 We selected these speakers through a seminar series that we had where we asked the GSEP students to come and give seminars and we picked out the best ones and we will hear from two of them today. And the first speaker we'll hear from is Jeremy Feister. He's a third year PhD candidate in the Yarmillo group in the Department of Chemical Engineering here at Stanford. His research is focused on designing reactors and catalysts to electrochemically convert carbon dioxide into fuels and chemicals. And today he'll give a talk entitled, Insights into the Electrochemical Reduction of CO2 on Tin Electrodes. So please join me in welcoming Jeremy. Thank you. Good afternoon. My name is Jeremy Feister and today I'll be presenting my research in the Tom Haremillo group in chemical engineering entitled, Insights into the Electrochemical Reduction of Carbon Dioxide on Tin Electrodes. And so just to give a framework for my talk, I'm going to first start with the motivation of why CO2 reduction is very important for what we're focusing on. And then we'll move on to identifying key questions that we would like to answer through this research, the design and execution of the experiments that we were able to conduct, and then also looking at an analysis of our results and finding and kind of summarizing everything into key findings and takeaways before summarizing the entirety of the talk. And so as you can kind of see in the lower left hand corner, there's a little diagram there to follow the progress of the talk. So I don't have to really tell you that we're all here because we know that the impact of CO2 and the impact of a lot of the things that in terms of our fossil fuel consumption will have an effect on our climate. What I would like to also present is that not only does this make sense for us to focus on CO2 reduction and trying to help clean up our atmosphere environmentally, we also, it also makes sense to do so economically. And so with CO2, it's not only just a greenhouse gas, but it's also the largest source of untapped carbon in our world. And as you can see here from some calculations provided from Opus 12, a startup out of our group that's currently stationed up at Berkeley, they've done calculations on point sources of highly concentrated CO2 and the potential value that that CO2 offers. And so from oil refineries, and that's only looking at just hydrogen production and separating CO2 from there. But then also with corn ethanol production from fermentation and natural gas fields where they remove CO2, there's only a fraction of that CO2 accounts for the 29 gigatons of CO2 that we produce per year. But it represents 30 to 40 billion dollars of potential value that's just left on the table. So again, just to iterate, it makes sense for us not only to look at ways to reduce carbon dioxide environmentally, but also it makes sense for us to do so economically. And so the way that our group decides to focus on tackling this problem is looking at the electrochemical reduction of carbon dioxide. And so the idea is that we will take renewable free or renewable CO2 free sources, energy sources, such as wind and solar, and be able to partner this with our electrochemical reactors to then convert carbon dioxide back into fuel. And the reason why we decided to take this approach is because it's an atmospheric pressure, low temperature. It's something that's very, very simple and it's a very straightforward process for us to be able to pursue. The idea is that we can either convert carbon dioxide back in the fuel, which would then create a carbon neutral cycle. Or we could also be able to sequester it in the form of chemicals, which would serve as a carbon sink. Of course, this technology is not quite there. We're still working on a number of areas in terms of improving our catalysis. And that includes improving the selectivity, improving our ability to make one particular product from CO2, as well as working on our activity, which is talking about the overall energy that you have to put into your catalyst or into your system in order to reduce carbon dioxide into something. Stability and energy efficiency are also two other very important metrics that we have to take into account, not only with a catalyst, but also with the reactor design as well. And so if we were to just kind of think about any products that we would like to make from CO2, where we, from here again from Opus 12, we just kind of picked five potential products that are very interesting, or that have been highlighted as very interesting. And between ethanol, propanol, methanol, acetic acid, and formic acid, on that plot to your left, you'll see on the x-axis is the overpotential or the energy that you have to put into each, into this, into the catalyst in order to get that particular product made. And then on the y-axis there of that plot, you also see the product costs. And so as you can kind of, intuitive, I guess intuitively figure out that the number of electrons that you need in order to reduce carbon dioxide to each of these products is going to have a very strong impact on the type, on the product cost of that particular product. And as you can notice here, formic acid actually kind of sits towards the bottom. So even as we put in more energy into our system, the cost of being able to make that formic acid only increases slightly. And that's simply because formic acid is a two-electron product. Whereas the other chemicals or the other products that you see on that plot require more electrons and more protons, and are much more difficult to do so kinetically. The plot on the right actually shows the difference between the market price for each of these products, as well as the difference between the