 Well, thank you so much for that. I'm really kind introduction. Thanks all of you for coming. It's really great to be giving the energy seminar. Um, I guess before I dive into the talk, I do want to make sure I acknowledge all of the graduate students and postdocs and collaborators who made this work possible. There are a number of people on this list. Oh, I'm sorry to bring a laser pointer. Um, does anyone happen to have one? If not, I'll just point. Um, but I especially want to give a shout out to Eileen Ko, the staff sciences here at Stanford, in charge of the transmission electron microscopes, and also Bob Sinclair. Oh, oh, thank you very much. I appreciate it. Ah, perfect. Um, so as you guys heard, uh, energy seminar take one, um, was supposed to be the first day of last quarter. I'm sorry, I couldn't make it then, but, um, we, uh, or I guess I had the chance to witness a different form of energy that day. Um, the birth of our second son, Hugo, and this is our first son, um, Marcus, meeting Hugo for the first time. He was pretty excited to become a big brother. Um, and, uh, I mean, I wish I was here to chat with you, but I'm really excited to have the opportunity to share with you my work a quarter later. And I think in giving birth and having a family, it kind of puts, um, the energy problem in perspective. And what I'd like to open up my talk with is, um, a quote from one of my favorite authors, Jack Kerouac, which says, the only people for me are the mad ones, the ones who are mad to live, mad to talk, mad to be saved, desires of everything at the same time. The ones who never yawn or say a commonplace thing, but burn, burn, burn like fabulous yellow Roman candles, exploding like spiders across the stars. And I feel like many people at Samford fit into this category, that's why I love being here so much. And I'd like to pitch the question just to open up the talk, what makes you mad to live? It may be science, it may be renewable energy, it may be family, it may be mountain biking, um, pretty much all of those things fall in the category of things that I'm mad to live about. And one particular thing that definitely makes me mad to live, um, is the fact that average temperatures will increase by more than 2 degrees Celsius, unless 90% of our energy is from carbon-free sources by 2050. And as many of you know, in 2016, renewable sources accounted for only about 17% of world energy consumption. So I find this to be a pretty shocking statistic and one that's hopefully a call for action, um, for all of us. Thankfully, I think if we look to the field of nanophotonics, there's been a lot of hope, and if you look at all of the speakers that this series has hosted in the past few years, they're really, I think, optimistic or, um, kind of hopeful picture for the future that I would agree with. So, just talking a little bit about what nanophotonics can do to help address the energy challenge, there's been a lot of really beautiful work, including by some speakers who have spoken at this seminar in the past, for example, Ilya Yablinovich, on light management for photovoltaics, where people are designing new materials and new methods to precisely control how photovoltaic materials absorb the solar spectrum, giving rise to structures that have near-unity absorption across a fairly large bandwidth and can efficiently convert photovoltaic material or convert sunlight into electricity with high efficiency. I'll just point out Harry Atwater has some of the most efficient solar cells demonstrated to date with his company, Alta, and he's used a lot of these tricks and techniques. Kind of a second area where nanophotonics has helped in enabling renewable energy generation in storage is in enabling more efficient storage of electricity in fuels. So, for example, people have used nanostructured materials, for example, nanostructured gold to enable efficient water splitting. And then a final area that I think has been discussed in this seminar series in years past is using new nanomaterials and particular plasmonic materials for water purification and also pollutant removal. So, a particular area of interest is, for example, in using excited carriers and metals, hot carriers and metals that can be injected into a semiconductor to enable new and efficient redox reactions in, for example, water purification. So, there's a broad spectrum of areas in which nanophotonics has helped address this challenge, and the two areas I want to talk about, in part, draw on these examples, but also provide kind of a new perspective to how plasmons can help. First, I want to talk about efficient renewable energy storage. I think storage is an important part of the renewable energy challenge. First, of course, is generation. And what I'll present in this part of my talk is a new institute microscopy technique based on electron microscopy and plasmon spectroscopy to understand how chemical energy storage depends upon nanomaterial and nanoparticle size and shape and crystallinity. And then in the second part of my talk, I want to talk about efficient solar energy conversion for either photovoltaics or for solar fuels. And I'll present a new up-conversion scheme that can take lower energy photons nominally from the solar spectrum and convert them to higher energy photons to enable more efficient solar fuel or photovoltaic conversion efficiencies. And this up-conversion scheme is based on hot carriers that take sub-bang at light and convert it to above-bang gap light. Okay, so let's start with the energy storage part of the picture. So probably all of us has one of these or one of its cousins basically in our pocket. And while we may complain that the battery life doesn't quite last long enough, it's pretty remarkable that in 2016 the weight of say an iPhone is about 0.3 pounds and only costs about $200 and the battery can last 300 hours on standby and it only takes about two hours to recharge your phone. And you can get about a thousand charge cycles before you need to replace your battery on the iPhone. If you compare that with the phones that were out in 1984, for example the Brick phone that Gordon Gekko used in his rendition of Wall Street, the phones weighed over two pounds. They cost about $10,000, $2015. The battery lifetime was only an hour on standby. It took 10 hours to recharge the battery and the battery only lasted for 100 charge cycles. So like I said we may complain about battery life here, but we've come a long way from 1984 to today and a lot of the reduction in weight and cost is due to advances