ChemCatBio Webinar: An Introduction to Programmable Catalysis for Chemical Energy Technology (Text Version)

This is the text version for the ChemCatBio Webinar: An Introduction to Programmable Catalysis for Chemical Energy Technology video.

LADY MIAH KANE: Hello, everyone. Welcome in, welcome in. I'm going to take some time to let everyone join in and get settled. Thank you for joining us today.

We will get started shortly. Just waiting for everyone to get settled in. Thank you so much for joining us.

Again, thank you to those just joining. Welcome in, welcome in. We will get started shortly.

All right, I'm going to go ahead and get started. Hello, everyone. And welcome to today's webinar, an Introduction to Programmable Catalysis for Chemical Energy Technology, presented by the Chemical Catalysis for Bioenergy Consortium, or ChemCatBio.

I'm Lady Miah Kane, the webinar coordinator for the consortium. Before I introduce our speaker today, I'd like to cover some housekeeping items so you know how to participate and learn more about the consortium. Next slide, please.

ChemCatBio has several resources available on our website, chemcatbio.org. You can find tools and capabilities, publications, webinar recordings, interactive technology briefs, and more. We regularly update this site so please be sure to bookmark it and keep an eye on the news section on the homepage for notices on ChemCatBio sponsored events and opportunities like this one today.

You can also navigate to our tools from the website, including the Catalyst Property Database and CatCost. Another great way to get updates from the consortium is through the ChemCatBio newsletter called The Accelerator. Subscribe to learn more about ChemCatBio events, fundraising opportunities or funding opportunities, publications, new technologies and research projects, research teams, and more. Next slide.

A little bit on participating today, you will be in listen-only mode during this webinar. You can select Audio connection options to listen through your computer audio or dial in through your phone.

You may submit questions to our panelists today by clicking on the Q&A icon in your Zoom bar. To submit your question, type in the Q&A text box, press Enter on your keyboard. You may send in those questions at any time during the presentation. We will collect these and, time permitting, address them during the Q&A session at the end.

Now if you have any technical difficulties, that's where I come into play. I want you to message me in the chat. So just locate that chat icon and you'll be able to send me either a direct message, or just put it in the chat box, and I'll make sure to help you out there. And remember that the Q&A panel is for the questions about the content.

Automated closed captioning is available for the event today. To turn it on, click on the closed captions icon in your Zoom toolbar. We are also recording this webinar. It will be posted on the ChemCatBio website in the coming weeks, along with these slides. Please see the URL provided on the screen here.

Now, a quick disclaimer. I promise it'll be quick. This webinar, including all audio and images of participants and presentation materials, may be recorded, saved, edited, distributed, used internally, posted on the U.S. Department of Energy's website, or otherwise made publicly available.

If you continue to access this webinar and provide such audio or image content, you consent to such use by or on behalf of DOE and the government for government purposes and acknowledge that you will not inspect or approve or be compensated for such use.

Next slide. All right, our speaker today is Paul J. Dauenhauer. Dauenhauer is a distinguished McKnight University Professor at the University of Minnesota and director of the Center for Programmable Energy Catalysis, a U.S. Department of Energy Energy Frontier Research Center.

PAUL DAUENHAUER: OK, thank you for the invitation. I'm happy to be here to speak to you about an Introduction to Programmable Catalysis for Chemical Energy Technology. I'm the director of the Center for Programmable Energy Catalysis. And I really want to speak to you about this concept of programmable catalysis and how we can use it to develop new technology.

Let me start by just a quick public disclosure of conflicts of interest. I have private research contracts with ExxonMobil, Dow Chemical, Casale, and the Minnesota Corn Growers Association, located here in Minnesota.

I'm also the co-founder of, have equity in, and have patents that are licensed by some of these companies or all the companies below in different variations thereof—Sironix Renewables, Activated Research Company, Carba, and Lakril Technologies.

Let me start by talking about how refining of fuels and production of chemicals affects every American. It's a key part of modern society. And in states in the Midwest, like my state of Minnesota, agriculture is a key part of our economy. And converting corn and soybeans and other forestry products into chemicals and fuels is a major part of our future technology portfolio.

At the center of all of this is catalysis. And so we're always looking for new technologies that can not only improve existing manufacturing capabilities but even look for disruptive technologies that could completely change the way we do it to make us more competitive in chemical manufacturing.

And I talk about disruptive technologies at the beginning here because I really want you to think about where the state of chemical manufacturing is in fuel refining. When we look at technologies like artificial intelligence or quantum computing, we're talking about technologies that could completely change the way those technologies work but also those industries operate.

What I want to argue today is that programmable catalysis offers a comparable ability to disrupt chemical and fuel manufacturing. It could change the way we do manufacturing of chemicals and fuels, both from an efficiency, throughput, and physical footprint, in terms of capital, of how we actually process materials.

So I'll say more about that later on, but let me start out very simply here. If we look at where we are in catalysis—and I'm mostly going to talk about heterogeneous catalysis. Many of these ideas translate to homogeneous catalysis and definitely to enzymes.

But in the past 20 years, the nanomaterials revolution has really pushed catalysis forward by allowing us to create and modify advanced materials. And of course, zeolites have been around for a long time, but enhanced synthetic capabilities and characterization improve them. This goes from MOFs to COFs and all sorts of incredible new materials that let us tune the active site for chemistry.

But at some point, I think where we're at right now is we're pushing up against the limits of what's possible in catalysis. And all three of these plots relate to really important energy-related chemistries. Methanol synthesis to convert CO2, ammonia synthesis, and the oxygen evolution reaction are all key chemistries that use a lot of energy.

All of these, if you look at the vertical axis, is their reaction rate or some metric for the reaction rate. And the horizontal axis here is a descriptor of the catalysts themselves.

And what you see is—and this has been known for 100 years for some of these—certain materials exist at the peak of the catalytic rate. This is called the Sabatier plot. And the idea here being that certain structures are at the limit of what's possible for catalysis, and we need to find ways to go around this inherent limit, this fundamental limit of catalysis.

