 Good morning from Stanford University. My name is Will Chu. And together with Itui, we are delighted to welcome you to the fall quarter of the Storage X International Symposium. Today, we have a very interesting technical deep dive into a very timely topic, which is the development of lithium metal negative electrode for the next generation battery technologies. And to get us started on this topic, we have invited two academic authorities on the topic, Professor Julie from MIT, and Jason Zhang from PNNL. So let me get started first by inviting Ju to the stage. Terrific. So Ju is a professor of nuclear science and engineering, and also material science engineering at MIT. He's actually our counterpart at MIT, where he leads the energy storage effort in the MIT Energy Initiative. So it's a good gathering of people today. And Ju has done so many things. If you go to a website, you can find his very diverse interest. He's an experimentalist, but he's also a computational material scientist. He works on almost anything that involves atoms and has made very outstanding contribution to the battery field. I think, Ju, I've never told you this. One of your early work in the battery field actually got me started in the battery field myself. In 2010 or 2009, I think 12 years ago, Ju published a seminal paper showing the in-situ T-M movie of a tin oxide nanowire undergoing battery cycling. It was the first of many papers that uses T-M to understand dynamics. So when I first read that paper, Ju said, well, this is exactly what we should be doing. And 12 years later, there are so many interesting applications of in-situ observations of battery dynamics. So Ju helped kickstart that particular effort. So that was a very inspirational and seminal work. Ju has received many awards, notably, he has received the highest honor in the US for an academic, which is for early career academics, which is the Presidential Award from the White House. He's also a fellow of the Materials Research Society. He has received the MRS Outstanding Young Investigator Award, the TR-35 Outstanding Investigator Award from Technology Review at MIT. He is a fellow of AAAS as well. So there are too many awards to list. So without further ado, let me ask you to get started and he will tell you about lithium metal. Ju, please go ahead. Thank you for that very nice introduction. And I will, it's really great pleasure and honor to start this semester's storage access symposium. So I like to talk about the topic actually that both Will and Yichui have done a lot of work and I took great inspirations from. And I took a little bit peculiar title, lithium metal engine, and just want to say that for solid-state battery, there are three well-known problems. One is you've got to maintain a contact, so ionic, electronic contact. And so that's the reason generally there is a few megapascals of stack pressure. And then there is a issue of fracture and from the Nernst equation, we know the lithium metal, each atom have this much volume and that converts to for every 130 millivolts over potential. If not relieved, then you can accumulate up to a giga pascal of pressure if you have a little surface flaw here. So lithium ion comes through the solid electrolyte and it's plated here, but it needs volume and so you have to squeeze the lithium back out, otherwise this kind of stress could accumulate, but generally the solid electrolyte could be a brittle material. And the last problem is electrochemical corrosion. And if you look at the electrochemical splitting window of well-known solid electrolyte like LGPS or ZO, et cetera, they are very usually not so stable at very ductive voltages. And first of all, this very nice work from a Gert Zetter's group is only on the bulk material. So if we have grain boundaries, then they would actually have distinct electrochemical splitting window from the bulk phase and they could be selectively reduced. And so generally two and three give you stress corrosion cracking problem. And this is a very nice recent work by Professor Meng Gu from China and you're looking at a crumbled solid electrolyte husk. So this was actually formed in ether-based liquid electrolyte, but what happens is if you take the lithium out, you see that this SEI, which is an ad hoc solid electrolyte shell, basically it crumbles. And so there's clearly kind of mechanics issues going on in this metal. So I took this somewhat strange title because our goal of doing battery research is basically defeat this beast, right? This is an internal combustion engine that use fossil fuels. And the working fluid of engine is air in this case. And so a provocative idea is when we deal with this in metal or alloy, the working fluid so to speak could be the engine of this solid state battery. Now, just to sort of illustrate what I mean, this is an experiment where people took gold at room temperature. They made a so-called mechanical break junction that tear it apart and then they just leave it in air for four months and gold is air stable. But then what they see is that the profile retracted by something like 20 nanometers over a period of four months. And what we have done is to show that this can also happen for sub 10 nanometer silver. So this is just a four nanometer silver nanoparticles. And it shows this kind of a liquid like plasticity even though at any time it still maintains a crystalline diffraction pattern. And at room temperature, it's really the homologous temperature which is normalized by the melting temperature is pretty low for gold and silver. So if they can manifest some kind of quasi liquid like behavior, the question is, what can happen for lithium? Room temperature is actually 66% two-third of the melting point, bulk melting point of lithium. And so about eight years ago in collaboration with Girei Shan and so at all, we have looked at actually another material tin with a pretty similar melting point to lithium. And this show that when we tear a tin ligament, the final stage looks like a liquid maniscus. You get this almost liquid like feeling. And in fact, when we really zoom in this is a diamond punch, there's some carbon here. But when we push this tin inward, there is a displacive plasticity. But when it recovers, you see this very smooth, gentle recovery which is driven by diffusion and not by dislocation motion. So in 2013, we have discussed this size dependent deformation mechanism. And so generally we have this smaller, stronger trend for dislocation plasticity. So the prediction is as you reduce the ligament size or grain size, the stress would actually go up in the whole patch relation where this is actually minus one half power. But what we have observed and said in that paper is once your size is below about 100 nanometer, then you would trigger this cobalt creep with a very different exponent. So instead of minus one half, the exponent is actually plus a third. So we predicted that smaller will not be stronger but will be much, much weaker. It's a very dramatic weakening trend. And this was later verified also in intermediate temperature materials like aluminum. So in this case, what we are doing is actually have a heated substrate at 400C and so the reduced temperature is from 70% to 50% for copper and with a cold tip, we can draw the metal wire across several microns and even in silver as well. So when we look at this deformation mechanism map that's developed by Ashby and Frost, what you generally have is the horizontal axis is this reduced temperature where you have T over T melt. And for lithium, we're saying it's actually in this region. And then the stress is the deviatoric shear stress normalized by the shear modulus. So lithium have a shear modulus of three gigapascal. So this would be three megapascal and this will be about one atmosphere, okay? So the question is, can we drive lithium metal motion with less than one atmosphere kind of pressure difference and give some kind of a creep like behavior. So called a liquid like behavior even though the bulk melting temperature is still above the current temperature. So let's say that that's a possibility but then the second question is, what can you use to contain lithium metal if it's a working fluid? Because it's a very corrosive fluid, as we know. It corrodes pretty much all the, our favorable solid electrolyte. And it also, even though we hope the stress is small the stress as I'm going to show can reach tens of megapascals. So we have to design for like 100 megapascals stress in the lithium liquid or solid working fluid. And so in this review with Professor Surin Zhang we basically say that because you have a limited choice of good high conductivity solid electrolytes which are absolutely stable against lithium metal as zero vote we have to think about other kinds of solids. And this actually I want to say of all set is not a new idea. Actually people have been doing this, Professor Yixue have been doing this for at least five years before we think about this even. But we're just sort of categorizing this behavior. So generally if you have a solid which is an electron insulator but a lithium ion conductor then that is in this quadrant. If you do not conduct lithium ion or any ion but you conduct free electrons then you have metal. And generally we know that solid state battery you generally have to have a metal current collector. You're working with lithium metal and you've got to have a solid electrolyte separator. But then there is another possibility which is you can have something which conduct both