in battery technologies. So if you look at a big picture of how a lot of energy storage devices, including batteries and fuel cells work, generally ions are intercalating into a host material, into an electrode, allowing that energy to be stored. And if you take a scanning electron micrograph of the interior of a battery, such as the Samsung battery, you'll find that the electrode is composed of a lot of nanoparticles. So ions aren't just going into a bulk matrix, allowing energy to be stored, but generally they're going into nanoparticles. And that's because it's been found heuristically that these nanoparticles first of all have faster kinetics. So you can get the ions into and out of the material faster, allowing your battery to charge up faster. So just going back to this slide, now you can see basically it only takes about two hours to recharge a battery instead of 10 hours. Further it's been found that they have an extended life cycle. So again, going back to this slide where we have a thousand charge cycles on our iPhone now compared to less than 100 charge cycles back in 1984, it's been found that if you nanostructure the electrodes, you can charge and discharge your battery many more times before it needs to be replaced. And then finally it's been found that these nanoparticles tend to have size dependent thermodynamics. So for example, you could charge up your battery at a lower potential, a lower chemical potential, if you nanostructure your electrode. But looking at this picture you'll notice that there are nanoparticles of a variety of sizes and shapes and crystallinity, and it's not really clear which particles are best. And that's part of what my talk is going to address today. Unfortunately lithium ion systems are kind of like messy and volatile systems to work with. There are some researchers on campus who are doing some really beautiful work actually looking inside battery electrodes to look at how lithium gets into these materials. But what we want to know is not necessarily how lithium is getting in, but really how can we design next generation battery electrodes by watching this process in well engineered nanoparticles and then saying okay in this particular shaped nanoparticle here's what happens, or in this particular sized nanoparticle here's what happens. So instead of looking at lithium intercalation into nanoparticles we're going to look at a much simpler system but one that has very analogous physics. It turns out the thermodynamics of the system are quite similar to the thermodynamics of what is thought to happen in lithium ion systems. So it's a system of palladium hydrogenation and it's a very old system. Basically studies of the system started back in the mid-1800s. It was one of the first phase transformations in bulk to be well understood. And in bulk here's the phase diagram. So I'm plotting the hydrogen gas pressure on the vertical axis as a function of the concentration of hydrogen per palladium. So the number of hydrogen atoms per palladium atoms in the crystal lattice. And you'll see that at low hydrogen pressures the system basically exists in the alpha phase. It's a dilute interstitial solid solution. So you have hydrogen gas in the environment. It gets catalytically split at the surface into two hydrogen atoms or ions that each intercalate into the host matrix and basically sit at those interstitial sites in the crystal lattice. Now let's just fix our temperature, say for example at 120 degrees C. So we're in the alpha phase and then as we ramp up the hydrogen pressure at one pressure the system can very abruptly uptake quite a bit of hydrogen entering what's known as the beta phase. So the beta phase is sketched here. It's generally where say at room temperature at 20 degrees C more than 60% of those interstitial sites are occupied with hydrogen. This beta phase is characterized by a pretty significant volume expansion. So in bulk the volume expansion is about 3% in lattice constant so about 10% in volume. So there's a pretty significant volume expansion associated with this phase change. And then I'll also mention that kind of in between the two phases you have this coexistence region in bulk so coexistence of alpha plus beta. So a handful of years ago people realized just like they did with the lithium ion batteries that in the simple hydrogenation based system when you start nanostructuring materials you get some pretty interesting physics to emerge namely faster kinetics, the size tunable thermodynamics and also extended life cycles. And here was one really nice study that was by Jeff Urban's group at Berkeley where they did ensemble studies of nanostructured palladium and they were able to see significant deviations from bulk behavior. In this particular study they saw that when they nanostructured the palladium there was a reduced hydrogen storage so they couldn't get quite as much hydrogen into the system as they would have liked and there was a more gradual phase transition. So you'll notice on this slide that the phase transition happens rather abruptly. So just kind of working through each of these electron micrographs and the associated isotherms so they're basically fixing the temperature and mapping out the phase as a function of hydrogen pressure. Here are larger nanoparticles that are 110 nanometers across and they're overlaid on an isotherm for the bulk material and you'll notice they look quite similar. One of these graphs corresponds to loading so they're increasing the hydrogen pressure and then they decrease the hydrogen pressure and this system shows hysteresis so these particles tend to unload hydrogen at a lower hydrogen pressure than they would absorb hydrogen. So this looks a lot like bulk and then as they nanostructured the materials these were all single crystalline nanocubes that had sizes of 32 nanometers and you'll notice that these isotherms became more sloped so you'll notice that for a fixed hydrogen pressure the amount of hydrogen in the system was less than what they had in the larger nanoparticles in the bulk system so hence the reduced energy storage capacity or hydrogen storage capacity and then you'll see the gradual phase transition become more pronounced as the nanoparticle size was reduced. So this was an ensemble study and it really intrigued researchers saying huh this is kind of unlike what we thought we might have for batteries we're in batteries we've seen kind of heuristically or by like check and guess that these nanostructured electrodes perform better here they seem to imply that perhaps