Now, if you think about what we do in terms of chemical manufacturing, the conventional strategy is what I'll call strategy one here. We find the reaction of interest. We find the conditions to raise the free energy of the reactants. And we let the reaction flow downhill over a surface that's engineered like a zeolite or a MOF or some supported metal or metal oxide.

What I want to propose instead is strategy number two. Instead of tuning the conditions of a reaction over a catalyst, to make the free energy flow downhill, we use a dynamic surface to create a local free energy gradient. In fluid mechanics, we would call this a pump.

So you can think of these like catalytic pumps, but the dynamics of that surface are what give us completely new capabilities for controlling the reactivity of molecules on surfaces. Now, this is not a new idea. This idea is hundreds of millions, if not billions of years old because biology found it first. In biology, these concepts are known as ratchets.

Ratchets exist as a form of proteins. They're what keep us alive. We use ATP to change the structure of surfaces. And they come in many shapes in many different applications. Our interest here is, can we adapt this fundamental concept of catalytic dynamic ratchets for the use of catalysis on metals, metal oxides, carbides, nitrides, et cetera, for the purpose of controlling conventional fuel refining?

So what we—if you really think of the concept here, I think if I really want to simplify this down, conventional static catalysis uses the approach on the left. We increase the free energy of the reactants and we let the reactants flow down an overall free energy gradient.

And you can think of the surface of the downhill skiing here as the catalyst. We change the surface of that free energy, such that the molecules can move downhill as fast as possible, or if we're trying to be selective, so that they move one way down the hill versus the other.

Instead, what I'm talking about is programmable catalysis, what you see on the right there. In surfing, there's a local free energy gradient, which is the wave, and that's going to move molecules forward. But if you look over a long distance, the surface is generally flat. The Earth is round, of course, but locally, it's relatively flat.

So to do this, if we want to create a local free energy gradient on a surface, we really have to think about how molecules interact with the surface. And if you've seen variations of this talk before, I'm going to talk about something completely new, which is how do molecules interact with surfaces.

Well, molecules interact with surfaces by electronic interaction. We have a surface, in this case, that's a metal. If you look at the diagram on the left, that green there is a ruthenium film, and we're adsorbing ammonia to that surface.

And so, in this experiment, what we're going to do, we're going to take—and this is the diagram, on the left, of what we're doing. We're going to take this ruthenium surface, which is actually supported on a device where we've grounded the bottom layer.

And in between the ruthenium and that conductive support, we have an insulating high-k dielectric film. Now when ammonia, when we switch from low concentration of gas of ammonia in the gas, which is state 4, to a high concentration, which is state 1, we go from a surface that has low coverage of ammonia to a surface that has high coverage of ammonia.

And when that happens, ammonia is donating electron density to a surface. Now, with this approach, if we fix the potential between the top and the bottom surface, we can actually measure the current that exists between the top and bottom surface in the ground upon adsorption of ammonia.

That's what you see here on the right. That's experimental data of a positive current peak upon ammonia adsorption. Now when we switch back to go from state 2 to state 3, we're removing the ammonia from the gas phase, and ammonia desorbs. And ammonia is taking those electrons with it, and we see a negative current peak that's equal and opposite.

So, in this case, if we do this peak measurement over many temperatures, if you look in the top-right here, you can see these are representative peaks of measured current upon the adsorption of ammonia at many different temperatures.

First of all, you can see the kinetics are different, as well as the total volume or the total area of current, the total charge that's been transferred, is different with temperature. Higher temperature in this case gives you more electron exchange.

And you can see there in the inset on the left that the adsorption-desorption is the same within experimental error. So why? What determines the extent of electron transfer? And why is it larger at higher temperature?

Well, at higher temperature, you get larger difference between these two concentrations of gas-phase ammonia of what's on the surface. There's more ammonia exchanged. Now we can take—using a kinetic Monte Carlo model developed by my colleague Matt Neurock, to determine how many ammonia molecules are going on and off the surface.

And we can compare that to the total number of nano coulombs of electrons that are removed and added from that surface. And when we do that, we get the data on the bottom right, which tells us we were adding about 6.5% of an electron for every ammonia molecule that's added to the surface. That's a lot.

Now we can do the same experiment and think about this conceptually of what's happening. On the device on the left there, you can see that the concept of the device is depicted for the materials ruthenium, the insulating hafnium oxide layer, and the conductive silicon support.

When we start out, we have a low coverage of ammonia, and the ruthenium and the silicon are in electrostatic equilibrium. Now when we add ammonia to the surface, we go to a high concentration of ammonia. Ammonia absorbs and contributes its electron density to ruthenium and raises the local Fermi level there.

That then has to relax because we fixed the potential between ruthenium and silicon. And so when it relaxes, those electrons flow out of the material, and we measure that as the current, and the work function decreases.

Now it decreases upon adsorption and then increases again once we remove the electrons. So you can actually understand this as a concept of accumulating electron density in the ruthenium. And then removing it, the titration aspect of this, is measuring the electrons that flow out to return the system back to equilibrium.

And we can do the same exact experiment with hydrogen. So, in this case, we're going between 0.5% hydrogen and 100% hydrogen. If you look in the top right, we see the exact same kind of adsorption and desorption peaks. And we can do those over many, many cycles.

On the bottom there, you can actually track the accumulation and de-accumulation of electrons in a platinum surface with time. So we can say we're adding electrons with adsorption and we're removing them.

In this case, if you look at the adsorption isotherms over many different—or over two different pressures of hydrogen, you get the plot on the left. And if you look at the difference in hydrogen with temperature, you can see that on the bottom left there, as temperature increases, the total difference in adsorbed hydrogen increases with temperature.

And that's exactly what we see experimentally. If you look on the right, as we add more hydrogen, we're removing more electrons. And the slope here, if you remember for ammonia, it was 6.5%. Now we're only contributing about 0.2% of an electron to the platinum surface. Or hydrogen, as you would anticipate, has a much lower electronic interaction with the surface.

