News

3Q: Improving solar cells

Michael Wasielewski on molecular approaches to solar energy conversion

Hannah Bernstein December 4, 2018 MITEI

Michael Wasielewski is the Clare Hamilton Hall Professor of Chemistry and executive director of the Institute for Sustainability and Energy at Northwestern University. Wasielewski is also director of the Center for Light Energy Activated Redox Processes, a U.S. Department of Energy Frontier Research Center. He spoke with MITEI following a recent seminar in which he discussed advancements in the study of molecular approaches to solar energy conversion.

Q: In your view, what are the best methods out there for improving solar cell efficiency? How can photosynthesis play a role?

A: The best methods for improving solar cell efficiencies will take advantage of understanding and applying new physical phenomena. For example, the absorption of blue light by solar cells results in most of that energy being lost as heat. The development of new materials that can be used with current solar cell technology to use blue light to generate charges instead of heat are important. These methods include multi-exciton generation in inorganic materials and singlet fission in organic materials. In both cases, absorption of blue light ultimately leads to formation of more than one electron-hole pair that can be used to provide extra current from the solar cell.

The primary light capture, charge generation and transport properties of photosynthesis give us a detailed picture of how the structural relationships between molecules can be exploited to optimize the efficiencies of these processes.

Q: What are some of the opportunities and challenges related to molecular-level approaches?

A: The biggest challenge for molecules is long-term stability. Fortunately, there are many molecules that are capable of enhancing solar cell performance that have been tested by industry in other applications, such as paint pigments, and have been optimized for long lifetime. An example of what is possible with good engineering is the organic light emitting diode, which began as a short-lived lab experiment about 20 years ago, and now is being incorporated into large screen displays, such as televisions, with estimated 100,000-hour lifetimes.

Q: Why is improving solar cell efficiency important for the industry?

A: Improving solar cell efficiency is important for the obvious reason that increasing solar cells’ power output can result in economic advantages as well as the ability to use smaller solar installations in locations that larger arrays that are less efficient would not be possible.  In addition, it is important to develop solar cells that produce electrons and holes at higher potentials that can be used directly to carry out energy-demanding chemicals reactions for energy storage and chemical and/or materials synthesis.


Subscribe:      Apple Podcasts|Google Play|RSS|SoundCloud|Spotify|Stitcher

Transcript

It is our pleasure to introduce Mike Wasielewski, who is here from Northwestern where he’s the Clare Hamilton Professor of Chemistry, also the executive director of the Institute for Sustainable Energy, and the executive director of the Solar Fuels Institute, and the director of ANSER, a DOE EFRC.

Mike has spent almost his entire life in Chicago area, unlike the rest of us, who’ve moved around, had to move around quite a bit. He’s born and raised, but he also did his bachelor’s, master’s, and PhD at University of Chicago. Aside from a year post-doc with Ron Breslow at Columbia, he then returned to Chicago to Argonne National Lab, where he rose from a post-doc to a group leader, and then in 1994 became a faculty member at Northwestern.

And so he’s been recognized with awards that are too numerous to list while him still making his flight this evening, but they include the Humboldt award. He’s a triple AS member, ASC awards, such as the Arthur C. Cope Scholar, the James Flack Norris Award in Physical Organic, and he’s also received the Porter Medal in photochemistry.

So Mike really has developed an impressive and broad research program with contributions in many areas, many in solar energy, which I think we’ll hear about today, from looking at solar fuels from light harvesting to charge separation to catalysis, solar electricity, using organic photovoltaics.

And he really has a long history of looking at photo-induced charge transfer and charge separation, some really seminal work in understanding the PS II reaction center, which is close to some of the research we do in my group, but as well as a lot of model systems to understand the fundamentals behind charge transfer and really demonstrating Marcus theory experimentally.

And so he’s also used a variety of model systems, like molecular aggregates to understand charge separation, charge transport. And there’s also a whole section of his group, I think that just illustrates the breadth that I just learned about on this visit, in fact, using molecular systems to create spintronic devices with the potential for quantum information.

And I think one of the best things that I can say about Mike is he is just a lot of fun to talk science to, because this breadth of studying molecular systems means that he has an impressive depth of understanding of molecular systems. And so it’s really been a pleasure having him here today. And I’m looking forward to his talk. So I hope you will all join me in welcoming Mike Wasielewski.

Gabriela, thanks so much. It was really kind, kind remarks and a great introduction. I appreciate it very much. What I’m going to tell you folks about today is one aspect of what we’re doing, but because this is a mighty seminar and focused on energy, I’m going to really focus on our work in the solar energy field. But what I’m going to do is essentially try to give you a flavor for three different areas that we’ve been exploring lately that are really important to the whole energy transduction apparatus for solar energy conversion.

So as the picture indicates, we are on the third coast in Chicago there, just North of Chicago. But in point of fact, our institute is actually sited– this is an old picture. Our institute is actually sited here where there’s nothing. So we actually have a new addition to the building there, which is where we live.

So that’s where we are. But going back historically, with regard to what Gabriela indicated, basically where we derive our inspiration from is essentially from natural photosynthesis, going back to examples from bacterial photosynthesis, years ago, to more recent examples from green plant photosynthesis. But what comes out of what I have on this slide are two things.

Number one is for a number of years now, we’ve been very fortunate to have X-ray crystal structures of the so-called reaction center protein complexes that drive the principal energy transduction and photosynthetic organisms. And the other thing that comes out of this is that each of these complexes is in fact a marvel of molecular self-assembly and a marvel of design with regard to not only taking advantage of symmetry, like in these beautiful circular light harvesting systems, but being able to stepwise, in a very efficient way transduce charge across a membrane, or alternatively in this molecular machine from green plants to split water and to generate oxygen and protons, something that the energy community would love to do in order to be able to ultimately generate hydrogen photochemically.