the nanostructured materials may not perform as well in these hydrogen based systems but researchers thought it could have been just an artifact of it being an ensemble measurement so while these were really well controlled nanoparticle synthesis where all of the nanoparticles in the measurement had a well controlled crystallinity and size it is possible that still you're averaging over the ensemble and individual nanoparticles have slightly different characteristics. So researchers turned to single particle studies and plasmas enabled the first single particle study starting in 2011 there was first this really nice study by the Longhammer Group at Chalmers University where they put small palladium nanoparticles on top of a gold nano antenna and then use optical scattering and spectroscopy to basically detect what the phase of the palladium would be if you just had the palladium on its own it turns out a single particle wouldn't quite scatter strongly enough to be able to figure out what the phase would be but by having this gold nano antenna on top they could get a pretty significant scattering signal and when there was no hydrogen they got the scattering signal shown here in blue when there was hydrogen in the environment they got the scattering signal shown here in pink and they were able to look at shifts you can see basically in the full with half max there's a slight change in the line shape to map out isotherms so the isotherms in this measurement looked quite a bit more like bulk regardless of the size nanoparticle they put on top of this gold nano antenna there was a related study by the Alivisato Group that was published shortly thereafter where they had kind of the planar version of this structure again there's a gold nano antenna they mentioned that these gold nano antenna support surface plasma and resonances that basically allow light to be focused onto the palladium nanoparticle and then allow kind of the phase to be transduced as an optical signal into the far field so in this planar analog they mapped out kind of something like an isotherm so hydrogen pressure versus wavelength here they were looking at shifts not in the full with half max but in the peak scattering signal and they saw isotherms that were a little bit more sloped upon cycling a couple of studies but if you look broadly through the literature there is a lot of ambiguity as to what might be happening in this particular system as you nanostructure the materials and right around the time this was happening I got the chance to go to Sweden where I met with a professor, Eger Zorich and told him we were trying to understand the thermodynamics and kinetics of the palladium hydride system and then he informed me, ah, if you can resolve this debate it'd be great because right now my question to God would be what are the hydriding thermodynamics of a single palladium nanoparticle so that's what, ah, hopefully the first part of my talk will address so the technique that we decided to rely on is, ah, in situ, ah, electron, ah, microscopy and spectroscopy and in particular we're going to rely strongly on a technique called electron energy loss spectroscopy and an environmental scanning transmission electron microscope so he synthesized, ah, it's kind of a mouthful of words it's actually a very easy technique to understand first of all we synthesize palladium nanoparticles we disperse them on an electron transparent substrate and then we come in with an electron beam you can think about the electron beam as kind of a broadband white light source but here just focus down and that allows us to get a high spatial resolution and then the electron beam passes through, ah, the nanoparticle and we can use the electron beam to image the nanoparticles but we can also use, ah, the electron beam to enable spectroscopy and in particular when the electron passes through the particle it loses some of its energy by exciting some of the electronic modes in the nanoparticle and it turns out that the various modes that get excited at different wavelengths are proportional to the permittivity of the material or essentially the dielectric constant so the electron energy loss spectra signal is, ah, proportional to the imaginary component of one over the permittivity so ILS is a really good signal of kind of the electronic signature the electronic structure of a material and that's something that you can imagine gets changed when there's either a new phase or when there's hydrogen intercalating into the palladium so advantages of using, ah, this technique are many but I'd say one huge advantage is that we have near atomic scale resolution so in the electron microscope there is an imaging spatial resolution of 0.07 nanometers subangstrom, ah, basically imaging spatial resolution so we can look, ah, ah, sorry for the typo here we can look both at and within single nanoparticles also since we're colloidally synthesizing our nanoparticles we have control over the sample shape and the crystallinity so we can synthesize a variety of shapes and sizes and crystallinities and then look exactly at the thermodynamics of a single nanoparticle to figure out which ones might be best to kind of maximize this energy storage capacity and then finally unlike many of the other single particle measurements that are out there that rely on nano antennas to kind of transduce the signal to the far field, ah, this is a direct measurement of the chemical and electronic properties we don't need to have any nearby nano antennas that may otherwise influence the thermodynamics or kinetics I want to show you, ah, just three different classes of nanoparticles, ah, basically a square, a prism, ah, and an icosahedron, the square and the prism are single crystalline nanoparticles so by looking at these two nanoparticles we can deconvolve the effects of shape and then this icosahedron is polycrystalline so we can look at how different crystallinities might in fact the storage capabilities so we disperse our nanoparticles on an electron transparent grid and then we identify individual nanoparticles that we want to look at more closely and like I said we can image each of these nanoparticles with very high spatial resolution and then we switch over to, ah, something called scanning transmission electron microscopy mode where we basically focus the electron beam down a little bit but we lose a little bit of imaging resolution there in order to be able to get higher spectral resolution so here's one nanoparticle it's 23 nanometers across, it's palladium and then