Below is the text version of the webinar titled "Materials Genome Initiative," originally presented on December 2, 2014. In addition to this text version of the audio, you can access the presentation slides.
Alli Aman:
—today's webinar. I am going to go through a few housekeeping items before I turn it over to today's presenters. First of all, thank you so much for joining today. Today's webinar is being recorded, so a recording along with slides will be posted to our website in about roughly ten business days. We will send out an email once those have posted to the website.
At the end of today's webinar, we'll be hosting a panelist discussion between all of the presenters, so depending on time, we will also try to answer any questions you have. So we encourage you to submit questions via the question function, and we will cover as many of those during that time at the end of the presentation. Any questions we do not have time to answer during the Q&A, we will follow up with you offline, so please submit your questions.
Since we have multiple speakers today, if you could indicate who your question is for, that would be very helpful. Also, I encourage you to check back to our website for future webinars, as we do host these monthly, and sometimes we host two a month. So I encourage you to check back to our website for that.
Also, if you have not done so already, you should sign up for our monthly newsletter. We send that out monthly, and it just keeps you up to date on what's going on and what the DOE has been working on in the hydrogen fuel cell industry. So I encourage you to sign up for that webinar—or for that newsletter.
So on that note, I'm going to turn it over to Eric Miller. Eric is a technology manager for the Fuel Cell Technologies Office. He is responsible for the hydrogen production activity. Eric?
Eric Miller:
Thanks, Alli, and welcome to all for joining the webinar. Today's webinar will focus on the Materials Genome Initiative, or MGI, and in particular, its use in materials research and development projects for hydrogen and fuel cell applications. Guest speakers from several of these really exciting projects will be featured during the webinar.
[Slide 2]
All right. As shown in the webinar outline, we'll start with a brief introduction to MGI and its growing application in clean energy technology applications. Then the main showcase will be on several of these exciting R&D projects in hydrogen and fuel cells, which have been really early adopters for the MGI methodologies. OK. After that, we'll spend a little time talking about specific opportunities and challenges in expanding the use of MGI across all of the Department of Energy.
[Slide 3]
OK. What is the Materials Genome? You hear MGI all the time. It's a Presidential initiative announced initially in 2011 specifically to promote innovation and materials research and specifically toward business development. The White House website on MGI includes a lot of further details about this initiative. It's shown on the screen. Feel free to visit this website to learn more. One of the important documents you'll find there is the MGI Strategic Plan, which I'll talk about next.
[Slide 4]
So this past year, a draft of the strategic plan was developed and made public for commentary. The final version of the strategic plan is actually slated to be unveiled this week at the MRS in Boston. It may or may not have happened already, but we'll double check on that.
But something to pay attention to: although MGI is sometimes perceived as fundamental materials research, really, the high level mission is to accelerate the materials-to-market process, as indicated in this slide. So what I'm—on a very high level, the mission of MGI is very applied, although it incorporates many fundamental aspects.
[Slide 5]
So as shown on this slide, which comes straight from the strategic plan document itself, there are four key challenges that the MGI is addressing toward meeting its mission. I won't read these out, but I do want to highlight a couple of points here. Specifically, look at challenge number two. If you read through this, you'll see that this really embodies what material science should be. It's really just good science, which is a synergistic approach that integrates the best tools in experimentation, computation, and theory. That's really the gist of what we're trying to achieve here.
Now if you look at the other key challenges, for example, challenge number three really highlights the important role of big data in such an ambitious undertaking. That's a huge challenge in MGI. If you look at a challenges number one and four, they are really key enablers that are needed to establish and sustain both scientific and even more importantly the human resources for a successful MGI.
So what are these resources? OK, what exactly is MGI? So that's the next slide.
[Slide 6]
This is the MGI. Now people working in different areas tend to have different views of what MGI actually is. For example, a lot of people view it as fundamental materials modeling. Often, people just think of it as big data repository of all materials properties. More recently, people are looking at ICME as a guiding principle for the MGI, and even more recently, we're seeing aspects of both combinatorial synthesis and high throughput experimentation coming to the forefront in MGI. If we really step back, we must realize that MGI represents a broader all of the above strategy—it's an all of the above actually "innovation ecosystem." That's a word you'll find a lot in the strategic plan.
So keep this in mind going forward. Even though you may have your own perception of what you're working on in terms in MGI projects, we are really looking at a broader innovation ecosystem here.
[Slide 7]
Now there's a lot here. I'm not going to go into the details of this, but what I want to highlight here in terms of this MGI innovation ecosystem is that it spans everything from fundamental materials discovery and design all the way through materials qualification for specific manufactured products. Now this is really ambitious. The scope encompasses a broad portfolio of multi-physics and multi-timescale methodologies, which is an enormous scientific challenge, especially in terms of bridging these various domains, as most of you listening probably are well aware.
I would have to say that early MGI adopters have focused a lot on developing fundamental tools which provide the scientific push. What we're looking for is more industry pull to focus the R&D efforts toward specific materials-to-market efforts, and that's one of the key things we'll be discussing, a key theme that we'll be coming back to again and again. How do specific energy— end uses help pull forward the MGI toward the marketplace?
[Slide 8]
All right. Now if we apply MGI specifically for clean energy materials, we can see that in the clean energy space alone, there are numerous applications where materials research and development can have significant impact. For example, if we look at MGI for clean energy applications, it would cover numerous materials classes, as shown in this slide, which makes it even a bigger challenge, because so many different classes need to be treated in different ways. But it does present a lot of opportunities for DOE, and that's, again, another recurring theme through our webinar.