So those are our inspirations. So the question is how do we approach the inspiration. So today I’m going to talk to you about three different topics. Number one is light harvesting. And the topic of light harvesting I’m going to relate is how do we get more for our money? In other words, in a typical solar cell, we’re essentially wasting the blue end of the spectrum in terms of generating heat, rather than electrons and holes.

So can we use essentially a form of multi exciton generation, but in organic materials, that is singlet fission in order to do better. So we’re going to talk a little bit about that. We’re going to talk about charge separation and transport. So once we’ve captured energy, we have to generate electrons and holes, and then ultimately get them to an electrode. We all know that more order is better. So the question is, can we use some sort of self-assembly, or molecular recognition to help us have molecular systems that actually do as well as, for instance, inorganic semiconductors.

And last but not least, if one is going to have charges, one is either going to generate electricity, or alternatively, one is going to make chemicals. And the idea of doing catalysis for useful things, for instance, splitting water generating hydrogen, or alternatively, reducing CO2 into a liquid fuel is really on a lot of people’s to-do list, and very important energy demanding reactions that we really need to deal with.

So let’s launch into this right away. First of all, if we take a look at singlet fission, singlet fission, as indicated in the slide here, if we take two molecules, put them next to each other, especially in a molecular solid, and this has been known now for– you can see the dates there. It’s 50 plus years at this point.

Basically, what happens is if you excite one of these molecules to its lowest excited singlet state, it, under certain circumstances will be accompanied by a process whereby one of the molecules is rapidly converted to its triplet state, whereas a neighboring molecule is up converted to its triplet state from the ground state. And so this is a spin allowed process. So it’s generally fast, as long as the process fulfills some requirements.

And the requirements are, of course, that the singlet energy is twice the triplet. There’s some minor issues with regard to multiple triplet states being of the right energy as well. But the really important one is how these molecules interact with each other to make this process happen. In other words, the devil is in the electronic coupling details, if you will, in this particular case. And of course you don’t want the system to go backwards.

Although in some cases, if you’re doing triplet annihilation for up conversion processes, you may want it to go backwards, but not in this particular case. So why the interest currently? Well, the interest currently has been around for about a decade now. And the reason for that is pretty simple. About a decade ago Art Nozik published a paper that indicated that if you do redo the Shockley-Queisser thermodynamic limit calculation, for this particular system if you pair this kind of system with a suitable red absorber to deal with the red photons, you can get an increase in the upper limit of performance of a solar cell from 33% to 45% if everything is absolutely perfect.

So the idea of getting something for nothing is appealing, but in fact, you really have to work pretty hard to do this, but it’s really launched a lot of good fundamental science, trying to understand this process in a much greater level of depth than had been before. So what are the issues? Well, the issues are from the standpoint of how’s it work. There are two primary thoughts in this context.

In other words, it can be a two electronic process, accompanied by the corresponding spin flips, and that kind of direct process or direct integral, if you will, going to the product state. And note the fact that even though you have two triplet states here, the overall orientation of these electrons is still an overall singlet. So it’s sometimes called a TT singlet state in this particular context.

You can do it that way, or alternatively you can do it like a lot of the community has been comfortable with for years, in other words, doing sequential electron transfers, for instance, either via the HOMO of the LUMO of the two molecules transferring an electron after excitation followed by a spin flip and then transferring the other electrons. So that basically, by doing it either way, in other words, either through this pathway or to the other electron transfer pathway, you can get to the same place.

Now, as far as electronic coupling and molecular geometry, a lot of work has been done on this. But I can kind of sort of summarize it in very crude way, by simply saying that if you go back to Joseph Nichols’ review from 2010, he had this nice little picture there. Basically, the various electronic repulsion integrals really either support or detract from the kinetics, if you will, ultimately through Fermi’s golden rule of singlet fission process, if in fact you have a geometry in which the molecules are in register co-facially right on top of one another.

There are many, many other geometries, mostly slip-stacked geometries and related geometries, which are really quite good for this process. And so there’s a wide space there. But the only one that’s really bad is if the molecules are literally in register. But it’s more complicated than that. In a paper that Greg Scholes put out now four years ago, it seems like it was only yesterday, he brought up the fact that there are electronic coherences possible in these systems as well, so that one might envision a system as involving this starting state, this singlet state, a CT intermediate, and this so-called TT singlet state that’s formed via singlet fission as being an electronically coherent object or a set of states that then lose electronic coherence to give you, essentially a spin coherent, TT pair, which then can actually transition to either a overall triplet or quintet spin configuration, and then that loses coherence and forms two independent triplets.

So then, what has to happen is these two triplets have to have enough energy to form electron whole pairs in order to make this whole process energy relevant, if you will. So the devil is in that detail as well. But I’m not going to really talk too much in detail about that at this point. OK, so we’ve done a lot of work in this area, just to summarize the last few years of what we’ve done, lots of different molecules, lots of different mechanisms, lots of different geometries.

But the one thing I want to focus on today, this one little segment is what are the role of these putative CT states. And so we’ve done some work on that. And one molecule that we’ve glommed onto, which is di molecule, which is called terrylenediimide, it’s a blue molecule, that is a cousin of the so-called rylene series that begins with naphthalene and perylene.

The perylene dye is quite well known because it exists in red car paint. This pigment is not necessarily used in car paint. But it’s very robust. It’s much more robust, for instance, than pentacene, which is also a molecule that has almost exactly the same energetics in terms of singlet and triplet energies, and yet is able to function as a singlet fission chromaphore as well.

So we’re going to use this chromaphore because you can beat on it and it hangs in there, unlike pentacene in this case. But we’re going to also look at its properties in this context. OK, so basically what we like about this molecule right off the bat is the fact that you do a femtosecond tranxene absorption experiment, what you find immediately is in near-infrared region of the spectrum, you actually have three sharp transitions, which are due to S1 to Sn transitions.

That’s not my cell phone interacting with the microphone, is it?

