 And do you have a way to time yourself so that you know when it is about 50 minutes or so, or do you want me to stop you somehow? Just remind me how many minutes I should talk? About 50. Okay, just remind me at say 45 or be fine. Okay, so maybe I would wave something like this. Okay. But it is five minutes. If I can see your screen. That's I can see your screen only pop up when you speak. I think the. Oh, okay. Yeah. Simone, you have to you have to say something to write that I will say five minutes. Yeah, we are online. Okay. So I think we can start. It's about three o'clock. Yeah. So maybe two minutes past. Okay. Okay. So, so my name is Simone Piccinin. I'm one of the organizers and I will be hosting this session. And the first speaker of today is a professor, June Chang from Shaman University. And Professor Chang got his PhD in Belfast at Queens University and then was a postdoc at the University of Cambridge for a few years before becoming a lecturer in Aberdeen, the University of Aberdeen, and then moved back to China to Shaman University, where he's now a professor. And his expertise is in the, both in the development and in the application of ab initio methods to the fields of electrochemistry, photo catalysis and solid liquid interfaces in general. So it's a pleasure to have you here and the stage is yours. Thank you. Thanks, Simone, for the kind of introduction and also really a pleasure to speak in this nice workshop. I've been enjoying the talk, the presentation since the past few days was really nice. It's also really an honor to give a lecture at ICPP. So I have background really in computational surface science and surface catalysis. Then I'm more of the interest moved to Cambridge when I started to look at the electron-tranded approach and maybe in solution. So in the last past 10 years also I start to look at the interface to basically combine what I learned during my PhD was the surface science catalysis and also the electrochemistry side, then we have to bring in the solution. So today I will present what I have been really focused on in the last five years also, starting to look at the structure dynamics at the interface, hopefully I'll show you. Those are really important in catalysis and also in electrochemistry as many speakers already emphasized in the previous few days. So I don't have to spend much of time in these slides, I guess, because you already heard a lot about this. So electrochemistry becomes very important because now the energy applications, we have fuel cells, batteries, so the Nobel Prize goes to the really three pioneers in the field of leading batteries and of course we also in the future maybe also just utilizing solar energy to convert to chemical fuels by using solar cells. So those are all very important applications in energy. So for this series we try to find what's the most important question in those, for example, the in electrochemistry. To me, electrochemical interface or electrified interface is really the key to all those energy devices I showed you. So electrochemical interface is where the electroconductor meets the ionic conductor. So certain Faraday electron or charge transfer event has to happen at the interface so that the carbon can flow across the interface. So if you look at the interface, we apply field. So that's a very important how to field polarizing interface. As I said, there are two sides of the interface. So normally we think of the liquid or electrolyte is always very dynamical. But the more and more we start, we realize that the electro material is also become very dynamical on a reaction condition. Of course, as I mentioned, charge transfer across the interface and particularly relate to the catalysis when people want to make useful chemicals. We also need to form chemical bond or break chemical bond on the surface. So that's the three points because our group really focused on in the last few years. So we try to model electric double layers. And later we started also look at the structure dynamics of the electrode. And then we also interested in the proton and electron transfer crossing the phase. So today I will spend some time to show you what we do at the first two topics, the electric double layer, particularly abinitial modeling of the electro double layer and also the structure dynamics and how that impact on catalysis. So I just quickly showed you why we want to choose abinitial molecular dynamics. To me it's a really good tool for atomistic modeling of the electro interfacial chemistry. The reason is the following. First of all, if we're interested in catalysis and also the interface, the electrochemistry key is the electron transfer and we also apply a voltage to drive electron transfer. So we have to dealing with the electron turning the structure of the electrode or the interface. So the chemisorption is also electronic structure effect that's closely related to the catalysis. And on the electrolyte side, of course, we know the solvent reorganize during the charge transfer and also dielectric screening effect, the interface. And as I said, the surface, the electro surface become very dynamic. So that's another reason I think we also have to look carefully at the dynamics of the interface on both sides. So as I just mentioned, I had background really in computational surface science or surface catalysis, the traditional methodology really using DFT and do statistical generalization so that we can, for instance, locate the energy minimum and even search for the subtle point so that we can calculate the energy barriers. That's a really standard approach already 20 years ago. And if we want to, for example, correct the, for the entropy, we apply certain models like the harmonic approximation so that we can also take into account the zero point energy and the vibration, entropy and vibrational entropy and so on. But that only works for really static or low temperature case. But in electrochemistry, we like things moving. And also the electrolyte solution, of course, the ions moving is liquid, we can't do water below zero temperature. So that's, we have to take into account the dynamic of the electrolyte. And in this case, the standard methodology of modeling the liquid, of course, is for example, the sampling method like molecular dynamics or monoclonal so that we can sample in all kinds of local minimum adding basically the configuration of space of the dynamical electrolyte. And in this case, we really have to calculate free energies that will take into account all kinds of configuration or entropy associated with the process you are looking at. There are in free energy calculation community, there are many different kinds of methods. And the one we're working on together with Nihil Sebrig at the Cambridge United Postal in his group, we develop methods, for example, particularly this particle insertion procedure so that we can calculate, for example, if the particle electron or proton or any ions, then we can calculate the results potential pKa and salvation free energy in aqueous solution. So I guess this, I don't want to explain this methodology here, but just to mention that in free energy calculation, we can't just calculate or sample or only calculate initial and the final state take the energy difference. We have to construct the reversible thermodynamic path, then we sort of calculate the force along the reaction path, then we do a integration, the force along the reaction path. So we have to sample the, along these, for example, any reaction quantity like or some, in this case, we have a McBean Hamiltonian, we have a coupling coupled with the carbon parameter, then we do really several MDs sampling corresponding to each point of this parameter that's where it's like reaction coordinate, then in this case, as a vertical energy gap, we calculate is like the force acting on this reaction path, then we do the numerical integration, okay, we get the reversible work for converting from the initial state to the final state, which is the free energy, okay, those are, I would say the theories will establish the free energy calculation community, we just implement somehow at the AMD level and combine with, for example, the particle insertion, different scheme for the proton and the electron transfer, lower work load on the proton insertion process, for instance. Okay, so that's the methodology part, we calculate the free energy, but back to electrochemistry, so I very often like to draw this using the photorelectrochemical cell, for example, so that's our photoanode, for instance, okay, that's our electrolyte, the pioneer in the field for the professor Norsey, already a long time ago, draw this type of level diagram, so that you can, so that you can really assess some of the dynamics or some chemistry of the electron transfer across in the phase, that's really the same also, Professor Gross present the other day in leasing ion battery when Professor John Good enough look at the energy levels in batteries, for instance, those are really the same, so in this case, for example, we want the reason to do this alignment for the band alignment for semiconductor, we have balance band, conduction band edge, so those levels need to straddle the oxygen evolution and the energy evolution level in the solution so that the hole have this sufficient thermodynamic driving force to oxidize the water and the electron can reduce the water, so this type of alignment really is important and have to mention that we only can get those energy level, because that also has the chemical potential of the electron, so we have to, that have to involve electronic structure calculation, so not just the electrostatic chemical potential as well, anyway, so it's very important to calculate or at least to calculate the potential or electrical potential for an interface so that we can compare the experiment, because any electrochemistry experiment, you always see this is certain measurement down at a certain potential, that's when we can, so that's really very important for the structure of the interface and if we want to compare to experiment, we better have the same potential so that we can compare the really the same thing with the experiment, so you see, just to say it's important to calculate the redox potential, so that's the but the key really to do alignment, the level alignment across the interface is really the reference, so we have to take the same reference for the electrode and the electrolyte solution, so that we can compare the levels and in electrochemistry people use reference electrode, most common one, the so-called standard hydrogen electrodes, there are many other silver, silver chloride electrode and caramel electrode and so on, so we really learned from them that we developed this so-called computational standard hydrogen electrode, in electrochemistry that's it's just a redox half reaction, say for converting a gas phase hydrogen molecule into a solvator proton and this electron in the vacuum, so that's basically this redox half reaction, this energy level associated or redox level associated with this half reaction is the standard hydrogen electrode, so then we calculate the levels both at the electrode and in the electrolyte and converting those levels to the standard hydrogen electrode scale, so then we really bridge our, say, automatic model to electrochemistry experiments, but just to mention to calculate this standard hydrogen electrode, all we need to do is calculate solvation free energy or protons, that's basically this species in our cell, so that we can convert all the energy levels with respect to the standard hydrogen electrode and the solvation free energy proton can be calculated by the methodology I just described using this particle or proton insertion scheme that Laura developed with me here quite a while ago. Okay, just some numbers, okay, we that's all the, that's the redox potential and pKa's calculated using adenisomeric dynamics in just that's also all in pure water, just to show you the pKa normally quite good in solution already at the GGA level, using hyperfunctional you got similar accuracy for the redox potential clearly GGA is the error is too large, the redox, the hyperfunctional really reduced half that the error, so that is important. The reason of course associated with the so-called delocalization error in the standard GGA functional and the hyperfunctional can really improve, but okay that's not the point today just to mention that we could use a high level theory it's important to get the band gap right of the water so that we can calculate the calculate the accurate energy levels, so the accuracy really can be improved with higher energy electronic structure theories in this case we tried RPA and the high double hyperfunctional can indeed see the improvement, okay, but in the following since we most of time dealing with metals, so GGA normally give very good work functions, so we don't have to worry about the error or uncertainty from the density functionals, so I will not mention that in the following. Okay, so that's the model, the platinum water interface model we used, so as you can see normally we have say in this case around 150 platinum atoms and roughly 150 water molecules, so we have a full periodic boundary condition so that we have two symmetry interface in our model so that you can think of this slab, this platinum slab also you have this periodic image here then you'll have two symmetry interface, so we could with current computational power we could do say 10 20 or if you really want to extend to 50 that's okay, but we can't really do too long because the computational cost of the IMB is really, really high. Okay, for metal it's also it's really expensive when we're using CP2K, so the wave function of semiconductor can get converted rather quickly, but metal tend to be much slower, so at that time that's my first PhD student at Aberdeen, when we do that we can't really do like full salvation free energy calculation of proton in a metal water interface model because it's just too expensive we have to think of some trick, so we're using another reference basically aligned for example the electrostatic potential in the water, bulk water so that's our interface model that's our for example 32 standard 32 pure water box model we do an extra alignment of the electrostatic potential okay calculated from DFT and this second alignment really helps because we then just using the salvation free energy proton in a pure water box okay using that free energy, salvation free energy combined with this correction or this second alignment in the electrostatic potential pure water so that we we can avoid directly calculating the salvation free energy proton in the interface box interface model okay that can save us a lot of time so we only need to run like 10 pic second the AMD of the interface model we can already do the for example calculate the level alignment in this case we don't charge the surface so that's corresponding to the potential of zero charge that's really a fundamental property of the metal interface and that can be measured by experiment so just quickly show you the number we have so that's the number we calculate for different metal surface the numbers in the bracket actually are the experimental value just to point out that we do check the in this case we take some snapshot for platinum from the PDE trajectory we calculate the hyper functional we don't see much of difference just steel a small difference but really that's within our statistical almost within our statistical uncertainty so this one and there's just sort of big difference for palladium and we check the reason for that is the error in the functional for the work function that's the the work function of metal in in vacuum right so there's a roughly a half EV error so that's really carries this error from the from the work function predicted by PDE but anyway so we also calculate the work function of metals using the same functional at the in vacuum so there's a if we converting this potential of zero charge with respect to standard hatching electrode to absolute scale okay that's this famous 4.44 so we can convert it to the vacuum scale then we take the difference okay that's essentially is this certified in electro chemistries called the voltage potential difference or outer potential difference that's when you the energy difference moving an electron direct from the metal to the vacuum just outside the metal surface in the vacuum of course or you can take another way from the electric removing electron cross the water in the face