electron and the lithium ion. And this will be a mix ionic electronic conductor and you can work with this to build your solid state battery. And there is even a force category which is an electron insulator and also does not conduct lithium ion. So we call that Eli material. And so the point of what we initially try to propose in the previous paper is to say that there are not that many good solid electrolytes which are absolutely stable against lithium metal but there are actually many, many mix which are absolutely stable. So they'll have no problem when you put it side by side with body center cubic lithium metal at zero volt. And they can even have electronic conductivity and some kind of ionic conductivity. In fact, the rule is that if you take a ternary phase diagram or quaternary phase diagram all the terminal and the member phases with a direct flight to lithium BCC phase will be absolutely stable. There will be no side reactions, no SCI in naked contact with lithium metal. Where of course any other compounds without a direct tie line they would actually decompose into these phases. So we all know Professor Zhang good enough. So my ex postdoc Yunmin Chen when he was working in Zhang's group developed this electro spinning method where he can make this kind of carbon hollow tubules with an inner diameter of about 100 nanometers and a wall thickness of 10 nanometers. And we're just thinking is it possible to first leciate this to the terminal phase and then guide use this as a rail to guide the motion of lithium metal and what is the nature of that motion? And so what we have shown you that if on this side you have a solid electrolyte on that side you have metal we can actually have lithium plating and stripping all within this tubule across a distance of 10 micron for 100 cycles. Now lithium has a very low Z so you don't see a very strong contrast but you probably can still see something well actually fast forward in the movie and this can happen like an engine piston for 100 cycles without any damage. And when we play with kind of normal speed and in this case is stripping we see something very peculiar which is we see actually this kind of fast setting which indicate to us that actually what's moving inside is a single crystal. Now of course in all these Institute EM studies you always worry about electron beam damage. So in the case of the team example we have a systematic studied electron beam damage and found that was insignificant. So what we have found in the case of lithium is first of all, we did see the body centric cubic lattice parameter fringes. And number two is that in a control experiment for lithium metal which is outside of the tubule with this kind of beam current it would occur amorphous from crystalline to amorphous in 86 seconds. But for the crystal that's encapsulated in the carbon hollow tubule it will stay crystalline for 20 minutes. So if inside actually protects the lithium metal. And when we vary the voltage this is the very first half cycle where we elicited this carbon hollow tubule. And then once we over elicited we can have this void volume start to pick up lithium metal and then we can have this piston like motion and generally the plating and stripping over potential that we can measure is on the order of a hundred milli electron volt. But of course that includes all the other losses in the system and not necessarily reflecting the local over potential because if it really is a hundred then we are going to have like a gigapascal local stress which is pretty challenging for the material. So the general idea is that if we can use a beehive of meek then a benefit is we can always keep the lithium metal in electronic and ionic contact. Then we will not have this dead lithium problem where it loses a contact with the current collector and with ionic percolation pass. So this will be maintaining the percolation and also because of the reserved space then we would relieve the stress and then there is no need for a few tens of atmosphere or pressure, a stock pressure to maintain the contact. And lastly is that if we use the absolute thermodynamic stable mix then they might be completely inert when you make a contact. If let's say there is some sliding motion then we wouldn't have SCI or SCI debris. Now later after the work from Samsung we have realized that it's also possible to have this inverted configuration. That is if you induce the lithium metal nucleation by let's say silver which actually have beautifully shown a few years ago as well as using a room temperature coarsening then the lithium metal can be lured away from the solid electrolyte side and can be mostly stored on the current collector side. And so that's actually the more desirable configuration. And then we use this big phase to coarsen the lithium phase away. And I also want to say that this meek surface can also be the site for the charge transfer reaction because you can have a reduced, let's say a lithium ion meeting an electron in the meek and having to reduce the lithium-ad atom. And this lithium-ad atom is not a BCC phase and it can do the surface diffusion and you can coarsen the BCC phase. So this can actually be part of the charge transfer reaction. So let's look at some numbers. The first question is what kind of stress are we talking about to have a decent strain rate? Because if you look at the classic deformation mechanism map these contours are creep strain rates and you have a very low strain rate something like 10 to the minus 10 per second. And you use a very low stress if you plug in the numbers but this is not something we can use, okay? And so if that's the case then we have to go to higher stress in which case we're going to have a hybrid diffusive displasive process where you're going to have dislocation glide and dislocation climb which are controlled by boundary or bulk diffusion. And that would be more stressful to the surrounding structures. So that's a possibility too. However, what we noted was that this diagram was for a coarse grained nickel. So that was for a 0.1 millimeter grain size nickel. And Professor Kobo have derived this creep strain rate to be inversely proportional to the third power of the grain size. And so if we can reduce the grain size or the patristic size by three decades from a hundred micron to 0.1 micron which is a hundred nanometers then this instead of 10 to the minus 10 will be 10 to the minus one per second. And that will be serviceable for battery because that just says you can extend by a hundred percent in 10 seconds and that could be useful. And so we plug in the numbers. We can have power law creep in the bulk lithium metal. We can have bulk nabar herring creep also in the bulk lithium metal. We could have diffusion in this MIG BCC interface which is only two angstroms, this incoherent interface or we can have lithium ion and electron transport in the mix on electronic conductor which is 10 nanometers. So all of these pathways are possible. Now, what we have asserted is that you do not need a dislocation creep if your diameter is a hundred nanometers. And the reason we see this is that if this side is a solid electrolytes and then we are stripping lithium then this region is vacuum. And so there's no way you can have dislocation power law creep here. And all the lithium must be sucked through the MIG or by surface diffusion to the solid electrolyte. And so it shows that it's not absolutely necessary to have sliding in the system. And another thing we have seen is very often these tubules have obstacles inside. It's not fully smooth but we see that this lithium can actually climb over obstacles as long as they're not fully closed. And so that again seems to suggest that there is a strong liquid like diffusive flavor to it. And when we plug in the numbers for the bulk concentration for the bulk diffusivity both in the beta phase, the BCC phase as well as in the bulk MIG phase as well as for the interfacial diffusion also for the surface diffusion. Now we don't actually know the interfacial diffusivity at that time we can model yet but we didn't quite know for sure. So we made this actually probably a bit optimistic approximation which is to say the interfacial diffusivity is similar to the surface diffusivity. And then use a universal surface diffusivity formula that in the previous works we have validated ourselves and other people have used this. And then we basically come to the conclusion that you can once this pore size is 100 nanometer you can have a interfacial diffusion dominant transport mechanism that overwhelms the bulk MIG diffusivity. And that's a pretty liberating fact because that means that basically all MIGs could work and because we don't rely on the bulk diffusivity we just rely on this interfacial diffusivity and if it's lessophilic then it has a chance to be a good enough channel. And so from the transport we have a design criteria which is okay let's say we want to have three milliampere per centimeter square of current density as well as three milliampere hour per centimeter square of capacity. Now this would require us to have a depth of 15 to 20 microns to make room for the BCC phase. And then we don't want to have a local over potential exceeding let's say 15 millivolt because that's going to over pressurize our working fluid. So that's going to force us to have a effective lithium diffusivity of 200 millisiemens per centimeter square. And then we have certain porosity so we plug in the numbers and see that these generally can satisfy this kind