[Slide 9]
So what we're looking at is DOE—what would a coordinated MGI approach in terms of research development and demonstration at DOE offer. Specifically, reading from the slide, there—it offers, one, synergistic leveraging of state of the art tools in computation, experimentation, and information management, that—that's a potential savings. There could be quite significant cost savings with shared resources, so people aren't reinventing the wheel over and over again. And there's an interesting opportunity for broader crosscut technology impact. We'll talk about that a little bit more later on.
What I want to highlight first is that many individual programs and projects at DOE are already employing a lot of the MGI methodologies in what I'm calling a local innovation ecosystem, specifically addressing a clean energy challenge of different types. One good example is in the hydrogen and fuel cell R&D community, and that's what we'll be focusing the rest of the webinar on as well.
[Slide 10]
All right. So this is really the heart of our webinar. This is the good stuff. Materials researchers both in the hydrogen and fuel cell technology development have been early adopters of many of the MGI methods, and what we'll do for the rest of the webinar is showcase ten projects, the ones listed on this slide, which cover materials research relevant to fuel cells as well as hydrogen production, delivery, and storage, and each of these projects employs some aspect of MGI that will be highlighted by the presenter.
[Slide 11]
And before we get started with the presenters themselves, I just want to thank them for participating. They all volunteered to be a part of this webinar. Each one was given this template shown on—the shown template, and asked to summarize their project in one slide, which is a big challenge for scientists to have to do it. I appreciate what they're doing.
But the key emphasis would be the mission—the DOE mission in fuel cells and hydrogen, how does their research apply to that, and what are the successes and challenges of their technical approach, specifically including MGI elements.
So to kick off this fun part of the webinar, I'll be asking Jacob Spendelow of the DOE Fuel Cell Technologies Office to tell us about fuel cell catalyst research at 3M. Jacob?
[Slide 12]
Jacob Spendelow:
OK. Thanks, Eric. So this is a project that we supported a few years ago with 3M that was focused on developing fuel cell cathode catalysts with higher catalytic activity, and the approach of the project had some MGI elements. So the 3M catalyst technology is based on sputter deposition, which makes it inherently higher in throughput than the more conventional wet chemical catalyst synthesis techniques that are often used.
So 3M, by using two or more sputter targets, they're able to produce alloy catalysts that have a composition gradient on the substrate. They used a variety of techniques to characterize these gradient catalysts, including XRD and electron probe measurements, but doing the characterization and getting fast and accurate measurements of catalyst activity, surface area, and durability was a challenge.
So 3M's initial approach to evaluating these properties was to form an MEA with a composition gradient catalyst, and then they evaluated it in a segmented fuel cell. And despite a lot of hard work, they weren't able to reliably measure catalyst activity in the segmented cell, so they ended up doing the testing in a rotating disk electrode experiment instead, and that took several times as long and a lot more labor than a full combinatorial approach should take.
The combinatorial approach was still a success, because it did allow discovery of a very specific platinum-nickel ratio that showed a much higher catalytic activity than the surrounding compositions, but from a throughput standpoint, it could have been a lot better.
The composition gradient approach could see a big improvement if a method were available that provided more accurate and more reliable measurement of catalyst activity. There's a lot of requirements for such a method, but the really key challenge is, at least in the case of the 3M project, is that it has to be able to electrochemically condition the catalyst in the same way that the catalyst gets conditioned during fuel cell operation. And this is really important because during fuel cell operation, a lot of changes happen, changes to the catalyst structure, to the composition, and to the activity. They all evolve dramatically and they all depend on the conditioning parameters.
So in summary, both the people at 3M and those of us here in the Fuel Cell Technologies Office are very interested in combinatorial approaches for fuel cell catalyst synthesis and evaluation, but we currently have some challenges that are preventing us from realizing the full potential of this method.
Eric Miller:
OK. Thanks, Jacob. That was great. Next, I would like to invite Debbie Myers to talk about her fuel cell catalyst work at the Argonne National Laboratory. Debbie? Unmuted?
[Slide 13]
Debbie Myers:
It—this project started approximately three months ago. And so we're attempting to address some of the things that Jacob talked about in combinatorial testing of the catalyst and fabrication of the cathode catalyst layer. So it's a two part project. One part is focused on developing cathodes for the polymer electrolyte fuel cells, based on PGM electrocatalysts, which address the two main issues for this technology, which is cost and durability.
And then the second part is completely replacing the platinum group metals with non-platinum group metals of cobalt and iron, and optimizing the—both the activity and the durability of these materials. So the first part of this project then is taking very high activity platinum alloy materials that have been demonstrated in ex situ rotating disk electrode measurements, such as the one shown in the upper right of this slide, the Argonne platinum three nickel nanoframe catalyst, which has on a mass activity basis, or amps per milligram of platinum, 30 times that of the commercial platinum on carbon catalysts.
So we need to translate this very high kinetic activity of these materials into high performing electrode layers. So that portion of the project is taking these high activity materials and optimizing the composition of the electrode to get the optimum transport of oxygen to the catalytic sites, protons to the catalytic sites, and rejection of product water.
So the methods that we're developing in this project, which is really in its infancy, is to develop a quick way to fabricate these electrode materials and to test their performance in an actual operating fuel cell. So one of the techniques we're developing is using something similar to the segmented fuel cell that Jacob just mentioned, where we can individually address 25 different electrode spots across the membrane, and this is actually a commercial apparatus from NuVant Systems, and it's been used very nicely to determine oxygen reduction activities of materials, but we would like to use this then to look at the performance of electrode layers at high current densities and on air rather than oxygen.
So the second portion of the project is to develop the cobalt/iron catalysts that have been pioneered by Los Alamos National Laboratory, and these are combinations of iron, cobalt, and polyaniline that are pyrolyzed. So we're using a free slave robotic system to make various combinations of cobalt and iron and the polyaniline, and then changing the pyrolysis temperature, and we're developing a method then to quickly test the activities of these materials in an environment that's very similar to a rotating disk environment, but so that we can test multiple electrocatalysts at one time.