and put in the vacuum just outside the water surface okay there are two vacuum here but those two vacuum have different potentials okay this difference is so called the voltage potential difference okay you can see that's the experimental value that's our value that's really close so that's a good sign our methodology is good easy also independent of of the functional we use so what caused this difference particularly if we look at platinum plating the difference is like is more than one EV okay but for gold and silver silver is smaller in experiment people think the water will take a certain orientation at the interface even at potential zero charge that's where you don't have net charge on the on the surface because the water already takes certain orientation then we look we look at the structure we have from the MD trajectory we analyze the water orientation we calculate the water dipole just water dipole profile so that's our surface that's really along the surface normal direction you can see up and down and the platinum gold platinum that's palladium and gold silver the smaller but we integrate over we actually got very small potential okay from this water orientational dipole so that that's potential really integrate out if you do but on the other hand we we check the the structure carefully that's actually common it's actually well known in surface science water can chemis of platinum or palladium right that's water chemis of on the surface you will have rather large water chemis option energy on the surface then of course chemis option is is like chemical bond right you will form a bond between metal and water you expect there will charge transfer or charge redistribution because forming this chemis of the water then we look at the so-called electron energy electron density profile okay the electron density profile difference okay before and after the water putting water on the surface then you can clearly see for plat for platinum and palladium okay that's it quite a large charge electron density maybe move from the surface to say from the water towards the surface we see a much smaller charge transferred on inner metal gold and silver okay and that that is not clear from this picture but if you also analyze the electronic density of state this water a that's that's all those are the chemis of the water you can see just really a density of state now goes above the Fermi level that's basically showing the this electron density here move to the Fermi level of the metal but in the bulk of water you clearly see there's a big gap right so that's really a a electronic structure effect and the that can cause a large potential shift even as a potential zero charge on platinum and palladium so previous that was a potential zero charge condition of course we want to model the electric double layer so in the electrochemistry we know that if we apply a potential difference from the potential zero charge okay you will start to build up charge on the surface so positive or negative then you will really attract the counter ions from the electric double layer so we we basically now in our calculation we still have periodic boundary condition in here we basically just insert some certain counter ions this case case sodium just near the interface so this blue balls here that's this sodium atom we insert the whole balls keep neutral and on the other side we we have the same number of sodium then because it's periodic boundary conditions is too in the face in the faces are really symmetric okay it's not anode and a castle that's really the same electrode I just want to hold now anyway in the electronic structure condition a electronic structure calculation if I have a put a sodium atom in the water solution they will have certain salvation structure then the ground state of the wholesale will will be the this will become a cut time sodium plus and the electron will automatically go to the metal surface that's really the ground state of the electron so then the wave function organization procedure in DFT can already be do the proper charge assignment for us okay that's as good we have the negative charged surface and we have sodium ion near the interface that will form a electric double layer okay in our case we can't simulate diffuse layer right now because the times the size and time scale really doesn't allow us to do the calculation to simulate the diffuse diffuse layer so we claim we are simulating the compact double layer okay that's so we because we put the ions just very close to the surface will we our mv just sort of equilibrate the structure near the interface at the time scale we don't see the time scale we have we don't see the the sodium ion move away from the surface so we don't see a net flux okay or a current say between this interface to the bulk water so that's sort of say we don't have this equilibrium between the the counter ion and the ions in the solution so that's sort of mimic the condition the high concentration limit that all the when the diffuse layer is suppressed so all the counter charge is compact at the home host plane okay near the interface so that's the model we have what do we do we we can change the number of atom or sodium ions we put in the solution then we have we will have different model corresponding to the difference different surface charge density then we using our computational standard hydrogen electrode we can calculate the electrode potential for each metal for each interface with respect to the standard hydrogen electrode okay so that we can compare to experiment anyway so that's our model we have then we we can assign the level potential of that model correspond to which experimental condition so that's collaborate with my colleague Yin Xia-men professor Jian Feng Li and Zhong Jun-tian they have this special technique they call the shiners this surface enhanced romance but uh technique so that they can really magnify the signal from really the signal from the water just add the interface okay then they can assess the romance spectrum of the interface water so we we sort of calculate the videos of the water at the interface water then we we compare but it's not the exact comparison comparison because we we don't calculate the polarizability so we don't have the intensities not there yet but the we only use the peak position but the point is we really can see that's the the roman shift corresponds to the OH stretching mode that's read the peak position so that's the experimental result okay as a function of potential you can see the roman shift okay from the goes to from the potential zero charge goes to the next more negative potential so the negative charge surface uh that's our result okay so it's not perfect but if you really look carefully we do reproduce these two transition okay as these two potentials clearly you see a slope change okay the students really they work together uh Jia Jia board that's um my my PhD student and uh Chao Yu who did all the experiments they really talk a lot and then Jia Bo did a lot of analysis on the water structure and the interface and I think they really present the convincing evidence that the structure change of the water corresponding to that two transitions in the roman shape uh I just quickly show you this picture okay the potential zero charge as I mentioned that the water at the interface more or less like flat you don't have net uh orientational preference so uh but if you start to negative charge your surface okay the water will turn around start to have the proton pointing to the surface so as you would expect because the proton is say have fractional positive charge so they were attracted by the negative surface uh that negative charge surface okay when you reduce say the potential get more negative then you have more water pointing to the uh the the hydrogen to the surface okay at some point that this potential that's the first the first transition I showed you in the roman spectrum in the last slide that's where you we see all the interface water pointing this hydrogen atom to the surface is sort of saturated at this point when the potential goes there's a potential window you get more negative the water just can't move is saturated at this point okay but when we close to minus two volt with respect to a point of zero charge we start to see the water make another term so that we have this two hydrogen down water start to appear okay at this the second transition the roman shift okay and in this case that of course if you have very negative potential the field is so strong the water have to align is diagonal with the against the direction of the electrical potential okay and in that case that's a really pure electrostatic I will say when we have one hydrogen down you can still form certain hydrogen bond with neighboring water so but in this case that will minimize the hydrogen bonding but of course in favor of electrostatic interaction with the electric field okay of course what we really interested also the those structures try to understand the dielectric property of the water and the interface so as I mentioned that we we can change the surface charge density we calculate the potential right so then we will really can calculate this charging curve take the derivative that's already give us the the capacitance of the double layer so actually that's the in this case we don't see much of threat is rather constant so uh ongoing around we get around 30 that's also what experimental see for for for many metal the dielectric or the ham hose capacitance or the dielectric differential capacitance of home ham hose double layer or the compact double layer is around 20 30 micro faraday per centimeter square so that's constant not quite so in the sense that is is flat right the constant the dielectric constant is is constant as as function as potential I I want to show you a really different case in platinum where we really do the same calculation I just show you for the gold uh we change the surface charge density we calculate the electrical potential but I I want to show you the the structure of the interface at the negative and a positive charged surface okay uh at the negative charges I also already show you for the for the gold or the hydrogen point that okay that's more less what we expect so at a positive charge surface we have many water with oxygen seat on top of platinum okay as I mentioned in the at a potential zero charge already have a fraction of water chemisorblants on the surface that's actually those are all chemisorbid water on the surface okay um if we look the say that's the surface coverage of the chemisorbid water as a function of potential we have this sort of S shape okay so previously I showed you the potential zero charge they are certain they're roughly 0.2 mononial water chemisorb that's kind of really give us if you remember around one volt interface potential due to the charge redistribution due to chemisorbation okay and when we get to negative okay since the the water need to take a configuration that's hydrogen point down to the surface that will really that's processed with basically force the water to dissolve okay then the chemisorbid water is gone you can also see the peak there's a graduation decrease but anyway that's a that's already clear here at a very negative potential you don't see uh any chemisorbid water when you increase the potential you now you would expect that the water will you will be get more water chemisorbed so uh so around one volt with respect to the potential zero charge so it's sort of saturated around the half mononial okay so that's the that's the chemisorbid water the coverage okay then that's also this charging curve I show you this is now structure is not like linear I showed you for the gold now you get a shape okay it's at the charge density as a function of potential is not linear anymore then you will expect you see you can see certain structure in the differential capacitance of this hemiphostal layer okay that's I already explained at the potential zero charge you have chemisorbid water that can lead to a around one volt potential drop due to charge transfer when you have chemisorbid water the negative potential the water now dissolved you don't have this contribution so this red line I showed here so you don't have this contribution due to this charge electron this charge effect this electronic density effect if you like so that's that's a potential zero charge when you increase the potential get more positive you have more water chemisorbid you actually have more contribution from here and if you look more carefully that this potential if you increase this potential from left to right you you increase the potential of the the the metal surface you actually see this potential increase increases due to the chemisorption okay not instead of like dielectric response you always try to minimize the effect the impact of the applied water that actually do increase not decrease then you will expect the sort of like negative so die like or negative capacitance behavior okay that's what we we so using this ham hose absorption sorry franken absorption item because this water can really chemisorbid water really can change with the potential so we we sort of develop a theoretical model to describe this process but the point is so we we now can view this ham hose that's really compact over there and I think that can be treated as two capacitors in series so one is the common this this seesaw this due to the more the dielectric response due to water orientation that's the standard one normally your in the case of very inert metal like mercury get rough almost flat around 20 or 30 I also show you for the gold case but there's another component due to the water absorption and desorption they just explained so that's another contribution okay so they are actually connected in series so that's this then the overall ham hose capacitance can be written in this form so that's in here this seesaw is shown in this flat this line I show is constant around 20 but for this chemisorbid water country this component due to water chemisorption give you another component okay that's actually negative and it's give you this shape okay this sort of new shape if we combine those two the overall capacitance now of course is always positive it has to be and you will have a bell shape and you have a maximum at slight at the potential slightly more positive a potential recharge okay this blue curve is actually what we can calculate we develop model and we fit in a few parameters that's the this black curve as I show in here can reproduce all the calculations so we really convinced that there's another dielectric response due to the water chemisorption is an electronic structure in fact I would like to mention and also the that's what the experimentally measured of course that's a very complicated impedance spectroscreen measurement they have to decouple the contribution from the pseudo compounds then this due to for example hydrogen absorption and so on anyway so they'll also have similar maximum this bell shape at the potential slightly more positive the reason why this pink is on the is more positive than the potential recharge the reason we would like to present is if using this uh fronking absorption is also a model so the maximum should have should occur at the say at the car the coverage of surface water basically taking half of the maximum so in this case the maximum surface water coverage at very positive potential is around 0.25 0.5 half mononair then the maximum should occur at half of half mononair that's roughly 2.0 0.25 mononair but the potential of zero charge we only have like 0.2 mononair that's that's the reason this maximum is more positive than the potential of the recharge okay we we think that this bell shaped compass differential capacitance is rather universal and we also look at the oxide water interface in this case we're using this final fuel method actually initially proposed by a stinger sporting and van der beel to treat the ferroelectrics in in solid and then we heal really see this connection with the electrochemistry and together with char they develop this nice technique to really model the the electric doubler in a sort of asymmetric model using final fuel method okay i will not explain to them he already explained that in the first day i just using this methodology and look at one type of oxide tin oxide water interface the reason we we look at the tin oxide because i know water really does a really lively proton hopping vendor occur at the tin oxide so water will