Now, one way we can think about this is if we think of the electron-sharing with a surface, how could we use this to understand the reverse process? Because what we want to do in our programmable approach is actually do the reverse.

Instead of adsorbates affecting the electron density of a surface, we want to control the electron density of a surface to modulate the molecules on that surface, such that we could do what you see in the picture on the right here.

I could take a catalyst in state 1, change it to state 2, and change it to state 3, such that the binding energy of the molecules is changing. And as I change the binding energy of the molecules, I'm changing the binding energy of the products as well.

And if those two don't change in equal amount, which they almost never do because they have different binding characteristics to the surface, you're going to change the transition state barrier. So very clearly, you can see, in the stronger binding case of state 3, the activation energy is much lower than it is in state 1.

So how do we do this? Well, if we can modulate the Fermi level of the material that we're working with, then we can actually alter the binding characteristics of molecules on the surface. It's the exact reverse process of what I was just showing you a second ago.

And that's what we do with this device that we call a catalytic condenser. This is the same device we were using a minute ago, only now we're going to do it in reverse. In this case, we have our hafnia insulating film in the middle. That's the blue-purple material.

It's supported on a conductive silicon wafer. Then on top of that, we put a sheet of graphene. It's very conductive. And then on top of that, we put our active sites, which in this case are alumina.

Now, in this case, we're not going to apply a zero potential. We're going to apply a potential that's either positive or negative. And to this capacitor device, we will accumulate either electrons or holes in that top layer.

Now when we accumulate electrons or holes, we're going to change the electronic characteristics of that alumina surface and, therefore, change its catalytic properties. We're going to use isopropanol because it's a well-known probe reaction, and we know a lot about it.

So if we go forward here, this is what the device looks like as a cartoon in the top left and a photograph on the bottom left. So that's an American quarter there. You can see the device. Looking down, you can actually see where the sheet of graphene is.

You can see that there's a little gold contact there. So we can make this electrical contact with the surface. My colleague, Andre Mikhoyn, at the University of Minnesota, has sliced this with a FIB system and then imaged it such that you can see, actually, the sheet of graphene and alumina that we've grown sequentially from the bottom up.

Now when we take this device, we can use this isopropanol dehydration reaction as a probe of how good the catalyst is as we change its applied potential. If you look at the top of the diagram here, you can imagine a reaction where we take our device and we put isopropanol on it at low temperature.

We then linearly ramp the temperature, and eventually, the isopropanol reacts to make propene. Now for conventional materials like tungsten oxide, alumina, titanium, zirconium, this experiment has been done for a long time. It's called a temperature-programmed surface reaction.

And we know that the peak of formation of propene happens at different temperatures, indicative of the performance of that particular material. So tungsten oxide and alumina are better than titanium zirconia.

In our case, we're going to do what's on the bottom. Alumina is the device that we've—the active site we've put on this device, in the middle. And presumably, if we remove electrons from that alumina, we're making it more acidic, and our alumina should give a peak of propene that slides to the left. So let's go to the experimental data.

So, to do—to understand that, we need to understand how many electrons we're pulling out of that alumina surface. In other words, if we pull electrons out, we're making it more acidic. And the number of electrons per active site we pull out is a metric of the acidity of that particular active site.

So here you can see we're measuring the electron density that's been removed. You can see the total number of electrons here is about 4 times 10 to the negative 12. And that, we can then divide by the total number of active sites.

And if you look over here on the right, you can see the electron density that's been removed in the red bar versus the site density. So we're removing somewhere around 10% of an electron per active site of alumina.

Now what is the effect on the experiment? If you look on the left here, what you're looking at is the temperature-programmed surface reaction, where we're measuring the formation of propane with a mass spectrometer as a function of the applied potential.

And what you can see at low voltage, zero volts, on the bottom there is we see a peak of propane formation exactly where we expect it to see it for alumina. But as we start to apply a positive bias, we're removing electrons from alumina and making it more acidic, and the catalyst slides to the left. It's become more acidic.

And what we can do then is take these peaks and interpret them and actually pull out an activation energy. And what you can see is the activation energy here drops from about 115 down to around 100 kilojoules per mole. So as I said, we've made this more acidic and we've made it a better catalyst by depleting it of electrons.

My colleague Matt Neurock at the University of Minnesota has taken this system and this particular active site and removed electron density from it, and compared that to the neutral site that we would expect.

And he predicted, as you can see in the data on the right, that we would change the binding energy of isopropanol and we would drop the activation energy about 0.25 electron volts or 25 kilojoules per mole, roughly, for about 12.5 electrons removed per site.

Now we removed somewhere on the order of 10% of an electron or less, and we saw about 15 to 20 kilojoules per mole. So we're in about the same order of magnitude of what's been predicted by DFT.

So you can back up and think about what we're doing here. If you think of all the possible catalytic materials that are out there, and we can design any type of solid acid that we want, we've essentially taken alumina and given it new properties with the turn of a dial.

We have alumina there in blue. It reacts when we're around 120 degrees C. And if we apply plus 3 volts to it, we pull out those electrons. And now it's reacting to give our peak propane formation around 70 degrees C.

So we can presumably do this, though, with any active site of interest because this is a platform device. We have this hafnium film, which allows us to accumulate electrons or holes in the top layer.

We have graphene, which distributes the charge horizontally. And then on top of that, we can decorate it with different types of active sites. I was showing you before with an oxide. Now I'm showing it to you with clusters of—I'm going to show you now with clusters of platinum.

But we can put different types of active sites depending on the different types of chemistry we want to do. And the metrics we care about are how many electrons I can add or remove because that tells me the extent by which I can modulate the catalyst active sites.

The other metric I care about is how fast I can move electrons across that surface. That's the corner frequency which I can measure. Most of the devices we work with operate at about 1000 Hertz before they start to see any degradation in the speed of electron transport.

So let me go ahead. I'm going to show you now some data for a platinum device instead of an oxide. So if I do the same thing now, only I exchange the alumina for platinum nanoclusters, then what I'm going to do is I'm going to either add electrons or withdraw electrons from the platinum nanoclusters.