of reasonable engineering requirement to make a bulk cell. So that was in our paper. Now there is also a issue of mechanical robustness because we initially just intuitively hypothesize that you cannot just have few layer graphene as this meek because it's too easy to crumble. So we need something like 10 nanometers to have and also this honeycomb geometry to have some kind of stiffness in compression and in tension. And one reason is in real manufacturing we're not going to have real vacuum so we're going to have some kind of inert vapor in here. And because this is like a piston it's going to compress the vapor maybe by factor of 10. And so we're going to have mega pascal level of pressure inside and also generally people may like to have some stock pressure. And so generally you need some mechanical rigidity to this open porous nanostructure. And I will talk about this later. So after our work is published there is this tremendous work from Samsung which used a nano porous carbon silver interlayer. And so that qualifies as a meek interlayer because silver would form inter-metallic compound with lithium and carbon is also a meek. And it regulated the deposition of lithium metal on bottom. Now this still requires a stock pressure. And also there is very nice work from my colleague Professor Yeming Chen, ingenious idea of using sodium potassium eutectic liquid as an interlayer. So both in both cases these are mixed on electronic conductors they conduct electron and lithium atom if you will. And they would relieve the stress, you know this is liquid, this is open nano porous and maintaining contact. So I do think that in the solid state battery you will use these meeks. There is an issue of if you use the solid meek how you are going to anchor it firmly in the solid electrolyte. There is what we call slippery footing problem because if the meek conduct electrons and the solid electrolyte conduct lithium ion then they could nucleate a BCC phase here if they can overcome the interfacial adhesion and you could have this basically an interfacial crack a wedge that's formed here. And then because this is very soft you know this could be pulled out of the socket. And so that is a problem. And so in our design we figured you may need to use this last type of solid medium which is an electron and lithium ion insulator and it's completely inert. And it just serves as a mechanical binder to the solid electrolyte and to the meek and it just holds the meek interface in place. And so in real engineering structures you can never have perfect alignment. So what if your solid electrolyte is taller than the Eli binder? Then in that case you could nucleate this at least the BCC beta phase here and it's going to grow, it's going to push out the solid electrolyte it's like a gingivitis, it's like a gum disease but then it's going to come to the root and hopefully it's going to stop there that adhesion crack hopefully should be stopped if this Eli SC interface is lithophobic enough and strong enough. And so this Eli is kind of like this insulators that you use to build our high-voltage grid. Most of this of course is metal to transmit electrons but without these insulators it's not going to work. So you could also have a case where the solid electrolyte is lower than the meek. So in that case you couldn't nucleate the beta phase and you need something like a spark plug using the engine analog or you could have some initial shorting between the meek and the solid electrolyte so you would nucleate the beta phase here. And then once you have this then you can keep going because then in all the other places now whenever you have the beta and the solid electrolyte interface that you can nucleate a new beta phase. And by the way, one reason that you could have the beta phase on the current collector side is that we have seen when you strip you are taking lithium first from here. So even though initially even though you are not nucleating here gradually it could migrate later to this side because you're always stripping from this side and the battery can still work. And we have utilized the materials project to search for Eli candidates. And so it turns out that there are these pens of compounds with band gap greater than three electron volt and have no lithium. And furthermore, they're not afraid of lithium. For example, when you put the brilliant oxide side by side with lithium metal lithium metal turns out is not able to draw up a brilliant oxide of its oxygen. So that's pretty interesting. And so you can have these as binders. And in our effort, initial effort to scale up we have made all kinds of this kind of nano porous open pore structures. We have tried anodized alumina using it as a template to make carbonaceous structures which are a few centimeter by few centimeter by tens of microns deep. We have also used micro fabrication to make the second mesh. Unfortunately, it didn't work. One is because it's too coarse but also that upon leciation is like a shock it crumbles and not really robust. But eventually we are successful in making these carbonaceous open nano porous structures which are reasonably robust. And also with inorganic ceramic type and actually very robust meek structures. And in collaboration with Samsung and now there is actually big advance in making this kind of open pores structures. And I'm not going to bore you with the details of the scale up effort just to say that the TEM work showed that we can have the alkali metal grow or strip in channels as a single crystal. And if it's below 100 nanometers and 20 microns in length. So that actually has a pretty long aspect ratio. It's like a gun barrel. It's like aspect ratio of 200. And you can actually have this kind of large area meek beehives which you do not need any stack pressure and can cycle up to 200 cycles like the Samsung colleague have shown with a very significant capacity like five milli ampere hour per centimeter square. You can maintain constant ionic electronic contact so you never have dead lecium. And also there is no SCI in most of the compact areas. And so there is no SCI and so there is no debris forming and the reserve space can relieve the stress. And then finally we're still working on this which is how do you have a good interface with the solid electrolyte. And in the last sort of two or three million I just want to come back to this mechanical issue which is if you look at, you know, search this word cavitation damage in pumps. It turns out that even water can damage stainless steels like this can cause all this kind of cavitations when you subject to the working fluid to tension. So we actually have a hypothesis which is that the reason that people need the stark pressure of a few megapascal is to keep the lecium always under hydrostatic compression or at least it's below the cavitation stress of the working fluid. Like we said, you know, a lot of time when you do the deposition so when the over potential is negative you would generate, let's say one millivolts would generate 7.4 megapascal of hydrostatic compression. But I think people paid less attention to stripping because when you apply a plus one millivolts then you can generate a tension of 7.4 megapascal like when you pull a syringe you can have cavitation in the liquid. This turned out to be a problem for trees. In fact, this may limit how the tree can grow how tall it can grow. The tallest tree is in, you know the beautiful redwood tree in California is a hundred meters and you're going to have when you come to the top by capillarity plus one megapascal tension in the sap of the tree, xylem. And so there are actually biologists who study liquid water cavitation in trees and they say that usually the trees are exempt from cavitation unless there is extreme drought or when the sap freezes. And so what they show here is like 20 different trees but when they have a 50% loss of conductance I find that very interesting because, you know, they also use the word conductance and it turns out that, you know this P50 about, you know minus five or five megapascal tension is when a lot of the trees would die because their xylem would actually fracture like this. And so if we plug in, you know some elementary calculations of the surface tension for lithium 0.4 joule per meter squared then if we have a nano structure diameter of a hundred nanometers then just this surface tension can create a tensile stress of 16 megapascal just from the Yang Laplace equation. You know, if this is zero then inside you're gonna have a tension of 16 megapascal. And people have seen that bulk water would cavitate at 20 megapascals. It's going to nucleate, you know bubbles inside overcoming the surface energy. And then because, you know bulk lithium metal is about 10 gigapascal bulk modulus so we roughly estimate that the cavitation tension for lithium metal could be a hundred megapascals provided, you know that your solid structure can even survive that stress. And we have evidence and this is in collaboration with Professor Huo Li-Yin that generally the solid structure is going to crumble at that kind of stress. So what he showed was that during stripping there is a critical war sickness. Now this is an epoch solid electrolyte is lithium oxide. And what he found was that there is a critical war sickness to a radius ratio above which is not going to cavitate. The cylinder is going to stay its original shape below which is going to crumble like this. And also shown previously in the cryo TEM showing that crumbled SCI layer. And from