[INAUDIBLE]

Oh. I get it, OK. I’ve had the cell phone interact with a microphone before. But notice those three bands are quite sharp and distinct, right, so that we’re going to look we’re going to use those with good effect as we go on here. This is just S1 to Sn transitions, right, so that they’re simply excited state transitions. The molecule decays. The excited state decays in about two nanoseconds, nothing special there.

So let’s take a look at what happens now when we look at a dimer. So my group produced this slip-stacked dimer with a linkage between the two terylenes that puts it in a slip-stacked orientation with a very specific geometry. And so what we found was really quite interesting in the sense that if you put this in a polar solvent and flash it with a laser, what you find is that if you probe the system sometime later, what you find very rapidly, you get a new set of four bands appear, three out here and one at 760.

And so it turns out that if you think about what those bands are, you don’t have to think too hard. All you have to do is do a little spectro electrochemistry. And what you find is that the 760 band is due to the cation of the TDI. And the other three bands are actually due to the anion. And so what we’ve done is symmetry. We’ve broken the symmetry. So we’ve done symmetry breaking charge separation in the system in the polar medium.

So it’s likely that solvent fluctuations are really driving the formation of a CT state. So the important thing here is we do see a CT state. But where’s the triplet state? In other words, do we see a triplet state there? Well, first of all, we have to know where to look. Turns out we have to look in this region of the spectrum, as I’ll show you in a spectrum here. So let’s take a look at that.

We can sensitize the formation of the triplet of TDI by using anthracene as a sensitizer, because TDI’s intrinsic triplet yield is less than a percent, so it’s not very much. So any triplet that is produced comes from some other mechanism, so that when we sensitize a triplet via anthracene what we find is sure enough, the band is in fact underneath the S0 to S1 transition in TDIs, which results in this kind of up and down transient spectrum here.

And notice there is absolutely nothing going on in the near-infrared either. The band is literally right here. So we decided let’s take the dimer and put it into a non-polar medium as well in toluene. And when you do that, lo and behold, what you find here is that you find that you get the band that is intrinsic to forming a triplet. And it happens pretty rapidly in a couple of picoseconds.

And what you end up with is what looks like the excited state decaying here. This thing is running out of me. The excited state decaying, and you don’t really see a lot of evidence of the CT state. But you do see the triplet come roaring in at this particular juncture. So what’s really going on here?

Well, it turns out that if you globally analyze the data, what you find is that the triplet yield maxes out at about 133%. So in other words, 200% is perfect for singlet fission. So the question is, why aren’t we perfect in this particular case, because the rate is so fast? Well, the reason why we’re not perfect is because we have putatively an excited state equilibrium going on. And that excited state equilibrium simply indicates that the two energy levels are pretty close, just from multiple statistics. And if you analyze the data further with that kind of model, this is what you get.

It’s about two picoseconds out, about four back, and two ways to get back to ground state. OK, so where is the CT state? , Well, what we feel in this particular case is that the CT state, if we are in a polar solvent is below these almost isoenergetic levels of the TT state and the singlet. And it acts as a trapped state.

But if you lower the polarity, you raise the energy of the CT state, so becomes a virtual state, so it engages in a super exchange interaction with the starting state and ends up being virtual, so you don’t actually see it. But the question is, when can you see it, or can you see the evidence of this particular state in the mix?

And so what we decided to do is to take a look at another variant of this molecule. And the variant is shown here. All we’ve done here is change the spacer, and the reason why we changed the spacer was in order to make this thing more soluble. And the reason why we needed more solubility, is because now what we’re going to do transient mid-IR spectroscopy on these dimers. Here’s a DFT calculation, which shows you the slip-stacked geometry one more time.

And you can see that the spectra of the dimer, the monomeric reference with the full spacer and just a plain monomer with nothing else there are pretty similar in this particular context. OK, so let’s take a look at this and see where it leads us. Basically, now, if we look at the IR spectra of these molecules, you can see there are bands in the carbonyl region, bands in the region of the CC stretches.

And if we calculate what the various intermediates should look like, whether they’re the ground state or the cation, the anion, the triplet, singlet excited stated, or singlet ground state, the immediate thing that comes out is that the anion has a distinctive band around 1570 wave numbers, and also the triplet state has a distinctive shift of its band near 1650 as well.

If we take a look, once again a form of spectral electrochemistry, using IR spectroscopy, reduce the TDI to it’s anion in dichloromethane, you find that sure enough, you get a large absorption here at roughly 1570. And the other bands match quite nicely to what one sees in the computed spectrum. So it looks like this is what we need to look for in this particular system.

And if we look at just the monomer, you see exactly the same thing I showed you before for monomere TDI, whether we’re in the a polar solvent, or whether we’re in a low polarity solvent. We’ve switched the dioxane here, simply because toluene has too many IR absorptions in the region we want to look. And so we did dioxane, because it has roughly the same dielectric constant as does toluene.

All right, so let’s take a look at the dimer. Well, it turns out that if you compare dicholoromethane with dioxane, once again, in the polar solvent, we pick up the extra four bands that are due to symmetry breaking charge separation, and in the low polarity solvent, once again, we see the triplet formed. We do see some hints here that maybe some of those bands are now a little bit more accessible in the system, and when we do the global fitting, you can kind of see that in this particular case.

But there’s no really distinctive view of the CT state, even in the dioxane case, whereas it’s absolutely totally blatant in the dichloromethane case. So you essentially pump probe spectroscopy in the UV vis and near IR is not going to help us very much. So now let’s take a look at the mid IR.

Now, if we look at the various monomeric references, you get a large absorption here near 1660 or so in the single state. And in the triplet state, we put this in the monomer here in iodoethane, in order to do heavy atom induced intersystem crossing in order to be able to elicit faster triplet formation.

Note the fact that there are now some bands in this region around 1640-ish in the triplet, and we bleach out the ground state bands, of course. But there’s no band here in your 1570. However, now, when we go to the anion, you see, what’s really quite amazing is that you see this 1570 band appear quite strongly. But interestingly, you don’t lose the triplet bands, which is quite interesting. They’re still there.