market price and the electricity cost, which typically tends to be the greatest factor or the greatest source of cost for this entire reaction. And so as you can also see here, formic acid retains much of its activity or much of its product costs here, even when you subtract out electricity costs. So if we're looking at formic acid as the product that we want to focus on, there's a number of catalysts that our field has identified as very promising. Of all of those different catalysts, 10 turns out to be one of the more active catalysts for CO2 reduction. And so there's been a lot of studies and a lot of good work done on 10 catalysts here at Stanford as well as internationally. But there's really kind of this one void that we haven't quite been able to answer, and that really is just more a fundamental question of why is so, why is 10 so good at making formic acid? And so that's really the key question that I would like to be able to focus on with my work today. And so in order to do these experiments, we decided to use our custom electrical nickel cell that the Harmeo Group has been able to develop. And so here it actually has an optimized surface area to volume ratio to allow us to be able to detect products and have very good detection limits. And then we use grass chromatography to measure our gas phase products as well as liquid as well as NMR for both protein and carbon to measure our liquid products. And so as you can see here on the plot to the left is our fairy day efficiency, or really you can think of it as an electron efficiency for the number of electrons that we're putting into our system versus the products that we're getting out as a function of potential. And that's being held relative to the reversible hydrogen and electron, which is just the standard metric that we use in our field. As you can see here, it may not work as well. But as you can see on the plot to the left, there's a maximum formic acid production right around negative 0.9 volts versus RHE on these 10 electrodes. And that really formic acid doesn't become the major product that we observe until about negative 0.8 volts versus RHE as well. We also were able to observe a little bit of carbon monoxide being formed on the catalyst as well that seemed to have a maximum right around that same potential range. And all of this fits very well into what we understand tend to do for CO2 reduction. On the plot to the right, you can also see our partial current densities, which really actually correlates to the products that we're making from CO2. So you can think of this as almost like a moles that we're being able to produce from each of these, from 10, from CO2. And as you can see here, the formic acid starts to tail off, it starts to plateau. And that's just really indicative of mass transfer limitation for CO2 transfer to the surface. And that's what's capping our formic acid production there. So we're pretty sure that 10 is being able to make formic acid. And it seems like it's doing it just as much as we would expect it to. But it still leaves out this major question of how does CO2 reduction on these 10 electrodes compare to other metals in the exact same system, in the exact same setup? And can we learn anything from that analysis? And so what we did was we actually partnered with some collaborators in the Norse scott group, which is another group in the chemical engineering department here. And we did density functional theory calculations or DFT calculations to be able to then compare experimental data along with theoretical data and calculations to see if we can gain any insight from this reaction. So for CO2, as it's going from this intermediate that we've done the calculations for, it's binding energies to the surface, which is a carboxylic acid type intermediate here, as you can see there. And then as that goes to carbon monoxide, it's just one potential pathway for CO2 reduction. You can see that this on the, this plot to the right here actually seems like it follows what's known as a volcatotrend, which simply means that the binding energy of this key intermediate to the surface cannot be, the binding energy cannot be too strong nor too weak. It has to have an intermediate binding energy in order to have an optimal production rate. And so as you can see here, a 10 seems to fit very well within this framework for carbon monoxide production, where it's on the weak side binding and the production rate that we're actually being able to observe matches very well with what we would expect theoretically. However, when we were to do the same calculation when we're looking at the same type of intermediate here, the carboxylic acid intermediate, but looking at formic acid production, as you can see at the very bottom, bottom left of the plot, the trend completely disappears. We're not able to see any volcano type behavior. And really what jumps out to you is if you look at 10 in the upper right region of that plot versus silver, which has a very similar binding energy for that intermediate, it seems like there's a complete orders of magnitude difference in the amount of formic acid that 10 is able to make versus silver. So this is something that kind of gave us a hint that maybe this carboxylic acid intermediate on the surface isn't the best metric to use to try to describe CO2 reduction. So we kind of took a step back, thought about it, and of course just made the problem as simple as possible. Carbon dioxide has two atoms. It has a carbon and it has two oxygens. So if we were to look at the oxygen binding energies as it was, and see how that correlates to formic acid production, maybe we can be able to then capture our volcano trend again. And that's exactly what we're able to