this we actually can infer an SCI a solid electrolyte strength of about 300 megapascals from this experimental study. And so just to end the talk is the idea is that, okay we're going to build at least in this approach without any stark pressure a solid structure made out of the current collector the solid electrolyte, but maybe a lot of meek and a lot of, I mean, and some are seeding agents like this Eli material to have a room temperature lithium metal engine with, you know, lithium metal in and out. And the goal of course is to defeat this beast which is, you know, burning fossil fuels. So with that I'd like to stop here and answer any questions you may have. Jew, thank you for the fascinating talk. Let's take a few moments to answer some questions. So Jew, maybe I'll start with my own question first. So you really highlighted the importance of interfacial transport or surface transport in the mix ionic and electronic conductor. So do you think at the end of the day it is possible even not to use a bulk mix ionic electronic conductor and simply rely on the surface for ionic transport? In other words, what is the relationship between surface ionic conductivity and the bulk ionic conductivity in these systems you're studying? Exactly. So in fact, those calculations we have put in the papers was actually trying to propose that. In fact, that liberates the choice of the bulk mix and all you need is some lethal felicity. So the interfacial transport or wetting with lithium metal you could have a monolayer of lithium that wastes the surface that becomes the key. And there is also a geometry design in this because it doesn't have to be 100 nanometers it could be 20 nanometers. As long as it's not closed the pores, so it's open pores. And in some sense we're proposing some kind of mesoscopic quote unquote artificial graphite where instead of the atomic graphite layers where you intercalate lithium atom we have these channels, long channels where you can intercalate this second phase which is the beta phase by exactly what you said this interfacial transport. Right, so I think this is very liberating because then essentially the backbone can just mainly be responsible for conducting electrons in the bulk as opposed to also lithium. And in terms of the transport mechanism for surface diffusion, I think you have implied it is more of a atomic species rather than ionized lithium. Is my understanding correct? It can be both. So you can have an electronic transfers number and ionic transfers number. Now when the T goes to the electronic dominance and you're more like a methyl and then when you go to lower to zero that'd be more like a solid electrolyte that you can actually have a gradient structure. So you can by tuning the transfers number you can actually have lithium ion let's say if you take a standard solid electrolyte and you cut the surface you can have lithium ion surface diffusion I think mostly ionic but then as you go towards more towards the current collector side and then you can do the electronic transfers number at some point that lithium ion will meet an electron and gets become a lithium atom and you can have your chart transfer reaction right on that gradient region when you turn from ionic to electronic. Right, so maybe just a further question on this. So you described the conversion of lithium ion to lithium by the ratio of ionic versus electronic transport, right? So you have the two meeting and that's where the electrochemical reaction happens. I was actually more- I want to stop here. I want to say that there are two things. There is the neutral lithium atom and there is also the BCC phase. So I want to separate these two concepts. So the lithium ion to a lithium add atom is the chart transfer reaction. But then that chart transfer that add atom can still diffuse many microns and get to a BCC phase. So there is that distinction here. Right, I guess the more fundamental question I'm asking is what is the surface diffusivity of a adsorbed lithium ion versus an adsorb add atom for lithium on most of these misconductors? Yeah, atom typically is faster. So if you take a random material, if it's this acidic then the atom typically is faster. Although I think it's unknown, our favorite solid electrolyte like lithium fluoride, the surface, you have a lithium saturated surface. What kind of lithium diffusivity is that? So it's actually unknown and unexplored. So that's a great actually research question. Great, and let me just ask a few questions from the audience. So how much versatility does this approach have with respect to the type of solid electrolytes used? Yeah, that's a great question. And I think in my view, we may not actually end up with just a single solid electrolyte. As you see in Gertzader's chart, you know, there are some like LROZO which are relatively stable on the low voltage side. So we may actually have two layers of solid electrolyte. But I will say that in our paradigm because we are picking the stress away from the solid electrolyte because we don't even want to have the beta phase near the solid electrolyte. And then if there is any motion, if there is any slippage we're trying to minimize, it's more with the meek layer. Then I think that does reduce the chemo-mechanical stress that does reduce the stress quotient cracking on the solid electrolyte. So I think this does help in other words. Terrific, Julian. One last question from the audience. So you talked about interfacial transport and this question concerns the interfacial adhesion as the lithium metal grows within the cavity. So can you briefly also comment on the adhesion aspect in addition to the transport aspect? Yeah, so even very basic things like a graphite, there are some people who say it's a lithophilic, some people who say it's a lithophobic. So we have a paper recently that discuss, this is very dependent on the partial pressure of the atmosphere and how you do the experiment. There's a lot of compact line hysteresis. It's a great question because usually the surface is dirty and you really need to have a reducing enough environment to really show the true lithophilicity. And there is actually a lot of room for engineering. You could have a meek just like Will said, which transmits momentum. So it's stress bearing. It also transport electrons. But then on the surface you could have a coating, just very thin coating, which is lithophilic. And you have the desired surface chemistry which conducts the lithium ion and also facilitates this charge transfer reaction. And so you can do a lot of surface engineering and nano engineering on these meek structures. Which I think Professor Itui already have done in a lot of his previous works, he's just not putting in this type of theory language. And it's a very versatile platform to very, very interesting. Maybe in the interest of the time, there are many more questions, but maybe we can switch to Yi who will introduce our second speaker. Let me, well, very exciting. So let me invite Jason to the stage. Let me introduce Jason Zhang. Jason is a long time good friend. He's very well known in the battery field, particularly for his development of electrolyte. For lithium sulfur batteries for lithium metal anode. And he has pioneered many concepts. I learned quite a bit over the years with Jason either through his paper or through the interaction with him. Now Jason is a PNNL lab fellow. Very prestigious title right there. And through the battery 500 consortium, Jason and I and certainly others have been working together quite a bit. Today is absolutely a great honor to have Jason to tell us about his work related to lithium metal batteries, particularly from the electrolyte designs standpoint. Jason, please. Good morning, everyone. This is Jason Zhang. And first of all, I would like to thank Professor Yixing and Vio Chu to invite me to give this presentation to have this opportunity. Today, I would like to, I mean, early this morning the Professor Zhu Liu has already gave you a very good instruction on their work for lithium metal anode, especially using MaxCorp studies simulation and also some sort of data studies. I would like to discuss the lithium metal problem, the lithium metal anode from another angle, mainly from the liquid electrolyte, non-acres electrolyte. So you may have a more complete view on the performance of lithium metal anode in this field. Here is the outline of my talk. First, I will give a short introduction about the opportunities and the challenges on lithium metal anode. Then we are going to discuss lithium battery deformation and the determination of lithium cooling efficiency. The third part is on, I will focus on high-efficiency electrolyte for lithium metal anode and lithium metal batteries. The fourth part is a quick overview of other approaches to improve performance of lithium metal anode. The last is a short summary. Here's a slice we prepared years ago on advantages and changes for lithium metal anode. So for lithium metal anode is one of the material which has the highest vertical specific capacity and the lowest reduction potential. It also has a very low density. In combination, it's one of the best or ideal anode for rechargeable batteries which can lead to more than 500 one-hour per kilogram at cell level, which is also the target and for lithium metal, for lithium, for better 500 projects. In fact, this 500, this 500 one-hour per kilogram, this density is first indicated by Professor Stan Wingham in his one of the review paper in 2014. On the other hand, there