 We're particularly interested in understanding how the electronic structure of oxides used in positive electrodes, in particular, the meteroxygen bond covalency, essentially, how the Fermi level relative to the oxygen P band of the oxide interfacing with the apodic electrolyte and how the electronic structure would influence the reactivity at the interface resulting to degradation or oxidation product. In particular, we have been utilizing a library of NMC materials to test and examine the reactivity between oxides and the carbonate-based electrolyte. And interestingly, well, through the work of Huber-Gestiger in Munich, they have shown for NMC-811 it has a much earlier voltage onset for oxygen evolution relative to other lower nickel NMC materials. And as we have shown recently, along with other people's work through X-ray absorption stratroscopy, nickel NMC essentially oxidizes or release oxygen at comparable oxidation state of nickel or comparable lithium de-integration amounts relative to other nickel-based materials. So then this raises a really interesting question. So essentially, they reach a comparable amount of lithium content around 0.2, 0.3, and then release oxygen. But before that, the materials of different NMC with various amount of nickel can exhibit a different degree of cycling degradation, in particular, as shown through the work of Hubert, two materials of NMC of 111, 622, and 811. Cycling to a comparable capacity or lithium amount, you see a significant greater degradation of NMC-811. So the question is, why? Why 811 would exhibit greater degradation? And our current thinking is that the surfaces can potentially exhibit oxidative dehydrogenation more so than other NMC materials. So we began this research. This is the research done by Livia, looking at the surfaces reacting with carbonate solvent through various reaction mechanisms proposed in the literature in the past few decades. So we examined electrophilic attack, nucleophilic attack, two different mechanisms because their nucleophilic attack is a popular thinking and also dissociation of EC molecules on the carbon surfaces and dissociation with oxygen extraction. And through our DFT calculations, we find for layered compounds, especially for late transition metal, the dissociation or oxidative dissociation of the EC or carbonate molecules is the most favorable. So know that you have an EC molecule. It come to the surface. And the surface is actually able to strip a hydrogen from the EC molecule and then essentially attach to a surface oxygen. And this hydrogen becomes proton with surface oxygen and simultaneously the transition metal of the surface is reduced. So we want to show some of the comparison from the calculation, in particular looking at electrophilic attack and this dissociation reaction. If you look at the energetics for these two mechanisms as function of transition metal, and you can see that for the dissociation the driving force essentially increases as we go to the late transition metal. On the other hand, for electrophilic attack, essentially it's independent of transition metal. In addition, that the dissociation mechanisms essentially when they dissociate on surface, there is exchange of charge transfer, roughly one electron between the EC molecule and oxide surfaces. Essentially the surface is being reduced. On the other hand, for the electrophilic attack, is there's very minimum of charge transfer occurring. And this also trend holds if we take lithium out of, let's say, lithium cobalt oxide as we essentially increase the covalency between cobalt and oxygen by the intercalating, we see the driving force increases for the dissociation reaction. So to put this simply, we propose a mechanism where on the late transition metal oxides with high covalency, also in the highly charged state, the intercalated state, where the EC or other linear carbonate can come to the surface and become dissociated and generate prodig species. And this prodig species absorbed on the surface oxygen can further react with the salt and generate HF. And HF and further react with oxides potentially leasing to the generation of cracks and the swallowing of the electrodes. So this is the thinking and mechanism from the initial DFT studies. So we want to see whether we can actually develop a unified thinking or descriptor that will control the driving force of this process. And we know that if we change the Fermi level relative to the oxygen P band of these oxides, which is shown as the horizontal axes, and the driving force increases. So essentially, going to the right, we're increasing the covalency of the oxygen. So that's essentially late transition metal, nickel rich, NMC materials at a high state of the charge. And the driving force, essentially for the oxide surfaces to strip a hydrogen from the molecule off of carbonates to generate a proton, the driving force, it can't be as high as a two EV or 200 kilojoule per mole. And this trend can be even generalized or simplified by just essentially computing the absorption strength of hydrogen on the metal oxide surfaces. And this is the work into collaboration with Janes Rasmizo at University of Copenhagen, where we're looking at absorption of hydrogen on various oxide implanted as a surface P band