naturally dissociate on tin oxide you will you will get a mixed structure with half water dissociate on tin oxide when we calculate the using the method the final fuel method calculating the differential capacitance we again see a bell shape differential capacitance but in this case at least the key there's many other factors but the key physics behind this is really i show you in here when we change the surface charge density that alpha is the water dissociation ratio okay so around point of point of zero charge that's when you don't have net charge on the surface it's also pdc but it's different from the potential zero charge metal so that's the ph core is the ph uh ph condition when the oxide surface carries no net proton charge anyway that's called a point of zero charge with a ph condition so we have roughly have okay it seems after you rush okay so we have like half um um the potential zero charge water dissociated when we change the potential change the surface charge density we really can the water dissociation ratio can really change so the water can get more water dissociated and less water dissociated on the surface that will once the water dissociated you will dissociation you will have a different uh dipole really correspond to uh that will contribute to the dielectric response of the overall water in the face that will give you this maximum okay at oxide in the face so there's a really a symmetry we think that's a fundamental for for oil in the face not just for metal also for oxide um since i don't have much time i have to really uh i really want to touch on this topic um i want to spend the perhaps five or eight minutes on this this part so uh as i mentioned that uh on the previous part we talked about the this really the structure change of the water of the water in the face of water now we also i also want to point out that the the the electrode really become very lively on the reaction condition but the point of is we we can't really simulate the whole interface right now so we we sort of shift back to the uh uh heterogeneous catalysis where just look at the the for example some certain metal clusters for now we we we don't include the solvent okay but anyway uh if you can see these slides it's two movies that's a tm movie you can see the metal cluster on the support they're really changing at ambient condition and there's a aimd simulation you can see the cluster really also very dynamical particularly if you have a co observer on this gold cluster but just to point out that this comparison is not on it's not really uh say helpful in the sense that experimentally the time resolution for a tm is around say uh one millisecond okay but aimd we say we have normally have like 10 pic seconds so there's a orders magnitude difference in the timescale so although they all always look very dynamical but they are really different timescale um if we look at the catalyst uh in in in industry every catalyst will eventually become dead okay it will deactivate depending on the microscopic time scale uh for for say one year or two they will they will have to replace that so the catalyst certainly have certain life steps span okay those are all the cluster will become sedentary in risk culture risk segregation just but that very often happens at the the timescale for example the experimental right now the tm can see so at sort of a millisecond or second timescale even hours but for the chemical reaction very often we look at so elementary step of always happen on the timescale of pic second so or any catalytic cycle say go to microsecond that's a clearly a separation of timescale okay then uh that's also justified why in indigenous catalysis people do calculate or static geometricization so that we can only basically see the atoms move on the surface we always sort of keep the metal surface rather almost like fix because they're really the the motion between these two type of events really have a very large separation in timescale but the underlying question is really what if the timescale of the dynamic evolution of catalyst structure overlaps with that of chemical reactions but it could happen just think about it could happen if we have it for example a small cluster that's now become rather popular these days in catalysis community is realized that many many active catalysts are a single cluster or even single atoms anyway but just to point out we we should also look at this possibility then in this case if we want to look at dynamical system we need to calculate free energy as I mentioned so we the methodology we use is just really standard so potential of mean force so we calculate the free energy profile for example in this case oxygen dissociation I go the certain cluster we don't fix anything allow everything move during the MD simulation so we calculate the free energy profile of oxygen dissociation and we can also evaluate the temperature effect so that we can calculate free energy profile at different temperature okay clearly you see the structure very different at the final temperature compared to that at zero Kelvin right but the point is I we really want to quantify the reaction entropy okay in physical chemistry we calculate free energy at different temperature then we take the free the basically the temperature derivative of free energy so that we can calculate free the entropy right so that's our free energy reaction free energy that's that's a reaction free energy barrier okay that's really similar so then I only focus on this red red curve that's a reaction free energy obviously overall reaction then you can clearly see the other feature right that's when we change the temperature uh and if we using just static say generalization and correct the entropy we don't see any any feature in here it's always flat now we see this big jump really at certain temperature then if we take the derivative it's really magnified in fact now we have a huge peak okay as this transition temperature around 300 400 Kelvin okay another point I want to point out this entropy is on the order of 2 000 joule per mole per Kelvin that's Gaikandi entropy change okay we were puzzled and if we then look carefully we realized that can only happen for phase transition so such an entropy large entropy change then we start to investigate the the phase transition behavior of those clusters um uh I so I just want to quickly show you so for the cluster you will have that's sort of that's the uh capacity to heat capacity curve we increase the temperature we calculate average the enthalpy or that's a total energy here we see the jump so that's of course when the the cluster started melt okay that's corresponding to liquid phase since it's a finite system we don't have the first order phase transition otherwise you will have discontinuity here is a so-called quasi first order phase transition so you will have a range of temperature corresponding to the solid liquid coexistence states um okay that's for the reactant when you have an oxygen molecule absorbed on the surface but just to point out at the product okay you actually can see um to say similarly we can see this phase transition behavior but the melting temperature is different from that for of the reactant state okay that's can also see from the heat capacity curve okay the maximum will give sort of the melting point you can see clearly they are different okay just quickly I want to summarize this in this figure um so just think about at a low temperature when this reactant product or solid state like okay they have both states have very low entropy okay then their difference reaction entropy will be small at a high temperature when both clusters melt they both have high entropy but they have small the difference the reaction entropy is still small however at the just a transition temperature range when for instance in this case the reactant steel is more solid state like so there's still low entropy but the product still to melt you can see this rise in the entropy then the consequence is really the reaction entropy will increase okay this point of course after that it will goes down but at that temperature range you would you can see there will be a huge increase in the reaction entropy just uh decrease in the free energy reaction free energy so that's really one state melts while the other doesn't and then you expect there's a anomalous increase in the reaction entropy similarly we see that in the supported model catalyst so that's the same trend I will not go through it again but just to point out that this really what could be a real catalyst okay almost finished so just that's the we see other cluster we see this similar that's the oxygen dissociation on copper 13 we we we see similar situation there's a phase transition but in this case that can we have a liquid to solid the phase transition will give a sort of reversed peak so that in this case there were these favor the dissociation reaction in contrast to the the example I just showed you and it also has size sensitivity I just want to pull now this entropy curve we even have this pause complex pause shape the reason for that is just the melting temperature range so in this case if it's very wide compared to a very narrow coexistence coexistence region they can give really complex entropy curve okay so with that oh that's my last slide just to point out really currently is really dynamical that's could be something similar to what in enzyme catalysis people think of the protein dynamics this key for the enzyme catalysis so I just want to point out we always think chemistry or catalysis rather lack local because the chemical bonding is local but you could couple with the non-local environment in our case the catalyst itself is can provide such a environment is a collection collective motion of all the atoms in the in the in the cluster really kind of fact the this local chemical bonding forming or breaking so I just wanted to point out that so our security conclusion I really want to thank the the group so I really enjoy working with my student they are they're really helpful and the discussions are always enjoy a lot and the and the collaborators particularly here is also in the panelist and many other people I enjoy the collaborations child probably also in the audience and that's the funding agency with this I think I will end thank you thank you thank you very much and we can go straight to some questions from the audience and I will let Arthur Agopian you should be able to to speak now if you unmute yourself so can you hear me ask your question yes we can hear you yeah okay hello professor Chang thank you for this nice presentation this is Arthur Agopian from Montpellier in France so I would like to know have you ever compute the zero charge potential in vacuum so I mean without any explicit solenoid molecule because maybe the difference in zero charge potential that you observe for platinum and gold for instance could not necessarily only be due to the water orientation but could also be an intrinsic properties of metals you mean calculate the potential of zero charge in vacuum right that's that's basically work function yeah that's basically yeah yeah yeah then the difference between the potential zero charge for an interface and the work function for the surface in vacuum that in experiments this difference is called just certify the water potential difference that's a quantity can be measured and we claim that it's due to an electronic structure in fact due to the water chemical particularly for the platinum I mean the difference the difference you observe in between the gold and platinum or the zero charge potential maybe it's not only due to the water orientation which is different from gold and platinum but could also be an intrinsic property of metals yeah okay it's not the water orientation I also said that it's due to the chemisorption of water yeah yeah yeah do I understand your question correctly or anyway I think we can move I mean sorry go ahead okay thank you oh yeah so maybe we can go ahead with the second question coming from Ali Hassan Ali so Ali I think you can talk okay thank you so this is Ali Hassan Ali from ICTP thank you for your talk your results on the change in the water orientation was very interesting so I was curious have you from both the theoretical and experimental side has anyone looked at the evolution of the dielectric constant in the systems from either the ab initio or or experimentally because you you might expect to see some significant changes there right yeah that's I would say in classical for example molecular dynamic simulation I mean people certainly look at those in fact particular for instance I mean I don't really think that many people look at the different combustions or any highly concentrated system there are many other simulations but they point I want to point out in here because because ab initio molecular dynamic simulation is rather expensive previous very rarely people look at the uh capacitance of the double layer in our work we would like to propose that the electronic structure is very important to correctly say capture this differential this bell shaped differential combustions because the water comes off and it is off that's that's kind of only captured by by the by the VAT business yeah so so so basically it would not be possible to converge the dielectric constant from the ab initio uh I think people working on this is still possible right we are also you are yeah yeah but of course that depends on in this case we as I said we don't have continuous layer right not fully okay okay thanks thank you and we have one final question from Professor Axel Gross so Axel yeah yeah June thank you very much for this excellent talk I have a kind of pretty technical question but which has a more fundamental background here you plotted at several slides properties as a function of the surface charge density now in electronic structure calculation charges are in fact hard to determine because you need to know where to integrate over so how did you determine your surface charge density so how did you define them okay thanks for the question yeah it's actually rather straightforward for us we only we basically count in how many kinds we put in right so that's also what say uh experiment electrical chemistry over the years so that's um the charge measure of the electrical chemistry is by color right that's the electron flowing or out the electrical uh through the uh internal circuit okay so we don't have to do these for this for instance the charge separation at the interface I only count in how many kind of ions we have on the surface okay yeah that's you say it's just compensated by the electrons at the surface yeah yes because the whole interface region have to keep neutral right then the electron flowing the system and then the color eyes uh diffuse the fronding electrolite both electrolite that's what the electrical chemistry measure is okay thank you very much okay so I see that there are more questions but we don't have much much time and we need to move on to the second talk so thanks again June and okay I just stopped here okay and there you go perfect so the second talk will be given by professor roel van de krol and he got his PhD at the university of technology delft and then he moved to to the us for postdoc at mit and then came back to delft as an assistant professor and then he became a professor at the technical university in berlin in germany and he's also the head of the institute for solar fuels at the helmholtz center in berlin in germany and his expertise is in the development of materials and devices for photo electrochemical conversion of sunlight to chemical fuels so professor van de krol stage is yours thank you very much thank you very much for the nice introduction um can you see my screen and curves are okay if you hear me okay yes everything yes yes I'm just trying to get rid of this do you also see this screen sharing thing I'm trying to get rid of that but is it not visible oh we don't okay perfect all right great okay so good afternoon everybody thank you very much to the organizers for inviting me to share some of our results and some of our thoughts on solid liquid interfaces in the work we do at the institute for solar fuels at the helmholtz center berlin and today I want to share with you some of the work we've done on ambient pressure hard x-ray photo electron spectroscopy on photo electrodes for water splitting and at the end also show you some results on liquid droplet trains and try to argue that that's also a potentially very