So the MGI elements that are incorporated are density functional theory, which is being used by Los Alamos National Laboratory to screen potential active sites in these non-PGM catalysts. We're also developing ways to quickly make electrodes and test their performance.
Eric Miller:
Great. Thanks, Debbie. I'd next like to invite John Gregoire to talk to us about some of the innovative combinatorial catalyst work going on at JCAP. John, are you there?
[Slide 14]
John Gregoire:
Yes. Hello. I'm John Gregoire, and I lead the high throughput experimentation project within the Joint Center for Artificial Photosynthesis, which is the DOE's Sunlight to Fuels Energy Innovation Hub.
My lab discovers new earth-abundant light absorbers and electrocatalysts using high throughput computation, experiments, and data science. The computational work guides and aids the interpretation of high throughput experiments and the ensuing data and data science ties together the different components of the project.
Something that has worked really well for us is from the onset establishing which properties are most easily measured by experiments and which are most easily calculated through theory. Along with this, we also determine properties that can be both measured and calculated to bridge together the different approaches. And we tie all of this together with an experimental and computational pipeline that avoids bottlenecks.
The top figure in the slide shows an example calculation predicting the stable electrochemical conditions for a given compound, which guides the experiments. The central figure is from the discovery of a new water oxidation catalyst. Our high throughput experiments not only measure the catalytic activity, but also a suite of other parameters to provide a more comprehensive understanding of the materials.
The bottom figure is the first public display of the web user interface to our database. We plan to release the interface in the next year, and one area in which we continue to struggle is how to effectively present high throughput data to the public. The data is collected with custom instruments that have unique noise and occasional artifacts, which we understand internally, but cannot be easily conveyed when releasing data on a million new compositions, which is what we've made and screened in the last couple of years.
So I think this is a universal issue with truly high throughput experiments that we'll be facing moving forward with the MGI efforts.
Overall, this HTE project is a successful demonstration of MGI strategies, and is directly relevant to FCTO through the renewable production of hydrogen fuel, but perhaps more importantly for long term FCTO benefit, the project has established high throughput catalyst discovery methods, including the invention of new electrochemical screening instruments that operate in—under fuel cell electrochemical conditions.
And I believe these techniques could meet the needs of many of the things discussed so far in this presentation. So with that, I'll thank the organizers and turn this back over to Eric.
Eric Miller:
Thanks, John. That's great. It's good to hear, and I think we're all taking notes at the same time. Next, we'll turn our attention specifically to photocatalysis and photoelectrochemical hydrogen production, and I'd like to invite Todd Deutsch to describe materials innovations, both at the University of Toledo and at the National Renewable Energy Laboratory. Todd, are you there?
[Slide 15]
Todd Deutsch:
Yep, I'm here. So first I'll discuss Yanfa Yan's project. So Professor Yan's project on theory-driven synthesis of new metal oxides for water splitting is just beginning, and seeks design materials that have the appropriate physical, optical, chemical, and electronic properties for photoelectrochemical hydrogen production.
The goal is to use theoretical calculations to guide the synthesis of these new materials that have the desired properties to allow efficient solar-to-hydrogen conversion and high durability. Since this project was recently awarded, there are no results to discuss yet. This project does include several MGI elements and uses fundamental design principles and DFT calculations to inform targeted synthesis. This step is followed by experimental characterization and validation, and as the process is iterated, each round of synthesis should result in material properties closer to those that are desired and targeted.
One particular difficulty encountered in this approach is achieving the properties that are calculated from perfect materials using typically imperfect synthesis techniques. The limited number of atoms that a DFT unit cell can accommodate don't always allow calculations for materials with trace level impurities, and semiconductor properties are significantly impacted by intentionally incorporating trace level impurities.
[Slide 16]
So transitioning to my project, which is surface validation on III-V materials for photoelectrolysis, which I collaborate with Clemens Heske at UNLV and Tadashi Ogitsu at Lawrence Livermore National Lab. And the goal of our project is to understand semiconductor photocorrosion for economical solar hydrogen. So achieving renewable hydrogen at a low cost from photoelectrolysis requires stabilizing the semiconductor electrolyte interface for a long time under harsh conditions. So our approach is to use advanced spectroscopy and theory to identify the corrosion initiation mechanism and design processes to modify the surface in order to mitigate it.
So one recent result is the correlation of spectroscopy with theory on nitride gallium indium phosphide, and this is based on the observation that nitridation led to the stabilization of the gallium indium phosphide surface.
So we've worked with UNLV and they measured nitrogen K-edge XES spectra at the Berkeley synchrotron, and then Tadashi's group at Livermore modeled the incorporation of various types of nitrogen impurity states in the nitrogen ion implanted gallium indium phosphide, and this established a base for predictive capabilities.
So the MGI elements incorporated obviously are theory, modeling and experimental, but beyond that, this has encouraged and enabled integrated R&D, where we have electrochemists, physicists, and theorists speaking the same language. And it's also enabled the creation of accurate and reliable simulations and supported the creation of accessible materials data repository in the form of a SharePoint site. It's also provided opportunities for integrated research experiences.
One particular challenge of this approach is getting materials that are pristine enough so that the spectroscopy is actually measuring the materials, and not the stuff that's on the sample. So we worked with UNLV and adopted a glove-bag extraction technique, so that we can get them samples in—as least contaminated as possible. And with that, I'll hand it back to Eric.