center. And very similar to what we have seen on the left, which is we're looking at the dissociative absorption of EC, there is a similar trend essentially going to more covalent compound oxide surfaces, the driving force to essentially oxidize hydrogen or hydrocarbon molecules or these small organic molecules increases. So we want to see whether this framework, this driving force as we increase metal oxygen covalency for dehydrogenation or oxidative dehydrogenation of carbonate or whether it can be supported by experiments. So we conducted in situ FTIR, this has been done by e-ray. So this is schematic of this in situ surface enhanced FTIR where we believe through this measurement and also through our measurements related to electrolyte catalysis, we are probing the surfaces or we're probing the electrolyte interface. So we can look at the CO double bound region of the EC. In the bulk electrolytes, you typically see two peaks. One is coordinated with lithium. The other one is free EC. You can also add let's say VC in the bulk electrolyte VC exhibit a unique signal. Now, if we charge the NMC materials in this EC only electrolyte as we increase the voltage a week from the differential spectra, we can see the appearance of a peak that had higher wave number relative to the free EC and lithium coordinated EC. And this peak corresponding to that of VC and this assignment is also further supported by the DFT calculations of various molecules of EC or dehydrogenated essentially dehydrogenated one hydrogen from EC or dehydrogenate essentially two hydrogen to form a VC and the wave numbers of computed and what we see matches very well. In addition, we could also form various oligomers due to these dehydrogenated species and we can observe the presence of the oligomers if we do the following experiments. So we first charge to 4.4 volts and we generate VC and then it's actually hold under open circuit conditions and you can see that with time increasing time the VC signal is decreasing, right? So this suggesting VC is soluble, right? So it's move away from the electrodes and with increasing time we see this peak at around 1813 increases and that corresponds to essentially what we computed for various oligomers formed due to the oxidative dehydrogenation of the EC molecules. So using this technique, not only you can actually see how the different species may form and also you can observe potentially what species actually can diffuse away and what species may stack on the surface. And we can contrast NMC811 with NMC111, right? So 111 surface is very stable, looking at differential spectrum and charge all the way from open circuit voltage to 4.8. We don't really see any dehydrogenated species and this is very much in agreement with the computed trend or descriptor of for dehydrogenation, oxidative dehydrogenation on the surfaces, right? So essentially as we go from nickel compound to a manganese rich compound we decrease the driving force for hydrogen absorption. So this is very much in agreement. We also know that this technique is very sensitive to the state of the surfaces for NMC811, right? So typically if we have NMC811 laying around in the lab, I will collect exposed to ambient conditions and on the surfaces it may form various carbonate hydroxide species, right? So if you heat treat NMC811, you see that it actually generates a different species relative to what we have in NMC811, right? So essentially you can see in heat treated the pronounced signal that's generated is the BC. And for the heat treated compound we have presumably removed large amount of lithium carbonate. And then if you look at only charging lithium carbonate in the FTIR cell and you see the presence of formation of oligomers with increasing voltages, right? So this is further confirms that tailoring the surfaces of oxide is critical to influence or control the reactivity between the electrolytes and positive electrode materials. To summarize what we have discussed so far we're essentially proposing a mechanism we call it dehydrogenation or activated dehydrogenation. It is the surface promoting the dehydrogenation or removing a hydrogen from the carbonate molecules and generate product species on the surface of oxide. And then these product species can further attack or react with salt, generate HF and further essentially edge away or leading to cracking of the positive electrodes. This mechanism we want to point out is different from what has been proposed for example, this attacking of, for example, singlet of oxygen proposed by Huber-Gastegger is also different from the electrochemical oxidation of the solvent. Because this oxidative dehydrogenation we propose happening on oxide surfaces as you can see from the FTIR the generation of DC can on the NMC A11 surface can start as low as 3.8 or 3.9 volt. So the question is, now knowing the mechanism how can we influence the reactivity? What will be the ways we can actually reduce the oxidative dehydrogenation reaction? So we can first control the surfaces or we can also influence the solvent, modify the solvent, modify the electrolyte to reduce the activity. So I want to show you two quick results. One is fairly straightforward. The main groups has shown that if you coat the surface of NMC A11 either with lumina