions. So an increase in covalency in this case actually causes the charge compensation to shift back onto the metal compared to this sort of more discreet iron oxidation and sulfide oxidation that we have an LI2FES2. So we thought that was a really nice sort of proof of concept to show that we're tuning the covalency. So what we find with the cycling stability is that you might expect that maybe this material is more stable because you have this more covalent metal anion bond maybe it's able to stabilize these oxidative structures in a better way. But actually what we find is the exact opposite. So if we look at the end number the sulfide the charge, the cycling data as a function of cycle number is shown in red. And then, as we put selenium into the structure you can see that the charge capacity retention basically tanks. So here's the selenium material. Here's the end number where it loses capacity very quickly. And so this capacity loss we think is actually due to the structural response to the anion oxidation so even though these materials are accommodating about the same number of electrons per formula unit on the charge compensation mechanisms are actually quite different, not electronically but structurally. So if we look at the operando x-ray diffraction of the LI2FES2, we see something very very different compared to what we saw with the LI2FES2. So we get a two phase region much earlier in the charge curve. So we see the formation of that new phase. And I think that's because we're actually involving the anions earlier in the charge compensation mechanism because we've made a more covalent metal anion bond. And what happens then is that these two phase mechanisms are less reversible than these solid solution type mechanisms. And you can also see that in the impedance. So this is the impedance spectroscopy here, representative Nyquist spots, and now I'm showing the charge transfer resistance as a function of oxidation for the LI2FES2. And here you can see that the impedance increases as a function of charge, but it does not recover like the LI2FES2 does. You can see that it maintains high impedance, even though we're reducing the period. So this is also likely a reason why the charge, why the capacity fades as a function of cycle number. So just to kind of summarize as part of the talk, we've been able to show that we can do anion redox and LI2FES2. We can tune the covalency by making anions all solutions by moving from sulfide to selenide and shifting the selenium P states up closer to the iron D states. And that allows us to systematically tune the potential of the anion redox, which is a very useful thing to do. And it also allows us to have a handle on covalency and understand how covalency might affect the reversibility of the anion redox processes. And so in the last part of my talk, I just want to put this into context and sort of a broader picture and understanding how pushing and pulling charge compensation from the anion back onto the metal might affect the mechanism of charge compensation, especially from a structural point of view, and also the reversibility. So to do that, here on the top, I just have a sliding scale of the number of electrons that are being donated by the anion. So how much is sulfur plain into the charge compensation. So on one extreme end of the spectrum we have, you know, zero to small amount of electrons associated with the anion. So this is where charge compensation is mostly contributed by the transition. So, these are mechanisms that we all know and love so this is just interpolation chemistry right so if we have things like LIT is to, and we do interpolate LIT is to, we end up with the IS to and lithium, and most of that charge compensation occurs on the titanium. However, some of that some of that charge compensation can be pushed on to the anion. So if you look at the beta charge analysis you'll actually see that that number is not equal to zero. So maybe there's some regime in here where we can maintain the structural integrity of the material, and do this topotactic the interglation circulation reaction, but still utilize some of the electronic states donated by the anion. And so this is really this interglation mechanism where you have reversible structural changes but minimal contribution from the anion. So on the other end of the spectrum, where we have a large contribution from the anion where all of it, only the anion is participating in charge compensation. And of course this is conversion chemistry. So this is doing something like sulfur to lithium sulfide or lithium sulfide to sulfide. I'm sorry lithium sulfide to sulfur. And so in this case, of course the anion is doing all the charge compensation, and there's a massive structural rearrangement we have to go from sulfides to these s eight rings. So we can change the phase on the initial phase looks nothing like the beginning phase, and that causes some issues associated with reversibility. And of course there's, there's a lot of interesting mechanisms that occur, especially in this chemistry depending on the electrolyte because there are soluble intermediates but I won't go into that right now. We can just think about the solid state structural transformations and understand that the anions have to change positions quite significantly in order to get from an essay type structure to a rock solid lithium sulfide. So the more charges compensation we push on to the anion it looks like we have more structural changes right. But now we sort of have this intermediate regime, the sort of hybrid mechanism where we can do both intercalation chemistry, but also have some structural rearrangements to accommodate this new bond that has to break and form as a function of anion oxidation. So we have a hybrid mechanism that lies somewhere in the middle. And, and in this mechanism, we can have some charge compensation by the sulfur, where