Eric Miller:
Thanks, Todd. Yeah. I think we're seeing another recurring theme in that we're not looking just at individual materials. We're looking at interfaces. We're looking at materials in their operational environment. I think that's a key point that's pushing the envelope for MGI, and it's definitely needed for getting the materials to the marketplace.
So—but we'll stay with hydrogen production for the next talk, but change gears and look at some materials development work in the solar thermochemical water-splitting arena. First, we'll ask Tony McDaniel of the Sandia National Laboratories to talk with us. Did I get the right one here?
Tony McDaniel:
Yep.
Eric Miller:
Tony?
Tony McDaniel:
Do we have the slide? I haven't seen the slide. There we go.
[Slide 17]
Eric Miller:
Ah, there it is.
Tony McDaniel:
We're using a [audio glitch] solar energy to provide process heat for a thermochemical water splitting cycle, and the cartoon in that first bubble on the top shows conceptually what a reactor might look like. But again, so the working material in these systems is a complex oxide that is heated to a temperature where it spontaneously reduces. The oxide is then taken off sun and the temperature lowered, and then that material, the reduced material, is exposed to steam. Oxygen goes back into the solid, and as a result, you split the water molecule and you collect hydrogen.
So conceptually, our thermochemistry is very simple, and in fact, our reactors aren't much more complex than that cartoon that you see up in that top—in that top bubble. Our need for MGI is to discover redox materials that have optimal thermodynamics. And so we know that to optimize essentially the end to end process efficiency, that the reduction enthalpy and entropy, they need to fall in a certain range, and that's indicated by that bullet there.
So we have, based on just, you know, experience with materials and oxides that we've chosen to run these cycles, we know a bit about what bounds the enthalpy and entropy reduction in this process. And basically, we're trapped between states where we don't want materials that are too difficult to reduce, it takes too much process heat to do that, nor too hard to re-oxidize.
And so in the past, if you go down to the middle bullet, we've done some very non-ideal, low throughput screening, again, based on chemical intuition, where we synthesize materials, where we have an idea of what we're looking for in terms of how labile the oxygen needs to be in the system that's typically found in oxygen separation membranes or solid oxide fuel cell anode materials or even electrolytes.
And so we have, just based on our chemical intuition, we've focused on perovskite materials, and we do a bit of synthesis and screening at a very low level. So here's an opportunity for DFT, in my mind. We know that the state of the theory today, it seems like predicting certain thermodynamic properties is fairly accessible. And so in this project, we're now taking a fairly pragmatic approach, baby steps, basically, trying to understand how a simple A site substitution in—let's see that material, if we go to the bottom bullet there, where a strontium manganite, a simple A site substitution of one element, cerium, for another element, zirconium, we notice that there's a dramatic difference in the thermal stability, if you will, of the material. That's—you know, basically, the onset of oxygen evolution. At what temperature the oxygen begins to evolve from the material.
And that change is huge. It's on the order of 500 degrees Celsius, by changing one atom for another. And so here, here's our initial—I would say sampling of what DFT may be able to provide in terms of building design rules for these redox materials. And I would say this project especially, and the materials that we're looking for, and what—a method for looking for those materials in discovery, it's probably amenable to not only DFT, but also a lot of the high throughput combinatorial methods of synthesis and testing that's been described in the previous projects. And with that, I'll thank Eric for the opportunity to speak.
Eric Miller:
Thanks, Tony. That's great. And we're going to follow on with Charles Musgrave, who will talk about related materials discovery and development work at the University of Colorado at Boulder. Charles, are you with us?
Charles Musgrave:
I'm here. Can you hear me?
Eric Miller:
I can.
[Slide 18]
Charles Musgrave:
Great. So Tony did a great job introducing the topic of solar thermochemical water splitting, and we are examining basically the same process. The overall goal, again, is to develop solar thermochemical reactors and processes that use novel materials to efficiently produce hydrogen at practical temperatures and process conditions.
Al Weimer is my collaborator, and he leads the effort on developing the reactors and the processes. I lead the materials discovery side of the collaboration, and we use rapid computational screening and prototyping and incorporate various aspects of MGI.
The specific materials we've been focusing on are redox flexible metal oxides that include spinels such as hercynite and perovskites. Our materials discovery approach, as I mentioned, incorporates various MGI aspects, and it's an ab initio computational screening and prototyping approach that couples with experimental characterization of materials and their hydrogen generation mechanisms. And the experimental effort, again, is done in Al Weimer's lab.
Over the last couple of years, we've been developing descriptors based on the ability to rapidly determine oxygen vacancy formation energies for metal oxide materials by calculating the materials formation energy, band gap, electronegativity differences, et cetera. And we want to do this because developing a direct calculation of thermodynamic properties can be quite time consuming, and if we can discover the descriptors that are simpler to calculate, we can rapidly screen through tens of thousands of materials rather than the slow approach of one material every few weeks or so.
In turns out that the oxygen vacancy formation energy is the key thermodynamic property important for determining the viability of these redox materials, at least in terms of the thermodynamic properties.
One thing that's maybe unique about us is that we tend to use more sophisticated methods, such as GW quasiparticle methods in our calculations that go beyond the standard DFT approaches. We find that although DFT methods are useful, they end up with too many false positives or negatives of candidate materials, and the compromise, of course, is that the methods we use are slower, but we think that's actually faster than experimentally examining false positives.
So some of the other aspects of the approach that we use is that we're developing descriptors for the kinetics of the process. For example, a key step in generating hydrogen with the solar thermochemical process is the hydrogen formation reaction. This is a surface reaction that tends to be rate limiting, and currently, there is no simple descriptor that relates to the kinetic barrier for the hydrogen evolution step. And so we've been working on discovering a descriptor for that, so we can quickly screen and prototype materials for good kinetic properties.