or with, for example, fluoride containing compound and as you can see from this descriptor as you go from chylicobinone oxide to a fluoride based compound of cobalt or nickel, the driving force for hydrogen absorption is very low. So essentially there's much reduced oxidative dehydrogenation reaction and this is in agreement with enhanced the cycling of fluorine coated NMC A11. In addition, if you modify the electrolyte, increasing essentially the concentration of the electrolyte, salt concentration of electrolyte, it can also increase the cycling stability. So this is the work of Riyoshi Tatara. So he made the various concentration of the electrolytes of lithium PF6. Essentially when you reach roughly three molar of a concentration there is no free EC or no free EMC. And by Gerrit of essentially the free EC, free EMC is seen much improved stability. And this improved stability can be explained through a DFT calculation of Livia where if in presence of free EC, the barrier to have dehydrogenation is minimum, either from direct to proton transfer or through a chemisorbed intermediates. On the other hand, if we have all the ECs are actually coordinated with lithium in these lithium EC complex, the barrier for dehydrogenation can be as high as 2B, significantly reduce the kinetics of dehydrogenation reactions. And for both approach, right, further using in situ FTIR, you can see either you modify the surface of oxides to reduce the surface reactivity towards electrolytes or if you essentially lower the activity of free carbonate solvent in the electrolyte. Essentially in both approaches, there is no generation, we detected no generation of dehydrogenation of species upon charging to 4.8 volts in a similar cell setup as we have seen previously. So I want to essentially relate the oxidative dehydrogenation of aepodic electrolytes. In our case, that increasing metal oxidative covalency is detrimental to the battery performance, but it can also be used to design interesting materials or design interesting processes. For example, there's a large field of doing selective oxidation of hydrocarbon or organic molecules. For example, if you want to oxidize methane to methanol, for example, how would you go about doing it selectively on the surfaces? And typically in this field of selective oxidation, there are three general mechanisms of deforestation, dehydrogenation, and there's a mechanism of oxidative dehydrogenation for selective oxidation. And this is schematic is shown essentially for oxidative dehydrogenation of methane. And you can see this is essentially through the same mechanism as we have discussed for oxidative dehydrogenation or oxidation of our carbonate molecules. So I want to essentially relate the surface reactivity of oxygen that degrade electrolyte can be also used for controlling oxidation or selective oxidation various molecules or either making fuels or for chemical transformation. So this thinking can be generalized by looking at the surface energetic of different azure baits applied at this function of Fermi-level relative to the oxygen P-band center. So the horizontal axis you can think of is a measure of surface oxygen activity. This is computed, but we have also measured it and validated to x-ray emission spectroscopy for a series of prospects. So what is really interesting in this graph essentially you can see by controlling surface oxygen activity, you can control surface vacancy formation, you can control CO2 absorption and H absorption and even methanol absorption. Essentially these absorption energetics are correlated and they're controlled by oxygen surface activity. On the other hand, if you absorb these species on the B side, they're weaker. So essentially further support the surface oxygen activity play a very important role. Of course, the surface oxygen activities controlled by the covalency of metal oxygen bond. And this is another example of the surface oxygen activity can control the NO redox. So how the NO can be reduced or oxidized to NO2 or NO3 is also controlled by oxygen P-band. And again, this can be applied to control catalysts for water splitting. So essentially this essentially generalize one of our approach that is really controlling the oxygen activity by tailoring the oxide surface or bulk electronic structure. We can essentially correlate and tailor activities for a number of reactions. So we can develop a unified framework for surface reactivity or catalytic activity design. So I want to move very quickly to our recent learning in trying to develop a comprehensive framework to design of apodic electrolyte with no background in chemistry or organic chemistry. So a lot of this work is really through working and learning from my colleague Jeremiah Johnson at MIT chemistry department. So we started this work about five, six years ago where we realized that there's no known electrolyte as stable against lithium in the lithium oxygen battery. So we have to design new electrolyte and trying to develop a framework. How do you design electrolyte, small molecules that can be stable against a superoxide or peroxide? So Jeremiah came up with this thinking that if you want to design stable apodic electrolyte, got to design stability against dehydrogenation reaction. So hydrogen removal