are many changes for lithium metal anode. First change is the lithium battery growth. This ICM image showing here is what Stan Wingham, Professor Stan Wingham and his colleagues appeared in the 1970s when they first work on the lithium metal anode. They found this kind of demographic growth and which eventually lead to very serious safety problem and the other one tend to lithium battery since the 1990s. With advance for new technology, new electrolyte, I think in these 10 years, more and more people start to look in at lithium metal anode again, try to use new technology to find a solution for lithium metal anode problem to resolve this problem. So to make, to enable lithium metal anode to work, we have to, it's very often to address what aspect of the problem, but to make it, to enable lithium metal battery to work, we have to resolve many problems at the same time. For example, we have to have a cooling efficiency of lithium metal for larger than 99.9 ideally. And this electrolyte also need to be stable with high voltage case out. And then they need for large cell operation, we need to operate at the lead electrolyte, see lithium metal and high capacity. At last, we have to address the safety problem and spreading of lithium metal anode. Among all of these challenges based on a little review and also our parents, we find that the electrolyte is the most critical parameter to enable stable operation of lithium metal anode and the lithium metal batteries. One of the reasons other component of the batteries have been used in lithium ion batteries and most time like high voltage case out, MC and other case out is well known and has already has a lot of work. So here's the overview of PNL's work during the last 10 years. We start from this all mind work on electrolyte. About 10 years ago, when we start to work on to address the lithium metal anode problem, we first start from baseline electrolyte used for lithium ion batteries. First problem we want to address is battery growth because that's where most people stopped on lithium metal battery 30 years ago. So we screened many different kind of solvent and salt was one of the activities. Later on, we find that when we use one more BCM-PF6 improperly carbonate, we can get most complete carbonate for copper substrate when we in a lithium copper cell. However, we still get a clear damage growth even in the best conditions when we use carbonate electrolyte. So the next step is we want to use some active fuel. So we developed this CCM-based active which can lead to no damage growth. However, the cooling efficiency of this electrolyte, cooling efficiency of lithium metal in this electrolyte is still only about 76%. Therefore, we switch to, we move to other electrolyte component. We, for example, we use a ECPC combination and with actives, we can get smooth lithium deposition and increase cooling efficiency to 97%. We will also figure back to this dual salt electrolyte which can also get a stable second of lithium with high voltage BMC. And the cooling efficiency is about 91%. At that time, we start, we know that the Japanese group, they are working on high-contrary electrolyte for graphite-based lithium-ion batteries and get pretty good result. So we developed this high-contrary electrolyte we try to use that one for lithium metal battery. The mine electrolyte we use is a four more for lithium FSI in DME, use this electrolyte, we can get a high cooling efficiency of more than 99%. This is the first time that we can get, in the future, this metal can be cycled at a cooling efficiency much better than 99%. So later, we will also develop some other electrolyte, further improve the cooling efficiency to 99.5%. One problem with high-construction electrolyte is high with costing and high cost. So to address this problem, we further developed a new electrolyte, a series of new electrolyte, we called localized high-construction electrolyte. And with this new electrolyte, we not only can get high cooling efficiency, but also it is also stable at high voltage k-salt. And later, we will replace solvent by non-flammable solvent and get similar cooling efficiency. So that's my roadmap in PNL on the development for new electrolyte for lithium metal anode during the last 10 years. So here, I want to give you a quick overview on one of our early work on supplies lithium dam drive. For this approach, we try to identify some electrolyte at-cube. Also, when we deposit lithium and control reduction potential, we pass lithium through the lithium copper cell and the lithium will be deposited at minus 3.04 volts. On the other hand, if we can identify another at-cube which has a reduction potential even lower than lithium, then this at-cube may accumulate on surface of lithium dam drive. When the at-cube is accumulated to certain degree, they will prevent further deposition of lithium. So the principle we base is according to this NERC equation. NERC equation included two parts. First part is baseline, the standard reduction potential at one more. But it's also included the second part which depend on the concentration of the at-cube, different component. Bumple, if we use one more for lithium salt, it will be reduced at minus 3.04 volts. On the other hand, although lithium has a reduction potential slightly higher than lithium, lithium ion, but if we use a much lower concentration from 0.05 more for lithium at-cube, the reduction potential can be reduced to below those of lithium. As a result, when we do the lithium definition, if we look at the figure on the left side, at first, if we look at figure B, at first lithium will depot it and form a small fluctuation because all the electrical chemical cell will have some fluctuation. We will depot it into a small boundary. And the figure C shows this boundary growth. If we control the electrical chemical reduction potential below minus 3.04 volts, but above, I mean, if we control the reduction potential below minus 3.04 volts, lithium will depot it, but if this potential is above minus 3.103 volts, cesium will not be depotted. This starts to show in the figure C and B. Eventually, if we can look at the figure E, when the boundary growth to a certain degree, the cesium at-cube will be continued acumen and they form an electrostatic shell. When this shell is strong enough, the incoming lithium ions will be repelled to depot it in the valley region of cesium deposition. Eventually, the lithium deposition will be smoothed out and this process will be repeated. Lithium deposition using different electrolytes. The figure A shows the lithium deposition without acting and the figure BCD is deposition with increasing amount of lithium cesium at-cube. As we can see, this lithium deposition become more smooth with increasing amount of the at-cube. Very interesting. We also find that if we use high resolution SEM, we look at surface and cross-section, we can see that lithium deposition is a kind of the nano rod. The diameter of nano rod is about 200 to 300 nanometer. The like a grass, they all grow together. So in the macroscopic point of view, it's very smooth. Although use this approach, we can get a very high, very smooth deposition. However, when we measure cooling efficiency, we find that the cooling efficiency of this liquid can only lead to a lithium cooling efficiency of about 76%. I think this is maybe related to high surface area of this deposition as we can show in this cross-section image. So here, I want to mention another work we did during the last five years. That's how to determine the lithium cooling efficiency. Because as I mentioned earlier, during our work, we have to screen a lot of electrolyte and solvent salt and add things. But what we noticed that is when we measure cooling efficiency, if we use a different condensate and all we use a different amount of lithium deposit during the process, the cooling efficiency measured will be different. So we need to identify a good protocol so we can measure cooling efficiency reliably and repeatedly. So in the figure on the left side, we show the cooling efficiency we measured under the using the same electrolyte. But with poly, we say condition using different amount of lithium from 0.5 milliamp power per square centimeter to 6 milliamp power per square centimeter. What we found is if we use low condensate, low capacity like 0.5 milliamp hour per square centimeter, we need to more than 15 cycles to stabilize cooling efficiency. On the other hand, if we cycle sample with higher capacity to measure cooling efficiency, in the first few cycles, we can get the cooling efficiency measured will be stabilized. Next result is if we increase the capacity we use during these cooling efficiency measurement is larger than 3 milliamp hour per square meter, the measured cooling efficiency will be stabilized. So based on this knowledge, we proposed a new protocol to measure cooling efficiency. In fact, this initial proposal is proposed by Dauern-Aubach in the 1990s when he measured cooling efficiency of lithium metal, he found that there are some certain layer like carbon oxide on the copper substrate, then he tried to deposit certain amount of lithium that fully strip it, deposit a larger amount of lithium that only strip part of that for lithium metal for cooling efficiency measurement. However, at the end of this measurement process, we still have to strip off lithium. In other words, the effect of the set direction between lithium and the copper still there and will be included in the calculation for the cooling