we, ah, here, ah, spread out the beam to probe about 5 to 10 nanometers across the center of the nanoparticle and then, ah, the electron beam goes through the center of the nanoparticle and we collect an electron energy loss spectra so this is basically eels counts or intensity as a function of energy or wavelength and when we have, ah, no hydrogen or very little hydrogen in the environment we just have a pressure of 4 pascals around the nanoparticles we find a spectra that has a number of peaks one of which corresponding to a little bit of hydrogen in the environment that's at about 12 electron volts one of which that corresponds to our silicon dioxide substrate on which we've placed the nanoparticles and then we have another peak at 8 electron volts that corresponds to palladium and I like to refer to this peak as, ah, the incredible bulk Plasmon resonance for those of you, ah, who have taken say like, ah, an optics class or an electronic materials class ah, you probably know that bulk Plasmon resonances are kind of these like spherically symmetric breathing modes of electrons in the nanoparticle you can almost think about this electron beam as being a point negative charge so all of the conduction electrons in the palladium are going to pile up at the edges because they're repelled from the electron beam and then when the electron beam has passed through all those electrons kind of relax back into the center of the nanoparticle so they're kind of breathing in and out ah, that's that peak that occurs here at about 8 electron volts when we introduce hydrogen this bulk Plasmon resonance shifts quite significantly by over 2 electron volts and if you compare that with some of the other nano antenna based studies where you had just a slight change in the full width half max this shift is ah, something that gives us a very incredible signal to noise ratio with which to map out that phase transformation so going from 4 pascals up to 100 pascals of hydrogen you'll notice this hydrogen peak has jumped up pretty significantly so we can use that as kind of a metric to know how much hydrogen we have in the environment ah, kind of a control to also double check just our gas valves and then we also have on the shoulder of the hydrogen peak the silicon dioxide peak from the substrate and then the fully hydrated palladium ah, hydride peak at just below 6 electron volts so we can take these sorts of spectra ah, at pressures so basically we're going to ah, over here increase the pressure from 4 pascals up to 584 pascals and then plot each of these spectra basically as a color map a horizontal slice on this map showing as a function of color the eels intensity and then we can decrease the pressure back down to 4 pascals so you'll notice that ah, originally the eel spectra shows a peak at just around 8 electron volts and then at one pressure there's a pretty abrupt jump ah, over to about 6 electron volts where the particle is fully hydrated it stays there as we're reducing the hydrogen pressure and then jumps back over to its ah, alpha phase so we've got the alpha phase and the beta phase we can fit the peaks of the spectra basically the map out isotherms I apologize that you need to do a little bit of mental gymnastics right now these isotherms are flipped from how I introduced the talk ah, so instead of having the hydrogen concentration plotted on the horizontal axis now I'm showing energy but just know that this corresponds to the alpha phase over here corresponds to the beta phase and just like I pointed out with some of those ensemble measurements these nanoparticles do exhibit hysteresis so they tend to unload at lower pressures ah, than they load you'll also notice that each of these isotherms is actually quite sharp so it's unlike some of the single particle measurements that were out there and unlike the ensemble measurements that were out there and also you'll notice that unlike bulk there are no points that occur in this coexistence region ah, so you'll remember in bulk we had a phase diagram we're in here we had both alpha and beta coexisting ah, but here when we're doing measurements in thermodynamic equilibrium ah, we don't see any small nanoparticles coexistence so we can do this study ah, on a number of particles here we're varying the particle size from about 13 nanometers up to just about 30 nanometers and we've now done these studies up to a few hundred nanometers but I just want to give you kind of a flavor of what we've seen first of all we've noticed that all of these nanoparticles down to about 15 nanometers in size exhibit these very sharp transitions from the alpha phase to the beta phase without there being really any coexistence ah, another thing we've noticed ah, I'll just point out that we can do eels maps to confirm the existence of only a single phase in equilibrium so when I collected the spectra I had slightly defocused the beam so that it was sampling over a large area of the particle if we want to take eels maps what we can do is focus the beam down to an even smaller probe size on the order of two nanometers and then scan it across the nanoparticle at different points so for example if we take an eels spectra focusing the beam down to this point here shown in green at zero pascals we just have the alpha phase and over here in pink we also just have the alpha phase and then if we increase the hydrogen pressure or just pure palladium in this case if we increase the hydrogen pressure to 250 pascals at each of these points you'll notice they both exist in the beta phase and what we can do is basically not just look at two points but look at hundreds of points within the nanoparticle to figure out what phase it's in and at low hydrogen pressures the system is entirely in its alpha phase at high hydrogen pressures the system is entirely in its beta phase so that was kind of one trend that we saw from the isotherms another is that if we looked at the loading pressures for each of these nanoparticles and compared it with the bulk loading of hydrogen and palladium these transition regions always fell well below the bulk alpha to beta phase transition so the bulk is shown as the white dotted line you'll notice that the transition from the alpha to the beta phase in nanoparticles is always below that line and another trend we saw is that if you plot the loading pressure as a function of particle size the smaller particles tended to load at lower pressures so it's much easier to get the particles that are smaller in the beta phase or basically they transition to the beta phase