got to design against proton removal, the deprotonation reaction, got to design special, removes special sites that can have a nucleophilic attack. And of course you can also screen for electrochemical oxidation and a reduction reactions. And so with this framework, we have to study a large number of compounds or small molecules of carbonate, Esers, sulfur containing, nitrogen containing, right? So we can first do is through the computation of shooting and later on balladism experiments of Minjun where we essentially looking at these compounds or these molecules in different groups where we can look at the deprotonation energy, right? So to look at which site, which atom in a given molecule is prone to deprotonation, which atom in a molecule is prone to dehydrogenation or nucleophilic attack, right? And then that gave us essentially design principles to essentially remove the sites, we remove them in the molecular design sites that prone to, for example, hydrogen abstraction or proton abstraction or nucleophilic attack. So through this design framework, a computation, we reply this framework to design new molecules. And these are the two molecules that propose and design by Jeremiah and Minjun, essentially these molecules, and one is Sophamid and Sophamid, both actually are free of, if you will, what we would consider a vulnerable site for hydrogen proton and nucleophilic attack. And with the presence of CF3 functional groups, we can further essentially using its electron with strong power to increase its stability or electrochemical stability against oxidation, right? So these compounds, they survived tests in peroxide, superoxide at 80 degrees C for three days, essentially further support these essentially molecules are stable against these reactions. And we also put them in the lithium oxygen cell, right? So, and then contrast the stability of these, Sophamid electrolyte with lithium TSI, with lithium TSI in DMSO and also a G4, Glyme. You can see that a cycling in lithium oxygen cells, let's say for a number of cycles, the DMSO would degrade to generate DMSO2 and the Glyme would degrade to generate formate. On the other hand, this new electrolyte, Sophamid solvent with lithium TSI salt remain stable. So this is helpful. It demonstrates this is possible to design using this framework is encouraging. Now with the work of Olivia, looking at surface reactivity of ECs and varying essentially the oxide surfaces where we're only looking at carbonate molecules. And Olivia came up with this framework. Can we relate essentially the stability framework of these small molecules against radicals in the solution? And looking at the energetics of dehydrogenation, deprotonation, these small molecules in bulk, can we relate the energetics of these small molecules against oxidation on the oxide surfaces? So essentially bring, essentially stability of these molecules in bulk to essentially to the surfaces. So first tries as you look at what's the absorption energy of all these molecules on the surface? And the surface uses a lithium nickelate and plotted this function of dehydrogenation energy, dehydrogenation energy. You can see there's absolutely no trend. On the other hand, see the absorption energy for these molecules actually scales with the CH bond oxidation. So there is a correlation. So the stronger drop onto the surface of these oxide surface meaning the small molecules being oxidized is correlated with free energy of oxidizing the CH bond. And using this thinking you can actually further design new molecules. So for example, then this molecule and then projected onto this diagram they have essentially reduced the driving force for surface absorption. And utilizing the same thinking, these would expect these sulfon emit salt would also have much reduced absorption strength and due to essentially their lower free energy for oxidize the CH bond. So utilizing this sulfon emits electrolyte in collaboration with Julie's group in particular Wei Zhang's work that they have tested this electrolyte with NMC811. And the relative to the carbonate solvent is shown much greater stability and this sulfon emit salt or electrolyte has much reduced nickel dissolution. And also this new electrolyte exhibit no evolution of CO2 electrolyte oxidation charging through high voltages. In addition, with this new electrolyte the particles of NMC crack less relative to what we have in the carbonate electrolyte. So the thinking is based on the mechanism we have proposed that carbonate easily can be oxidative dehydrogenated, generates product species, HF and essentially attack and etch the particles within along potentially the green boundaries on the other hand, but this new electrolyte is much more oxidative stable against the oxide where relatively speaking less product species therefore is in agreement with relatively intact particles. And then Julie's group and also with the creativity of Wei Zhang they were able to test this electrolyte in the more practical conditions with a lower electrolyte carbon ratio and also a low negative positive light ratio. You can see also this sulfon emit electrolyte that can last much longer than the carbonate electrolyte. In addition to its positive effect on the positive electrode we also know that it appears the new