interesting system for studying solid liquid interfaces um and now I try to go to the next screen yeah um okay so I don't think I need to explain to you the let's say the the need and the interest also in producing hydrogen green hydrogen for instance with with sunlight and water and when you think about that that's basically one of our main goals to do this and when you think about that you have to think about what the current state of the technology is the current state of the technology is of course shown over here it's just simply coupling a photovoltaic solar cell to an electrolyzer both are commercially available and then you may or may not have to put a dc dc converter in between so so this is how renewable green truly green hydrogen uh can be produced today already at a quite large scale um so so there's a couple of issues there of course there's many components right there's there's at least three different components all these have to be packaged have to be wired and so on and that means that the cost is still relatively high so the cost for solar electricity is is low at the moment but but putting all this together is uh represents a significant capital investment the good thing is that you can optimize each of these components individually right so it allows you to get the maximum amount of the efficiency and if you use commercial electrolyzers you can also produce the hydrogen at pressure and that is for instance important if you want to you know fill tanks you usually have to do this under pressure um one of the or let's say there are also some problems here right so the problem is that alkaline electrolyzers that's the 100 year old dominant electrolyzer technology uh they are not really compatible with the intermittent nature of sunlight so they work fantastic if they are kept running at 100 all the time but if the sun doesn't shine at night these electrodes they basically corrode away that's just iron and nickel and something like six smaller k o h and that's not stable if you don't run this this machine by putting a current through it you can address that by using p m electrolyzers a polymer electrolyte membrane electrolyzer but those those always need noble metals right they they are working acid uh under acidic conditions and there you need things like platinum and they're medium to make it work and of course that's in terms of cost that's not a big issue but but one wonders if if it's possible to scale this up really to a terawatt uh skill uh technology and then finally and that's a little bit underappreciated perhaps is that heat also reduces the efficiency so as you I think you know that that solar panels heat up when you put them in the sun and that reduces their efficiency they don't like this and that's actually bad because you you want this heat in the electrolyzer so it would be great if you could somehow transfer the heat from the solar cell where where it where it's bad towards the electrolyzer where would actually help to increase the reaction kinetics and one way to do that is to try and integrate these functionalities and that that's really that the core of of the work that we're doing in my group so the idea is to have an absorber material maybe a top absorber and a bottom absorber I'll get to that in a minute you simply put it in water you put some catalyst on both sides and then you split water under sunlight and that's of course conceptually much easier and much simpler and potentially also much cheaper than than this kind of complicated system the only challenge I have to be almost there of course if you do this you have to collect sunlight over the large areas right so you basically couple the area where you absorb the light to the electrochemical active area and that might also present some challenges and also increase actually the balance of systems called so that's that's something we have to think about what is certainly good about this approach is that the current density that you have is about a hundred times lower than in commercial electrolyzers so we're typically talking about the current density here corresponds to the solar current density which is in the order of 10 to 20 milliamps per square centimeter instead of the I don't know 0.5 or up to two amps per square centimeter you have for commercial electrolyzers and that might just enable you to use earth abundant catalyst that that you can really scale to a to a terrible scale and the other big advantage of course is that the heat that there are also absorbers here and they also heat up when sunlight is absorbed but here there's a very close contact between the absorber and the catalyst so the heat is very easily used to accelerate the reactions and vice versa also the water that you have here it can also be thought to cool the absorber and therefore avoid the efficiency losses that you have when heating up a normal solar cell so there are some reasonable arguments to try and do this and then if you want to take it a step further you can maybe even directly use the hydrogen that you produce maybe you don't have to collect it separately maybe there's some way with either molecular or biological systems to maybe combine it with CO2 and to produce a hydrocarbon fuel all in the same device by by coupling different catalytic processes but I won't say much about that but that's that's an interesting thing to keep in mind for future applications well if you want to drive electrochemical reactions you have to look at what you want to reduce and what you want to oxidize I think that's nothing new right so and the interesting thing from this is that all the reduction reactions of interest are all at about the same potential range they're all at around zero volts versus RHE slightly below that and the oxidation reaction usually we prefer of course to oxidize water that's the most abundant product and the easiest reaction is just oxygen production so if you subtract these potentials it turns out and if account for over potentials you need a voltage in total of about let's say at least one and a half volts maybe up to two volts and commercial electrolyzers need to work at about two volts now that is very difficult to get with a single light absorber and for that reason we have to and I think there's there's nowadays a more or less universal agreement in the field that if you want to make an efficient solar fuels device to to produce hydrogen with sunlight you need to have at least two absorber materials so and if you then there's you can easily do some calculations and it turns out that something like silicon would be a an almost ideal bottom absorber and then what we're missing at the moment is a stable and efficient top absorber so that you that's the absorber that is directed towards the sunlight where the light is absorbed first and that should have a band gap of about let's say 1.8 electron volts would be ideal and such an absorber 1.8 electron band gap with silicon would then give you a pathway towards solar to hydrogen efficiencies of up to 20 percent if you do everything right and if you can minimize all the losses so that's something that we're trying to do okay so what what is of interest then is that usually in most configurations it's not necessarily always the case but the most configurations this top absorber is done in direct contact with the electrolyte so we are interested in understanding what happens at the semiconductor electrolyte interface and we've seen in the previous talks today the previous speaker but also earlier this week already many examples and many band diagrams so I don't have to spend much time on this I just want to remind you that the actual work the electrochemical work that can be done is given by the splitting of the quasi-firmly levels at the surface of the semiconductor that gives you your driving force for the electrochemical reaction and in this case this is the optimistic picture so here I show that one absorber can basically do the entire water splitting process in practice that's not the case usually this absorber at the surface can only oxidize water to form oxygen and then typically in a tandem cell device we have another absorber that we placed at the back here and that boosts the energy of the electrons a little bit more and that then can do the hydrogen evolution reaction. All right well we've also shown seen yesterday what we learned from the talk from Anja Bebela for instance and also some other speakers mentioned it is that this looks simple but in real life these solid liquid interfaces are pretty complicated right and one of the issues is these surface states that you can have right surface states can be bad they can act as recombination centers for the charges so for instance they can capture a hole and then subsequently capture an electron and then act as a recombination center but they might also act as intermediate electronic states that are involved in the charge transfer process and then they're a good thing to have. Okay so and that was sort of an introduction what we do in my group we have different main topics so one of our biggest topics is to develop new light absorbers so we're working all kinds of oxides that absorb visible light usually they're complex oxides with two or more metals we're trying to make thin films out of these and see how stable they are how large the photo currents and the photo voltages are that they that they give and try to optimize the materials chemistry. So we do thin film deposition like pulse laser deposition atomic layer deposition sputtering all these things and then our main characterization technique is photo electrochemical measurements so illuminate the sample measure the current as a function of applied voltage and so on but we also do ultrafast spectroscopy so James Durant earlier this week gave a very nice overview of that he mostly uses optical spectroscopy what we do mostly we look at the same time scale so from femtoseconds basically to two milliseconds if necessary but we mostly look at the electrons themselves at the carriers themselves so we use techniques like time-resolved microwave cometivity and terahertz spectroscopy we also do a lot of surface and indication phase chemistry of sort liquid interfaces and that will be the main topic of my talk today and then we also have a more recent activity where we develop real tandem devices and also think about scale so do we photo electrochemical engineering because we've learned that it's one thing to make a nice demonstrator device at less than a square centimeter that's what almost everybody is doing in the literature and you can report some nice efficiencies but putting this in a real device with an area of let's say 50 square centimeters that's a whole other business and you need to worry about things like mass transport and shadowing and bubbles and all these kind of things but I won't talk about that today so in the past we have published at least two of these real integrated water splitting devices so these are two of our highest efficiency devices one is a 5.2% solar to hydrogen efficiency device that is based on a top absorber of bismuth vanadate it's a yellow colored material and a bottom absorber of a double junction of amorphous silicon and this double junction gives them an extra bit of voltage and together this device splits water with a with a 5% efficiency a few years later in collaboration with the Korean group Jason Lee we supplied the bismuth vanadate for this and they make this iron oxide so this is actually a triple junction a triple absorber so first part the high energy photons are absorbed by the bismuth vanadate slightly lower energy for photons by the iron oxide and then the really what's left over after that is just absorbed by a normal silicon solar cell and that gave us them close to 8% solar to hydrogen efficiency which was one of the highest values as far as I know still for for an oxide based device how do we get to this point so especially here right this was not so easy to make this bismuth vanadate in such a way that it was that it gave us those kind of photo currents we had to do a couple of tricks to do that and I'll discuss them in a minute so want to also show now at this point the outline of the rest of my talk so we I will start with discussing what we've learned about the chemical nature of some of the surface states that we that we see in bismuth vanadate then we'll look at the changes at the bismuth vanadate electrolytes interface that we studied with ambient pressure XPS and then at the end I'll talk about the droplet rays okay so one of the things we had to do to optimize our initial bismuth vanadate already many years ago was we had to deposit an oxygen evolution catalyst on the surface because the intrinsic activity of the bismuth vanadate seemed to be quite low and it didn't give us the photo currents that we wanted so we use this this co p i this cobalt phosphate catalyst that was reported at that time quite recently by by the nosero group then nosero's group at that time at MIT published in science and this is this nice cubic like cobalt phosphate structure so we we simply electro deposited that on our bismuth vanadate and we see this this is the increase in photo currents so if you turn on the light with the co p catalyst we get about five times higher photo currents than without the co p catalyst so that was a very nice result that that was one of the things that allowed us to to get to the five percent efficiency but but one of the things bothered us and and and that is that we also tried to use a traditional option evolution catalyst in this case a ruthenium oxide and one of the things we measured at a long time ago but we didn't really understand this if we took bare bismuth vanadate and put the ruthenium oxide on it it actually decreased the efficiency and this was not due to uh let's say uh because we made the layer too thick and it absorbed a lot of the light it was simply yeah we just didn't understand it just really gave us a much lower photo curves and a few years later so I described in this paper over here and from 2017 we learned that actually the cobalt phosphate has a different role in the bismuth vanadate than we expected so it turns out we we used a technique called intensity modulated for the current spectroscopy and with this technique we could distinguish between recombination currents in our material and charge transfer currents and it turns out that if we have bare bismuth vanadate so without the cobalt phosphate there's a lot of recombination happening at the surface via some kind of unknown surface state and then if we put the cobalt phosphate on there we saw a clear shift we saw that no longer the recombination pathway was the dominant one it was the charge transfer pathway so what seemed to be the case here is that the cobalt phosphate the main role of the cobalt phosphate on bismuth vanadate is actually to passivate the surface defects that that cause recombination that is not to say that the cobalt phosphate isn't acting as an oxygen evolution catalyst we actually think it is but the main reason why it improved the performance of bismuth vanadate in particular is because it passivated the the defect states well that was a nice result and that helped us understand things a little bit better but you know we we fell into the usual trap that everybody working on photoelectric chemistry does right you see that something doesn't work well and you blame it on surface states that's basically the standard reaction that every photoelectric chemistry does you know stick a semiconductor in water see it doesn't work and blame the surface states and that's a little bit unsatisfying of course we really want to understand what those surface states are no we can see them we also saw that in the talk of ania biblia yesterday we can we can measure them we can even determine their energy level we can determine their surface density and so on but but really understanding the chemical nature of that state is is a tricky thing so we wanted to get more insight in that and in order to do that we used a bismuth vanadate single crystals so at that time we had a collaboration with the Institute for Kristalsichting sort of Institute for crystal growth in berlin and these guys were able to make a really nice undoped and molybdenum dope bismuth vanadate single crystals they're they're actually not single crystals they're holy crystalline crystals but the individual crystal domains are relatively large so we don't have a lot of grain boundaries in between there we then were able to cleave these crystals and make nice lead patterns