at lower hydrogen pressures and one way of rationalizing what's happening there is that when hydrogen gas catalytically splits into hydrogen atoms at the surface hydrogen can basically sit at those interstitial sites kind of saturating the surface sites and I told you that when we transition to the beta phase there's a pretty big change in volume a 10% change in volume so right at the surface these fully saturated sites are kind of lattice expanded and they're starting to pull on the core of the nanoparticle where hydrogen has not yet entered so there's essentially tensile strain at the surface that's acting to kind of pull apart the nanoparticle making it easier for hydrogen to get in there if the particle is smaller because those surface layers have more of an effect on those smaller nanoparticles and shortly I'll kind of show you some videos showing this process happening in real time to kind of help confirm this hypothesis okay so I mentioned that we're going to talk about a couple different shapes so how does nanoparticle shape affect thermodynamics if we look at cubes pretty much regardless of size like I said as long as they're above about 15 nanometers we wind up with these very sharp transitions here now I actually am plotting the concentration of beta on the horizontal axis so cubes load abruptly and have sharp transitions it also turns out prisms regardless of size have these very sharp transitions with no points in the coexistence region but if we look at icosahedra their isotherms are quite a bit more slow so almost more like those ensemble measurements right now and you'll also notice that once we're in the beta phase we don't actually have as high a hydrogen concentration as we would have in these single crystalline nanoparticles so first let's take a look at the prisms let's do an eels map and just confirm that we don't have phase coexistence here's the eels map for prism and again just like with the cubes these prisms exist either only in the alpha phase or in the beta phase in thermodynamic equilibrium there isn't coexistence but if we do that for icosahedra here at low hydrogen pressures or no hydrogen we essentially just have the alpha phase in the core but then when we move up to an intermediate hydrogen pressure we have the alpha phase still existing in the core and the beta phase in the shell so we have parts of the nanoparticle that aren't transforming to the beta phase and it turns out that occurs even up to the highest hydrogen pressures that we can access so our hypothesis is that if you consider what an icosahedra is composed of it has 20 tetrahedra that don't quite fit together perfectly so if you want to colloidally synthesize one of these nanoparticles it turns out that all these icosahedra or tetrahedra pack together but they leave kind of a strain gradient where the core is compressively strained and the outer part of the icosahedra is tensely strained so the tetrahedra kind of need to pull apart at the surfaces to make one solid nanoparticle so this compressive strain at the core makes it really hard for hydrogen to get in there kind of regardless of how tensely strained the shell is it's going to be very challenging for hydrogen to make its way into the core and that's why we think we're seeing coexistence of phases in the icosahedra also I mentioned that there were these sloped isotherms for the icosahedral nanoparticles and we wanted to investigate that a little bit further so we turned to diffraction and dark field imaging here what we do is we take a diffraction pattern of our nanoparticle and then we know that different spots in our diffraction pattern correspond to different crystallites in our nanoparticle so what we can do is center our aperture over one of those diffraction spots and then image which part of the nanoparticle that spot was coming from so for example spot number one which is very close to spot number two corresponds to this particular tetrahedra within the icosahedra that's diffracting to that point and then what we can do is look at how individual tetrahedra or pairs of tetrahedra in the nanoparticle are transforming so here are isotherms for in this case an individual tetrahedra all the rest of these are pairs of tetrahedra and you'll notice that the transition pressure from the alpha phase to the beta phase is occurring at different points within the nanoparticle so this kind of helps explain why there's a little bit more of a sloped transition basically different parts of the nanoparticle are transforming at different hydrogen pressures and what we can do is take our diffraction data and our illus data and combine that together to make a video kind of showing in three dimensions how these polycrystalline nanoparticles are transforming so to kind of sum up this part the core of the icosahedra is compressively strained preventing hydrogen storage there and basically reducing its energy storage capacity and then grain boundaries are essentially decoupling all the crystallites within the nanoparticles leading to mosaic loading or a loading where different parts of the nanoparticle basically transform at different pressures okay so all of that was in thermodynamic equilibrium I told you that we had this kind of hypothesis for why say smaller particles were loading at lower hydrogen pressure so we wanted to know can we visualize non-equilibrium states and here's one video ironically captured on an iPhone 6 where different colored regions of the nanoparticle correspond to different phases so again we're in stem mode scanning transmission electron microscopy mode and we're taking advantage of both the different electronic or ill scattering spectra plus the different diffraction patterns of the two phases so white here corresponds to the beta phase black corresponds to the alpha phase and hopefully you saw kind of this triangular front move across the nanoparticle as it transitioned from this kind of stripey phase into the pure beta phase so here we started off with the nanoparticle and kind of a coexistence between alpha and beta so we're fixed at a set hydrogen pressure and then there's this diagonal phase front that kind of moves in from the corners and eventually takes over the entire particle when it becomes beta so when we first captured this movie we're like oh this is really cool it shows that unlike those equilibrium measurements that I just presented these non-equilibrium states show a clear coexistence of both phases so in equilibrium if we have a single crystalline nanoparticle we don't have phase coexistence but