And similarly, when we go to dioxane, we get the triplet accentuated, but we don’t lose the anion band. So what’s going on here. Let’s take a look at this in a little more detail. In fact, let’s do it once again global fitting so once again, if we compare the dimers in dicholoromethane and dioxane, you can see that the triplet signal is accentuated, but the anion band never goes away.

And here’s just a pure triplet signal. Did you see here with the band here, now it’s 1640. So it looks like all of the states are there all of the time. So that’s interesting because one would surmise originally from Greg Schole’s arguments, that we may actually end up with producing some sort of mixed state configuration here, where all states are possible and in the mix, but which ones dominate depend on environment and conditions, in this case solvation.

So the model we’ve come up with, and here’s comparing these various situations with the triplet state, the triplet state monomer and the dimers both in dioxane and dicholoromethane , and you can see that in the cases of the dimers, the anion band is always there and so is the triplet. So the three situations you can run into are that OK, well, let’s say we have a situation, we can depict this by using a vector diagram with the three states being on the orthogonal axes, so as we go across in time here, the red vector is the total population, so that the total population changes as a function of time, but we can either favor the triplet formation with only a little admixture of CT, which would mean a trajectory that would go in this direction.

However, if we then have the opposite situation, we can start with a state that’s mixed CT into the singlet state, which we see the CT state essentially at time zero here, so that it’s there as quickly as we can look. And so that state itself can then go over primarily to the CT state. This may be the case, for instance, in a very polar solvent with no admixture, if you will, of the triplet.

However, what we really think is occurring in this molecule is you do start with this mixed state that has CT character mixed in with the singlet, but as time evolves, this vector rotates. And so what you get is a significant admixture of triplet, and then ultimately, where you end up in this CT triplet plane, if you will, depends on conditions and solvation.

So this kind of model I think is one where we can visualize what goes on in this particular system. So let’s just summarize this part of the talk. And that is we can use femtosecond IR spectroscopy to show that electronic excited states of TDI have this mixed character. At times less than 300 femtoseconds, we already have CT character in the mix, even in low polarity solvents. And the degree of state mixing depends on the solvent polarity and changes as a function of time.

What I don’t have time to tell you about today is work we’ve done using 2D electronic spectroscopy to essentially certify that this picture is correct, and in fact, it corroborates this view quite nicely. And so that’s up and coming at this particular point. So let’s move on to the second thing.

So if we can maybe harvest light better, we can maybe generate some charges, but we’ve got to move those charges around. So let’s think about how we might do that. One way we might do that is by taking advantage of the order one can elicit from molecular self-assembly. Imagine this is a donor and accepter, and these are a couple of electrodes. If we could get these molecules to stack and the segregated stack to arrange, in other words, all the acceptors on the acceptors and donors and donors, we would have something interesting, because we could actually use a covalent molecule in order to control the rates of charge separation and recombination and in each individual monomer, if you will.

And then if we could get these to segregate, stack in a segregated way, if the rate of recombination was slower relative to the rate of charge hopping within the segregated stacks, we might be able to get the electrons and holes to the electrodes before recombination occurred.

So this kind of control would be important for getting much more order than you would, for instance, in a typical bulk heterojunction organic solar cell. So this is kind of the dream, if you will. So let’s take a look at a really simple system first. One donor, one acceptor. And we’re going to use some trickery involving the side chains on the molecules in order to elicit this kind of assembly, where you have segregated stacking.

And so we use two of our favorite molecules. We as a porphyrin, and we use one of the red car paint molecules, the PDI molecules, with some alkyl tails on one hand and peg chains on the other end to really desymeterize and to drive the segregation. And sure enough, when we do this, let’s just focus on one of these with the shorter chains. What you find is that if in solution, you simply get a sum of the individual spectra. In a disordered film that’s put down by spin coding from dicholoromethane, you don’t do very much better. The spectra still look pretty much like the solution spectra.

But we can make an interesting viscous mix of this material N-methyl pyrrolidone and what that does is it actually elicits the formation of aggregates and you can see that the spectra changed quite dramatically here. And you can see there’s some red shifting, and the PDIs are doing some blue shifting here. So that there’s a lot of interaction between individual molecules.

But what’s really going on? Well, turns out that if you take a look at the IR spectrum of these aggregates in the CH stretching region, it’s well-known that ordered alkyl and pegged chain aggregation results in IR shifts that are particular magnitude and particular directions, which these, in fact, have.

But that’s not particularly good evidence for anything. I mean, it’s just sort of indicative. What we did is we resorted to going to our local friendly synchrotron at the Argonne lab and doing small angle X-ray scattering. So when you do this with these films that are generated from the NMP viscous mixture, what we end up with is a small angle region, intensity versus scattering vector squared of minus 2, which means what we have essentially have are ribbons, ribbon structures.

So this indicates that we perhaps have actually done what we think we’ve done, but we can go further. We can do grazing incidence wide angle scattering, and we see really quite good order. And the order, I won’t go through the detail, the order is not the way you want for a solar cell. In other words, it’s edge on from the surface, so that you have layers that go this way, but the direction of charge transport would be orthogonal to the plane of the substrate.

But nevertheless, we have good orders. So we can see we can see what effect this order has on charge transport. So if we compare the shorter chains with the longer chains version of this, first of all, let’s take a look at just those shorter chains in solution. Nothing special going on. You get charge separation to form this anion radical band here at 700 nanometers when you flash it with the laser. And there’s about a 2 and 1/2 nanosecond lifetime for the charge separation. End of story.

However, now when you flashed the film, the ordered film, what you find is that the band has now broadened to a large degree, and what you have is a component that lives a very long time. In other words, it’s out past seven nanoseconds at this point. So what’s going on here? Well, it turns out now if we look at the magnitude of this change, it turns out in the ordered system about 30% of the sample is living long.

But what is really long in this context? Well, first of all, we worried about things like, well, you know, homogeneity. We worried about temperature dependence. It turns out that these fast rates of charge separation are not temperature dependent at all over this temperature range. So we’re getting consistent data.