see. So with the y-axis still being our experimental data that we were able to show for all these different metals that we tested in this setup at negative 0.9 volts versus RHE, and then our x-axis being theoretical DFT calculations of the OCHO binding energies, which corresponds to an intermediate that looks something like that to the left, you actually are able to completely recapture the volcano trend. And what's the greatest part about this particular figure is that 10 is at the very top of this volcano. And so it has this optimal binding energy for this key intermediate that then results in it being able to make formic acid. So not only just for 10, but when we were able to kind of do calculations for other metals that are also known for making formic acid, they seem to fit very well within this framework, which gives us just more confidence in thinking that perhaps it's the oxygen binding that results in formic acid production for CO2 reduction. And so to try to kind of put this all together in terms of a pathway, if we start from CO2, we kind of have the initial carboxylic acid type intermediate on the surface that leads to carbon monoxide that many people in the field and that we're very comfortable with. But then also there's an additional pathway of oxygen binding to the surface for 10 and for some of these other metals that results in formic acid being the major product being produced there. And as a result, it seems like this OCHO type intermediate on the surface is one of the key intermediates to determining what your activity for formic acid will be for CO2 reduction. And so just to summarize, we were able to thoroughly investigate 10 for CO2 reduction as a function of potential. We were also able to then identify the primary factor for high selectivity on 10 electrodes and really be able to answer that question of why 10 is so good at making formic acid for CO2 reduction? And of course the key being that OCHO binding mechanism. And then also we were able to then propose a mechanism that encompasses both carbon monoxide and formic acid production as a result for CO2 reduction. And so with that I'd like to quickly summarize and to give a quick shout out to the Harmia Group for not only being able to help with the scientific aspect of this group or this research, but also just being a good group to be able to work with as well as our collaborators with the Norsegov Group and the Conyello Group here on campus. And of course, funding sources NSF for funding me and GSEP for funding this project. And with that, I'd love to take any questions. Thank you. What was the electrolyte you were using here? So I was using 0.1 molar potassium bicarbonate. And so when we bubble CO2 through that electrolyte, we were actually able to get a pH of about 6.7. And you think you're reducing the bicarbonate directly or through the CO2? So from what we've seen, I think that it's probably directly from the CO2. Of course it's gonna form some equilibrium there, but I think that from just from what we've been able to see with varying the CO2 flow rates and then also with other experiments and other reactors, I think the CO2 is probably the reacting species. So there's a history of CO2 reduction via the anion radical that occurs very efficiently, even on mercury, which I think has... It's very, very low activity. Low activity, well, activity in terms of over potential, but 100% current efficiency because you're not making any hydrogen. Yeah, so I wondered if the density functional theory stuff looked to something like mercury. Yeah, so we've done some... There have been a few calculations done for mercury. Essentially, it is the over potential that ends up kind of eliminating it as a potential catalyst that we wanted to be able to look at for this experiment here. But you're absolutely right that mercury, as well as lead and a few other metals, are very, very active for formic acid production. Do you have any idea about the conversion efficiency? For example, you put the electricity up there and how much of the fuel you generate up there and can you predict for the futures what kind of efficiency you can realize? Yeah, so that's a good question. And actually from what this study was really designed to focus on was not necessarily optimizing the conversion efficiency for CO2, but more so trying to understand like a fundamental, essentially what we can do with CO2 conversion on tin and being able to understand what products we'll be able to make. I think that conversion efficiency is going to be critical and that it would probably require a change in our reactor design in order to be able to get much higher conversion efficiencies. But for this particular design of this reactor, it's not very high and that's just simply because it's meant for fundamental studies. These I think on understanding the binding sites for CO2 on a lot of different clean metal surfaces usually under the UHV conditions. Does your volcano plot correlate with any of those studies? So I think that there is some correlation actually. I think I'm actually, I'm pretty sure I'm thinking of the right papers that you're referencing. So there is some correlation, but there is a little bit of a difference and that discrepancy is really due to, they were studying in under completely different conditions than what I'm studying it under. So under UHV, the catalyst surface will probably be much, much different than the dynamic surface under an electrolyte, you know, atmospheric pressure and the temperatures we're at. So there is some correlation, some of the trends do hold with the DFT calculations, but there is a few of the discrepancies there that we noticed. What I'm shooting for in studies on the electricity now