electrolyte also give rise to a flatter lithium surface with less porosity. So with the sulfon emit electrolyte the lithium thickness metal thickness around 10 times less than that cycle or generated in the carbonate electrolyte. So this is something I don't understand and we'll be further exploring and try to see how the electrolyte of compositions would influence the morphology and also the kinetics of lithium plating. So I want to end the with I think it's a really exciting topic of trying to understand the complex processes at the electrolyte interface and to understand how the oxides will react electrolyte and how the ICI make form. So I want to thank the co-workers and the collaborators at MIT and also outside MIT and our financial support. Thank you so much for your attention. Jan, thank you so much for the terrific talk and the deep dive. So now we're ready for some questions but before we do that I thought I would just make a personal observation. I have followed your work since when I was a graduate student. I still remember meeting you when we invited you to give us this symposium maybe 15 years ago at Caltech. And I remember back then you were working on very different things. You know, in your own thesis work you were studying microscopy for battery materials then you moved to work on high temperature fuel cells and electrolyzers then you moved to aqueous electric chemistry looking at electric catalysis the lithium air batteries and now looking at interfacial reactions. So I'm just amazed at the breadth of the work yet they are extremely unified as you pointed out today especially in your mechanistic understanding computational methods, experimental methods. So I thought maybe you can share for a moment with our audience sort of your guiding principle when it comes around to navigating these very different and sometimes very disconnected fields and communities. Thank you very much. I don't know what to say because this is not the plan, right? So I believe when we talk about work in the past where things happened in the past we talk with certainty. Now if we look forward there is no certainty in what we do and so therefore I could not comment to say we do this by design but rather than I think maybe it's the fortune of running to colleagues that can really inspire us get very excited and or easy to work with or willing to teach us new fields and so we essentially is in some ways and pull into these areas and just by curiosity and then of course what I think will there's a lot of work maybe did not fit. I did not talk about, right? So it doesn't mean that everything fit is really by design because it's the certainty of the past. Well, that's a wonderful lesson I think if I think if I were to extract some wisdom from what you just said it's definitely be willing to listen to new people talk to new people and explore new topics and perhaps don't have too well defined of a strategy as to have a tunnel vision. So I think this is a really important message especially to many of our younger audience and aspiring and developing scientists. So thank you Yom very much for that insight into your journey. So maybe coming into the specifics of talk so there are a number of questions and I thought I would start at the higher level. So you discuss a lot about stability of solvent in the positive electrode side but obviously when you start to play with the solvent they will have effect on the electrolyte transport in terms of diffusion, stability at the negative electrode. I know that you didn't have time to talk about it today but what are some of the general consideration when we start tuning one of the knob to modify the property how you would have trade-offs for the other component of the battery? Yeah, fantastic. So in fact that this is something that we are being working on this in terms of looking at the scriptures for iron mobility. And we just begin to understand or have some understanding in this regard. And this is really, I think is very interesting because one can potentially unify the two field of kinetics or transfer kinetics. For example, Marx's theory to also utilize the stinking looking at transport or iron mobility in the liquid or polymer or in ceramics. So I think the connection is there. So it is really how do you design a certain stability or reactivity but at the same time and how would you design essentially the barriers or the free energy landscape to facilitate iron mobility? They're not the same but can in some cases correlated. On the other hand, I think this is the second part of your question is how do we design stability against lithium? And this is something is very new. We just started maybe a year or two years ago. I would say I'm really ignorant that there's a lot of people whom I'm working with and learning from collaborating with Shirley Meng, working with Baytar Gallant. And also there is a large group that's been led by Cleo Amin and also Martin Winter this US-German collaboration on this lithium electrolyte interface. So there's a lot of expertise but we don't have much to say in that regard but rather than I think it's just fascinating. I think it would be really cool to understand why and how the electrolyte can change so much the mythology and the kinetics of lithium stripping