And to evaluate the effect of doing that, I'm going to absorb carbon monoxide to that surface as a probe. And you can see here pictures of the platinum nanoclusters. My colleague has taken this again and sliced it.

And it's got a few layers of platinum clusters, but you can see they're about 2 nanometers thick. They sit on this carbon film. And we're going to absorb carbon monoxide to those active sites.

Now, what you can see are two data sets. I've done this—I've made this device—the postdocs have actually made the device in two different ways. On the left, you see the condenser, as I was just showing you in the TEM images, where we had two- to four-nanometer platinum nanoclusters on the graphene surface.

In panel b, you can actually see this with a very thick, continuous film of platinum. Now, if you look on the left first, what you can see is that I've taken our conventional device that I was showing you earlier and applied zero volts to it.

And then on—I have also applied negative 6 volts and negative 3 volts, and on the bottom, plus 3 volts and plus 6 volts. And what you can see if you back up is you can see that this peak is shifting from left to right.

When I apply negative 6, I weaken CO binding and it comes off at lower temperature. And when I apply positive volts, I'm removing electrons from the platinum. It binds stronger. And many of the sites are desorbing CO at higher temperature.

Now, if you look in panel b, when I make the platinum really, really thick, the accumulation or de-accumulation of electrons that's happening in the platinum is happening at the interface on the bottom, and the chemistry is happening at the top.

So all of this effect is shielded when I have a very thick platinum film and there's no effect whatsoever. So this is a good control to understand why is this effect happening and do we know it's happening at the platinum surface interface.

In this case, if I have a small amount of platinum and it feels this electronic effect, I can shift the binding energy about 20 kilojoules per mole. Now, where do I want to go? I want to make devices that are more powerful and faster so I can more effectively modulate the chemistry.

And one way to do that is to swap out hafnia for an alternative material. And there's many we're looking at. I'll show you one that we just published last year called an Ion Gel. You can see here this is a printable material.

So I can actually coat these and make these in large surface area. It has the benefit that it gives me much higher capacitance, but it has the negative that it's only stable at up to 200 degrees C.

Even that, there's a lot of chemistry we can do at 200 degrees C or below. And what we can do is we can again quantify with this device, that instead of hafnia has an Ion gel—you can see the cartoon at the top left—I'm again going to put carbon and platinum on top of that. I'm going to use CO as a metric.

Only this time, I'm not going to do this vacuum mass spec temperature-programmed desorption experiment. I'm going to do infrared spectroscopy. So I'll take this device. I'll put it in the chamber you see down there on the bottom left.

And I can bounce light off of it and track the formation of adsorbed CO on platinum. In this case, I can change the temperature of the device and I can change the gas phase concentration.

Now, in this case, I'm going to fix the gas phase concentration of CO. And I'm going to measure how much CO2 is on the surface as a function of the temperature and the applied voltage. And that's the data you see on the right.

And what you can see is I can change the adsorption energy of CO by about 16 kilojoules per mole. Only this time, instead of using 12-volt range, I'm using about a 1-volt range. And that's because this device is accumulating many more electrons or holes at lower voltage because it's a higher capacitance device.

OK, so if I summarize what we're doing, we've talked about molecules adsorbing on surfaces and transferring electrons to the surface. And then we talked about, can I do that in reverse using a condenser? Can I modulate the binding energy of molecules on surfaces by applying a voltage?

But if it's possible to do that and to do that with time, as fast as 1000 Hertz, I got to ask the question, for any chemistry and combination of active site, how would I like to change it with time? Would I like to change it very quickly, very slowly? Would I like to send signals of high amplitude and then low amplitude?

There has to be some insight as to how to modulate a catalyst with time. And that input is the idea of a program. I'm going to give it some sort of perturbation with time that's tuned for that particular chemistry active site combination. And that's what programmable catalysis is.

So what can I use to think about the types of perturbations, the types of input catalyst programs, to give me high performance? Well, this is actually where this research started about six years ago, actually, longer than that.

But anyway, you can see here a diagram that is plotting the turnover frequency of a reaction as a function of a descriptor of the catalyst, in this case, the binding energy of one of the surface intermediates, B star in this case.

You can see here the classic Sabatier plot, where there's a peak material, the best absolute material, exists at a binding energy of about 0.05 electron volts. It's in red. And then other materials with different binding energies of B, those descriptors will give slower overall rates on a log scale.

Now each side of this Sabatier plot corresponds to a rate limitation. But those rates actually continue to increase, as you can see in the dashed lines. But of course, the catalyst can only go as fast as the slowest step.

Now, if we could change the binding energy of molecules on the surface, I could move side to side, as you see here in purple. And if I could do that fast enough, I could actually exist on the dashed lines for small moments in time and I could achieve catalytic rates of, in this case, almost 10 to the 2 or 100 turnovers per second, or several orders of magnitude faster than the about 10 to the negative 1 turnovers per second in this particular simulation.

So that's the concept. And what we're then doing is if we're changing the binding energy of molecules with time, our free energy profile is not a single profile anymore. It's a single material that can exist in many different states—state 1, state 2, or anything in between.

So it's moving up and down, similar to those animations I was showing you earlier. Now you can think of this mathematically, but you can also think of this conceptually. In state 1, the transition state barrier to go from A star to B star is lower than it is in state 2.

But in state 1, the desorption energy to go from B star to B in the gas phase is higher. Alternatively, in the state number 2, in brown, the transition state barrier to go from A star to B star is really big, but the desorption barrier is now small.

So you can think of what it's doing is the program that's good for this reaction changes the catalyst, so it's good for each sequence of steps in the catalytic reaction itself.

So you can think of this now more mathematically by doing kinetic modeling of these types of systems. On the top there, you can see we're changing the catalyst between two states in a square waveform.

So it goes from strong binding to weak binding, to strong binding to weak binding. And in the different cases of strong binding or weak binding, we have different molecules that cover the surface, in this case, A star or B star, as you can see down there on the bottom right.