so we know we have a pretty well-defined surface of bismuth vanadate and if you want to really learn about surface states using a single crystal and having a very well-defined surface is a I would say a pretty good place to start and this was work really done by the team that we have Michael Favaro and David Starr there are surface chemistry team and they they did a lot of this work so we first what we did first is to do the valence band XPS spectra so we took the one model one percent molybdenum dope bismuth vanadate and recorded valence band spectra so here you see the valence band edge and of course we were a little bit naive we have hoped to immediately some some kind of surface state in the band gap and of course we we didn't see anything right so the thing is that that doesn't tell you much right because you you don't know if if if the surface state is really not there or whether you just don't see it because the surface state concentrations can be really low right so they they are known to affect behavior even at densities of 10 to the 12 per square centimeter right which is a only fraction of a monolayer so you you can't possibly hope to see that with a technique like XPS if you don't do anything special so we thought about this and we figured one way to try and enhance the surface state is to do resonant photo emission spectroscopy and I'll try to explain how that works now it doesn't seem to yeah now it goes to the next page okay so in in normal valence band spectroscopy what you do is you have an incident photon on x-ray photon and you excite a species in the valence band directly to to the vacuum and you measure the the energy that you have there and normally that that excitation is not very element specific right so any energy level here that's occupied will be emitted with different capture cross sections but still there's no element specificity in there what you now can do you can also do an excitation in the in the valence band in the LS region or for instance vanadium so you can do by tuning the energy of the incident photon you can make a resonant excitation from a vanadium 2p state to a vanadium 3d state and what happens then is that that electron can recombine with the core hole again and if that happens we get an oj emission from another state but now the interesting thing is that this oj emission is much more likely if it comes if the electron that is excited is resides in the same atom as where you have the original absorption right so the oj emission from a vanadium is something that that would have a much larger intensity over here so this resonant photo emission can be used to to enhance certain features in the valence band that are associated with specific elements by by doing this resonant emission so if you want to then do this you have to understand a little bit the electronic structure of the bismuth vanadate so this is an attempt of that we made based also on a lot of literature data to try and make some sense of it so we start of course with the atomic orbitals of the of the oxygen the vanadium and the bismuth and then we can make a boiling and empty boiling combinations in a molecular orbital diagram and we can try to correlate that on real measured value so these are this is a combination of the valence band photoelectron spectra in the lower energy regime so this is the valence band density of states and then we can also probe the conduction band by doing a nexops experiment so x-ray absorption so that's over here and here you can see the different contributions from the different hybrid states over here and what we can now do is we can do a photo emission into one of the from one of the core levels of for instance vanadium in this case into one of these states here and then see how that affects the ejection of a lower lying valence band electron an oj emission into into the vacuum next slide i have trouble going to the next slide here it goes all right and this is then an example of the of the spectra that you can see so this is then the the energy of the of the valence band or sorry of the of the l three edges and this is the binding energy so on this scale this is the energy of the electron or of the photon that we that we put onto the system right so this is the the core excitation in the vanadium atoms the l-h atoms and this is then the binding energy and to make a long story short if we choose a an incident photon energy of 514 electron volts so that is not that is a non-resonant excitation over here we basically just see the valence band as we see now states over here but if we then tweak the energy to go to uh if i remember 517.4 e v so that's somewhere over here then we get all of a sudden this state over here and this state is now a resonant state an electron occupied a filter energy state in the in the band gap of the material above the the valence band maximum so this is the valence band maximum this is the Fermi energy and this is then the the energy of that state and to make a long story short we're pretty confident where that state comes from it's the resonant absorption so to speak at 517.4 is not really imprinted to the bismuth vanadate but we did these experiments on molybdenum dope samples and the molybdenum is a six valence dopant and it substitutes for the vanadium this is something that we could clearly see so we have ml4 tetrahedra in the system and that molybdenum brings an extra electron with it and that electron is localized or will localize on a vanadium on a neighboring vanadium 5 plus species and with that it forms a vanadium 4 plus species and and that electron is pretty well localized here so it forms a small polar roll and and that's in turn induces a distortion of this vanadium l4 tetrahedrum and that distortion is the resonant state that we excite here in this resonant photo emission spectrum and the response that we see we can just correlate that in the uv of a vo4 species where the vanadium is reduced to a 4 plus state by the electron that comes from the molybdenum dopant and the energy level of that state is about one and a half even below the the Fermi level so it's about 0.8 EV above the above the valence band maximum well that's described in this paper what's not shown in this paper is that we also exposed the material to water to a i think one probably less than one molar layer of water so we expose it to 0.05 torus of H2O and then we see a shift of this bake so we see a broader peak and we could resolve that in two peaks and that is actually this state over here so it's a little bit closer to the valence band maximum and this we assign to a vanadium tetrahedron at the surface that is now terminated with OH groups instead of just pure oxygen and that's how we get to that is this is something we suspected we've worked together with the Julia the group of Julia Agali in Chicago and she did they and her team don't have time to show that today but they did some calculations that indeed confirm that this is a very likely mechanism that that takes place so that's great we now have two candidates for surface states in Bismuth vanadate they're related to vanadium so we have a good understanding of the nature of those surface states but to be honest we we don't really understand yet how this now affects the pc performance because doing pc measurements on these kind of rather thick single crystals that are also not super nicely conducting that that is not an easy thing right but at least we now have an idea of what kind of chemical surface state could form at the surface and and how it could change even in terms of energy when you when you expose it to to water that's also again something interesting right it also illustrates really that the pressure gap you always have in catalysis right so a single crystal EUHV gives you different surface states than a single crystal that you put in water right so the electronic structure of the sample really changes and of course now not a surprise but it's it's good to remind ourselves so once in a while that this is indeed changing the picture okay so the single crystal stuff is nice but we want to get back to a more realistic sample and realistic condition so that will be the next part of the talk so we wanted to do XPS measurements on real electrode materials and this is something that you can do with the so-called dip and pool technique that was developed in Berkeley by Miguel Sommarón and others and what you do here is you you have a sample holder here with a counter electrode and a working electrode reference electrodes you dip this in a beaker of water you pull it up and they have a meniscus over here and if you position your XPS analyzer now in the right place you might be able to find a region of water where the water layer is really thin and where you're able to get photo emitted electrons where you're able to detect photo emitted electrons from the sample through the water layer into the analyzer so in order to do that if you if you put a beaker of water in a chamber then at room temperature that gives you vapor pressure of about 20 millibars of water vapor and of course your XPS analyzer doesn't like that so you have to differentially pump this and these are now commercial systems that you can buy right so we have one of these systems Spex and Sienta both have ambient pressure electron analyzers so the electrons need to be able to travel through within a layer of water that's a few tens of nanometers and in order to be able to do that you need to have high energy photons right so with low energy photons soft actuate photons no chance they really need to have let's say energies in the range of above two kilo electron volts and for that of course you need to have a single front to do that yeah so at that time there's already a few years ago Marco and David went to the ALS at beamline 931 the team of Ethan Cromland they have developed something like this and this is then a picture that you see here so you have the analyzer nozzle over here you have the working electrode you see the the water interface here and you see that the analyzer nozzle is a few millimeters or maybe even a centimeter above above the water so the nice thing is that you can use this technique also for non-single crystals so these are just spray-deposited samples these are this spray deposition is basically just airbrushing abysmal vanadate solution onto a heated substrate plate so they're very I would say compared to a single crystal these are very messy electrodes but they work they give very high photocurrence up and beyond three milliamps per square centimeter and we just dip them in here and then measure that the electrolyte that we have is 0.1 molar of potassium phosphate and maybe just as an interesting aside you can calculate the bias screening length that's about one nanometer in a 0.1 molar solution so that basically tells you that if you have if you go well beyond one nanometer in the electrolyte layer thickness the electrolyte really behaves as a bulk electrolyte so there is no funny constrained electrolytes effects over here all right then immediately to some of the results that we got so this here shows the oxygen 1s spectrum this is in the dark when we start we recognize the water the oxygen signal from water in the gas phase water in the liquid phase we have here the the oxygen in the in the abysmal vanadate bulk phase and we have a species here that basically can be due to an OH species or to the hydrogen the oxygen in the hydrogen phosphate that we have in the electrolyte now if this is in the dark if we then illuminate the sample with a solar simulator via an optical fiber we see an increase in this signal this goes up quite a bit and if we make it dark again it goes back again so we clearly see a reversible change here and we also looked at oh no before I go there yeah one a few other things that you can do if you do a very careful analysis of the of the ratio of for instance this liquid water peak to the abysmal vanadate peak you can actually and if you do some calibration you can actually determine the electrolyte layer thickness which in this case was in the order of 20 nanometers and you can actually determine that's that quite accurately so we know we have a 20 nanometer layer of electrolytes in between the abysmal vanadate and the nozzle the gas phase okay so back to this peak here that we see increase could be either OH or the PO4 peak and then we looked also at the abysmal 4F peaks and we also see a shoulder over here right so there's no shift in the peak but there's a shoulder developing under illumination and if we make it dark again it goes away and to make a long story short that is basically consistent with the formation of bismuth phosphate so what we have here is the formation of a very thin bismuth phosphate layer on top of the of the bismuth vanadate okay so how how can we now try to explain this and this is this is not easy this is somewhat speculative I have to warn you but it describes the observations that we see in the XPS spectrum so what we think is happening so we start out with a bismuth vanadate layer we have the electrolyte phase over here this is all in the dark and if we then illuminate the sample what we think happens is that protons leave the surface and then are buffered by the by the potassium phosphate buffer and that explains the increase in the H2PO4 versus HPO4 ratio and that's something that we measured with IR measurements at another beam line over here so if you see here the blue curve is in the dark and if we then illuminate it we see a decrease in the HPO4 and we see an increase in the H2PO4 and that's consistent with protons leaving the surface and being buffered by this then the next step what we think happens is that phosphate groups on the electrolyte they specifically absorb at the surface and that would also explain the negative charging so we see change in the open circle potential of the sample which we which we measure of about minus 30 milli electron volts so we know that the surface becomes slightly negatively charged and then we think what happens is that as a result of that other phosphate groups are repelled from the bismuth vanadate electrolyte interface to a region closer to the electrolyte vapor interface which is 20 nanometers upstream so to speak and that would explain the increase in the H2PO4 signal that we saw on the previous slide so it's a little bit tentative but it gives some feeling for what's what could happen and what's interesting to note yeah and so this is then the layer you've just seen over here if you illuminate it this is the bismuth phosphate layer that you form at the surface of the of the material and again I emphasize that these changes are reversible you can also look at the time evolution of how the peak ratios of the different peak changes and we see that for instance if we look at the HPO4 over H2O ratio that's this one over here that changes over a time span of about five minutes and it relaxes back when we make it dark again over a slightly longer time scale and and this is such a long time scale that this cannot be any direct electronic process it has to do with some kind of chemical transformation of the surface that we that we see so the question is a little bit what does this bismuth phosphate layer really do and to do that well we were again looking at maybe we thought there might be a surface state over there right so what we did is this is a work by Marco Favaro so what he did was he he looked at a certain potential range and in a very narrow region in a potential range made very fast back and forth current voltage sweeps and what we see here is basically the magnitude of this sweep depends on the on the scan rate and that gives you a measure gives you a way to measure the capacitance of the sample at that particular potential and if you don't plot the capacitance over here it's shown over here what you see here is that in the dark those are the the filter triangles we see a clear evidence of a peak in the capacitance of the presence of a surface state if we illuminate the sample it goes away and if we make it dark again the surface state comes back so there's a capacitance peak at about 0.8 volts versus rhe that that seems to indicate the presence of the surface state and then what happens is it disappears under illumination so and under illumination we also know that we get this bismuth phosphate layer so it seems to passivate the surface state right so maybe we thought there's maybe a similar role as the cobalt phosphate that I shown earlier maybe the bismuth phosphate simply passivate surface states and that