and plating. Well, on a personal level, I just think that a little bit, yeah. Go ahead, Yi. Yeah, so a young great talk and speaking of that, so particularly based on your results of dehydrogenation this is proton generated. I'm also thinking now the cathol side or generation of proton that will diffuse and go to the end now and have you looked at this communication, the coupling effect from cathol going to the end now? We have not looked at this. So, but I think this is really interesting question. If I recall correctly, I believe Hubert has been looking at some of the defects but this is something I think will be really interesting to look into. Yeah. Yeah. And also related to that, you have this beautiful, you keep monitoring these reactions, the EC. How do you activate this molecule? I really like your view from a really catalytic point of view to study this process. So then this species, once you pull this away, certainly CEI is the layer. People try to study, it's very controversial. Whether the CEI right there or just not CEI, what's really there, once you look at it, sounds like it's there, it's not there. So, I mean, it's very complex. Yeah, I also want to pick your thought a little bit about CEI. Right. So this is something that is really interesting because the CEI is also perhaps like CEI is dynamic. For example, like this highly concentrated electrolyte salt, we did, we thought, okay, if we cycle once and then generate a stable surface, we should be able to cycle in a dilute electrolyte, but in fact, it doesn't work. So then it means somehow the surfaces and resume back. And of course, simultaneously we've seen that if you cycle in these carbonate electrolyte NCA11, you will see metal oxyfluoride and nickel-axymethylfluoride forms. But this, it appears to not the passivating, it's still developing, right? Maybe this is imparted related to generating new surfaces, right? As we continue to cycle. In contrast to if you quote, it appears to be more effective in terms of slow down the dehydrogenation or generation of product species. Well, back to you. Yeah, thank you. That was actually the question I was gonna ask about CEI, but maybe let me just make one observation. I think the co-optimization of all of these different things EU mentioned, the diffusion of species to the negative electrode, I think actually makes a very fascinating scientific problem. It's a very difficult problem. So maybe coming to the CEI, one of the thing that I thought it's very difficult to understand when it comes to electro-electrolyte interfaces is heterogeneity. So, Jan, you appreciate from the electric catalysis community, say in aqueous systems, that there are a range of active sites and so forth. And when we look at cathodes, I think one of the most strange things is that folks have done all kinds of coding and some of them work really well, but almost all of those coding are not conformal completely. They're always exposed sites. Sometime it's coded in a very non-conformal manner. It's particulate, but they work really well. So can you give us some insight to how to think about the coding or whatever decomposition you have, which does not occur conformally yet, it tends to say protect the entire cathode particle. I think this is really one of those really strange mysteries, at least to me. Yeah, I think that's a fantastic question. So I don't know the answer, but I can speculate. I don't know. For example, I think the process is really dynamic, right? So as we've seen that transition model can resolve in the presence of HF and transition model can form a metal fluoride and a metal oxy fluoride. And we know if we coat with fluorides, the stabilizes. So I think it's really a dynamic process, right? So let's say you coat alumina, right? It's like you said, most of cases really not monolayer full coverage. But the other hand, it generates codex meshes, right? And so then I think there is essentially this two processes, potential competing processes, one is dissolving the metal at the same time, it's also forming a protective layer. And I think it's really a matter of how well it can cycle. It really depends on which process it potentially is faster. Yeah, go ahead. Please, I was just going to say that a related question is also you talk a lot about how the metal oxide can affect the electrolyte. But of course, the opposite is also true. I think you hinted this, for example, in the microstructure degradation. So one question is on can the solvent also affect things like densification at the interface? And you just mentioned, for example, dissolution, which leads to the formation of metal vacancies. So this kind of a forward-backward interaction between the electro material and the electrolyte also seems quite interesting. Can you comment on sort of the processes and mechanism in which the electrolyte affects the electrode? Yeah, so I don't know if I fully understood your question, but if I try to answer, essentially look at this image here. So if you have this NMC cycle in the carbonate and with time after hundreds of cycles, you see cracks, right? And presumably this gave rise to the swelling that Jeff presented earlier, right? So essentially when you have a carbonate electrolyte generates a product species, it appears to be able to