where there's a great deal of detail and very rich body knowledge to tap into. Absolutely, I agree. So we're, there's actually a few studies that we're kind of currently undertaking to try to do something in between of we're able to do some beam line experiments up at Slack to then kind of start to probe some of the space in between the actual like, you know, relevant studies that we're looking at here but then also with the things that could give us a lot of really basic and intrinsic information about our catalyst surface. So we'll keep you tuned. How it goes. Thank you. Okay. Our second student speaker for this session is Jeff Lopez. Jeff is a fourth year PhD student in Jenin Bow's lab in the department of chemical engineering. His research is focused on developing and studying novel polymers for lithium ion battery applications and Jeff received his bachelor's of science degree in chemical engineering from the University of Nebraska. We worked with Ravi Saouf on enzyme biosensors and today Jeff will be talking to us about polymer binders for high capacity lithium ion batteries. Please join me in welcoming Jeff. First of all, thanks for being here everybody. Really excited to share just a little bit of the work that I've been doing on trying to improve the capacity of, or the cycle life, excuse me, of high capacity electrode materials for lithium ion battery applications. And the reason we're working on battery applications trying to improve these technologies is that we see really an area that we can impact in reducing CO2 emissions through greenhouse gas production here in the U.S. And most of that comes from either electricity generation or use in transport. And so what we can do is through either using these batteries to store energy from wind or solar, which are intermittent sources, so the sun only shines during the day and the wind only blows certain parts of the country and at certain times of the day, we can store that energy and then you can have the constant power supply that you're used to with the current grid we have set up. Additionally, the other thing that we can do is we can use electric vehicles to replace our gasoline powered vehicles. And what this allows us to do is reduce all of that CO2 production that comes from burning that gasoline. The problem is though, and this is really the application that the batteries that I'm studying focuses on, is that these electric fields are quite expensive and so they're pretty... they're not widely used yet. They're definitely out of my price range as a grad student. And really the huge cost of these comes from the... you can see the battery costs more than about half the car and that's actually the entire base plate of the Tesla model assets, about 8,000 lithium ion cells. And so there's two ways we can think about approaching this problem. One is we can improve the capacity and then we can either decrease the number of cells and have the same range on our EVs with the lower cost or we can keep the number of cells the same and then increase the range on our EVs and that makes them more attractive to replace gasoline vehicles. So either of these things requires us to improve the battery technology that we have. So that's a question that people have been asking for a long time and one of the answers is silicon. It has a theoretical capacity 10 times higher than that of graphite which is a traditional electrode material used in commercial cells today and it's also really cheap and quite abundant and we have really a great infrastructure for producing silicon for the semiconductor industry. The problem is and the reason we're not using it today is that it undergoes huge volume expansion when it charges with lithium. So you'll see actually it goes... there's a lithium phase 4.4 lithium atoms for every one silicon atom which results in about 300% volume increase and then when you de-lithiate through to amorphous silicon and it shrinks but silicon's not a balloon, it doesn't just blow up elastically and so what instead happens is you have fracture and pulverization of your electrodes and that leads to cracking you'll either have pieces of your electrode mechanically de-laminate or you'll have new surface area that forms and there's an electrolyte decomposition reaction that happens due to the really low potential of these silicon electrodes and so both of these, those are the two problems that people are aiming to solve to make these cells have more cycle life which is what we want. One of the approaches, one of the most widely used ones and really exciting one is to do nanostructuring if you structure the silicon below about 150 nanometers that's really the critical dimension they found it stops breaking and so it'll just expand and contract anisotropically along different crystal planes but it won't fail and so there's been numerous ways two works here, the silicon nanowires and these core shell structures or other GSEP work done here but the problem with these nanostructured materials is that sometimes they're expensive and sometimes these processing techniques don't really amend themselves to scale up and so we have a hard time imagining these being realized on an industry scale and so what I focus on and I think the second approach that people are using is to just play with the polymer binder so an electrode is made up of really three components one is the active material, silicon which I'm talking about today two is some sort of conductive additive that facilitates electron transport through the material and then third is this polymer binder even in traditional electrodes commercial electrodes today there's some sort of polymer binder and so you can play with this, they're usually quite cheap and