Another aspect of the work that we're doing is that we can use our descriptors to determine the various driving forces for the reduction and oxidation processes that split water and re-oxidize the material, and this allows us to computationally prototype the process itself and put that into a reactor design module.
The third panel on our slide shows computational results. It examined a solar thermal water splitting process using hercynite where the mass spec high temperature XRD and EDS was actually carried out in Tony McDaniel's lab at Sandia, and they had seen that using the high temperature XRD, for example, that the mechanism seemed to be a mechanism that really didn't change the structure of the material. It couldn't be a displacement mechanism.
And our computational results agreed with that. We predicted an oxygen vacancy formation mechanism, and also, we were able to rank the materials, various materials, based on their hydrogen evolution capability and at what temperatures would they reduce. And so that was kind of a nice example. And we're working on a co-authored paper with Tony McDaniel on that.
So we see that the biggest opportunity, at least with what we're trying to do, is incorporating kinetic filters into the computational screening approach, and to use corresponding design rules that could come out of those kinetic screening descriptors to develop materials that are both thermodynamically and kinetically viable as solar thermal water splitting materials. With that, thanks, and we'll move on to the next person.
Eric Miller:
Great. Thanks, Charles. That's—it's good to see both thermodynamics and kinetics coming to the table. I think everyone in this—all the researchers on this panel working in fuel cells and hydrogen production technologies and delivery and storage technologies all have that on their mind. Thermodynamics is certainly important, but the kinetics can be one of the larger challenges.
What we'll do next is we'll look at several other products that are looking at materials development more related to delivery and storage of hydrogen. First, we'll hear from Brian Somerday at the Sandia National Laboratories. Brian, are you there?
[Slide 19]
Brian Somerday:
Yes. Thank you, Eric. As Eric mentioned, I'm going to be talking about a different type of material class compared to previous view graphs. And specifically, I'm going to be talking about structural materials. And these structural materials are used across a range of components for high pressure hydrogen delivery and storage. And the specific phenomenon that we're interested in with the collection of projects that we have at Sandia focused on structural materials is the phenomenon of hydrogen embrittlement.
And I wanted to acknowledge my colleague here at Sandia, Chris San Marchi, and I also wanted to acknowledge that we have active collaborations with the International Institute for Carbon Neutral Energy Research, which is based at the University of Illinois, and Kyushu University.
And the collective objectives of these projects at Sandia focused on structural materials are to enhance materials testing efficiency, to establish an interactive materials database, to enable materials selection for these hydrogen delivery and storage components, and to develop predictive models, ultimately, of hydrogen-assisted crack growth. And these objectives support the Fuel Cell Technologies Office goals to facilitate the deployment of safe and cost effective components for hydrogen delivery and storage.
One of our differentiating approaches here at Sandia for studying hydrogen embrittlement is to apply our core capability in this subject area, and one of the key components of our capability is a specialized laboratory where we can do testing of structural materials in high pressure hydrogen gas, and in select cases, we try to apply these results for materials testing in concert with modeling to interpret the underlying mechanisms for the measurements that we're making, and also to determine what are the governing variables for the trends that we're measuring.
And I just wanted to highlight one of our—one of our activities with the graphics that are on the right hand side of this view graph, that the top plot is showing measurements of the key crack growth rates for a pipeline seal in a reference environment which is air, and also in high pressure hydrogen containing trace amounts of oxygen. Starting in the lower left corner of this plot, there's a dashed black line which represents these crack growth rates that are measured in air, the reference condition. In relatively high purity hydrogen gas, which is represented by the red symbols, hydrogen will accelerate these fatigue crack growth rates, starting at relatively low stress range levels, but as the concentration of oxygen is increased in the hydrogen gas, there's a delay in the onset of these accelerated crack growth rates, as the oxygen concentration increases.
And so we wanted to understand, what is the basic mechanism for how oxygen could be inhibiting the onset of these hydrogen accelerated crack growth rates we're measuring? And so we partnered with Kyushu University and the International Institute for Carbon Neutral Energy Research, and specifically, I worked with a theoretical chemist named Professor Alex Staykov, and his specialty is applying density functional theory modeling.
And so Alex modeled the approach of a hydrogen molecule toward an iron surface which already had pre-absorbed oxygen on that iron surface. And what he found was that as hydrogen—as the hydrogen molecule was approaching that iron surface with pre-absorbed oxygen, that the activation barrier for hydrogen dissociation—so to dissociate the hydrogen molecule onto individual hydrogen atoms—is greatly increased by the presence of oxygen on that surface. If that surface was a pure iron surface, the activation barrier for hydrogen dissociation is only 0.1 EV, but with the pre-absorbed oxygen, it's as high as 0.6 EV.
And so what we essentially are finding here is that oxygen serves as a catalytic poison to the hydrogen dissociation reaction, and without the hydrogen uptake in the material, ultimately hydrogen embrittlement cannot be activated. And this is a really great example of applying a modeling technique, which is essentially looking at the behavior at the electronic level to understand macroscopic measurements.
And so one of our successes in this collection of projects focused on structural materials is the integration of experiments and modeling, but we also have an evolving database for mechanical properties of materials that we're measuring in high pressure hydrogen gas. And some of the challenges that we've identified is if we're trying to understand the basic mechanisms of hydrogen-induced damage, we need to make sure that we're posing very specific questions and then applying the right tools to answer those questions, both experimental tools and modeling tools.
Eric Miller:
Great. All right. Thanks, Brian. We're going to continue on with Sandia National Lab and hear from Jonathan Zimmerman, who will be telling us about efforts to develop high performance steel specifically for hydrogen applications. Go on to the next slide. Hold on. Let me try to get your slide up.
[Slide 20]
Jonathan Zimmerman:
Oh, thank you.