in order to get from one phase to another phase we do need to have coexistence and it turns out that coexistence is actually within the bulk of the nanoparticle here's another movie where we wanted to capture the beginnings of ion intercalation we didn't just want to capture the process kind of halfway through let me see if I can find the video to play this so here again white corresponds to beta and hopefully you saw it kind of growing in from the corners of the cube it kind of sprawled out to form these fingers and then eventually forms a straight phase front that moves across the nanoparticle and then the old phase gets pushed out and then eventually the particle is fully transformed so here are some snapshots of this video here you saw the mouse moving around because we were using kind of a separate technique to confirm which color was which phase so here are snapshots of that video where right when the video starts or shortly thereafter the beta phase the new phase kind of grows in from the corner you can see it's also kind of wetting the surface of the nanocube and then eventually this beta phase kind of all aligns itself to be along a low axis of this nanocube it kind of marches across the nanoparticle moving from 272 seconds to 324 seconds here we've got the beta phase the contrast has kind of switched I can explain that in the Q&A session if you're interested to understand why and then the beta phase grows across the nanoparticle eventually pushing the alpha phase out through a corner of the cube so in addition to looking at the stem movies that kind of tell us how hydrogen is intercalating into the nanoparticle in a nanocube basically entering through a corner and then reorienting itself from this diagonal interface into something that's straight across and then pushing the old phase out through a diagonal we can also look at diffraction patterns and I mentioned that this was a single crystalline nanocube and when we introduce hydrogen that new phase has a different lattice constant than the alpha phase so we can rely on diffraction so the innermost circle corresponds to beta so basically as a smaller radius corresponding to a larger lattice constant and then the outermost circle corresponds to the alpha phase so right at the beginning of the transformation we have coexistence of alpha and beta but in this diffraction pattern you'll notice that the spots are very circular corresponding to essentially a very good crystal there aren't many defects or imperfections in this crystal but as the transformation occurs you'll notice here you'll notice that the diffraction spots get more diffuse and elliptical meaning that basically imperfections are formed basically there are lattice rotations and potentially defects that are forming and then at the end of the transformation you'll notice that the diffraction spots kind of tighten up again implying that the nanoparticle is trying to heal its imperfections and if we go back and take a diffraction pattern of the pure beta phase nanoparticle it's again back to a perfect single crystal so really importantly these small single crystalline nanoparticles even though they exhibit phase coexistence during the transformation have the ability to heal themselves at the end of the transformation okay so what we've learned basically single crystalline nanoparticles regardless of shape can maximize chemical energy storage basically you can get the particle to fully transform and if you have smaller nanoparticles you can get that to happen at smaller chemical potentials so smaller particles store hydrogen more readily i.e. at lower hydrogen pressures particles with defects or strains such as the icosahedra tend to have a reduced energy storage capacity so essentially we're not able to get any ions into the core of those compressively strained nanoparticles and then also we found that single crystalline nanoparticles can self-heal so they have the ability to push out defects and lattice imperfections leading to extended device life cycles so that kind of helps to explain why these nanostructured electrodes and batteries give you more charge cycles so a thousand compared to say a hundred from 20 or 30 years ago but this I guess a kinetic study also opens up the platform for kind of rationally designing next generation battery and energy storage device electrodes so what's next for those of you who haven't seen the electron microscope on campus it's basically one building over I think we can use this tool to help unravel other important nanomaterial transformations for example vanadium dioxide it's a pretty common phase change material used in some electronic devices or bi-metallic alloys which are used quite frequently in different redox reactions for example for solar fuels or electro-chromics for example for smart windows like helping to design nanomaterials for various electro-chromics and then I'll also mention that what we've been working on lately is finding a way to couple light into and out of the microscope I kind of started off this talk with a pitch for nanophotonics and then I focused a lot on electron microscopy so being able to couple light into and out of the transmission electron microscope I think is really exciting because we'll have the ability to probe photochemical reactions in an environmental cell whether that's a gaseous cell or a liquid cell okay so I know I only have a few more minutes and I want to make sure I leave enough time for questions but since I started off with nanophotonics I do want there to be some light in my talk and I just want to transition over to kind of quickly introduce to you a new up conversion scheme for utilizing sub-band gap photons in solar fuels and in solar cells okay so what is up conversion and what can it do to help us address renewable energy if we take say just a single junction solar cell we know that it can only absorb light above the band gap of the material and that means that depending upon the band gap of the solar cell somewhere between about 20 and 50% of sun's energy can't be absorbed it's basically lost to transmission so the idea of an up converting solar cell or solar fuel is that you would place the material behind the solar cell that can take the lower energy photons and combine them together into a higher energy photon that then can be absorbed by the cell material or by the cell above it contributing to photo current and unlike a multi-junction cell the scheme doesn't require say lattice matching so you don't need to grow the up converter behind the solar cell you can