Every time we flip the catalyst between states, though, we're depleting the surface and readily absorbing B star in the weak binding state, and the turnover frequency instantaneously spikes. So, in the middle data set there, you can see the rate of reaction all of a sudden goes up and then depletes.

So you can immediately tell that there's some frequency here that is optimal for this to get as many of these spikes as close together as possible. So that is this idea you see here on the left in this plot.

This is a very complicated plot, but the one on the left I just showed you a little bit ago, in orange there is the Sabatier volcano. You can see the dashed lines are the extensions of the rate limitations. And in purple is the tie line of this resonance frequency.

On the right now is actually the simulation output. On the black is the—on the right is the turnover frequency that we observe. And you can see that in red there is the maximum catalytic rate we could ever get.

And in black, you can see that as we increase the frequency at which we're switching the catalyst between states, the rate starts to pick up around 10 to the negative 4 Hertz, very slow oscillations.

It surpasses the Sabatier peak and then eventually reaches its maximum at about 10 to the 4 turnovers per second. Now there's another metric we can use here called the effective rate, where we multiply the turnover frequency by the efficiency of our system, and that gives us the effective rate.

And that exhibits a maximum at what we call this resonance frequency. This is where we've matched the applied frequency of the catalyst perturbation to the natural frequencies of the catalyst itself.

OK, so putting this all together, this input program, if we use it with a condenser, we're switching the condenser between two or more states, we can envision the integration of this chemistry in each step as it changes with what's happening.

In this case, we put in a positive charge to make A convert to B under strongly acidic state. And then we weaken the acidity by putting the electrons back in, and that's when B comes off the surface.

So this is where we started thinking a lot about how do we design input programs. So if we have a device like we have in the middle, that's one way to do this. Other ways to do it include dynamic strain or dynamic illumination of surfaces with different frequencies and wavelengths of light.

We can start to think about what types of inputs are really important to control the chemistry the way that we want. We want simple sinusoidal plot frequencies. We want combinations of frequencies or something way more complex.

So we wrote a whole paper thinking about the philosophy of how you can design catalyst input programs to control for rate. We can also control for selectivity. We can use dynamics to push reactions in one way versus another.

I'm not going to—I'm going to show you one small data set for that. But the other strange thing that comes out of this is the idea that we can use programmable catalysts and input perturbations to control conversion. And this is really unnatural for heterogeneous catalysis and catalysis in general, but let's go focus on that.

If I run a simulation now of just A star reacting to B star and A in the gas phase converting to B on a surface, I can run that simulation, for example, in a batch reactor, and run it for a long period of time until it comes to a steady state.

So the data you're seeing on the left here is a simulation, where A is reacting to B. And I can start with any composition in a batch reactor that I want. And the catalyst is oscillating. It's got an amplitude of 0.2 electron volts. So a relatively small perturbation similar to the ones I was showing you earlier.

What you can see for this reaction, we've selected the parameters such that equilibrium conversion is 40%, but the reaction on a dynamic system does not go to equilibrium. It goes to a new steady state that's 30% higher in conversion.

And no matter what composition you start with, it goes to that steady state. Then ultimately, if you turn off the dynamics, as required by the laws of thermodynamics, this system will relax and it'll go back to the equilibrium, as it has to.

If I turn the dynamics back on, it goes to this new steady state. Well, first of all, that's really exciting because the idea that I can control the steady-state conversion of a reaction that's beyond or different from equilibrium is a new way we could control chemistry and control the extent of conversion in reactors or how we pick the conditions to operate.

The question is how does this happen? This comes back to this concept of the ratchet I was talking about earlier. If you look at the picture on the right, this is an extreme example I picked to really show how this can work.

If you have a catalyst that's in the strong state, then molecule in red can easily traverse the barrier in a downhill reaction to form molecule in purple. And then when I flip the catalyst state to a very weak binding state in blue, the molecule in purple can readily desorb, but it can't react backwards.

And this is the idea of the ratchet. It's very easy to move from left to right but very difficult to move from right to left. So the concept of the ratchet fits with this concept of what unidirectional movement. And that's only possible because we're putting in this energy input to move the purple molecule from the strong-binding green state to the weak-binding blue state.

So the question then becomes, well, which way is my catalytic ratchet facing? Is it moving reactions to higher conversion than equilibrium or lower conversion? And it turns out that they can do both, depending on the parameters you give it.

If you give it a certain input program, you can either make it go forwards or backwards. And what you're looking at here is a simulation where we've actually changed many of the parameters. We've changed the perturbation frequency, but we've also changed where the amplitude starts and ends.

We fix the amplitude, but we slide it left to right. And it turns out you can make reactions either promote a reaction forwards or backwards. This is something we learned very early but we didn't understand it.

And just recently, this past year, we wrote a paper where we developed a descriptor—you can see there, lambda—that tells you if your input program is going to make a reaction go forwards or backwards.

Now, putting this all together, you can start to imagine what this does for real chemistry. Real chemistry is not a single elementary step, most commonly. Especially the big reactions that are important for energy, they're a sequence of steps.

So A will adsorb, then A star to B star to C star to D star to E star and F star, and then F star could come off for example. So every elementary step here is a ratchet. It could have bias on the kinetics and the overall conversion of reaction.

So we have to come up with some way of thinking about every single elementary step. And so if you look at this example here of C star to D star, you can imagine a scenario like you see there.

Every single step has a forward and a backwards reaction. So it has a reaction constant k1 and k minus 1, for example. But it is in two states. So there's two rate constants corresponding to each of the two catalyst states.

Now if you think about what's happening here, of the four rate constants, inevitably, one will be faster than all the others. That's not always the case, but in general—but it frequently is, and it's not a requirement for this to work.

But you can see here, in this case, it's very fast for orange molecules to react to purple, but very slow for purple to go back to orange. You can also see that in the data down below.