could also explain some other behavior and phenomena that have been reported in the literature so maybe to summarize what we understood at that point is that we could also estimate the thickness of the bismuth phosphate layer it's very thin right um a mechanism of formation is a bit unclear it's we think actually now it's it's fault the corrosion followed by precipitation of bismuth phosphate that would also explain that it's a self-limiting growth mechanism right I mean the more you grow it the less there can be of the underlying bismuth vanadate can dissolve so you would have a self-limiting growth of this layer by by reprecipitation the material itself the bismuth phosphate is actually transparent optically transparent so the block doesn't block the light so it's not bad to have it there and there are in fact indications in the literature other papers that show that this kind of junction between bismuth phosphate and bismuth vanadate actually appears to you know have some favorable photocatalytic properties all right so that was work that we've done at the ALS a few years ago and we were so excited about this and we're so intrigued by all the new insights that we got is that we also decided to develop this activity in Berlin at our busy singotron and that's part of a longer story so we work together with the Fritz Haber Institutes the Institute of Professor Schlögel who you heard earlier this week on Monday and we have are now building up this Berlin joint lab for electrochemical interfaces so this is planned to look like this so we have three beam lines over here there's the ISIS hutch that is operated by the Fritz Haber people we have our beam line where that we are building up and currently that we will operate and then we have another soft x-ray beam line over here plus the chemistry preparation lab so this is all being planned it's a very long story but we hope that to have this hutch built by the summer shutdown of next year and then of course everyone can use it all right so a little bit more about the end station that we want to operate we've actually built it already it's based on the modular concept this modular concept was also conceived by the Fritz Haber Institute so we follow basically their their recipe for making such a modular system so one module here the red one is the xps analyzer itself with the differential pumping and then we can put different modules in front of it and the one module that we've built is the dip and pool which i'll discuss in a minute and then later i'll show you the droplet train margin so what happens in this chamber with the dip and pool is this is a photograph what happens so this is where the x-rays from the synchrotron come from this yellow is the bismuth validate sample this is the analyzer and this is the beaker of water so we basically dipped the sample in here we position the analyzer closer so the the x-rays coming at a near glancing angle and the photomated electrons are detected at almost normal to the to the surface and this whole ua3 chambers of course in equilibrium with water at room temperature so there's a background wave progression of 20 millibars of water and the team that operates this and a lot of experiments that i've shown over here so david and marco are permanent staff members in my group and they run this effort and here is the postdocs that they work with two of them have already left dip and michael with rocella and maryland are still there okay so the analyzer is an analyzer that's involved to 10 kev we can have images resolution better than 20 micrometers at this energy range and we can go basically from slightly over 200 or sorry 2 kev to about 8 kev with with the beam lines that we're operating the same as i mentioned before that the hutch that we have that we will build for belkem it's not ready yet so we're currently doing these experiments on another beam line this is the kmc one beam line at at besi and here you see the system you recognize the analyzer and then over here is all the stuff that we need to to drop the sample and is the the dip and pull margin that we have over here okay so when that end station was ready we decided to revisit the work that we've done before at at berkeley with the bismuth validate samples and we basically had a closer look at two of these samples one with the old recipe that we had where we annealed the samples after sparing deposition at 450 degrees c and then one that we annealed at higher temperatures so we see here for this sample that at the lower temperatures we see a little bit less regular surface structure i would say a combination of smaller and larger grains and only when you anneal at 500 degrees you get i would say some of the pictures that you also see in the literature with i would say a more regular bismuth validate structure so that's what those samples look like so they look pretty messy but but this one looks slightly more nice than than this one perhaps so we first yeah so and we did we repeated the the dip and pull measurements next slide and and we started out by just looking at this tokiometry of the surface so this is not dip and pull yet this is just looking at the bismuth to vanadium atomic ratio at the surface as a function of probe that so basically as a function of a photon energy that we put in here and one of the things to note here is that for sample a that was annealed at lower temperatures we actually see a dip in the bismuth to vanadium ratio so in other words this is a vanadium rich surface that we have for these i would say imperfectly annealed samples and when we anneal at 500 degrees we see that the stoichiometry is i would say more repaired there are no longer vanadium rich they even seem to be a little bit bismuth rich now the question is how is that how does that affect the local bonding at the surface of these vanadium rich films so we looked at that with x-ray absorption measurements so we made sure that the penetration depth of these measurements was the same as we we do for actual xps measurements to really have a result that we can compare and here you see the comparison of the two of the two samples sample a the blue one annealed at 450 and sample b the green one annealed at 500 degrees and you see some differences in the spectrum and you see that there where there's a positive difference of the annealed sample there seems to be more seems to be more related to a vanadium oxide layer and to make again a long story short we did a linear combination of these signals to try and get a better match and then we figured out that what happens probably is that part of the vanadium oxide segregates out and sample a right so in the imperfectly annealed sample we have some vanadium oxide segregated out at the surface and that gives us the vanadium rich signals and indeed if you do this linear combination you see indeed a much a quite nice fit of the spectrum okay now the interesting stuff right so we did the experiments in the in the electrolyte we tried to repeat the experiments that we've done at at Berkeley a few years ago and then here we were happy actually to be able to reproduce also the change in this in the bismuth 4f signal right so i don't show the option but i just saw the bismuth and we see here the shoulder here we cannot fit this any other way down with the shoulder and that is indeed indicative of the bismuth phosphate that we formed so that's entirely consistent what we saw what we saw before now the interesting thing is that if we look at the sample that was annealed at much higher temperatures we we don't see that right we don't see the shoulder anymore and in fact i can tell you we don't have any bismuth phosphate that is being formed under these conditions so the bismuth phosphate only forms for these i would say poorly annealed samples and it doesn't form for these nicely annealed samples what we do see here is a shift in the peak we we don't see that here so here the bands do not change there's no band flattening in this case when the bismuth phosphate forms but here we do see a shift and we see a flattening of about 0.3 electron volts which corresponds to more or less complete band flattening of the of the material at the surface so have another look at the surface states again we were happy to be able to reproduce these surface states for the basically the the old sample under the old annealing conditions imperfect but we also saw this surface state and the disappearance of the surface state under light and its reappearance again for the sample at which there is no bismuth phosphate right so we know we don't have any bismuth phosphate but we do see passivation of the surface state so the only conclusion that we can draw from that is that and that is different from our earlier conclusion is that it's it's whatever this bismuth phosphate does it is not the thing that that passivates these surface states because the passivation also occurs when there's no bismuth phosphate right so it seems to be an unrelated phenomenon which was a little bit disappointing but that's what it is okay so we find that it only forms on non-perfect surfaces and there are some explanations for that right so certainly the less perfectly annealed samples are partially amorphous also so they might easier more easily dissolve and then reprecipitate we also saw that they're vanadium rich that also means they might dissolve and reprecipitate and we think also because of some of the leaching of the vanadium oxide the vo4 groups that might actually leave reactive bismuth 3 plus surface sites that combined the phosphates from the electrolyte quite easy and that might also cause the bismuth phosphate formation but this is actually a quite complicated story right we've also seen Francesca Stelma's talk earlier this week and she did a lot of work also on understanding the corrosion mechanisms so she found for instance in her very nice 2016 paper that bismuth and vanadium dissolve in the same ratio and that the rate of the solution depends well partly on pH but also on the strength of the electrolytes we saw in a in a later paper that came out last year we saw also that the type of electrolyte that we have this was work we've done in collaboration with Carl Meyerhofer and Christina Archer here in Germany we saw that also the type of electrolyte whether we use a borate a phosphate or a citrate also gives us different dissolution behavior of the different species of the vanadium species and the bismuth species so it's quite a complicated story to really understand the dissolution so the good news is however if you make your bismuth vanadium exactly the right way and you make sure it's crystallized you can get uh uh stabilities of operation of over a thousand hours so that was published by for instance the Domen group in Japan and also Kim Shin Choi at the University of Wisconsin-Madison they both published papers one year after each other where they showed really exceptionally stable bismuth vanadium behavior for samples that were really made in the right way so they don't they don't suffer from all this dissolution behavior at least not not to the extent that we that we and Francesca could show here all right a few words on how does this now affect the photoelectric chemical response right so first of all if you measure the open circle potential so basically you measure the Fermi level of the bismuth vanadate you see for the imperfectly inhaled samples if you turn on the light the Fermi level drops down but it slowly goes back to its equilibrium situation and if you turn the light off relaxes again so but this is basically not so favorable right there's no persistent change any change in the OCP seems to relax back to its initial value and that's basically you cannot directly correlate open circle potential to to real photo voltage that's internally generated that's that's a difficult thing to directly measure uh but they're also not entirely unrelated and this I would say is a bad sign for a sample producing photo voltage a much better sign is that what we get for the better anneal samples that look more regular there we really do see if you turn on the light we see a drop uh in the potential which means an increase in the Fermi level so band flattening that we also saw and then if you turn off the light to slowly relax back but we see a real OCP developing here perhaps more interesting is the photochrome densities right so these are the photochrome densities for sample A and B and what's interesting is that if you look at the potential of 1.23 volts which is sort of I would say uh uh often used as a key performance indicator for this material you see actually that the steady state photocurrence for these samples is exactly the same even though we know that the surface actually behaves very very differently so for sample A we have the Bismuth phosphate for sample B we don't but they give us the same photocurrence and this tells us and that's that's also a lesson that we knew already but it's good to re-emphasize is that photochrome by itself is actually a very poor indicator of photo electron quality right you at the very least you would also have to look at the photo voltage or perhaps as an approximation of that the OCP and so on right so there's no word of warning all right final few minutes on the liquid droplet trains um uh liquid droplet trains are often used for studying uh liquid vapor behavior what is a liquid droplet train well it it looks like this this is a photograph and these are all little droplets that come out of a nozzle and that fall down right and these droplets are made by putting a fluid through a capillary here and the nozzle and then there's a vibrating element here a piezoelectric element and that breaks up these the stream of liquid that you would normally have in two droplets well this this type of experiment is quite common in for instance atmospheric sciences where they look at gas exchange solid liquid solid vapor sorry liquid vapor interfaces and very often this is done with a liquid jet in which they don't break up the jet into droplets but a liquid stream but the problem with liquid jets is that very often they break up but they always break up due to two way and stabilities what you now can do by by vibrating this this element over here you can deliberately break up and see that over here break up the the the liquid jet into a stream of of droplets and that's nice if you do this in an atmosphere that has that is an equilibrium with water vapor you can actually do xps measurements on these droplets you can put the analyzer very close to these droplets and do xps measurements under conditions where the droplet is in equilibrium with the environment right so the equilibrium vapor pressure of water at room temperature is about 20 tours or so and and that means that if these droplets come out they're in an atmosphere of water vapor at 23 degrees c and that means they're in in equilibrium with the environment so it's it's really equilibrium atmospheric pressure xps that we do here can choose the droplet sign I will show that more often sorry later in the talk and what's interesting is now that because every time you measure such a droplet right the droplet goes down with the speed of a few meters per second so what's nice about this is that these droplets don't have time to form impurities at their surfaces because every time an impurity forms because of absorption from the gas phase for instance yeah you measure already the next droplet it's also nice a nice way to avoid beam damage right because even if you have such a high photon energy that you that you would let's say evaporate this droplet you know no worries there's another droplet following a few milliseconds later so it's a really nice way to do xps on liquid vapor interfaces so David and our group are my group leader in this area he explored this already quite a few years ago when he was a postdoc at Berkeley and he built this droplet train machine and it worked here you see actually photographs of this is the nozzle the xps nozzle these are the droplets over here and just to give you a feeling for the scale right these droplets are in the order of a hundred microns right and the distance between the droplet and the opening of the nozzle is something like 0.3 millimeters so it's very very close and he often told me stories that this is these are actually very difficult and frustrating experiments to do because if your droplet train is not so stable and that was actually the main issue that he struggled with if your droplet train is not stable then one of these droplets might hit the nozzle and