we're going from sulfide to per sulfide, but it's reversible and we can go back to the sulfide. And we maintain some registry in the material to allow us to get higher degree of reversibility so the anions aren't moving quite as much in conversion but they're definitely moving more than they did in terms of intercalation because there's new bond breaking and forming reactions. And so I think the question now is, how reversible is the sort of hybrid type mechanism, and how far along the sliding scale can we push it until we start to run into the reversibility issues of conversion chemistry. And we just submitted a perspective on this. So, hopefully that will be out soon. So I have to thank everybody that worked on this. I think the students and postdocs at the beginning of my talk but I have a wonderful group of people that work with me and I wouldn't be able to do any of this course without them this is all of their hard work. This is work that was done in our EFRC scaler. So we have several collaborators that we've worked with to get this done at UCSB slash UCLA and USC and I highlighted them here in these green boxes so so many things to them. So thanks to our other funding sources and thank you again for the invitation to be here today. Kim, thank you for the very nice fundamental study works, the talk. There are a number of questions flowing in. So the first one is about the wattage hysteresis and the lithium iron sulfide back and forth. What's the origin of that wattage hysteresis is it related to iron migration. When you have software software bond formation forming the dimer so what's the nature of that when hysteresis has a little bit not not too big though. Yeah, that's a great question so the iron actually maintains its tetrafedral geometry throughout the charge and discharge curve and we see that through the XSF and also through the XR absorption so that sort of initial pre-etch feature tells us about that tetrafedral coordination. So it's not migrating to an octahedral site like you might expect if you were forming something that looks like pyrite. It's definitely not doing that. But what we do see is that. Okay so the iron doesn't migrate, but that hysteresis we think is actually associated with the breaking and forming of these sulfur sulfur bonds and how kinetically accessible that is either on oxidation or reduction. We have some work going on now to try to understand the charge compensation mechanism on the reduction, which will tell us more about the hysteresis but what you can think. So simply speaking the hysteresis is a path hysteresis so the charge mechanism is different than the discharge mechanism. And we think that's because initially you start with a sulfide and you end up with a per sulfide. So that difference in making that bond versus breaking that bond is what's causing the hysteresis in the charge and discharge curves. Okay, so so will you also have a question. Yes. Wonderful talk him. I was looking at your li to FES to result in the mechanism reminds me a lot of the, the venereal phosphate system. Is this a sort of a comparable thinking as well just due to the corner sharing nature and the flexibility to dimerize reversibly. It's very similar. I think, I think for us, I think the driving force is making that sulfur sulfur bond. And that's probably very similar in the vanadils so but yeah that's an interesting point I haven't thought to correlate those two before. Certainly, the, the generation of vacancies facilitates the structural distortions that we need in order to access the anti redox right so like, without the vacancies we can't form the sulfur sulfur five. So, so that that's a really key factor. And that's, that's also true true in the corner sharing materials you're talking about to. Yeah, very interesting and another the related thing I've been sort of, you know, one of these intuitive guiding rules and layer oxides is that the, you know, the edge sharing is what really keeps the layer sliding from happening. Is, do you think that's what's happening in your XRD where you see after the first cycle and it really sort of amorphizes is it because of layer sliding or is this something else. There's definitely some layers like like there's some stacking faults because, you know, if you think about the degree of the interpolation to you know it's one, we can actually we can get 1.85 electrons out of this thing. If we try hard at making our electrodes. So we're taking out most of the lithium. So there's really no reason for those layers to maintain any kind of registry between each other. So that I think there's lots of stacking faults and we see this in other materials as well. Like the light to TIS three family, for example, where you have octahedral metals, we see stacking faults in those materials when you do the oxidation. Do you think it's a trade off. So, you have more flexibility to dimerize the software from the corner sharing, but you have a little bit less cohesive energy to retain the layer structure, compared to the edge sharing is that a sort of a that you're seeing here. Yeah, I think that's a that's a good take home message and I think that's going to be true probably of any of these alkali rich type chemistries where you know you're really you're breaking down the the framework of the material by deintercleting it so it's it's sort of a miracle if you think of like LIT is to that you can get to TIS to that you can pull all the lithium out but now we're going further than that right like we're pulling lithium out of the metal layers to if you want to correlate it to that right that's not iso structural the light to FES to but you can think about it that way. So yeah I think that's a once you start doing that. It's going to be difficult but actually I think there might be a way and this is something that we're trying to study now of using the these new bond forming reactions