It's very quick. The green state allows for a very rapid conversion. And the blue state allows for very slow reverse reaction, and that's this ratchet concept. Now, obviously, the frequency I apply to this matters because if I back up a slide here, if I was to give this enough time, the purple could go all the way back to orange.

So at some frequency that I'm applying, there's plenty of time for the orange to go to purple, but not enough time for the purple to go back to orange, and that's the idea of a cutoff frequency.

If you look at the surface coverage that I'm predicting for these different types of ratchets—this one, in particular, is a simulation—you can look at the surface coverage that's averaged over time, and that's the heat map you see there on the right.

And what you see there in yellow, yellow corresponds to essentially a time-averaged equilibrium where the ratchet has no kinetic bias. If you run a reaction under those frequencies and temperatures in yellow, it'll go to equilibrium.

But at some point, at lower temperatures, when the kinetics slowed down and I apply a faster frequency, one of the two directions is allowed to pass and the other one is not. That's beyond this cutoff frequency when I say the ratchet is in the on state. In this case, it really biases the system to very low coverages of C.

This is all predictable by this cutoff frequency, which is actually determined by this metric here identified in yellow—in red. So all of this is definable. And we can put this together to understand big, complex chemistries.

So let me put all this together. When I talked before about disruptive approaches to catalysis, I want to think of them like we think of disruptive approaches to quantum computing or AI—the ability to do things that were previously unimaginable before.

Running reactions at orders of magnitude faster than conventional materials is unimaginable because we don't know how to do it. I think what's great about catalytic resonance is it gives us a strategy for how to do it.

If you look at the left there, we can imagine making materials now because we've demonstrated that we can change the binding energy of molecules and surfaces. And if we can do that with enough amplitude and enough speed, we could achieve these rates.

Number two is we could drive reactions to conditions—to conversions that were not possible before. You think about a lot of the things we do, like ammonia synthesis, a very important fertilizer for the manufacture of all sorts of agricultural products.

We operate that at extremely high pressure and temperature so we get the rates that we need and at least sufficient conversion that we can recycle the reactants. That's a very expensive process to build. We need so much capital and it has to be extremely large to make it efficient.

But if you could operate something like a high-pressure ammonia process at low pressure, you could potentially use these in different locations and for different applications, like storing energy for deployed soldiers or something like that.

The one I didn't get a chance to talk about because I want to finish on time here is the third category, which is if I have much more complexity in the reaction, which is almost all chemistry, for example, A could go to B or A could go to C, you can start to use—think about using dynamics to push reactions one way or the other.

If I can tune the bias of elementary steps, I can actually push reactions away from the products that I don't want. And what we show in this simulation here is that we can take products that were, before, never possible to be selected for and push a reaction all the way to almost perfect selectivity using the right frequency and amplitude of catalyst perturbation.

That's a very exciting area. There's a lot of potential to make highly selective catalysts using that approach. And I think it's a big part of the economic potential of programmable catalysis.

So let me end by putting together this with two examples and a vision for where our center is going. So our center is the Center for Programmable Energy Catalysis. And we put all of these ideas together using a team.

So on the bottom left, we're using these advanced mathematics, but also we use artificial intelligence and machine learning to understand catalytic resonance theory and the complexity that we get with big complex models, as long as fundamental concepts.

We make, synthesize, fabricate different devices, and we also have a light team. And we use spectroscopy at national labs to understand what's happening on those surfaces with time and how the surfaces potentially restructure.

On the top right, we're using data science to understand—this is one of the data plots from our recent papers, where we're trying to understand the complexity of reacting systems, but also combining that with advanced reactor design that allows us to probe what's happening on surfaces as we modulate catalysts.

And on the bottom right, we're also exploring surface reactions, both in kinetic modeling but also in computation, to understand how molecules on surfaces change as we accumulate electrons or holes. All of this together comes together in our center, where we converge on these focused scientific questions of how programmable catalysis works.

So let me end here with two examples that convey what I think is the disruptive capability of programmable catalysis. This is work I did with Dion Vlachos. At the University of Delaware, we were studying the reaction of H2 and N2 to make ammonia.

And this is a complicated reaction that it can happen on multiple surface sites. The model here happens on a terrace site and a step site. And so there's two parallel pathways where we're adding hydrogens 1, 2, and 3 to get ammonia on the surface, and then that desorbs.

In this particular model on ruthenium, you can see one of the simulations here on the left were reacting at 320 degrees C and 50 atmospheres. Now what we're doing is we run this on a static ruthenium surface, and then at some point in time, we turn on a dynamic strain perturbation.

So we're oscillating between positive and negative strain. And what you can see is at different frequencies, we get different extents of steady-state conversion. At 350 hertz, the reaction goes backwards. But at 1,000 hertz, it actually goes forward.

And by 5000 hertz, we're pushing this almost to 20% higher conversion just by applying this surface perturbation at 5000 hertz. And again, if you turn off the dynamics, the reaction comes back to equilibrium.

So we haven't violated equilibrium or thermodynamic rules. We're just giving it—putting in energy through the surface to drive it in the direction that we want. Let me end with one final example, which is a very important reaction for producing reductants, and that is splitting water.

If you want to take water and produce—break it apart to OH and then eventually evolve hydrogen, you have to also evolve oxygen. And you can see here, this has a very famous volcano that you can see in panel C.

This is actually presented as an overpotential volcano, most commonly. And you can see the most famous material here is one that's not—that's quite expensive, iridium oxide. That's the teal material you see there at the peak.

So what Dr. Sallye Gathmann has done with her Ph.D. is take this system and apply a kinetic model to it. So the first thing we did on the bottom left there is we recreated the overpotential volcano as a rate volcano, which in this case is a current volcano. You can see it on the bottom left there.

And then we were oscillating, using a modeled condenser, the binding energy of molecules at two different overpotentials, 250 and 350 millivolts. And what you can see here is you can actually accelerate the reaction such that you could achieve the DOE current target by applying a lower overpotential.