What’s interesting, though, is what the charge recombination times are. The charge recombination times are quite long. You can see this is out close to a microsecond at this point, or it’s beyond a microsecond, and the kinetics are fit to a model, which indicates that we’re kind of trapped limited with regard to recombination here. So we have to kind of get out of defect traps in order to recombine.

So we have a long lived species here. So can we learn more? Well, the question is we can take a look at these molecules in a different context, using some magnetic resonance spectroscopy. And we can do time resolved EPR spectroscopy. In other words, when we do a laser flash, but we do in this case, CW microwaves. These are not pulse microwaves.

And so these spectra are not like your garden variety EPR spectra. They’re not derivatives. They’re actually direct detection. So an upside down signal is actually an emissive signal, not at absorptive signal. So why is the signal completely emissive? Well, it’s telling us something about the intermediate range of distances in the electron hole pair that we’ve created. And what it’s telling us can be gleaned from this energy level diagram here.

Basically, when you make an electron hole pair, and you turn on a magnetic field, you get a Zeeman splitting, obviously of the three triplet levels that constitute the spin pair that you’ve created. And so there’s also a singlet level here, and there separated of course by the exchange interaction 2j. But there is a crossing. If the 2j value is big enough, they will cross somewhere here at a magnetic field that’s corresponding to crossing, let’s say that t plus 1 level here.

So at the 3,500 Gauss field we’re using, this indicates we are populating the mixed state that results from mixing the s and t plus 1, levels because we come in with all of our population from s, and those are the only two that mix, because from those mix levels, when we induce microwave transitions down to the lower triplet levels, you get an upside down spectrum, because all the populations up here. So you get an emissive spectrum.

What is this telling us? It’s telling us that because j is so big in a kind of relative sense, that the distance between the electron and hole, in this particular experiment that we’re sampling, is only on the order of 15 angstroms. So this is sampling a subpopulation. So we know we’ve gotten out at least that far.

But what about living long? Well, it turns out we can actually turn off the magnet and use our spectrometer as a way of measuring, essentially, time resolved microwave conductivity. And when you do that, it’s really a dielectric loss experiment, so that putting dielectric material, like for instance, something with charges in a microwave resonator, essentially diminishes the intensity of the microwave. So consequently, you get, as you create the charge pairs, you lose a lot of them rapidly, but a whole bunch of them hang around a long time.

Note the time scale here. We stopped the experiment at 50 microseconds. These guys are hanging in there a long time. But what, of course, is the yield in this case? Well, the yield is about 30%. So we actually are generating electron hole pairs in a reasonable amount for a long period of time. But we can do better. And we can do better in a number of ways. We can take a bio inspiration approach.

And this is an idea that came from one of my research professors in the group, [? Yulin ?] [? Woo, ?] who talked to our colleagues who are into framework materials. And so what we were doing was saying, well, gee, is there a way we can use– and we’ve also been into a lot of electron transfer and transport it in DNA structures. So we thought well, gee, if we have guanine, we know that guanine, in nature, hydrogen bonds to itself in G-rich regions of DNA.

And in the presence of alkaline metals, like sodium and potassium, forms the so-called G-quartet structures, which can then stack upon one another to form G-quadruplexes. And these are well-ordered structures that you can control. And in fact, this just shows a cartoon of what a G-rich region of DNA looks like with some of these layered structures. But of course, this looks a lot like what we wanted to do with regard to our self-assembled structures.

In other words, the structure doesn’t have to be kind of rectangular. It can actually be annular as well. And so we thought we would take a careful look at this. And so it’s five years ago now, we published this paper on putting one G with a very strong accepter, the perylenediimide that can actually act as a linking agent, in other words, an organizing agent, which is passive. Or since G is the base that gets oxidized most easily of the four DNA bases, you can act as an active donor as well.

In this case, was an active donor. We formed a double layer G-quadruplex from this particular system, and what we discovered was that in fact, when we pulsed this with a ultra fast laser pulse, we generated ion pairs, and the charge recombination of the monomer occurred in 10 picoseconds, whereas in the quadruplex, this lengthened out by a factor of 100 to 1 nanosecond.

So we thought gee, that’s pretty interesting, but we really want to explore how this quadruplex structure really helps, if you will, extend the lifetime of this charge separation. 1 nanosecond is way too short for EPR spectroscopy, so we wanted to go a bit longer. And so we did a fairly extensive study on this triad molecule. Once again, the G is an active ingredient in here. And the chromaphore is this actually naphthalene imide, amino naphthalene imide in the middle, and the internal acceptors in the napthalenediimide.

And so we actually generated a quadruplex from this system as well, and not only did EPR, but did femtosecond stimulated Raman and all kinds of other techniques in order to figure out what was going on. So when we separate charge here, the thing we found out is that the hole that’s in the quartets that form the quadruplex, that hole is moving around not only within the quartets, but also hopping back and forth rapidly between the quartets that constitute the double decker quadruplex system, so that this fact that the hole is strongly de-localized is helping extend the charge separation in this particular system.

So we thought well, OK, if you can do this with just a simple quadruplex system, the thought occurred, because you know, we sort of had frameworks on the brain, well, why not put two of them on something interesting, and maybe you can actually make a two or three-dimensional material, and sure enough, that’s what we did. We decided to take two guanines and put them on opposite ends of a system, in this case just a simple chromaphore.

If they formed a quartet, you’re going to get this. If the quartets get together, you’re going to form a 2D array. If these guys stack upon one another, then you’re going to form a 3D quadruplex framework. So that’s where we went with this. And in fact, this is very much analogous to what we were trying to do before.

So how do you do this? Well, you have to start with a guanine that’s fully protected. You can do some cross coupling chemistry that’s well in hand and generate whatever system you want in this context. But the trick, if you will, in all of this is, how do you de-protect the system in a way that elicits some degree of crystallinity.