Eric Miller:
Is that it?
Jonathan Zimmerman:
Yes, that's it.
Eric Miller:
Very good. Thank you.
Jonathan Zimmerman:
All right. Thanks very much. So this project is really about the problem that Brian was just talking a moment ago, looking at structural metals that are used for hydrogen infrastructure, and in this case, for fuel cell balance of plant components.
The idea here would be we want to make a material that is resistant to hydrogen embrittlement, or we want to identify at least them, but we'd also like it to be low cost, so that it could be used fairly freely by industry. We have a kind of working hypothesis that one of the elements that makes it high cost is of course nickel, and nickel gives it superior performance because it has a high stacking fault energy. This is an atomic scale property which basically measures how dislocations are structured in a material, and also how they'll interact with each other.
This is a new effort. It's just starting this year, and in fact technically hasn't even started yet. We kind of got added on to an existing experimental program that was started this year, and we will start in January, but our approach is to use density functional theory calculations to provide reliable estimates of stacking fault energies.
When we're doing this, we're working off a hypothesis that we really want to evaluate this correlation between stacking fault energy and metrics of hydrogen embrittlement, whether it's how does hydrogen embrittlement effect ductility or fatigue life or other properties. We want to look back and forth using information from the literature as well as information from an experimental component of this project to see can we connect the materials that have high stacking fault energy and not necessarily high nickel to higher resistance to hydrogen embrittlement.
If we can verify that correlation, then we can go explore our alloy composition space to look for low nickel but high stacking fault materials. This approach is a little different from the standard high throughput computing approach that you've heard about in other talks, because we're trying to use optimization algorithms to kind of efficiently explore the alloy composition space. The idea here would be not just to—first of all, to confine in terms of what's a realistic alloy composition that could be manufactured, and we'd go get guidance on that from our industry partner, Carpenter Technology Corporation, but it's also the case that we want to be able to see are there connections between the two, and really look through the space wisely, not just in a high throughput fashion.
If the correlation doesn't verify, and we can't make the connection between stacking fault energy and embrittlement, then we're going to look at alternative atomic scale properties, things that can also be calculated in DFT, and then maybe can be more firmly tied to macroscopic performance.
So the MGI elements, again, are in the steel discovery through first principles calculations, in this case DFT, and again, the key is guided high throughput computing. We want to integrate experiment by looking at calculated properties, experimentally quantified metrics, looking for correlations, looking at uncertainties and variations in these materials to see how much overlap there is, and use sophisticated algorithms to explore the composition space wisely.
We also intend to develop data analytics, measures of uncertainty with our estimates, as well as maybe composition rules, so that material—certainly our industry partners and perhaps even our research and development partners don't have to do this level of calculation every time. They can use design rules that we'll develop to know for a given composition what is their measure of stacking fault energy.
Eric Miller:
Great. Thanks, Jonathan. We're seeing quite a scope of materials innovations here, ranging from functional material systems to structural materials, and all of the above. We have one more presentation. Last but not least, we'll be hearing from Brandon Wood at the Lawrence Livermore National Laboratory, discussing his innovative approach to investigate chemical hydride materials for hydrogen storage, and let me try to get your slide up. There we go. Brandon, are you there?
[Slide 21]
All right. Let me try to make sure he's unmuted. OK. Anyone—that's good, we got to the last one before any trouble. This is not too bad.
Alli Aman:
Yeah. Of course. And it looks like—he's unmuted, but he's not able to speak. So Brandon, I don't know if you can hang up and call in really quickly again.
Eric Miller:
What we'll do is we'll ask Brandon to hang up and try to call in to see if his connection has been lost. It really is an interesting project that really does tie the time and length scales across the board, which is one of the early themes I mentioned on the MGI, so we hope we can get him back soon. I have a few more slides I can finish up until—
Alli Aman:
And it looks like he's trying to call back in right now, so yeah, if you want to come back to him.
Eric Miller:
Yeah, let me continue, and when Brandon comes on, we'll come back to his slide. How's that?
Alli Aman:
Perfect.
Eric Miller:
All right.
Brandon Wood:
Can you hear me now?
Eric Miller:
Oh, OK. You're back. We'll go back to your slide.
Brandon Wood:
All right. I'm sorry about that. I don't know what was wrong. OK. So this is a new project that we just got underway looking at chemical hydride materials, so basically replacement materials for hydrogen storage that could get you to much more sort of compact and lower pressure hydrogen tanks.
And there's a lot of materials out there that are—that are good candidates for this, actually, in many ways, but the real issue has been kinetics. And kinetics is really the focus of our project, is to try to understand kinetics, understand the limitations, and then be able to develop strategies to really overcome these limitations.
So our basic approach involves integrating theory with advanced synthesis and characterization techniques, and on the theory side, a large part of it has been bridging length scales. So what we want to do is develop kinetic models that treat everything from atomistic scale properties, things like chemical bond breaking, all the way up to transport properties, and ultimately sort of particle level properties that are very difficult to treat atomistically.
And so a lot of the effort has been to sort of stitch these length scales together and use techniques like density functional theory and combine them with techniques like base field simulations. In addition, in order to actually develop a modeling framework that can accurately predict kinetics, we really need a lot of information from experiments. And so this is tightly coupled with synthetic effort and the characterization efforts to do that, and that's led by Sandia.
Fortunately, we do have some clues that are out there in the literature, and these clues are really what's been guiding this effort so far. So for many of these materials, it turns out that if you change the particle size, so basically change the geometry, or if you add certain chemical additives, you can actually change the kinetics quite a bit.
And it's not really known I think at this stage why that happens. There are some theories that are out there. And what we would like to be able to do is pursue these directions to really first of all understand what are the levers that are out there that one could potentially pull in order to improve the kinetics, and can we use those to sort of develop a viable strategy?