efficiency and will affect the accuracy. So what we propose is add one more step. We do initial deposition, deposit 5 milliamp hour per square meter of lithium. The reason is we want to be larger than 3 milliamp hour. We know, based on previous slides, if the copper substrate will be stabilized, then we fully strip all the lithium. In this case, we believe the copper substrate has already been fully passivated. Then we start to deposit another 5 milliamp hour per square meter of lithium, then strip only part of that. The specific protocol can be seen in figure B, which shows current and figure C, which shows a multiple profile. You see this profile, we find that we can measure cooling efficiency much more reliably. So it's done depends on person, done depends on how do we treat substrate. So in all further work, we also find that protocol not only is important in determination of lithium cooling efficiency, but was also important in the second stability of lithium metal batteries, and was one even included anode-free lithium battery. On the top, we show a few figures. First figure A is a typical lithium ion batteries. We have used a thick graphite anode. For lithium metal batteries, the thickness is much less, less than half of the thickness of graphite can be used. For anode-free batteries, we can eliminate the lithium metal anode. If the cooling efficiency of lithium cycling is high enough, the lithium metal can be removed and we still get reasonable cycle stability. The figures showing in the bottom is what we get from a copper lithium ion phosphate cell. If we use low, the solid cycle curve is when we use low depletion lithium depletion and the high rate of lithium stripping. At 100 cycle, we can get a cooling efficiency of 99.8%. But at low current density, if we use low current density for both lithium stripping, we can only get a cooling efficiency of 98.8%. So this is a clear indication that the cycling protocol is also important for stability for lithium metal batteries and also the anode-free lithium metal batteries. So in the next slides, we will show our work on high efficiency electrolytes for lithium metal batteries. As we mentioned earlier, we have done a lot of work using conventional carbonate-based electrolytes from one more lithium-plated 6-in-PC with all these all active. What we find is, as I mentioned earlier, we get this dielectric growth. Sometimes if we include active, system active, we can get no dielectric growth, but cooling efficiency is still low. So on the right side, we find that we show the lithium depletion using a high concentration of electrolyte. For this high concentration of electrolyte, we get this large nodule growth of lithium, which has a much smaller surface area. This much smaller surface area not only will largely reduce the lithium loss during cycling. On the right side, we show this lithium cycling for more than 1,000 cycles. This cooling efficiency is still larger than 98%. At a low current density, the cooling efficiency can be 99.1%. For high-contrast electrolyte, we still have some problem. For example, like high cost, we have to use four to five times of lithium salt. We also have a high-risk cost, as a result of high concentration. So we try to design a new electrolyte, which can retain all the advantages of high-contrast electrolyte, but reduce, try to avoid the problem with high-contrast electrolyte. Our idea is, first, we need to find a base solvent, which can have a high solubility of lithium salt. So the base solvent we use is either GME or GMC. And they have a high solubility of lithium salt. And for salt, we find that the lithium FSI is most stable with the lithium metal anode. So the most important part is the part C. So to avoid high-contrast solution problem, we add this dilute such as BTFE or TTE or other dilute. The most unique feature of this dilute we selected is they have a very limited solubility of lithium salt, but they can still fully mixable with base solvent. As a result, from a macroscopic point of view, it's a uniform mixture of the electrolyte. But if we look at, if we have our eye to look at the put on the lithium metal, what they can see when we apply an electrolyte field, the lithium will only move around the cluster of lithium salt dissolved in the base solvent. And this lithium metal is blind to the addition of this dilute. As a result, we can retain the odd advantage for high-country electrolyte, but we get low cost and low viscosity, also reduce, also a slightly increased conductivity. So the figure on the right side is our proposed mechanism. We have several ways to, we use the theoretical simulation and other approach to prove this point. On the figure on the left side is a calculation of the room manager for low concentration electrolyte. At low concentration, the solvent, like DMC, has a room manager lower than salt. Therefore, when we add an electrolyte field, the solvent will be decomposed first to form an organic reach, ACLR, which is not stable with lithium metal. On the other hand, if we use high-country electrolyte, the room manager of this salt will be shift to the lower room manager. Therefore, this salt will be decomposed first during lithium diffusion, and which will form an organic reach, ACLR, and can largely increase the cooling efficiency of lithium cycle. On the other hand, if we use this on the figure on the right side, we show the simulation calculation of the room manager for this localized high-country electrolyte. In this electrolyte, we find that salt will still be decomposed first. Therefore, in this localized high-country electrolyte, salt will be decomposed first and form a stable ACLR. We also calculate the radiation distribution. For example, we calculate the distance between DMC and the lithium salt, which is less than two ounce strong. On the other hand, the distance in atomic level, the distance between BKTE and the lithium FSI is more than is about five ounce strong. In other words, the salt is closely bonded with a base solvent and push away the dilute. We also use a Raman spectra to show that addition of this dilute does not significantly shift the original salt and the BMC bonding. Therefore, does not affect their base structure. With this kind of fundamental knowledge, we did intensive study on the electrolyte to a chemical property of different localized high-country electrolyte. First one is, first localized high-country electrolyte we studied is based on carbonate. We use DMC at base solvent from SEM picture. We can show, see that we can get large decent definition. From cross section, we can see the decent definition is much denser than when we use baseline electrolyte. And we also get more than 700 cycles at a high current density of two milliampere per square centimeter. In addition to DMC, which is a carbonate-based electrolyte, a base solvent, we also look at the other base solvent, especially the ether-based base solvent. Once the solvent we find work best is TTE based. Here we use 1.2 molar for lithium FSI in DME TTE electrolyte. And the cell is stable up to 4.5 volts. At the liquid efficiency can be up to 99.3%. And also this SEM image also indicate that it's very stable and the decent definition is very dense. So another optimization of all electrolyte is we try to identify our turn to dilute. Because our original BTFE solvent dilute we use is BTFE, which has a low boiling point about 63 degree centigrade. Of course, this kind of dilute will not be stable at higher operating temperature. So we are looking for alternative for BTFE. We find that one of alternative is KFEO. This fruiting of the bondage has a boiling point for about 130 degrees C. We work with the Professor Yi Chi's group in Stanford and they help out to use this Crayol EM, the electrical microscope, to measure the nanostructure of SEM layer. We put it on lithium surface. On figure B we see that very uniform definition of SEM layer formed on lithium metal non-rod. On figure C we show that this SEM layer is about 10 nanometer. It's very uniform. And what's wrong is one thing surprise us is this SEM layer is monolithic instead of the multi-faceted or multilayer structure. It's also amorphous. This is the first reported SEM layer with this kind of monolithic solid-electron interface. With this kind of very unique and stable SEM layer we get a cool efficiency of up to 99.5%. And the electrolyte will also lead to very good discharge read capability. So in next slide we show our work use a non-flammable localized high-concentrate electrolyte. Here we replace the DMC by a non-flammable solvent TEP. And as we show in the figure on the right side is the optical image figure A is the one we use recent PF6 in baseline solvent in carbon solvent. It's very easy to burn. On the right side we show the flammability of this TEP based electrolyte is very poor. And basically we can use TEP as a mine solvent. We noticed that TEP have been used in early 2000 by Dr. Kang Xun of Army Research Lab. At that time we find that the TEP can lead to exploration of graphite. Although it's a good solvent to surprise flammability of the electrolyte, but we can only use a small amount normally less than 10%. Here with this metal we don't have problem with graphite exploration. Therefore we can use TEP as the mine solvent. And this electrolyte also can lead to very stable cycle more than 600 cycles. In next step, after we find this look the idea or concept of localize the high country electrolyte can lead to very good cooling efficiency and the stable cycling. We try to