to stabilize the structure so. So you know if you just had conventional deintercalation chemistry where there's no bond breaking and bond forming reactions per se with the metal and the anions. Then you're sort of stuck and you just kind of have this open framework now and your layers can slip and all of these things can happen, and you know there's a lot of changes between like a no three structure and a no one structure. So, but if you have these bond breaking informing reactions maybe you can actually leverage those to stabilize those transitions so that you can control it in a better way so you like, you blew it together with these per sulfides, and then you unglue it when you reduce it. Maybe that's something we can leverage right we're not there yet but that would be very interesting. Looking forward to hearing more. Thank you Kim. Well, so Kim, you know, moving from oxide system to software system. Of course software is heavier. So, you show very nicely and the effect soft by adding selenite selenium right there right to tune the voltage. So it's still a thought of going the other way instead of adding selenium adding oxygen. Oxygen is very different from of course from software so what's the challenge, challenge in doing that right so then you can ship the voltage higher. It's actually better give you a higher voltage give you more energy so what was the thinking of adding oxygen. Yeah, so we have several people in my lab who are now trying to ship the voltage in the correct direction which would be up. One of our thoughts is to use oxy sulfides but as you mentioned, making oxy you can't really make solid solutions of oxygen sulfur. So if you look in the in the literature there's very few examples of that so if you have an oxy sulfide type material those those tend to face separate and you get or site separate into their own sites so you have these like anti perov sky like structures where you can put the selenite sulfur sits on its own site. And in fact, that makes it really difficult to understand what's going on with the Anna redox. So, we can't do this like solid solution type thing that we did with the selenite. But I agree with you if we if we can find a way to form stable oxy sulfides as solid solutions it would be great. And we're working on that but I don't know that we're going to get there I think, you know if you look at the sulfides and the oxides, they're often they're not iso structural for the two end numbers. So it's going to be really hard to stabilize those in a solid solution but we have other ideas on how we might be able to do it. Yeah, I agree with you because oxygen and period table period to my god it's atomic radios compared to next one down. Oxygen is different so big so it's a very, very different chemistry compared to solid sulfide selenite. It's challenging right there. Yeah. It's extremely hard and sulfides are much. Yeah, well, I'm thinking along similar line so you look at iron so that's great iron is abundant. It has these oxytocins they are two plus three plus. There are another transition metal iron combined with sulfide other than iron sulfide and titanium sulfide you look at you look at this to today. And other transition metal iron that might look attractive. Yeah, we have been able actually to put other metals into the light to fs to space. We have to actually change more than just the metal to get it in. And those are much less reversible than iron face. So we don't fully understand why that is but I think that there's some conversion that happens when we do the oxidation that we don't get in the iron phase, because we have participation of metal throughout more parts of the charge curve than we do with iron. How is what made him so far. There are some really interesting vanadium sulfide phases and I've actually stayed away from vanadium because vanadium is so promising. Yeah, yeah, yeah. But yeah, there are some really interesting lithium vanadium sulfur phases that that might be interesting to look at for sure. Yeah. So just for brainstorming purpose came. Now looking at iron sulfide is a one iron tool software. So it's still a possibility to have something I understand it's probably not a stable phase right there. The software ratio to iron ratio go higher. For example, you have one iron, but so far it's almost like I am so far to mix out with some software, then you solve a month goes high then you increase the capacity of your and iron that part, even more to access a million month program, much higher capacity. Yeah, that's an interesting idea. I'm not, I'm not familiar with any thermodynamic phases that have a higher stoichiometry of sulfur to iron than one to two. But if you look at phases like, I mean one way to think about this I guess would be, you know, conversion of FES to so if you do conversion of FES to eventually get to essay. And at some points along that curve right you're going to have high degrees of sulfur sulfur sulfur bonds compared to iron. So, I guess that's one way to think about it but from a thermodynamic perspective that there's there's nothing there. That's that's stable. Yeah, okay. Well, I think this is great. I think I should probably invite also Wolfgang back to the stage together with team, we can have a panel discussion. And I want to maybe ask the first question we'll feel free to talk at any time. Wolfgang and in King, two of you are you are in a, in a relative sense right early career of stage of your career and anything. I'll just say anything you can share with a young students in postdocs. In the, in the benefit space, like the chemical energy storage space. As one question we often asked to our panelists or any otherwise to the students and postdocs. I would say, I guess specifically in the battery space, I would say be wary of trends and try and do things that aren't along the lines of what everyone else is doing. Because I think that