go through attack along the green boundaries and generate new surfaces and fresh surfaces. On the other hand, if you through electrolyte design, presumably with this new electrolyte, we have much less product species because we have minimum solubility of the transition metal and then you see the particles, the polycrystalline particles retain largely intact. Yeah, I think that's absolutely the right thinking here. And additionally, I was also thinking, can the solvent affect the thermodynamic energetics in the cathode? So if you change the solvent, can you for example, change the migration energy or can you change the stability leading to reconstruction of the interfaces, just borrowing a lot of the similar concept from electric catalysis that the catalyst is also modified by the solvent environment as well? That's a great question. So that actually is really exciting to study in the chemistry that involve involving redox species that are soluble, right? So let's say if you look at the redox couple of oxygen to superoxide where superoxide is soluble, if it's something soluble, then the energetics free energy can be modified by solvent molecules by ions around forming complexes. And that in the past work, we have seen as varying solvent, you can tune the thermodynamics, the energetics up to 0.2, 0.3 ED. On the other hand, you can also tune the lithium redox potential up to 0.4 ED, meaning the lithium typically has a stronger interactions with the solvent, right? So typically if you go to a higher donor number solvent, it would solve the lithium more. And so essentially you can build a lithium lithium battery of 0.4 volt, right? This is the same lecture, but two sides of immiscible solvents, for example. That's generally, I think it's pointing to your question, but typically that's a very small amount of energy that one can capture. Yeah, I agree they're small, but I think they can still be appreciable. So I think this would be an exciting area to think about. Maybe one more technical question before we have maybe a broader discussion. So in the beginning of your talk, you show some really beautiful computation of absorption energetics, dissociation energetics of carbonates on metal oxides. Later on, you also showed how these energetics are modified when you go from say, NMC-111 to nickel rich compositions. So the question is how do you think about the effect of the oxidation state of the transition metal? So when you change the nickel content for the fully lithium state, the oxidation state is changing when you change nickel content. Is that the dominant driving force for affecting the energetics in terms of the oxidation state? And related to this question, when you remove lithium, right? So when you charge the battery, then you also have the oxidation state once again changing. How does that change in the state of charge then affect the energetics? So this is a function of composition basically, an oxidation state as a function of nickel, as a function of lithium content. Great, great question. So this is what we aim to explain with this descriptor of oxygen P band, from your level relative to the oxygen P band, because they take into all these factors, as you mentioned, right? By changing the D electrons and also the oxidation state of this D electron, of the D metal, right? So you can see that as you vary the surface P band and you can see that essentially as you take more lithium out, the oxygen P band, you can see it's plotted relative to the formula also shifted to the right, right? And as you go to later transition model, it shifted to the right, right? So essentially there's number of factors, right? You can actually, they're compounded and then they essentially shown essentially this type of linear correlation with absorption energy of hydrogen on the surfaces as function of oxygen P band center, right? So then if you say, okay, what does it mean, right? So why doesn't you have this correlation where you go from early transition metal, fully litigated to the late transition metal, fully dilated while there is a driving force of two EV of difference, right? And you can think of this difference is that the surfaces of nickel O2, right? So nickel O2, let's say if there's ever existed a nickel four plus, right? It really doesn't want to be nickel four plus. So then when you have absorption of hydrogen or absorption of EC molecules, essentially the transition model wants to be reduced to essentially three plus or two plus. And that's what essentially the process of oxidative dehydrogenation, right? So it is either through a hydrogen absorption to generate a protons that electron that hydrogen lost is actually given to the transition metal. And that essentially is the really the driving force. This is- Can I interpret this linear relationship here as that the electronic effect is dominating here because you can explain everything basically in terms of the band center and oxidation state essentially. Yes. This is very exciting finding. Thank you. But E back to you. Yeah. So Yang, thank you for sharing about the insight. I think this we touched upon quite a bit already but let me ask again maybe from just very big picture every time I think about electrolyte this is a really multi-parameters space optimization need to