optimize different component. For example, we try to, we select, we fix the dilute at TTE try to compare different base solvent. All investigation indicates that when we use TTE, when we use DME as the baseline solvent we get the best cooling efficiency of 99.5% and the lead to very good cycle stability. Another component we optimized is dilute. Here we try to fix the baseline solvent is we use DME as base solvent. Then we choose five different kind of dilute. From a cooling efficiency point of view we find that TVEO based electrolyte can lead to best cycle. On the other hand, when we use TTE with the electrolyte, it's cooling efficiency is very close. Another concern is when we use this electrolyte for pot cell, this electrolyte also need to have a low viscosity and also can operate at the linear electrolyte conditions. So from a cooling cell point of view TVEO based electrolyte is the best but when we compare using the pot cell we find that TTE based electrolyte is still the best when we combine all the properties required. So here's what Professor Jun Liu, the director for all battery 500 project reported in this year's review meeting. And during the last five years we work on these metal batteries using different electrolyte. And at first when we use carbonally electrolyte the battery can only operate for about 10 cycles. So in 2017 when we switch to our localized high-contact electrolyte, we get more than 50 cycles. That with improvement on the electrolyte the cooling efficiency continue to increase but only if we only rely on the improvement in the electrolyte is still not enough. Our pot cell team led by Dr. Jie Xiao and her team did significant work and tried to optimize all the process, the match between anode, kysode and try to optimize all the ratio and the other component in the pot cell. So by the combination of all this effort right now we can get more than 600 cycles with close to 80% capacity retention. And we are continuing to improve this performance. And we believe in the next five years we can reach our goal for 500, we can move close to our goal for 500 one-hour program. So here I was one point, mentioned that in addition to look like the high-contact electrolyte, we developed the MPNL there are also significant work in this field on new electrolyte and the coatings. In this figure we summarize the cooling efficiency for lithium copper cell. Myriad using different electrolyte by many other groups. This is a combination of different electrolyte. The best cooling efficiency reported is 99.8% reported by a professor changing one's group in University of Maran. This is obtained by the optimized electrolyte and the optimized substrate. Here I was one point, I was one to mention several other efforts. In addition to the electrolyte effort, there are also other approach should be used to improve the performance of lithium and iron. For example, the professor is his group is Stanford during the recent years, they developed all kinds of several different kinds of lithium host including this graphene based for folding and also the encapsulation of lithium. And professor Julian Bao and his group developed many kinds of lithium coating on the lithium substrate. This is just an example. And professor Jun Liu and also Chao Jiang Liu of PINEL, they developed this kind of the self-smoothing like I know by functionalize the carbon film then the combined lithium and copper in the weighted functionalized carbon film which can also lead to very high cooling efficiency definition. So another approach is proposed by the professor Shirley Meng and they propose this kind of the idea recent definition scheme. So if we control the pressure ideally we should get the ideally we won't get the definition scheme like showing efficacy and in this kind of the non-rot structure we'll get the very smooth definition without dead lithium. So another method which lead to the best cooling efficiency is proposed by is reported by professor Julian Wang on the top line on the right side top line is the definition of lithium in a lithium-phobic copper foil which can lead to significant growth of lithium dead lithium. On the other hand he tried to coat this copper substrate by combination of graphite and bismuth in the case it forms lithium-philic layer on copper substrate and on the other in the bottom is lithium-philic so lithium will form uniform distribution for nucleation on the copper substrate on the other hand there was a choice high-efficiency electrolyte which will form lithium flower rich and lithium-phobic ACI layer so the address in the bottom is lithium-phobic in the top is lithium-phobic by combination of these two physical properties for materials we can get lithium dead lithium growth and with very high cooling efficiency up to 99.83% so here's a summary of my presentation lithium metal anode is the IDI anode for the next generation of high-end density batteries and among all parameters we can control we believe the electrolyte is most important factors determine the performance of lithium-philic anode and localize the high-efficiency electrolyte and other fluorinated electrolyte are highly effective in lead to high-efficiency lithium-philic anode and with our new electrolyte and optimized cell structure we can demonstrate the high-capacity part cell with more than 350 one-hour program and more than 600 cycles with about 80% capacity retention and last I want to mention that to further improve the performance of lithium metal anode and lithium metal batteries we can combine good electrolyte and all other operating and preparation procedures procedures and to further improve performance of the data in a higher cycle life and last I would like to thank Yo-Yi's battery 500 program for support and thank our PNL team members Dr. Wu Xu, Xia Chao and many other collaborators thanks for the 500 PIs and team members thank you for your attention Jason thank you for the very exciting talk really great data from your lab and also some from battery 500 consortium let me start by asking the first question I really like the idea of localized high-concentration electrolyte you have been leading the development in these areas so and you show the table comparing you know the advancement of putting in different component right there so what's still really needed for localized high-concentration electrolyte to go to the next level so what are the things you are looking into I think you touched upon doing your talk a little bit so I want it to be maybe more clear to the audience very good question what we found is once we get the electrolyte which can enable a cooling efficiency for more than 99% many other parameters are less effective right now even increase 1% of cooling efficiency is very difficult and the first element is still continue to optimize baseline solvent and especially baseline solvent because as you can see the my advantage of localized high-concentration electrolyte come from high-concentration which directly related to lithium salt and base solvent so right now we are continue to optimize base solvent and we have some new result will be reported later with alternative base solvent we can increase cooling efficiency by another 1% but eventually if we want to further improve the cooling efficiency we need to combine all kind of the parameters for example we need to control the temperature control the pressure control the testing protocol if every component including materials selection and operating parameters if one of these can increase upon 1% then that's very significant so what I mean is we do need to combine all the parameters to further improve the performance so if I look at 99% cooling efficiency now so what's that 1% loss the detail analysis of that could be important I'm looking at the past of the advancement of cooling efficiency when you have 80% cooling efficiency you have a lot of dead lithium loss when you get to 95% the dead lithium loss becomes less it becomes SCI formation and then when you get to 99% of course every bit of loss that is SCI or a little bit of dead lithium all contributes so any insight about 99% what's still the loss SCI of course always has the loss and whether the dead lithium Jason still having a little bit so what's your thought on that I think that to stabilize to get the high cooling efficiency first the most important thing is to get stable SCI layer and I think that's where the new electrolyte including all localized high-conference electrolyte and other group developed different kind of highly fluorine electrolyte that's the first step to get a high very stable SCI layer but after that once we get a good SCI layer the stability of this SCI layer under extreme under operating conditions such as high current density and also different temperature different pressure they all affect stability of SCI layer so during the second we the lithium had to deposit on the streets and the bottom of lithium continue to change during this process part of the SCI layer will be damaged so that's why I said the electrolyte itself can only do so much if we want to further improve 1% to 2% we need to find some way to control the process and prevent damage for SCI layer due to operating that's a good point so I always have the question is because lithium matter is a deposition stripping process and can we fully utilize the past SCI you already built not growing the new SCI because there's no strong evidence saying lithium matter will plate it back absolutely under the initial SCI already for maybe the initial SCI