often, I don't know. I think this is probably true of a lot of fields but I think with the battery field in particular, you know, we can get really focused on on like one thing, and it's probably good to try and try and deviate from that sometimes be courageous. Yeah, I would have gone along the same line I think like it, I would have used the word volatile. And so do what you find fun and it'll work out. In a sense, both of you are showing examples today. Yeah, of, you know, yeah, I appreciate particularly the key discussion that I'm telling for you. So we are doing this beautiful fundamental study on a system has in a lot of beautiful science right there. I think this really, you know, the software system software selling that open up a new understanding that's that's just been great. Well, do you want to chime in. Sure, sure. Thanks, he came again great, great talk. I want to ask maybe just on a broader note as well before diving into more specifics and both of you come from the solid state chemistry field. You know, it's such a rich space. But I think the battery community has sort of focus on a few important say crystal structures for example. So I'm curious in sort of looking forward, what are some of the new really cool solid state chemistry concept that you think we can learn from, say materials for batteries. Something I'm really excited about are these anionic solid solutions I think Wolfgang is to I think there's sort of a, you know, I don't know maybe this is wrong because I haven't been around very long but it feels like most of the solutions are, you know, cationic solid solutions and there's a lot of good reasons to do that. But I think there's sort of untapped potential in these anionic solid solutions and we're really learning a lot by doing that and I think not only from charge compensation point of view but like what Wolfgang showed and and looking at the the sulfide versus the selenium selenide materials and how that can affect these lattice dynamics I think is also very cool. Yeah, I would, I would add on to that because what. I think along those lines when you when we need to think about local structure of materials because when we're saying or we're making these solutions, what you're really doing is, at least in our case you have a PS4 unit and you're substitute the selenium, you're actually creating a variety of polyhedral species, you get then PS3, SE1, PS2, SE2, PS1, SE3, so you had really a nice, in terms of Rama it looks like a Bernoulli distribution of tetrahedral symmetry C3V and C2V symmetry polyhedral that are clearly somehow distributed through a material. And so whenever one just thinks of oh it's a solid solution from an from an anion this locally means that you have all sorts of different polyhedral that will have locally entirely different effects. But for one one needs to understand this from a local structure, and then figure out, can we even average over things. Like we're, I think as structural chemists were automatically averaging over everything. The way we're doing our like the fractal fraction is really just a global average but is this enough. And this is something that I that I find quite exciting these days actually. And Kim I couldn't agree more in fact I was one of the question was going to ask next is, you know, I trained in the point defect chemistry community so we also think in terms of average right so you put all the defects in and then it has an average transport property or average thermodynamic property. But fully agree with the both of you that these are modifying the local structure wherever the defect sets and you show them beautifully for substitutions or for point defect like vacancies. And this distribution of local structure really makes it confusing to establish in a roughly what you said about descriptors, or Kim your efforts to understand the redox potential for example because it's really hard to know exactly what which one of those local motifs are really responsible so and I think from an experimental perspective and perhaps it's easier in theoretical approaches but experimentally this is also quite challenging. So I was wondering if the both of you can comment briefly on, you know, what are the next steps as we go from this more of an average structure perspective into a local structure perspective. What are the technical breakthroughs that we need in order to really think about these things more as a collection or ensemble of local structure as opposed to averaging them and taking them as what you pointed out, Kim. Um, I think, you know, of course we have our local structural probes, like PDF and xfs, which xfs isn't you know xfs you're still averaging so you know you really need something like PDF. And you can have other tools and and you can have also other local structural probes like NMR can actually be a good reporter on local structure. But but it's hard to measure so one thing we've been doing. And sort of actually in the context of an Android ox is installing very diluted reporters into the material so that you can say okay, I can you know, just look at that reporter and that will tell me what's locally around it and because it's so dilute I can assume that I can see the local structure around that reporter. So that's something that we're sort of trying to do. But it's, it's hard. It's really hard. But I, yeah, I agree that it's incredibly important to understand where defects are I mean what we've seen especially, you know, in, especially in these like these lithium rich materials you have multiple locations where you can form a vacancy, like we really know where the lithium is coming from. And, and that's hard to figure out right. So, I think it's even more complicated right because even if you look at these local structures and the defects or let's say, maybe vacancy clustering all of these classic classic things that we know is this is out of