just say colloidally synthesize and up convert and then spray cast or spin coat these nanoparticles on the back of a cell and then also importantly the up converter can be electrically insulated or isolated from the cell so you don't have to worry about things like current matching like for example in a traditional multi-junction cell that's essentially limited by the worst cell in that stack here the up converter can only help to boost solar cell efficiencies so if you look at calculations at how much an up converter can improve a solar cell with no up converter if you do a thermodynamic calculation the peak efficiency is about 30% for a single junction cell and if you add in an up converter that peak efficiency jumps up to about 44% for a significant improvement in cell efficiency of course most up converters are not ideal in fact none of them are most of them only absorbed say over a narrow bandwidth of the solar spectrum so obviously as you can increase the bandwidth of where the up converter is absorbing you're going to get a larger boost in the solar cell efficiency so ideally we want a broad bandwidth absorber for up converter and one that has a high efficiency so let's take a look at some current up converting materials I think two of the more promising ones are the bimolecular systems which are quite good at up converting visible light so for example they could find use in photo electrochemical cells and then lanthanide based systems are also quite common they're quite good for near infrared to visible up conversion so if you assume these materials are ideal solar cell efficiencies can be improved by up to two percent for these bimolecular systems these would of course be higher junction band gap solar cell or higher band gap solar cells with these lanthanide systems solar cell efficiencies can be improved by up to six percent so this is quite encouraging especially for something like a silicon solar cell where it's hard to make a tandem design but these calculations while they take into account the narrow absorption bandwidth of the up converter they assume essentially unity quantum yields and since it's two photon process that means they're assuming that these up converters are fifty percent efficient and it turns out that these are generally nonlinear processes with typical quantum yields of less than five percent so what my group has done is worked on kind of a new scheme for up conversion and it's kind of based on some insights from plasmonics where we have these metals that can act as nano antennas and kind of historically what people have done is use metal nanoparticles near an up converter to kind of more efficiently funnel light into the up converter and then get the higher energy light out of the up converter but what we wanted to do is not just use plasmonics to enhance these sorts of light matter interactions but enable new up conversion schemes so here is kind of a layout of the up conversion scheme we have a metal, a nanostructured metal near a semi-conducting quantum well and when two photons come in they excite electron hole pairs basically from the Fermi level up to higher in the band diagram and from kind of deeper in the metal band diagram up to the Fermi level and then this hot electron and this hot hole can get injected into the semi-conducting quantum well and basically get trapped there until they radiatively recombine to emit a photon that's of higher energy than either of the incident photons so importantly this is a linear up conversion scheme I can talk you through in the Q&A session why it's a linear scheme it works with incoherent illumination so you don't need the electron and the hole to be temporally correlated to get them trapped in the semi-conducting quantum well and it's also spectrally tunable and more broadband kind of based on how you nanostructure this metal so let me just quickly talk you through some calculations if you take for example a metallic nanoparticle on a semi-conducting substrate the absorption efficiency of this nanoparticle tends to peak at a given energy the energy can be shifted around just based on the metallic material you're using and the size of the particle so here this is something called the Plasmon Resonance where the particle is absorbing most strongly and what we need to know first of all is how efficiently this nanoparticle is able to generate hot carriers and then how efficiently it can take those hot carriers and inject them into the quantum well so the hot carrier population if you illuminate this nanostructure tends to peak right on the Plasmon Resonance so we get a large population of electrons and holes that can be ejected over the shock-key barrier and then also if we calculate the hot carrier injection efficiency that's something that kind of follows the electronic local density of states but depending upon the size of the nanoparticle it can also be quite high on the order of 80% so the conversion efficiency is basically the product of how efficiently we can generate hot carriers and how efficiently we can inject carriers and for example in this particular 5 nanometer nanocube you'll notice that we get quantum efficiencies of upconversion that are over 20% so these small metallic nanoparticles provide generally for more efficient upconversion than existing material schemes so we can get efficiencies of 25% rather than just less than 5% in generally more like 1% and like I said here I did all the calculations or my post-doc guru did the calculations at visible frequencies but we can also tune these materials to work in the near infrared so can we demonstrate this technique the answer is yes this was worked by my post-doc guru and an undergraduate student Alex Welch as kind of a first proof of concept they took small gold nanoparticles and placed them on gallium nitride indium gallium nitride quantum wells when they illuminated the structure in the ultraviolet and detected the photo luminescence they got a spectra that looked like this some of the oscillations come from Fabry-Perot resonances in the nanostructure but you'll notice that it peaks at about 440 nanometers and then we can shift our illumination wavelength to be kind of below the bandgap of this quantum well structure so here at 40 nanometers and when the pillars were 100 nanometers across we saw clear upconversion that kind of paralleled the photo luminescence we could also tune the upconversion spectral intensity based on the size of nanostructures so 75 and 50 nanometers and then if we had no gold basically there's no upconversion