then you have a liquid water droplet sucked into your expensive xps analyzer and that means you have to stop the experiment open the analyzer clean it put it back and before you know it you know another two days lost so this is not these are not easy experiments what's nice actually and this I want to I forgot to mention that in the previous slide you can position the nozzle of the analyzer at different points below the nozzle opening and that gives you a different timescale right so the lower you are the longer it takes for the droplet to travel from the nozzle for the opening towards the point where where the analyzer nozzle is and that means you can use xps to follow for instance chemistry you can for instance put two liquid layers liquid streams together to try and form a nanoparticle and then you can see by positioning the analyzers at different points here you can look at the xps data of the nanoparticles that are formed at different points in time by just changing the z position over here that's quite interesting so David as I said before David's first attempt more than 10 years ago stability was a big issue so he tried to improve that now and one of the ways to improve that is to use a technique called flow focusing and that basically means that this is the nozzle over here where your liquid comes through and now you have another opening under here and this is a closed volume where you put gas through so there's a gas pressure difference between here where there's a low pressure of gas and here where there's high pressure of gas and you use that gas the gas flow over here to focus the the the liquid stream of of your solution that that goes here and that gives you a big improvement in stability so David and his post hoc PIP data together they built this flow focusing liquid droplet train system so we have here the capillary where the liquid goes through this is the piezo ring that sort of vibrates and breaks deliberately breaks up the droplets and then over here the gases flow over here at a at an over pressure and this over pressure basically make sure that you focus the liquid flow here in a way that that improves the stability and actually as an added advantage is also reduces the risk of of clogging this is then a photograph so this is the outer area where the gas flow goes this is the liquid outlet this is the muscle liquid comes out the droplet is focused here and then underneath here that's really the opening into the rest of the vacuum chamber where the droplets come out this is a photograph what it looks like so this is illuminated by an led it's a it's a fast blinking led and the blinking of the led is synchronized with the the excitation of the piezoelectric nozzle so if you would look at this these droplets seem to seem seem to stand still and this is where they come out it's of course the the image is a little bit tilted but of course this beam goes down perpendicular to gravity so it goes down and this is the analyzer nozzle and here you can measure the xps spectra of this droplet and this is where the x-rays from the singleton come from the x-rays goes over here fall to emit electrons into the analyzer and then you can measure it so this is a little movie and I hope ah we tested it but now I don't get the response so the movie doesn't seem to work anyway it's probably the most boring movie that you will ever see because if I were able to play it you wouldn't see any changes and that's simply because the flashing of the light the background light is synchronized with the droplet so these droplets seem to stand still if you will see the movie and by doing this in the right way you can actually position so to speak this droplet exactly at this point in the nozzle and then you can measure the spectra so a few experiments that we've done a few permission experiments just to see if we can get any signal from that this is a series that we measured as a function of incident photon energy and here you see a 2 kilo electron volt basically you don't see anything right you just see the gas phase water peak over here you don't see the the liquid phase water it's too small but if you go to higher photon energies then you're able to for the photo excited electrons to reach the analyzer nozzle they have enough kinetic energy and then you see a signal over here so that works and it also illustrates the need for sufficiently energetic photons so you have to have a singleton for this second experiment is to play with frequency so by playing with the case of electric frequency you can control the droplet size you can also play with the flow the millimeters per minute and the frequency together determine the droplet size and here you see for instance 220 micrometer diameter droplets spaced a certain distance from each other and then if you play with the frequency you can reduce the the droplet size to 155 micrometers and they are also spaced more closely together and this and that's shown in this picture what's not so interesting is just tells you basically that in this situation you get a better ratio of the liquid water to the to the gas phase water peak of the xps so that ratio goes up if the droplets are closer together and if you think about that that that makes sense so we also have a delay line detector that means that we can temporarily in the time regime decide when the photo excited electrons are analyzed by the analyzer so we can do time correlated experiments and this is shown over here so in the the blue line over here is when we measure only 30 percent of the time so 70 percent of the time we don't detect any photoelectrons coming up at 30 30 percent we do and we have now time synchronized the experiment so that the the signal that we measure here is measured in between two drops right so this is the 30 percent that we measure in between two droplets and we only see the gas phase signal if we now shift that to if we do the phase shift now we make sure that we only measure when there's a droplet in front of the analyzer and there indeed we see that we get the liquid phase signal over there so we can really do this in a nice way and this opens up a really nice set of experiments because for instance what we could do is we could have positional laser upwards let's say higher in the droplet stream and do maybe a photo excitation on a photo catalyst nanoparticle that's dissolved in that droplet and then we can maybe photo excite every other particle and then measure the differences between a non-excited droplet and an excited droplet right and that allows you to measure different spectra which goes a long way in really determining what the different photo excited excitations what they're resulting all right i think i'm almost at the end so i have to hurry a little bit the final experiment that we did is the i'm almost at the end also is to put nine nanometer silicon dioxide nanoparticles in the solution and we try to measure and try to see that so we see the gas phase water we see the liquid phase water we don't see the oxygen from the silicon oxide here that's just too small a signal but we can also go to the silicon one s peak and then we do see the signal and this was measured by averaging a net amount of averaging of about 15 minutes right so and the droplet train is stable for several hours right so we can even get much better signal to noise ratios and since in this in such small nanoparticles large fraction of the atoms is at the surface that means also that we will be able to study solid liquid interfaces with this kind of technique with this liquid droplet train so this is what it looks like this is in the sample so this is just a liquid droplet train module that we then hooked up to our analyzer these are the different parameters so we can look at droplets between let's say 130 and 280 this should be this should be micrometers not nanometers sorry this is micrometers and the time window that we have from the top of the droplet train to the bottom and we can have stable droplet trains up to at least 50 centimeters is from 250 microseconds to 100 milliseconds and the droplet train that we've looked at we can routinely get more than three hours of stability right so we can do a lot of interesting science with that all right and then finally and it's too bad that this video doesn't work but at the end of the experiment basically what you have to do is you have to remove all the water that has accumulated in the bottom of your UHV chamber and you just have to open the the tap and then water flows out and that's of course something that is you know for a surface scientist it's a very odd idea to tap liquid water from your UHV system all right now it works I hope you can see it so there's liquid water coming out of our very nice UHV system all right so some final conclusions so I think I've shown you that the crystallinity and stoichiometry have a huge influence on the dynamics of the surface I don't think that's anything new after this exciting week that we've had in this series of seminars but it's yeah it's it's amazing to see some of the effects that occur we could find this very nice with XPS this two nanometer bismuth phosphate layer I haven't said that in detail but to be able to see a two nanometer layer that only forms under in the liquid and only under illumination I would challenge you to do that with any other experimental technique as far as we know ambient pressure XPS is one of the very few techniques with which you can study such films that are formed I think we also saw I didn't go in that detail about this but we've seen it in previous talk there's a lot of reger solution of ions at the solid sum up on the per liquid interface and you can you also use this these nub Huxbass studies to look at them here I would say we have now a better understanding of some of the of the chemical nature of some of the surface states that we see in this one right but the challenge is really to try and correlate this to PC performance right because there are so many things changing right we have the materials chemistry itself that's already difficult to understand and and then when you put a material like that in water and all the dynamic change that you have there it's a I would say a very rich but also very complicated field of study that I think will keep us busy for for many years and I hope to show that this liquid droplet train is an interesting idea and then next slide finally saturating you with these systems with water is is not only fun to do but it's even useful okay many people to thank especially the the span tax team with the name of our of our ambient pressure station that David star and Markov of Ireland did the bulk of the work a really fantastic crew many others also contributed this and thanks also to collaborators and thank you very much for your attention thank you thanks a lot and we have we're running a bit late so yeah sorry we have two questions from from the audience so we'll let Deepak Kumar be able to talk you should be able to talk now hi am I audible yeah thank you for the nice presentation so I'm curious like on slide 15 you have mentioned about the lenses so can you make a comment on the specialty of the lenses you used slide 15 you say huh so let me see if I can go there but the electromagnetic lenses you mean yeah yeah so I'm god I'm not an expert on that but but the idea is of course if you collect your photo electrons through a small nozzle that that will go of course all the way and and you capture if you wouldn't do anything special you would capture only a small fraction of them right so that's why you use this electromagnetic lens to make sure that you refocus the ion beam and get a high enough fraction of the of the electrons actually into the analyzer so so we basically just bought this part from from specs in this case right so they have now developed these electromagnetic lenses in these in these differential pumping stages they're all integrated and that's just you can basically just buy them and work with this does that answer your question so so you have used two lenses right so can we use one I mean instead of two can we can it solve the problem probably you can but to be honest that that's just depends on the electron optics right apparently I'm not even sure if in our system they might even use three I'm not sure whether they use two or three but but typically you you need different stages to really get from a pressure from 20 millibars to let's say ultra high vacuum right you you need at least three stages for that yeah and in every stage of course you'll lose because there's this little capillary you lose so in every stage I think you need a lens the less lenses you have here the more signal you lose so yeah thank you I have one more curie like in your general of applied physics d work I mean you have shown two types of samples one in green color and another in the blue right so so that that color is something I mean showing some signs or it is a sake for the representation that sample yeah in the green and the blue so yeah it is just so it is just for the sake of representation you have assigned green and blue color or something some property you are studying like for something no no no this is just a random color to make sure that we recognize them right and to correlate them to the different to the different curves that we have here so there there's no physical meaning of the color it's just distinguish so so this like in the blue color sample so I can see that this grain so these are on different sizes right while in the green they are they look uniform so why just so like so why did so like in the blue color sample you can see they are small and they are big I mean grains so can you comment on that I mean yeah I think that's basically we know for bismuth vanadate if you you know even if you make a perfectly flat layer of bismuth vanadate for instance with a technique called a post laser deposition right we've also done that it turns out that if you heat it to a high enough temperature it always forms these worm-like structures that you see on the right so it's it's I would say almost impossible to make a well-crystallized flat bismuth vanadate layer they always seem to curl up in these kind of structures and that just has to do with the mobility of the bismuth also especially that just moves all over the place and then you get the formation of these structures and of course yeah depending on what temperatures and what times and what the original morphology of your film was yeah if you don't wait long enough to get these structures yeah then you get basically a mixture of structures that that haven't really evolved into their equilibrium state at that particular temperature so so is it possible like wait wait wait a second many questions to allow also other people to ask a question okay okay thank you thank you yeah so José Carlos Conesa please you you can you can ask a question yeah well the question was answered partly in the last of the slide but in the in any case Dr. van der Grohl how do you envisage adding powder to the droplets this would require that the liquid layer on the powder is very small yes so that's actually a good point right so I would say the picture that I showed right for the wait a minute this picture that I show here is a naive picture it doesn't really work like that right it this is kind of the idea that this meniscus if you position the the analyzer at the right point that the meniscus will be thin enough to see what actually happens if you put these kind of substrates in water is that also due to capillary forces the the electrolyte just creeps up a little bit right and if you then choose the right point then you you get a part to part of the surface where the liquid layer just happens to be in the order of 20 nanometers and then you can get your your signal does that answer your question no my question is about adding powder to the droplets adding powder to the droplets oh in the liquid droplet rain experiment yes that's what oh sorry oh gosh I am not sure I think they just bought a bulk of this silicon oxide powder and just put it in a you know just disperse it in a solution and just flow it through the needle I think it's just you just have to make some kind of soul gel process or you buy particles that are already dissolved in the in the solution or you make them yourself what I mean is the liquid layer on the on the powder should be very small yeah so here I mean keep in mind that the droplet is really big right compared to the particle size the droplet is really big so you really have a droplet of let's say 0.2 