And so each system is a little bit different and in getting rid of these protecting groups and actually forming crystals turns out to be the biggest challenge. But we’ve been able to do this in a variety of very simple ways that one simply needs to be patient in some cases.

But important part is that the crystallinity is high, these are PXRD patterns, for instance, a phenyl, a naphthalene diimide and a perylenediimide, and the distances correlate very nicely what you would expect from the frameworks, and we’ve done the usual kinds of things of taking a look at what the BT surface areas are and such. And everything correlates quite nicely with a nice framework solid.

So what can we do with these? Well, first of all, we can put a few electrons in them. So you can do a little bit of a chemical reduction on these systems. And look at just standard CWEPR and see where those electrons are and where or they might be moving. And it turns out that in each case, whether in naphthalene diimide or in perylenediimide, the lines narrows substantially in the EPR spectrum.

And if you go back historically, to the late ’60s, some worked on by Sam Weissman at Washington University along with Jim Norris, showed that the line width is going to be proportional to the square of number of molecules that the electron is visiting in this particular case.

And so indeed, what we find is that the line narrowing in this particular case is indicative of the electron very rapidly visiting at least six or seven. Note I haven’t said de-localization. It could be simply hopping at a rate sufficiently faster than the EPR scale, which in this case, is something on the order of 10 to the ninth per second.

We do know, however, that if you go very, very fast, let’s say faster than a picosecond, then they’re not moving that fast, OK, so somewhere in that regime. OK, so let’s take a look at some other chemistry here. If we once again, use a laser pulse to generate electron hole pairs, you can see both in the naphthalene case and the perylene case, you get a big bleach, and then in perylene, you can actually see the anion peak here.

You do get a long lives component of the kinetics, indicating that there are some charge pairs that are living a long time. If we look at that same EPR experiment I did before, the time resolved EPR, you can see both spectra are upside down. They’re both emissive. So we’re sampling that same distance regime. So we’re getting the electrons out there. But more importantly now using time resolved microwave conductivity, we now raise the overall yield of ions to about 60%, especially in the case of the perylene, a little under 40% in the case of the naphthalene in the system.

So we’re getting there. In other words, we’re generating systems that have the ability to separate charge and move that charge over long distances. In this particular case, the holes are propagated by the G-quartets that are stacked, and the electrons are propagated by the perylenes or the naphthalene systems.

OK, so let me just summarize here that these ordered solids, whether we’re in these simple minded, simple porphyrin PDI systems, or we’re in these more and more complicated frameworks systems all have ultra fast photo driven charge separation that result in independent charge carriers that last for more than 10 microseconds. So this should be long enough if we have the proper orientation to get us electrons and holes to an electrode, independent of the ability or certainly in competition with the ability to recombine.

So let me now jump to the last topic, so we’ve talked about harvesting. We’ve talked about moving electrons and holes around in ordered solids. Now, let’s talk about a completely different, but yet related topic. In other words, if you have any of these systems where you are generating charges from solar flux, can you generate charges that sufficient potentials to actually do the kind of energy demanding chemistry that water splitting and CO2 reduction represent.

So that’s what I’m going to talk to you a little bit about now. Now, there’s a fanciful cartoon here we have of what really the goal is, to have some sort of solar tree that’s generating liquid fuel, which you can then just kind of drain off at your leisure. But the reality of it is that people today have taken this approach of doing– two half cells basically with doing photoelectrochemistry on both half cells, in this case with the dimolecules and catalysts to split water, and then you have the protons go across a membrane.

And they are then reduced by a catalyst, which is driven by another photo sensitizer on the companion electrode. So in principle– and then there are many embodiments of this in the literature, and people have done this with varying amounts of success and varying yields. But the question, of course, is are there any other approaches one can take?

Well, our overall strategy is pretty simple, actually. We like to do spectroscopy. We like to understand dynamics. So what we’d like to do is look at catalytic processes are intrinsically fast, because a fast catalyst means it’s generally a good catalyst. But if you have a fast catalyst, that means all the intermediates aren’t going to last very long. So you need some tools that are going to actually be able to address those intermediates and see them in real time.

But we’d also like to step through mechanisms one step at a time. That’s really important as well. We’ve done a little bit of this work historically here. This goes back to a few years where we used a water oxidation catalyst and were able to pull electrons out of it with one of these perylene dyes.

But you can see that you pull the electrons out, but they come right back fairly quickly. We’re able to surmount that by actually using, once again, the triad approach that’s intrinsic to photosynthesis, where you have this chromophore pull an electron out of the catalyst, and then goes out to this other easier to reduce species, giving us much, much longer lifetimes. What was useful here is we’re able to go to the synchrotron, and use time resolved X-ray absorption spectroscopy to do studies on the edge spectroscopy of the iridium, so that we were able to tell which oxidation state and which ligation state the iridium were in, so we could certify that we actually did the oxidation chemistry we hypothesized, because the problem with metal complex is the dirty little secret is frequently their electronic absorptions aren’t very distinct.

And frequently, they’re buried under the organic stuff, basically. OK. We’ve also done work with, you know, putting these kinds of systems on electrodes like TI02 and protecting them through atomic layer deposition with layers of aluminum oxide to preserve them their lifetimes, et cetera. So we’ve done a lot of that work.

But one thing that occurred to us is that we forgot about one thing that we did a long time ago, and it was time to remember it and do something about it. And that is this. All of these little molecules we’ve been dealing with, way back in 2000, we did a whole bunch of fundamental studies on there excited doublet states. We were interested in if you had an anion radical of these guys or even a dianion, for instance, what did their excited states look like.

And so they’re very easy to reduce. Here’s the ground state spectrum of this NDI anion radical. And you can see that it absorbs all the way up to about 780. And when you pump a photon into the system and generate the excited doublet state of this system, you find that it has a lifetime of about 140 picoseconds, not really that impressive in terms of lifetime, if you’re going to use this as a, for instance, a photo sensitizer. But still, not inconsequential as yet.