the experiment. We don't know how any of these things behave once there's an applied potential. And when there's actually things moving around. Maybe averaging is good enough. So I think in principle one would have not just at the local structure but also under load. And while doing some things are really operando and this is difficult to really map out an iron jumping and then locally resolving what it's doing with the structure. Maybe with like, I mean free electron lasers are coming up more and more so you have much faster timescales, maybe one can sort of design electrochemical experiments while using a free electron laser. PDF is probably still averaging too much over it. So TEM would be nice in our case the TEM just shoots away your sulfur and your TEM collaborators don't like you anymore because they have sulfur all over the place in the chamber. I was going to say TEM but exactly the beam damage it's kind of condition is challenging and then you can do quite a lot of those imaging what the right electron detector but once it's quiet, nothing will be moving anymore. So it's a bit of challenging. Yeah, problem. This is something that just because it's challenging now doesn't mean that we can't solve it and are able to do this routinely in 30 years. I mean now look at this. There's lab PDF defectometer, there's lab XSAPS, back then you needed to go to a beam line, you don't need that necessarily anymore. So there's a lot of things that hopefully in 20 years are not too problematic anymore. And so we just need to figure out how to get there. We just need more money. It's such a great vision you're painting because you know 20 years ago operando anything wasn't really a thing. And now it is routine, right. So I think in 20 more years and certainly things we don't think it's possible will also become routine. I think we're all children that just have expensive toys and maybe in 20 years, there's new toys. Hopefully the price will have come down they're not as fair enough. Yeah, I want to ask both of you, you know, we have seen the explosion of lithium iron to the market space. And that brings the challenge of online and some of your, I think slides is we need to look for really abundant elements. We need to look for the features that can get us to go beyond lithium iron, but lithium iron will still be there, you know, for extremely long time. What will be the type of chemistry. I think they can enable, you know, and a different end in Catholic chemistry, and Kim you're looking into an iron, but, you know, look down to the past can matching and go beyond the lithium iron both performance in the cause. It can be the one right there for us to, to, to think about I know Kim you look at, you know, software and I and redox is a soft and I read out the, the one continue down to the path. Wolfgang your solid state right right there. What will be the system if it's for solid state what will be like for life. Oxide right here this all type of software family availability but if we use lithium iron as the reference right there. What could be the, the one we should look into more people first. I know it's a hard question. I should not ask myself anything. Not to increase the stakes you're also, you know, on on video and on record. So, so maybe then I'll start with the easy answers and I give the hard ones to Kim. You know, I think I think for one sodium is very attractive from an abundance perspective. A lot of the challenges that people said, we're talking about five years ago with terms of sodium, iron batteries, they're sort of slowly being mitigated I think CTL is now trying to put those into the packs together with lithium iron. So, sodium solid state batteries work. Not well yet, but they might get there. Going away from lithium so my, my typical answer answer of lithium sulfur solid batteries that won't work. There's there's fluoride iron batteries and there's fluoride iron solid batteries, but you know what maybe in 30 years we just need to go back to the oxide iron that is moving around because that is very abundant and that would just bring us to fuel cells. That's a fail you know speculation. I guess a little more political about this. So, I think it depends on the application I think I think you know of course the application will define the energy density metrics and the cost metrics that you need so it makes sense to me to for us to pursue a variety of different chemistry is because we're going to need a variety of different types of batteries for the applications that we want to use them in. So I think, you know, I, like you said, I don't think we're ever going to replace lithium ion, but what we need to be able to do is be able to not use lithium ion and applications where we can use something else. I think that's really important. So, so that's why we've been really interested in these multi valent ions I think that I think that the, the hope of using metal multi valent type batteries with metal anodes is really exciting. I don't think using sulfur I think there is a future for sulfur, whether it's elemental sulfur or sulfides. I'm not sure, but I think there's, there's a lot of interesting things to do there. So, Kim, I don't know if I'm allowed to ask a question because I always get when I say well sulfur, there's more than enough sulfur in the world to use. I usually get the answer, well, you know, the most of the sulfur comes from the oil that we're pulling out of the earth, right? If we don't do that anymore, then where do we get all that sulfur from? Are we then just drilling for oil to get sulfur or? Well, one, I think that there's plenty of sulfur from oil refining. And two is that if we do run out of that then good for us and we can start looking in volcanoes. So, if you've seen, I mean you've seen these images, I don't actually know the reserve numbers in volcanic areas, but like in Indonesia you can walk into