they're easy to switch in and out of the materials and they have quite a significant influence on the properties of your battery and so there's really the three here there's a carbohydrate polymers that people have been using they're water soluble, quite nice to work with very cheap as well also conductive polymer binders what you can do there is you just replace the conductive additive in your material and so you're actually reducing the weight a little bit as well there and then finally people also play with kind of the chemical and mechanical properties of these binders by cross-linking them what we do and what I'll be talking about today is a self-healing polymer coating and it's a concept that Chow Postdoc in our group came back in 2013 and how we think it works we're still doing some inoperando studies up at Slack to really elucidate the mechanism but what we understand right now is that in a traditional binder you'll have large silicon particles that'll lithiate and then after many cycles they'll fracture and your electrode will fail but with our self-healing polymer we design it to be highly stretchable and a little bit flowable at room temperature as well we think that as the silicon expands cracks will form and the self-healing polymer will allow those cracks to heal and it'll fill in any new surface area from the silicon so you're really eliminating both of the mechanisms of failure in these silicon electrodes one again is the electrolyte decomposition reaction on new surface and the other is just mechanical failure of the electrode we wanted to check this out just to be sure we were it was doing what we thought it was doing and so here are some SEM images this is after 20 cycles in the lithiated state so in the expanded silicon state you see there's some really large cracks in the electrode that form which is what we would expect in a material that has expanded so significantly during cycling but then if we wait 10 hours put it back in the SEM and look at it again we see that some of the smaller cracks of the field and some of the larger cracks are smaller than they were before and so we think this really shows and illustrates that the self-healing mechanism is doing what we think it's doing and now really the important piece of data in cycling is just the cycle life so you can see this is charge stored and then cycle on the bottom so how many times we've charged and discharged it and comparing it you can see Alginate and CMC are two of the carbohydrate polymers that I was talking about before and then PBDF is a binder that's used in graphite electrodes commercially and we have really much much better cycle life than any of these with large silicon particles and all of those studies I talked about before are binder studies, they all use small nano-sized silicon particles as well not nanostructured particles but still small particles that don't fracture because they're below that 150 nanometer critical dimension and so what we're doing here instead is using about micron-sized particles and these are about an order of magnitude of two cheaper than the nano-sized particles which is why we're excited about them and that we can get cycling stability that's comparable to the nano-sized particles but with a much cheaper material and so moving from that background into my work the question that I've been asking is why is the self-healing polymer so good? What exactly about this material makes it a good binder for these micron particles and once we understand that can we engineer specifically new polymers that are even better and so what I've done is taken if you're interested in the chemistry of the self-healing polymer that's here we start with a diacid and a triacid fatty acid material so they're actually derived from vegetable oil we functionalize them with amine and then functionalize them again with urea to give us that hydrogen bonding that allows us to have the super molecular self-healing capabilities that we think help us do so well on the batteries and all of the data I've showed before is with this 80-20 mixture so mostly a linear super molecular network and what I have been able to do is play with the chemistry and basically I can have any sort of concentration of tri-functional group between 20 and 70% and what that allows us to do is really change the mechanical properties and so we can probe exactly how much is this viscous flow important and what about the self-healing as well here we're looking at a frequency sweep on a rheometer so you just take two parallel plates and you oscillate them at a low strain and you change that frequency so at slow times at low frequencies you're probing more of a liquid characteristic of the material and then at fast times you're probing it more like an elastic solid the closed circles are the storage modulus which is really the solid-like properties of the material and then the open circles are the lost modulus which is more of like the liquid part of the material and where they cross over that's what I've highlighted that's where you can kind of say that the material is transitioning from a liquid-like characteristic to a solid-like characteristic and what you can do from those times is you can calculate out a relaxation time, a characteristic time that the material behaves more like a liquid or more like a solid and we use that to quantify these different binders that I've made and then compare them the other thing we've done is a stress relaxation experiment and what you do there is instead of oscillating the plate you just step strain and then watch as the stress decays and again you can use some modeling to extract the relaxation times here and what was exciting about that is that two things, one the