So why would we want to use this as a photo sensitizer? Well, if you look at the energetics, you can reduce this molecule at about half a volt, versus saturated [INAUDIBLE]. That’s easy. It’s reversible. It’s nice. However, even if you consider using the near infrared photons here, that’s about 1.3 volts. So that you just simply add them up, and what you’re going to end up with is an excited doublet state that has a reductive power of about minus 2.1 volts, which is really out there and can reduce a lot of things, including CO2 reduction catalyst.

So we thought, well, maybe, we can try to use this. The other thing is that its cousin, the perylene dye is very similar. It too has all these near-infrared absorptions with the terminal one, the lowest energy one being at 950. Same thing. You can generate the excited doublet state here for sort of coincidentally, it has a very similar lifetime.

But the important part is here, you can even get like minus 1.7 volts as well. But let’s focus on the naphthalene one, just for a moment. So basically, we can use these as super inductance. And so the idea is couple this guy to a catalyst that requires electrons that have really negative potential, something hard to reduce. And we can use multi-step electron transfer just like photosynthesis does in order to be able to get the electron to where we want and have it stay there basically.

So we’re going to go back and use a catalyst that is a kind of benchmark catalyst, this rhenium one catalyst has three carbon groups and a bipyridine, and it’s been used for 30 plus years, and a lot is known about the mechanism of CO2 reduction. It’s a catalyst that has a non innocent ligand. In other words, you reduce the ligand, and then you reduce to rhenium from rhenium one to rhenium zero. And once it’s in that state, it will bind CO2 and reduce it to CO. So this is well-known.

So what we’re going to do is do this, at least get the first electron in at really these fairly negative potentials, but almost minus 1.3, 1.4 volts. And so see whether this kind of strategy is going to work. And as I said, it binds CO2 at this point, right? OK, so what we’re going to do is we’re going to use a triad strategy, and we’re going to try to get the lifetime as long as possible, because we’re going to have to have a diffusive encounter with CO2, right.

So that’s important, so that we’re going to look at two molecules. We’re going to put the naphthalene on one end. We’re going to use this spacer in the middle and the rhenium complex on one end. It’s linked in two ways. In one instance, it’s actually linked as a ligand to the rhenium. In the other instance, it’s actually covalently linked to the bipyridine. Why is that important? Well, we wanted to understand the electronic coupling difference between these two situations.

And it turns out that in this case, this is not going to a particularly good catalyst, because for instance, the chloride ligand in this case falls off during catalysis. And this pyridine is going to fall off. So the whole sensitizer will fall off after the first event. But we still want to look at coupling and see how that goes.

OK, so electrochemistry is straightforward and easy. It’s beautiful electrochemistry. Almost every transition is reversible except for the rhenium one to rhenium zero is a little flaky. But nevertheless, as electrochemistry goes, that’s not too bad. So what we’re going to use the spacer in the middle for, and this has been optimized over several iterations, is we’re going to use this, we’re going to reduce this anthracene spacer at minus 1.85 volts.

It’s going to provide a tunneling barrier for the way back. So we can get the electron out here and sort of try to prevent its way– put it far enough away with a barrier in between to prevent its recombination. So that’s what we’re going to do. And so here is the chemical reduction of this species once again, gives you a clean NDI minus spectrum. There’s nothing else absorbing out there, so you can pick any place you want out here to excite, and be sure you’re going to excite only the photo sensitizer.

And so we can do that and try to elicit these two reactions to happen. If you look at electronic spectroscopy, sure enough, you see the NDI minus bleach out, and you see the NDI zero come back in, but you don’t see any indication what’s going on with the catalyst, because once again, these catalysts have really terrible electronic absorption extinction coefficient. So you’re not going to see anything in this particular background.

So that’s where we resort once again to transient IR spectroscopy, because these carbonyl groups are absolutely amazing with regard to being indicators, because their out where nothing else absorbs. And so they’re out in this region near 2,000 wave numbers. And so we can see, we can bleach out the ground state, and we immediately come in with two new absorptions for the carbonyls.

There are actually three absorptions. Two are overlapping in this particular case. But the important part here is that it’s about a 40 wave number shift to lower frequency. And what that means is that the electron’s sitting on the bipyridine. It’s not sitting on the rhenium in the first case. So this certifies that that’s really what’s happening.

Because if you put it on the metal, it’s well known from the literature that it results in about 100 wave number shift in that case. So we got the electron out there. So how long does it take to get out there? Well, it takes about 21 picoseconds. Actually the second step is faster than the first. It’s inverted kinetics. But it’s out there now for 43 microseconds. So this thing really just hangs in there forever.

But as I mentioned, this is the less desirable one, because if we want to turn the crank and do this again, this ligand ultimately is going to fall off on us. All right. Let’s take a look at the other one. The other one, the electrochemistry is almost as equally nice. I mean, once again, the rhenium is a little bit irreversible. But nevertheless, the potentials that we measure are more or less the same as the other example. Same experiment, again.

In this particular case, again, reduce the system and then pick a place to excite. When you do this again, once again, you get bleaching out, now you can actually resolve the three carbonyl bands, and then you get the three carbonyl absorptions growing in. In this particular case, we get out there in five picoseconds, and we cannot resolve the two steps in this particular case. But it only hangs in there for 24 nanoseconds.

So we lost three orders of magnitude in a lifetime by attaching to this stronger coupling spot, but nevertheless, we can see if we can actually utilize this. Turns out that if we’ve recently attached it to this six position here, adjacent to the rhenium, and you restore at least two orders of magnitude of lifetime by doing that, there are reasons because of twisting of bonds and such that will get you there.

OK, so one thing you always worry about with any of these systems is it still a catalyst after you’ve added all this stuff to it? Right, if you change the electronic structure, so it’s [INAUDIBLE] a catalyst. Well, in this case we can measure its electro catalytic activity. What doesn’t reduce CO2 to CO, sure enough it does, if you go out to negative potentials, the presence of CO2 relative to argon, you see a big catalytic wave.