And unlike a lot of the presentations you've heard so far, we're not really focused on screening multiple materials. We're actually looking—focusing only on one materials class for now, but digging very, very deeply, and trying to understand all the relevant processes in that materials class. But because the framework is intended to be universal, then ideally, we could be able to use this exact same approach for multiple materials, as other materials become available.
So what are the MGI efforts in this project? I think the most significant one is the integration of theory with experiment in this very tightly coupled feedback loop. And actually, the theory requires inputs from experiment, and also the theory is intended to basically guide specific experiments that should be done, specific length scales, specific geometric and chemical compositions, and so forth.
Another very important aspect that's been mentioned before by some of the other speakers is this getting toward real materials. So we're not really relying on very idealized models. What we're trying to do is get some of the sort of real world conditions in our models, to look at things like non-equilibrium conditions, different temperature/pressure conditions, different fluxes of hydrogen, and so forth, in addition to simple sort of idealized models based on ab initio calculations.
So the other thing is I mentioned that we're trying to make the framework as flexible as possible, so a large part of that is trying to at some point in the future ideally integrate this with maybe a materials database approach, where we could take this same approach that we've been developing and apply it to multiple materials. So that's it. Thanks a lot.
Eric Miller:
Great. Thanks, Brandon. And thank you to all of the panelists. I think we've seen a lot of interesting materials innovations and approaches. And keep in mind, MGI, in my definition from before, is all of the above. All of these things are a part of the MGI, and how do we keep track of all the data, and how do we better utilize our resources for faster development to the marketplace. We'll come back to that in a minute.
[Slide 22]
I do want to give you some motivational slides before I come back to the panelists and give them a question to ponder along those lines. And one thing I want to bring to your attention is that MGI is starting to find its way into commercial application, and I wanted to at least put up this one example, such as Dow Chemical Company, that's been integrating some of the MGI methodologies to accelerate some of their products to the market.
They've been—Dow in particular has been an innovator in this area, as illustrated in the example shown in this slide, specifically related to the olefin block co-polymers. Instead of focusing on that specific case, when you talk to the Dow researchers, which have a broad portfolio of materials-related products they're bringing to the marketplace, they will stress, and they have stressed, that the success of this type of approach depends not only on the availability of the resources that they have to implement these various experimental, computational, and data information management systems, but really, even more importantly, it's about establishing the right mindset for using these resources correctly and effectively within a real innovation ecosystem.
So we're seeing successes on the large scale when the resources are available, but the resources alone are not sufficient to be—to ensure success, as we're seeing. But the right approach, the right mindset, is showing great successes. And the question then becomes what can small companies do and other entities that don't have the access to those resources in the first place. And that's what we're thinking of—that's what we're looking for options, perhaps a government-enhanced collaborative approach, where shared resources could broaden the access of these capabilities to more and more applications.
[Slide 23]
I won't spend a lot of time on this very colorful slide, but this is one of the interesting envisioned approaches to a clean energy MGI that—what the innovation ecosystem might look something like, if there were some core capabilities that were developed, and they were being developed not only in functional materials and structural materials and extreme environment materials.
We realize that this is a very ambitious project, and that different materials for us relevant to energy applications really would require different sets of tools. But the interesting—can we develop some common core capabilities that could be developed in time through their application in parallel research efforts in all of these thrusts. And that's something to think about, from your point of view, how we might construct something like that.
[Slide 24]
The last summary I'll give before I open the panel discussion is I want to discuss some key considerations moving forward with an MGI mindset specifically for technology development and clean energy applications. I've included them on this slide. The first, again, we'll be coming back to this idea that industry pull really needs to be balanced or needs to balance the expansive growth and the scientific push. We're seeing a lot of new tools being developed, but being able to get them up into the industry with the right curation, the right organization, that's something that has to be done very carefully.
Developing the right combination of foundational technology resources and expertise, human expertise, to really man this equipment, is very important. We can't—this is not a plug and play. No component of MGI is plug and play. Everything requires very specific and skilled labor and expertise to really make it work. So keep that in mind, and just—we're pretty much in the weeds of the science generally, but we have to keep in mind that business-friendly mechanisms need to be incorporated from the beginning into this innovation ecosystem in order to really achieve the mission of accelerating materials to the marketplace. So that's my summary background.
[Slide 25]
And what I would like to do next—first of all, that's me surfing from Hawaii to Washington, and thank you very much for paying attention.
[Slide 26]
But now I'm going to put the spotlight back onto the panelists. And if we have a few minutes, I'd like them at their leisure to ponder and possibly contribute some of their thoughts on the main opportunities and challenges that they see in a broader adoption of MGI principles, specifically in commercial development of their technologies.
And if we run out of time, I'll actually ask them to submit their thoughts in writing to the webinar, so we'll have that published for public dissemination as well. So hopefully everyone is unmuted. Maybe we can just go down the list to get thoughts on this from—in the order of the presentations. Will that work? I don't know, should Jacob—are you on still? Do you want to give some ideas on this? He may have signed off, actually. OK, who is next on the list? Debbie, are you still with us?
Debbie Myers:
Yeah, I'm still here.
Eric Miller:
Any thoughts?
Debbie Myers:
So I see one of the challenges would be the integration of the computational aspect with the experimental aspect, and keeping them in the same timeframe, and making the computational aspect relevant to the experiments.
Eric Miller:
Mm-hmm. OK. And I think we'll just go quickly through, and I think we'll run out of time if we don't. And maybe John, are you with us still? Could you give us some final thoughts?