consider melting point, the boiling point, the viscosity, electrochemical stability the ability to solvate lithium salt the annual reduction stability the cathodic station stability. You have been working in the area for a long time. How do we, how do you go about to start to think about well, how do I design the electrolyte? You know, what do I need to look into first? You know, step one, two, three and how do we do that? I think the audience right here will be interested in this question, you know electrolyte sounds like an area you kind of, you know, trying by arrow, trying a lot and then try to make a sense out of it much, much harder to make progress. But after so many years of learning, you know certainly at lithium ion we're still in the carbonate that few molecules are solid in that, still in that regime adding some a lot of additives. So how do we go about think to design the electrolyte? So can you share some of the thoughts we saw? I know it's not an easy question. That's why it's at the end of the panel discussion. Thank you for this great question. Yeah, so we've been thinking about how to design electrolytes for some time maybe five years, but we haven't had any publications in this regard just because I think there's overwhelming amount of composition and processing and then performance. And we're trying to develop some sort of a universal descriptors for ion mobility or ion conductivity for liquid, polymers and electrolytes. And I would say we have, we are still in the process to try to take what we, the little bit we learned and apply and trying to design new electrolytes. So I want to say that I believe many of the properties that we talk about, for example, viscosity, melting point and ion mobility, they're actually, we believe that they're physically correlated, right? So they're actually could be potentially governed by the similar energetics that gave rise to similar viscosity or similar viscosity trend. So very often you see the conductivity trend to make it correlated with viscosity, right? And so what we try to understand that this time, I should mention that the work we've been actually doing is supported by Toyota Research Institute for some time. This is in collaboration with Jeff Grossman, MD person, Arafa Bambra Raleigh, and he's a machine learning person in Jeremiah and Adam Willard. So essentially we are taking a combined high throughput and also machine learning approach, not only generating our own data, but mining the data from past literatures and trying to see if there is a correlation or a map that we can see there's some directions that we can learn about physical intuition or tricks to design electrolytes. And I would say we're still in the midst of doing it, but it's really, I think it's really fascinating, especially we're interested in looking at how to think about the ion mobility, essentially ion in liquid or polymer when it move is also activated. And how do we think about design the barriers through really the salvation structure and the dynamics? And how do they even correlate with, for example, of what we typically think about in catalysis or electrolyte catalysis in terms of designing the activated, designing the barriers. So that's really the direction we're interested in. Hopefully we'll have something interesting to say in a few years. Yeah, thank you for sharing. Back to you, Will. Thank you, Yi. Yeah, I think our time has come into end here and I think Yang Nosa has a hard stop. So I just want to, again, I'm so impressed by the talks today because we went from devices to microstructure to molecular chemistry in just a little under two hours. I certainly learned a lot. I think it also points to the complexity of lithium ion batteries in general. I'm actually surprised the battery works at all. This is so complicated. So I think this is all a mystery. I'm sure Stan Wintingham who is usually in our audience can appreciate this. Justin, if I can have the slide. We have a very exciting event coming up in two weeks two weeks from today, a Friday, 7 a.m. Pacific. And the topic is going to be storage X where X is equal to long duration storage. We already had one such symposium featuring Micah Z's at Harvard and also George Crabtree at Oregon. And now we're going to have a second one and the second long duration storage discussion will involve colleagues from industry. And specifically three very promising startups working on different aspects of long duration energy storage. We will have the co-founder of Energy Vault, Andrea Padareti, who is looking at mechanical energy as a way to store electrical energy in the grid. We will have Yorick Hanomen, who is the CEO of Intervenu, looking at chemistry for long duration storage. And then we will also have a very different talk from the co-founder of form energy on the importance of understanding the grid. Because only by understanding the electrical grid then you can understand the value and the cost of energy storage. And that will be given by Marco Ferrara. So I hope you will mark your calendars for two weeks from today for this, what I think to be a very different and very exciting discussion around long duration storage. And again, I'd like to wish everyone a great new year in 2021 and thank you very much for joining today.