so I think this is a pretty cool scheme for taking a non-upconverting material and making it upconvert simply by adding a thin metallic nanostructure on top pardon me? yep this is the last slide so we can look at the power dependence and also the upconversion spectral dependence and kind of the main takeaway message from this slide is that the upconversion emitted power is linear which is unlike most other upconverting systems where you have kind of this x squared power dependence so this linear transfer curve can only be explained by hot carrier injection and basically allows us to get more efficient upconversion with very low illumination intensities or illumination powers so all of these preliminary results were obtained with a laser our ongoing work is now demonstrating hot carrier upconversion with incoherent sources and then moving on to show infrared to visible upconversion okay so I promised that was my last slide kind of summaries first of all hopefully I've shown you that plasmonic methods from you know various like field enhancements and meta surfaces to hot carriers can help enable next generation clean energy technologies my particular work has focused on how plasmonic methods can allow us to better visualize some of these energy storage devices and what we found in looking at a relatively simple and straightforward system that has many parallels to lithium ion batteries is that single crystalline nanoparticles tend to exhibit this pixel switching where they're either in one phase or another but they don't coexist in equilibrium whereas these poly crystals exhibit kind of mosaic loading where like parts of the nanoparticle don't transform to the new phase and also I've shown you that the single crystalline nanoparticles can self feel and then in the last couple of minutes I showed you a new scheme for renewable energy conversion that can use more of the solar spectrum it's based on hot carriers being injected from a metal to a semiconductor then enables tunable, linear and efficient subband gap frequency conversion so with that I'd like to thank our funders and also all of my group members who made this work possible many of them were pointed out here Tarun graduated earlier this year he's now a postdoc at the University of Maryland he helped pioneer a lot of the in situ electron microscopy work along with Andrea Baldi he's now a professor at the FOM Institute Differ and then I'll also point out Guru Nek who helped enable much of the hot carrier work so I'm happy to take any questions thanks so much for your attention wow that was a lot any quick questions from the audience? we just have a few minutes here seeing none I'll ask one somebody like me can take the results almost immediately from your up conversion work and put it into like an economic model see that is a big deal all those percentages are worth gold I can't quite do that for the battery, the storage one I can imagine that the percentages are big enough that it's going to help a lot have you done that or do you have anybody you work with who's done that working on a startup? yeah yeah so we haven't done any of the economic modeling but we're definitely interested in collaborating with people who can kind of figure out what would be the cost associated in taking say conventional electrode material and then developing synthesis to nanostructure say into a single crystalline nano-cube or a single crystalline, you know, say rod where you have kind of this long aspect ratio to enable like a larger surface area to volume ratio to kind of figure out that cost and then say yep like is it worth it to pursue these synthesis and put these materials into electrodes? yeah so at least you could scale up to in between my level just really accurately kind of marketable device so question on that upconversion, that's really hard to see so what is kind of the pathway to making devices and to do that and then how do you see those competing with some of the other schemes like tandem props guides or props guides with silica and then yeah yeah that's a great question so I guess I'll start with the second question first which is how do we see this upconversion scheme competing with other schemes like the tandem props guide so I mentioned that upconversion is especially promising for say silicon solar cells where it's a little bit more challenging to make a multi-junction cell there are people here on campus who have done some really nice work with kind of tandem props guide silicon architectures I would probably refer you to them to kind of discuss the most up to date research there but my impression is that the props guides still have some challenges with stability and they're working on new encapsulation techniques but I would say that this scheme first of all doesn't have that problem with stability and secondly I think even if those cells work really well that's actually great because we can easily put an upconverter behind the cell and still get an added boost over some of the record cells that have been made here so it's quite complimentary to that scheme in terms of marketing this right now we had just kind of a proof of concept of say gold nanoparticles on a gallium nitride, indium gallium nitride quantum well structure I don't think or I know for sure that's not going to be the sort of structure you want to grow behind your solar cell the economics of that just aren't going to scale but I have one student who's working on colloidally synthesizing these nanostructures so making various air stable quantum dots that can be decorated with metallic nanoparticles so you could just paint these upconverting materials behind the cell great last word over here yes how far away is commercial fabrication now what are they doing versus what you're doing in the upconversion scheme or in oh okay so commercially there are of course companies that make upconverting materials either by ion acceleration or by colloidal synthesis there are no companies that I know of that are adding upconverters behind the cell and that's simply because the efficiencies not there yet like I mentioned most of the conventional upconverting schemes have efficiencies that are too low to yield a significant boost in cell efficiencies like I think the maximum boost people have seen is on the order of 0.01% but that was under concentrated sunlight so hopefully this scheme that promises higher efficiencies could be something that makes it a scalable technology great we're kind of out of time so Jen thanks for sharing just a little bit of your magic with us today