So it’s every bit as good a catalyst as we started with. No problem there. So we decided– my students got really excited, said, hey, let’s make some CO. Well, basically, I told them, look with a 24 nanosecond lifetime, you’re not going to do very well. So they chose a spongy carbon electrode, which had a huge surface area, used red light on this system, and it did manage to actually get some CO out of this thing. It is above just loss of CO. In other words, for instance, its turnover number is more than three.

But nevertheless, because diffusion is killing us here, it’s not as good as we’d like. So where do you go from here? Well, where you go from here is take advantage of the other pigment that I told you about. And that is, for instance, what you really want to do is attach this thing to p-type semiconductor. You can use the perylene dye, and you can have two for one here.

In other words, this chromaphore can act at both ends of the spectrum. For instance, if we excite this molecule with green light, it’ll pull an electron out of a p-type semiconductor, nickel oxide, whatever. And you’re going to have this anion sitting here now, and this anion then can be excited to it’s excited doublet state with 950 nanometers, and push the electron out to rhenium. We know that’ll work. The potentials are good enough, which means it puts the PDI back in its ground state, which you can turn the crank one more time and get one more electron out into the rhenium.

So you can actually make this work in a real catalytic fashion. And since it’s attached to the electrode, nothing to worry about with regard to diffusion as well. So that’s the advantage here is you can use both the visible and infrared photons in making this system go. So that’s where we’re going with this particular approach.

OK so with that, given the time, and given how long I’ve gone on, let me just summarize here. Basically arylene diimide radical anions, you can reduce them reversibly. They’re very robust. They have really nice, well-behaved spectral properties that you can access using their infrared photons, you can access these, very energetic states, and these excited states are powerful reductants that can drive energy demanding reactions like CO2 reductions.

Also there’s a commensurate whole topic on the oxidative side we’ve been able to use cation radicals as super oxidants if you will in that side, but I don’t have time to tell you about that today. So let me just thank my group, who are shown here. Some of their names are on the slides, but basically Jose and Nathan were responsible largely, and Joe [? Christensen ?] for the catalytic work. The singlet vision work I told you I was largely Michelle Chen here, along with see– who else is here? Along with Ryan and to a large degree, the work on the self-assembly work is Jenna’s work, Jenna Logsdon. And [? Yulin ?] [? Woo ?] was the driver behind the quadruplex work.

So with that, I thank my colleagues. We have a very collaborative environment. And so they all helped out and thank the DOE for support. And thank you once again for the invitation, the opportunity to tell you this story. Thanks very much.

We have time for a couple of questions.

[INAUDIBLE] just looking for a clarification. Were you hitting it with a pulse or a continuous beam, or how were you doing it?

I mean, for the dynamics we hit it with a [INAUDIBLE], but you know, you can use a lamp for the steady state turnover.

That’s what I was getting at, yeah.

Yeah, it’s not a problem, because the extinction coefficients of the various absorption bands, even in the near-infrared are pretty substantial. Unlike some systems where the near-infrared bands are very, very weak.

[INAUDIBLE] I’m just curious about, first of all, just sort of size elements. Is your sample large?

Yeah, it’s actually– the film is freestanding film in a resonator. It’s macroscopic. I mean it’s about two or three millimeters wide by a centimeter or two tall.

Yeah. And then, of course, your photo exciting some fraction of the chromaphores, right? So are you calibrating the measurement, sort of making an estimate of how much–

Yeah, we do blanks and things like that. We do have calibrations for this, because TRMC can be a little bit flaky, if you don’t do that.

So roughly, how hard do you have to excite in order to make a measurement like that?

Not hardly at all.

I see.

I think we have– I don’t know I think it’s a couple hundred micro joules on the sample in a fairly big area.

Yeah, so a small fraction is fine.

Right.

[INTERPOSING VOICES]

Some people do microwave conductivity without a resonator, just in a cavity that’s not resonant. You get better time resolution, but you get lousy signal.

Yeah.

So we do any of the resonators, so we throw away some time resolution, because we don’t need it. But you know, you get incredible sensitivity, so you don’t have to hit it very hard.

yeah OK. Great.

So I had a question about the G-quadruplexes [INAUDIBLE]. How long is that charged transport length just in terms of how large it is and what limits, the length of that wire?

Yeah, I think– well, I mean defect site’s limited. But I think we’re estimating that the length of the wire should be somewhere between 100– before we get a defect, the estimate is somewhere between 50 and 150 in register there. So I mean, it’s kind of a rough estimate. But if you think about it, what you really need in order to do a solar cell that would be absorbed 90% of the light and put the electrons in the holes in the right direction, you’d need about 300 molecules in the layer.

So we’re getting there. But we’re not quite there yet.

I’m curious from those quadruplexes, has anyone characterized these in the terahertz regime? Because I know it’s been studied in proteins.

No, they have not been characterized in terahertz regime, but we’ve done just about every other thing you can imagine on them, everything from stimulate Raman to EPR techniques and whatever.

All right. This is the last question because Mike has to go to the airport.

In your system, you create a very powerful [INAUDIBLE], would it be possible that you’d get to direct reduction of CO2?

That’s a super question. In fact, I’ve heard for weeks, actually for months, I was encouraging my group members, why don’t you just bubble CO2 in the solution and see what happens? It turns out that the excited state lifetime is not long enough, even at 0.3 molar CO2 and DMF to actually see meaningful amounts of CO. But if we– one approach we’re doing is we’re actually in collaboration with [INAUDIBLE]. We have an NDI MOF, which actually– we’ve actually published a paper. We can actually reduce dicholoromethane and sort of mitigate chlorocarbons and solvents, but we’ve yet to try that experiment with a high concentration of CO2. We think that might actually do it.

Great. So let’s thank Mike again.


Download the slides

Basic Energy ScienceEnergy EfficiencyRenewable Energy
Share:

Press Inquiries: miteimedia@mit.edu

Subscribe to energy news & events: (Preview)