So by achieving faster catalyst, you can apply lower overpotential to the reaction. And you can see that depending on the parameters of the model—we actually tried to make this model account for all possible scenarios—you can actually reduce the overpotential significantly.

So let me put all this together. I think what's exciting about this is it's a strategy for producing the next generation of catalysts. A lot of times, we think of catalysts as like, I got to find a magic material, a material that has a certain particular active site that gives me better performance.

And the problem a lot of times is synthetic people don't know what that site would be, or if we can predict what a site would be, it's not stable. But in this case, we really are thinking about changing a material with time and giving it that input such that we can achieve the performance that we want.

The benefits, of course, I've shown you, but the targets are really big. If we could do selective natural gas and conversion in fertilizer production, energy storage, plastic precursor synthesis, all of these things are really important for the American economy and for catalytic manufacturing.

If I end with just one thought, it's that this technology is just getting started. It's only about five years old. The ideas and the materials have only been synthesized and tested in the past few years.

And on top of that, it's a platform approach so it's not limited to any of these particular applications. So let me just end here by acknowledging that I presented a lot of material, and that's because we have a large team working together.

This work was funded by the Department of Energy—Energy Frontier Research program, which is the Office of Science. And it funds a large number of postdoctoral scholars and graduate students who are working to develop these new ideas, and its publications as a team. So thank you for your time. And I'll leave it there for questions.

LADY MIAH KANE: Well, thanks, Paul, for that interesting and informative presentation. We do have time for a few questions. As a reminder, you can use the Q&A box to submit your questions.

And Paul, I'll kick things off with a warm-up question as folks begin typing in their own questions. Is this approach more or less energy efficient than conventional manufacturing methods?

PAUL DAUENHAUER: Yeah, I think that's a great question. If I back up to here, I can say, first of all, we'll never know until we are actually doing this and we can take into account all of the aspects of the capital and the cost to operate these systems. But you can imagine scenarios like this one, where we could operate at significantly lower pressure than a conventional system.

If you think of the compression costs for taking hydrogen, increasing the pressure to 150 or 200 atmospheres, in this paper, we've actually shown you can reduce the pressure down to 20 atmospheres, which reduces the energy input required to compress a system like this.

So there's all sorts of opportunities for faster catalysts or more selective catalysts to significantly reduce the energy costs, but we're not going to know until we put this all together.

Another aspect of this that's very energy-intensive is a lot of times, the catalyst defines the separations that have to happen. So if you have a catalyst that's only partially selective to what you're looking for, there has to be a very demanding and expensive downstream train of separators to tease out the molecules you want and do something with the molecules that you don't want.

If we can make these things super selective with this dynamic approach, we save energy on unit operations we don't even need. And so there's a lot of potential for different aspects of significant efficiency improvements.

LADY MIAH KANE: All right, thank you for that. And we have another question here. Do you work on syngas to sustainable aviation fuel?

PAUL DAUENHAUER: Yes, so sustainable aviation fuel is obviously really important across America. But in Minnesota, it's something we talk about all the time because we have lot of soybeans oil, we have natural sugars from corn, forestry products.

And so a lot of the technologies we're talking about here are exactly the type of catalytic technologies we need for the production of sustainable aviation fuel, whether it be reduction or reforming catalysts, all of these types of materials, or selective carbon-carbon bond splitting. These are all things we could use for sustainable aviation fuel.

LADY MIAH KANE: All right, next one here, how can we adapt this strategy to typical thermo-heterogeneous catalytic processes? How can we tune temperatures, oscillations, and allow heat dissipate faster than the TOF?

PAUL DAUENHAUER: Yeah, that's a great question. So there's many aspects to this. I think where you focused a lot on the fundamental science, that's what we do. But in parallel, we're also thinking about the science that leads to technology at the end of it.

So one aspect of this is how we could manufacture—let's see if I can find this quickly here. How we could manufacture a device that's something we could use in a real system.

So one of the reasons we looked at these particular materials is we wanted to find devices that would be scalable. So in that regard, we could manufacture significant amounts of the material at high surface area. And they would have this form factor, where I could coil them up into what's called a jelly roll confirmation, which gives me a lot of surface area per unit volume.

So some approach like this, that's why we did on this—I didn't talk about this, this flexible substrate, Kapton. This is the way a lot of these things are done. You could imagine rolling these things into pellets and then having these daisy chain Christmas lights, such that we can move charge around, but we can also use them in a fixed bed like a conventional reactor.

That's one of the possible visions for the future. There's many others. I think right now, we're still focused on a lot of the fundamental science, but we're also thinking ahead to how we can do this.

LADY MIAH KANE: Thank you for that. Next one, targeting classical reactions, for example, ammonia, et cetera, many technologies have tried to beat Haber-Bosch and failed. What are the expected implications on process techno-economics?

PAUL DAUENHAUER: I think that's a great question. If you think—I think we were talking about a little bit of that just a minute ago, which is if you think about these types of reactions—take ammonia for example.

Ammonia is very interesting concept reaction here in terms of the fundamental aspects of it, but also on the more applied side, what it means for us. The reason we like ammonia conversion at low temperature, because it's thermodynamically favorable.

And which means you don't have to recycle as much hydrogen and nitrogen and separate out the ammonia. But at low temperature, the reaction is too slow and the reactor would have to be too big. So we have to push the catalyst to high temperature.

And at high temperature, the reaction is fast but the conversion is slow by thermodynamics. So the benefit of these approaches is if you can push something beyond equilibrium, you could operate at higher temperatures and achieve high steady-state conversion that's a non-equilibrium conversion.

Now all that put together means you could potentially make a much smaller reactor, a much smaller separator. All of that would add up to a more efficient overall system. Now take that and think about applying that to all sorts of different applications.

A variation of this is, of course, you could improve the efficiency of large-scale manufacturing. You could also do something like distributed manufacturing. So you could potentially control things more selectively in smaller scale, with higher throughput, in smaller distributed locations.

There's all sorts of places. I think a lot about rural America applications because I'm from Minnesota. And agricultural applications, there's all sorts of things we would like to do on the farm, for example, that you just can't do with conventional approaches. People are trying to do it, and it's a big challenge.