data lined up and so it was nice to know that these experiments agreed and two materials that range over about two orders of magnitude of relaxation time so the synthesis was successful and we can actually probe what we were trying to probe which is these different mechanical times so what we see this is exciting for two different reasons the first one is that there's almost no change and that doesn't really sound exciting but from an industry perspective if you're working with a material that has a really wide processing window so we can change the capacity of this material change how you can process it and that doesn't change the performance in the batteries at all we see that for the first about 16% through 57% tri-functional groups in the polymer that they all cycle to about 175 180 cycles before they reach 80% capacity which is better you'll remember than the data I showed before and also about the same and then the second thing that I was happy to see is that there was at least one binder that didn't perform as well the 70% tri-acid material and that only cycled to about 80 cycles so even worse than the batteries before with all of our device improvements and things like that and so what that says I think is that really the viscous flow of this material the self healing capability of this polymer is critical to the way we've designed these electrodes and it really allows them to function well with the micron size particles and so with that just kind of a conclusion in wrapping up a big picture we've seen from our work and from literature that it really is important to have a strong interaction with the silicon surface through hydrogen bonding or some other interaction we know now for our self healing electrodes that we need a fast relaxation time for that healing to be effective and then we also think we want really good adhesion in electrolyte interaction which comes from data that I haven't been able to show today and then zooming out even further I really do believe that if we can improve battery energy we can help to reduce greenhouse gas production here in the U.S. and around the world so just like to thank my advisor and professor Eastway who we collaborate with quite closely and then to postdocs who were really critical in this work and also just helping me my Ph.D. GCEP for funding NSF for funding and the whole bow group you can see there and then if you're interested in hearing more there's actually three posters from our group on this project that you can come check out right after this so thank you. So with this silicon anode technology what do you think is the ultimate capacity increase you could get in a full cell? So in a full electrode if you didn't change the cathode so I said 10% increase and that's just for half of the battery so you've still got to think about things like current collectors and casings and electrolyte and separator and all of that so if you just replaced graphite with silicon changed out today you'd get about two or three times increase or so with the best performance of silicon and then if you switch to like a sulfur cathode or something I think you can get up to like six times increase. Right you do have to... That's true yes, yeah. It ends up... I guess depending on how you calculated it it varies but it is you get percent increases up to I think the maximum you can get is two or three times but it maybe more practically ends up being 150% or something. A couple of quick questions. Sure. In your last data chart you showed what looks to be about close to a 50% reduction in capacity initially. To what do you attribute that? So this first we actually cycle I didn't mention these first three cycles are at C over 20 rate which is one charge cycle per 20 hours and then discharge cycle per 20 hours so a 40 hour and then we switch to C over 10 and that's a standard cycling procedure that's been put out by a couple of the DOE programs on battery stuff so that some of the capacity just comes from charging it more quickly so kinetic effects and then the other part that first cycle column of efficiency for us is about 83% or so and that's probably some lithiation of the polymer a little bit and then also that forming of the initial electrolyte decomposition layer that happens and then any the other drops that other part is probably just more electrolyte decomposition and then a little bit of silicon that's getting isolated probably. That actually leads to the other question I had what do you think would happen to the self-healing system if you went to a 2C or a 5C charge right? So this it's actually terrible I have the data once you go to so 10C works 5C you see it at about 1200 or so and then if you go to 2C or 3C it's below 500 so that's really a problem with this material this wasn't specifically designed to be used in batteries we had it in the lab and it was a good idea and it worked so what I'm doing now is working on the chemistry to see if we can increase that rate yep so the relaxation time seems to be an important property when looking at these different polymers what sort of relationships can you draw between different structures and different properties compared to relaxation times? You mean the like functionality of the material or branching or yes effectively when you change the polymer chemical structure how does that affect relaxation? Sure yes so when I increased the tri-functional groups present in the starting material what I effectively do is increase the molecular weight of the material the final super molecular polymer and so really I think it's that molecular weight effect that's affecting the relaxation time and probably a little bit of entanglements as well from that molecular weight increase thank you