John Gregoire:
Yes. I think an important distinction in this question is whether we're commercially developing the high throughput technology or just applying it to commercial problems, because for small companies, I do see that there's going to be—the initial investment may not be possible. But yes, having the kind of national expertise you talked about in core centers where small companies could come to perform high throughput experiments I think could work really well. And in particular, I think it's the only way to really do high throughput experiments and computation where the end use experts are there working with the fundamental researchers to really make sure, as was just mentioned, the experiments and theory are directly related to the ultimate technology.
Yeah, so I think it's a challenge and hasn't really been exercised yet, but it can certainly be done very effectively.
Eric Miller:
Thanks, John. Todd, are you still with us?
Todd Deutsch:
Yeah, and I think I would echo what Debbie and John just said. I think the challenge is getting the theory and experiment to work together in a timeframe that's kind of where they work—I guess where they work together so that they're on the same time scale. Often, the experiment can be very kind of dirty, and you may find something out that leads you to figure out the theory was going and calculating the wrong thing. So while it's—it works well to—in theory to have them—you know, one leading the other, they really—the symbiosis is a little trickier in reality, so . . .
Eric Miller:
OK. Great. Tony, still on?
Tony McDaniel:
Yeah, yeah. So I would further add that I think if you're going to try to create something like a national asset that people can access, is you have—you have to come up with a way that can create maybe general tools, or really generalize certain methods, because you've seen just the great breadth in these ten examples. We all have different problems. The only thing that's the same in any of them is the word material. But all of the problems are different.
And so in my mind—and then the information you seek and the experiments you might run and the code you might—it's—if you're trying to create something that's going to move toward some national asset, how do you generalize that? That would be my comment.
Eric Miller:
OK. Charles, any thoughts? Are you still with us?
Charles Musgrave:
Yeah, I'm still here. I mean, we're philosophically totally in agreement with the MGI idea, and have been advocating this for the last 20 years, basically. So we're glad to see it coming. There's long been this promise that computation cannot just explain experiment, but that it can help guide experiment, and be tightly coupled with materials, process, design, development. And so there's—we think that it's really about time that we started doing this.
Of course, there's enormous challenges that some of the other panelists have already just mentioned, but the opportunity is there. Some of the challenges are that as you go from one phenomena to another, the computational requirements of the different theory methods that you use change enormously. And so knowing how to best use the computational methods and know when they break down and when they—when they give you good predictions that you can actually trust and hand over to your experimental collaborators is very important.
Eric Miller:
Very good, and we're getting some pretty good insights here. Hopefully, these are being recorded, because we'll be needing these insights again, and we'll call on these panelists to elaborate further. Brian, are you still there?
Brian Somerday:
Yes, and Eric, I thought I'd offer some remarks that are specific to my topic area, which is hydrogen embrittlement of structural materials. And having worked in this area for a while, I often hear a perception that the community has been working on this topic for decades, and so don't we know everything. And the answer to that is no. And for example, we really don't know how to develop low cost hydrogen compatible materials. There aren't any predictive models for hydrogen-assisted cracking.
And these MGI concepts offer new opportunities to answer some of those questions through the application of advanced tools and techniques. And to me, the key is to make sure that we're posing the right questions and applying the right tools and techniques toward answering those specific questions.
Eric Miller:
Good. Good comment. Thanks. Jonathan, you still with us? Can you add on to that?
Jonathan Zimmerman:
Yeah. One of the challenges that I see is a combination of something that John mentioned during his talk, as well as Eric, you just brought up a few minutes ago, which combines kind of the big data storage and access question with the maintaining foundational resources. I sit on a number of advisory boards for ICME-type institutions, and one of the challenges there is the—even if they do all the work and get a big load of data both from the experimental side and from the computational side, the question remains, is it going to be there in five years? Is it going to be there in ten years? Do they really have a financial plan for maintaining it? Or can they partner with some government institution to make sure that information will still be available?
And I think one of the challenges is managing the data, managing the access to the data, but making sure there's a support system in place so that that data is there when we need it in the future.
Eric Miller:
Great. Good. And down to Brandon. We still got you?
Brandon Wood:
Yep. Can you hear me now?
Eric Miller:
Great. Any final thoughts?
Brandon Wood:
No. I mean, certainly echoing what everyone else has said, I would add one thing that I think wasn't brought up, which is this moving toward sort of more realistic operating conditions. I think one of the challenges in MGI is if you're looking at really taking an idea and moving it to the marketplace, is making sure that it can be actually synthesized, make sure that it can be synthesized scalably, make sure that it has the properties that you would expect in the actual working device, as opposed to in sort of a very idealized framework. So this involves things like—well, certainly kinetics, but also looking at interfaces, looking at surfaces, looking at defects, looking at sort of the messiness of real materials, and I think that's a pretty formidable challenge, but something that we really need to address.
Eric Miller:
Great. And that kind of emphasizes the industry pull. We need to specify that up front, what the real operating conditions are, and what the product looks like in its manufactured state.
And that's a perfect ending. I think we did a great job. I really want to applaud the speakers who were presenting today. And now I'll hand it back over to Alli for some housekeeping.
Alli Aman:
Thank you, Eric, and thank you to each of the speakers. You guys did wonderful. Just a few housekeeping items. Just a reminder that we did record today's webinar, so a recording along with slides will be posted to our website in roughly ten business days. I will be sending out an email once those have posted.
And then we did get a couple questions, so thank you so much for that, and we will follow up with those questions offline. So on that note, thanks again everyone, and I appreciate you joining today's webinar. Have a wonderful holiday season, and we'll see you in 2015 for our next webinar. Thank you.
Eric Miller:
Thanks, Alli.
Brandon Wood:
Thank you. Bye.
Eric Miller:
Thank you all. Bye.
Alli Aman:
Bye.