LADY MIAH KANE: All right, thank you for that. You might have answered this already because the question disappeared, but I copied it. Can you also please comment on process and reactor design implications?

PAUL DAUENHAUER: Yeah, I think this is a great question. So how do we—is there a path to scale this up? Of course. And if I go back to—see if I can find it again.

If I go back to this picture right here, this is what we're looking to go towards in the future. And I think a lot of the technologies you see here, the catalysis people, these solid-state electronic devices, they feel very foreign.

I think the concepts of electron transfer on surfaces is something we feel more comfortable with, with the idea of electronic devices. But I'm a benefit because I'm in a department of chemical engineering and materials science. It's all integrated together. In the electronic materials space, manufacturing by roll-to-roll processing these types of materials is very common.

And so the idea that we could scale this up and print these in large volume, like printing a newspaper, and then rolling them up and putting them in reactors is something that at least is feasible in terms of technical capability. Where it ends up in terms of manufacturing, that's something we still have to question because we're just not there yet.

LADY MIAH KANE: All right, we have a lot of questions in the chat—I mean, the Q&A box. If we don't get to your question today, we will try to follow up with you the best way we can. We'll get to a couple more before we wrap up for the day. So one here, what is the effect of the large frequencies of changing polarity on the catalyst material integrity?

PAUL DAUENHAUER: Oh, that's a great question. So one of the projects we have going on is if you imagine—when we put down the platinum clusters here, when we accumulate a lot of electrons in that material, you can imagine that the facets that are exposed might not be the most stable anymore.

So there's all sorts of things we're doing to look at this. So we have an imaging team that's trying to study what's happening on the surface under applied bias for electron accumulation and the depletion of electrons.

It's not part of our center, but we also partner with another researcher who can study exactly that question of what is the preferred thermodynamic structure under accumulated electron structures. But then there's another aspect to it that would be, what structures are stable?

There's also the question of the kinetics of the changing of the structure. So if you imagine that applying accumulating electrons in a structure would make it want to grow or form smaller clusters, if that happens slower, significantly slower than the dynamics that we're applying, these structures might be frozen in time.

And this is one of the questions—this goes to, what is the dynamics of change versus the dynamics we apply? There's going to be a cutoff frequency for every type of change of the active site of the material and the clusters or whatever it is we're working with.

This is all a future area that we're very interested in. There's too much for us to do, and so even within our center, we continue to work with others to try to understand these problems.

LADY MIAH KANE: One here, it says—first, it gives you a compliment. Fantastic work. Do you see opportunities for using this approach to control materials synthesis?

PAUL DAUENHAUER: I don't know how to answer that. I think it's a great question. I can tell you this—these types of architectures are used for so many different things. They're used for sensors, for example, in the reverse process.

You could—I don't know how to answer that because I'm trying to imagine how you could control synthesis of materials on that top layer. I'm sure there's things you can do with the ability to modulate electron density, but we're not doing it and I haven't thought about it. And it's an interesting question, though.

LADY MIAH KANE: All right, one more here. How long does it take for your team to study a reaction and suggest if this is a feasible way to improve the turnover frequency for the particular reaction?

PAUL DAUENHAUER: That's a great question. So it's what we're—one of the things we're driving towards. So to get there, we have metrics. So we use our computation and modeling team to predict what performance we need to have a measurable effect in a flow reactor.

So some of this, I've shown you. I've shown you devices that are achieving about 20 kilojoules per mole perturbations. We're targeting materials that can achieve 40, 50, 60 kilojoules per mole as our target for performance.

I'll tell you, we're basically there and we're moving into that space right now. But we're not going to do an experiment that the modeling doesn't say will be a detectable change.

LADY MIAH KANE: OK, maybe we can do one more. So I see the acidic process of developing the catalyst. Has any consideration been given to the potential degradation of the equipment used to process the final product if any?

PAUL DAUENHAUER: I'm imagining the question is related to—do these devices degrade with time, or if it has an effect on the reactor or something like that? I think this is a good question.

I think this relates to, what happens on the surface with charge accumulation, and how robust are these devices? I think these are all good questions. It's going to depend on the conditions you react at, what the reactants are, how aggressive you are in terms of the amplitude of charge accumulation.

These are all things we don't know the answer to. It's something we're definitely looking at because when we look at—this slide is perfect for this. This is a platform device with all of these variations.

And there's all sorts of different, for example, dielectric film compositions and structures we're looking at. One of our metrics, like I said, is overall charge accumulation. Another metric is stability over time, but it's something we're looking at.

LADY MIAH KANE: Perfect. I think I can squeeze one more in. Are you interested in moving towards using even thinner material—thinner metals, atomically thin, for reduced charge screening?

PAUL DAUENHAUER: That is exactly one of our current projects. That's a great question. So I was talking before with—just so everybody understands here. Wrong way here.

This slide right here says if I have a very thick material in panel B, the effect is completely screened. In panel A, when I get down to these small nanoclusters, the effect is observed. And so the logical thinking is if I get down to single atoms, the effect should be maximized.

That is absolutely—I'd say that's a great question and it's absolutely something we're focused on. So different ways we can enhance the effect and be as efficient as possible. That's a direction we're going.

LADY MIAH KANE: All right, looks like that is all the time we have today. Thanks, again, for everyone who joined us. And a special thanks to Paul for sharing a bit about his expertise and research.

As a reminder, a recording of this presentation will be posted to the webinar section of the ChemCatBio website as soon it is available. If you have any questions, we encourage you to contact ChemCatBio through our website, chemcatbio.org, or you can email Paul.

I'd also like to make one last plug for the ChemCatBio newsletter called The Accelerator. This is a great resource to keep tabs on any further updates or events from the consortium.

I have posted a link to the subscribe—to subscribe in the chat. With that, we will take our leave. Have a great rest of your day. Remember to stay tuned for future ChemCatBio webinars. Thank you all so, so much. Have a great one.