a volcano with a shovel and come out with chunks of sulfur. And it's very bad for you, you should not do that because there's lots of H2S, but that's people do that. So Wolfgang, even though we don't burn oil, but we still need plastic so this will come from very likely for oil. We're still pulling out solvents, they have sulfur coming from there. Can I also add my predictions? Yeah, please. So it's just like everyone said, you know, lithium ion for sure will be here for a long, long time. And just because the industry is just above critical mass in unprecedented ways. But, you know, I do think if I look at the history of lithium ion, why did lithium ion emerge? It was really because the aqueous chemistry, electrochemistry didn't deliver the voltage that was needed. And that was the step jump that was really essential. But I think if we look at it, a lot of the challenges with lithium ion also come from its voltages. Safety, electrolyte compatibility, and I think you show beautifully when you go down in voltage a lot of interesting things can happen. So my prediction maybe in the 20 year horizon is I think we're going to see maybe a shift back to aqueous, electrochemistry for energy storage and there will be a more emphasis like what Kim you're doing is to just increase the capacity more. You know, we didn't really have the tricks 20 years ago when the research for nickel metal hydro battery stopped that we have today for lithium ion. But some of those I think principles can be applied much more easily when you're at the one volt range. And I think manufacturability will definitely be a big topic sustainability will be a big topic. You know you don't have to work with NMP that's going to be an advantage so I think a lot of these price that we pay for lithium ion will become higher and higher, and that may shift the balance back to aqueous, which I think will be kind of a cyclic thing will be kind of fun to watch to see that was really the case. Well, that's a great point. Probably aqueous for I think what you mean is for stationary more for stationary storage, instead of portable. I think it's possible that if we can make the advancement needed on the on the end on the capacity side, and there may be a pathway to getting a closer it probably would be never as good as lithium ion because three axis is really hard to get in capacity. But maybe two x is possible, and then you can be sort of at the levels of lithium iron phosphate. But I think the safety of manufacturability trade offs will become increasingly more important, even for transportation that I think, I mean, I already like to FESU is is competitive with definitely LFP, for sure. And I think the question in my mind is balance of plant though so as you go to these lower voltage materials you have to stack more right so it's kind of a question of how much that affects your energy density on the pack scale. It's not clear to me what what's going to happen there because we don't, we don't really know what that material is going to be. So actually came to that point, we have a very interesting session coming up in a, in a few weeks where we will have a two speaker talking about sell to pack technology. And this is looking basically at the volumetric utilization on the pack level. And I think most people agree that the safer the battery, the lower the voltage, the more you can pack. And we also had Nicola companion from McKinsey talk a few weeks, about a month and a half ago, and he basically told us that the state of the art, you know, nmc 622 battery in China at the pack level is only a few percent more energy dense than the state of the LFP battery, just because you can pack more density. So I think, like you mentioned completely agree that the system level consideration may be paramount, because at the end of the day for transportation is the pack that matters not not the individual cathards chemistry. Well, I probably should let you conclude today's session. Sure. Do either of you want to save like a one minute thing to finish us up today. Parting words. Hardly words. Well, I would like to be in California right now to partner with. No, I think from a science perspective. And this is something I came already said there's there's always these trends and and will also point out I think a lot of things do come in waves and in the end. I don't think it's really the question of which system will survive. It's, we need to get better on a fundamental understanding, and then like sort of push this into an application and it's just a lot more to do over the next time of our career at least. And I think that's what's exciting. Yeah, I would definitely echo that sentiment and I think on top of that I think there's a lot of interesting reasons to look at new chemistry, which is really exciting from a fundamental point of view which is what we all love to do I think, and really I don't know why things work or why they don't work, but it's also really exciting to think about what the future of battery chemistries might be because there are so many different options. So that's, it's a very cool time to be working in this field I think because we can start to shape what what the future might look like. And thank you so much on, and on those note I'm, let's go ahead and close up the session today. So as I mentioned in two weeks, we will have a professor Chaoyang Wang from Penn State to talk about sale to pack technology in the context of thermal regulation at the battery pack level, and we're also delighted to host a Dr. Jeff, the CEO of one, and they're developing sale to pack technology so it is quite different than the talk today, but I think this is the fun part of storage seminars that we're really going from Adam's the systems. So I'd like to thank everyone again for tuning in King off game. Thank you so much again for joining us. And I look forward to seeing everyone in two weeks.