already for having some function right there how do we promote the process of utilizing just like graphite graphite once you form SCI on the surface the SCI got reutilized every time so to be stable and just any thought on that maybe this is also a good opportunity if G is still online let's bring back Julie into the discussion I think we're going certainly more on the solid state but I think that I call that as a hose structure having a hose right there if the lithium can grow into the hose you know you can reutilize the SCI already formed so it will be good Jason to see your and also G's input on this actually among the ideally we get this kind of the column structure when we strip lithium all the lithium will be removed and the but SCI layer will not damage that's the ideal situation but basically it's very hard to realize in the case of graphite they have this graphite SCI layer on the surface of graphite so graphite itself can withhold the stability of SCI layer in the case of pure lithium metal there's nothing to withhold this kind of mechanical strength so ideally we have to find the structure similar to graphite if this kind of structure can be withholded by physical principles similar like this kind of the self healing mechanism without physical frame that will be the idea you have any comments? I agree with what Jason said there was some recent crowd study from my collaborator and he showed that for a few cycles at least indeed you can reuse the old SCI shell in fact when you strip there is this collapse of the SCI shell there is a complete wetting almost like a two-dimensional lithium metal sheet that refills the old SCI at least for a few cycles but you cannot exceed the previous capacity for that chamber and I think a very important aspect is also maybe we should not just think of one cell Jason showed I think we should think about a framework structure where you have multiple shells mechanically reinforcing each other and indeed the liquid electrolyte has shown very interesting mechanics and that is very important also I think for the solid state battery design if I think about combining today's two talks it will feel free to come back and jump in as well you are having this mix-iron, electron-iron conductor right there and Jason is having this electrolyte if this could be combined this mix-iron conductor right there you have lithium going into this shell also have liquid electrolyte sounds like this could be an interesting opportunity to address the problem of lithium metal and that lithium metal is that deposited inside outside is this mix-iron conductor, electron conductor so that position have an inside but outside could be stable facing liquid electrolyte that could be a possible route to address this problem I think in fact you and also Professor Hongli have basically been developing similar to a semi-solid battery to control lithium metal I think there might be a very good sort of common science in all of these and I think liquids are very good for maintaining contact without any stuck pressure solid have some problem doing that and I think this issue of lithium metal and cavitation crumbling of whatever is containing the lithium metal and there is almost like a fatigue issue going on in these structures so these are all related issues yeah so I ask one more question maybe Will you feel free to ask others this is more for Jason but feel free to also share your thoughts as well you have done some work in this area and looking at the whole electrolyte space over the years this many exciting comes are coming in you know if I look at Jason you talk about localized high concentration electrolyte this purely high concentration electrolyte right they're not localized one right and there's also new type of fluorine contain electrolyte to make SCI more stable it will make efficiency higher and what you're thought and compare all these ideas you know I mean you can go to high concentration to the extreme it becomes a solid and solid right what Chen has published in a number of papers on that and then what are the new concepts might be coming in Jason and I think all kind of new electrolyte recently in recent years for to stabilize this matter I know has a common point no matter what you call you can have a different name but the common point is they will form a lithium fluoride and lithium oxide this kind of organic rich that's the common point you can have a different formula but the most time is a lithium fluoride rich and lithium oxide rich so in order for the SCI to be lithium fluoride rich it had to be highly fluidity I think that's one common point so Jason is that lithium fluoride rich or is an iron rich I'm thinking because the lithium fluoride rich might be still debatable and several electrolyte system because a lot of XPS study looking at the global scale of composition not necessarily having lithium fluoride in the immediate SCI layer this is what we find out using bioEM but it looks like a good electrolyte having a common more common feature is there is an iron rich SCI and the lesser solvent composition compound in the SCI layer is that a fair statement yes I think more accurately we can see your kind of species rich yeah I think from application point of view this depends on we're talking about cell phone battery or automotive battery I think a lot of the liquid electrolytes are already good enough if you just want to have a lithium metal battery for a few hundred cycles but for automotive you need thousands of cycles and I think there is still more development that's needed and new ideas okay Will do you have questions sure thank you E I thought since we're coming to the end of the seminar I would ask a provocative question as I hear about all these new approaches for the next gen lithium based batteries I can't help to think about the industrialization challenges and bottlenecks as you try to implement this at extremely large scale and the two things I'm thinking about of course is the cost of manufacturing and the variability of manufacturing and the two of you actually have presented very different approach along that axis of something close to being a drop in electrolyte and something that requires a new manufacturing methods so I'm curious in designing and world mapping your research how do you think about the industrialization aspects and how does that set your research agenda so this is a high level question I think many of our industrial colleagues will be very interested to know how you make those decisions in your in your groups that's a great question so I think the liquid electrolyte because it's a dropping solution the best important thing is bulk cost so reducing the cost to accept the level is very important so the localized high concentration electrolyte that's a great idea and you also need things like flammability viscosity and wetting of the electrodes so I think to mitigate environmental sort of carbon reduction these are the workhorse technologies we need in the next 10 years the solid state battery I think there are many manufacturing challenges for example the stack pressure I think it's to me very difficult maybe to to achieve in a cost efficient fashion in actual large format batteries and so I think for solid state battery to come in and you know to be having a real impact on climate change I think it probably wouldn't happen within 10-15 even 20 years and this is more what I call a type B or like a baby technology kind of research so I think the liquid electrolyte work would have immediate industrial impact fortunately I think the problem is much longer term than this so plenty need for both A and B solutions Jason? I have one comment about the lithium metal anode in Chinese martial art they said that the best skill is not skill and I think in field of lithium metal batteries the best lithium anode is not lithium anode in practice you mentioned that the manufacturer ability as you know the lithium metal itself from material possible is not too expensive however to make very thin lithium foil is expensive and especially if we want to make the lithium foil to less than 50 micrometer or even 90-10 micrometer the thinner ideally we use less lithium but in fact the much thinner lithium is not less expensive it's much more expensive on the other hand there are also we need dry room to handle lithium metal so ideally we can have a very high quality of lithium cycle and we can minimize loss from all components and we can use the configuration for lithium battery but in practice this may not be realized we still need for at least in the future lithium metal battery is still more practical than anode free battery but in the search in the exploration of anode free battery it can further improve the quality and all the components and can help to improve the production of lithium metal lithium metal battery so I'm going to go ahead and conclude today's seminar Jew and Jason thank you very much for participating today and E for co-hosting of course so Kaylee if I can have the final slides so we have a couple exciting seminars coming in the coming weeks November 5th we're going to have a thermal storage panel we're going to have a special session on battery circular economy featuring colleagues from Northvolt and BSF so I hope you will register for these events as well so on that note I'll end out don't forget to connect with us and so you can stay up to date on the latest progress at Stanford Storage X and thank you very much for tuning in today and hope to see you next time thank you so much