millimeters or 200 microns and dissolved in that droplet are all the nano particles so the problem is then that the liquid layer on the powder should be very small yeah exactly so you will only you know that's a good point so you you will only see those particles that are close to the liquid vapor interface close to the surface of the droplet that's absolutely right okay you will only see those particles yes for sure okay excellent so I think we can stop for a while have a coffee and come back in about 10 minutes or so at 5 30 yeah for the third and last talk so thanks again and I'll see you soon okay I should um just try my talk I think someone's gonna help me do that yeah that's right there should be uh yes madame I'm here thank you waiting for the questions can you share please your yes yes I can share um do you have any audio or video in your presentation no I don't um please launch the slideshow mode the recording has come are you using a double screen I am using a double screen but I I'm just using do you see my screen actually we do but we also see all the the comments the next slide we see in fact the presenter way oh no how do we do this you you should unplug physically the second screen block it like close it no no unplug the cable oh I can't I don't know uh but no my that's my computer I can't unplug the cable but let me let me try again let me try again do you see it now yep yes now it's good okay all right it's just a matter of order of operation can you can you change a couple of slides please okay it works thank you can I just leave it like this then yeah yeah absolutely okay thank you pronounce your last name as chuk it is it is it is that's right thank you welcome okay see you later yeah see you in a bit Tanya I see you studied at Princeton got your bachelor yes yes I was there too for grad school oh really yeah yeah in chemistry in chemistry uh huh whereas your bachelor is in engineering is that right electrical engineering yeah it was somewhat applied physics but have you been back yeah I go back I um yes I'm given some talks there I I um yes and and otherwise do you never been back since I graduated no I know that the chemistry department moved moved somewhere else there is a yeah they they changed from I guess it was Frick I still remember that and yeah and they have a very nice new building it's quite good yeah I heard they got a really interesting kind of architecture and it's next to the physics department right yes yes it's not so yeah things are closed there I'm not sure now but yes it's it's a very nice open architecture but um also they have their offices but there's a lot of light a good amount of lab space and it's just very pleasant cool it's right next to the sports fields where they are playing a football and stuff like this yes remember the Princeton Tigers and these kind of things they had to expand in that direction that's right have you been back for us I have been there I think last year with Roberto yes oh okay yeah I've seen Roberto and Annabella Saloni there yeah so Ralph was a postdoc there and I was I was supposed to talk with Roberto yeah oh I see yep and both Annabella and Roberto they come quite often here to Priesta so I can imagine yeah I yeah I had a nice dinner with her I don't think Roberto is there but we had a dinner there and Princeton together yeah so but you were with the Giacinto no or did I understand this from me I was with who Giacinto Scoles or no no George Scoles no no oh okay then I'm messing up things okay yeah I was at Albionel with Heinz Frey as a postdoc before that my PhD was actually in transition metal oxides also but in super connectivity in uh yeah but I Greg's schools I I know quite well and I've seen him many times does he come to Italy Giacinto no oh I thought you said Greg's schools I thought no oh no no no no I understood wrong no no no no man Giacinto Scoles I think we had some other speaker who was a student or a postdoc with Giacinto a product yes I don't remember um there was somebody who in my in my on Tuesday I think George you're really good yeah sure sure sure yeah okay yeah Elena Magnano who's in Italy and she's at Electra she was here in Colorado for a little while with me just died she was she was here with her husband but we did some work together okay so how about starting yeah 32 past okay and all all the participants are there right yeah okay so I think we can start with the third and last talk of today's session and the presentation will be given by professor Tania Chuk and she graduated in got her PhD from Stanford University and then she moved to Berkeley for postdoctoral position and then became an assistant professor at the University of California Berkeley before becoming an associate professor at the University of Colorado where she is right now and her expertise is in the the use of time-resolved spectroscopies to investigate solid liquid interfaces or a variety of catalytic processes so Tania the stage is yours thank you very much for the invitation to speak here so I'm going to be talking to you about how we're time-resolving a catalytic reaction at an electrode surface and the um the focus will be how we're connecting experiment to theory experiments primarily kinetic and at the heart of theory is thermodynamic so how we're making those connections I can't advance now it was just advancing fine I have to stop and share again I think this is my group at the University of Colorado Boulder and I'm going to just highlight the people who have done the primary work this is a current research associate Ilya Vinogradov this is a previous research associate Eritre Mandal he is now starting his own group in India these are two more senior materials and science and engineering students Hannah Lyle and Suryansh Singh and Michael Polino from physics has also helped I will also be discussing past work by Jihan Chen he is now starting a his own lab in Suzdeck China Dan Ashefemmer was a previous postdoc and a theoretical collaborator Das Kamaraju so what we study as we've heard a number of times already is the water oxidation reaction and that is because the sunlight energy is stored here on earth in chemical form through this reaction using a transition metal oxide catalyst so this is the heart of solar to fuel generation and we of course want to do that artificially and this is an example of an integrated system where you have a photovoltaic that takes that solar light and converts it into charge and then that's charges fed into a catalyst that does the water oxidation reaction so solar cells are pretty well developed the problem is integrating this with the catalyst and really understanding what is this catalytic process that's happening and how can we best integrate it so this bottleneck here and the reason it's so hard to understand is that it's inherently a molecular and dynamic process so what we do in the lab is this water converting to oxygen and we're trying to put a lens on the intermediate steps of this reaction and this will be at a transition metal oxide surface the way we do this is using light spectroscopies where we have a pump that initiates the reaction and then probe that interrogates the intermediates and we do that with a range of light spectroscopies so catalysis we're used to is talking about as a catalytic cycle I have the intermediates of this catalysis highlighted here there are four listed essentially because there are four proton and electron transfers but in spectroscopy we would like to map this to a potential energy surface such that each of these minima represent these metastable intermediates of the reaction and then we can probe their electronic and vibrational states so you'll also notice that this reaction is shown as spontaneous so that the potential energy surface is fully downhill and that's also something that's quite advantageous that's how we run our reaction because when it's fully downhill essentially this spatial reaction coordinate is analogous to a time axis so we can sequentially resolve the intermediate steps that's just to say that catalysis should be causal and so we should a time would be a good probe of it so I'm just going to give a little background just as a framing of the oxygen evolution reaction or the steps we might look for so this is in a single site mechanism where we have the electron and proton transfer steps individual and sequential each has their own delta g this would be a charge transfer step because we're taking electron and proton away from the water and creating an oxidative intermediate this is the first oxidative intermediate and later on those electron and proton transfer steps are also coupled with chemical steps for example o o bond formation so if we are to drive this reaction fully downhill what we need to do is put a potential on the electron and that's shown here in this delta g versus reaction coordinate diagram this is the downhill reaction and you can see that it's flat or leveled here and that just means that one of the steps here the delta g three is the rate limiting one so there are a number of assumptions here one this is a single site mechanism and we have many sites on an oxide surface another big one is the computational hydrogen electrode where these are in equilibrium nonetheless the north scoff school has been able to use this mechanism with scaling relations to to essentially compare materials activity and those scaling relations come from the fact that one material or one metal site can only do this if it does this reaction very well it'll limit how well it does for example this reaction step because you're coming from the same metal site so with those scaling relations they can then make a reductive plot so not one of all four delta g's but one delta g on the x-axis here and this is of the first oxidative intermediate and what they're doing here is that essentially when the delta g one is high this is the rate limiting step and that's called the weak binding side in other words once you've done this electron transfer this oxygen is weakly bound to the metal site on the other hand when this is low that's not the rate limiting step and something else is and for example it can be suggested that delta g3 is because now it's strong binding of oxygen to the metal site and so it's hard to create the oxygen oxygen bond and so you get catalytic activity of these materials on the weak binding or the strong binding side and in between is the sweet spot well this is nice and quite reductive which is good for connecting experiment and theory the problem there are two major problems one is that this x-axis is largely computational so we don't have experimental handles on it of this metastable intermediate the other one is when we do compare to experiment we're comparing kinetics to essentially thermodynamics and that's true whether we're doing the electrochemical current or time result spectroscopy so when we're looking at experiments we're looking at activation barrier heights for example the rhenius law instead of these thermodynamic delta g1s of the minima if you do know something about that transition state you can frame it in terms of transition state theory and an activation free energy if we can separate out these electron and proton transfer steps so that you're just looking at an electron transfer step you could use marcus theory and that would essentially relate these kinetics to delta g1 using a reorganizational energy for example the reorganization of the metal oxygen bonds around the intermediate now one of the most general ways to relate kinetics to thermodynamics is using equilibrium constants their ratio should be the equilibrium constant and what i'm going to propose is that essentially time results spectroscopy is suited to do that because we can isolate these steps all right so with that i'm going to tell you about our experimental setup what we have is a semiconductor it's a strontium titanate single crystal and that's sitting in water and when you do that you get a nice electric field a shocky barrier in the semiconductor and you can use that with light to initiate the water oxidation reaction so that's been done a lot in the past we simply do it with an ultrafast light pulse such that we can get the time resolution and we get this is a band gap light pulse we get electron holes and we get quite good charge separation even though we're doing this with pulse light you measure steady state current and this is an example of it this is the current versus voltage this is with the light on this bending here is indicative of this band bending here of the shocky barrier we do our experiments essentially at zero volts versus scd so everything to show you with current is there so this tells you about the charge separation we can also measure the oxygen and the electrolyte using a Clark electrode and we get approximately faraday oxygen evolution of a four volts two one two one oxygen so with that we combine it with a pump probe setup this is the steady state version we initiate that steady state with this pump and then with a delayed probe beam we interrogate the reaction that pulse sequence looks like this so we have kilohertz light pulses and so there's two milliseconds apart because one is used for the the unexcited state in between we have the probe to interrogate the reaction we use a full electrochemical cell so this is the working electrode this is the other half reaction with a reference electrode we are constantly we don't we do do measurements with fluence to some extent but we try to keep it as much as possible to a single fluence that's one light pulse can only excite as much as two percent of the surface sites and that's essentially to keep these interfacial energetics the same all right so now i want to locate this semiconductor photo driven surface within the oxygen evolution reaction because that's usually thought of as a potential and a metal electrode and the essential thing that i'm going to point out here this is now the Gibbs free energy versus the reaction coordinate is that we first put the sample in the electrolyte and it equilibrates that a cologne can change the surface hydroxylation and then we have an instantaneous light pulse that creates those valence band holes and that's what drives the reaction downhill these essentially this is one of the things that's different and it essentially separates out the electron and proton transfer step into two separate reactions the first is a proton transfer step that occurs in the dark and i'll be showing this to you during the talk so the proton transfer occurs in the dark i'm going to write it that proton as a that's usually as a product here as a reactant hydroxyl anion that's because we're going to be changing the ph of the solution in basic conditions but into the standard state this would be a simple proton transfer reaction where you have a water absorbed surface hydroxylated here plus a proton it's the surface metal pka it is also what defines for a semiconductor the shift in the valence band edge with ph so after we do that we that's the explicit equilibrium in the dark we then shine light and instantaneously create holes in the valence band and they can trap preferentially on these hydroxylated sites to create the first oxidative intermediate i'm going to be calling this tioh star this can also have several different geometries we're going to be looking at one of them those geometries are based on where the oxygen is that's been oxidized and where the proton ends up so that can be somewhere on the surface as well so we use we actually learn a lot from the papers by michael sprig and jen chang they have calculated this free energy difference for titanium dioxide surface and that free energy difference is between a delocalized valence band hole and a trapped hole on the surface they did that with dft and amd calculations and they also very nicely put it within the context of marcus theory since this is a simple electron transfer reaction now i should say i now wrote that electron what's the product here is the reactant hole because that makes sense in terms of a photo driven reaction okay so already i've probably complicated the idea of a computational hydrogen electrode but this is very nice experimentally because we can use this dark equilibrated surface to understand what the photo driven surface looks like so if we're going to do that we better be able to measure it and you get got a good introduction to xps in the last talk we use ambient pressure xps this is also at berkeley at the beam line 11 we did that for our strontium titanate samples and this is for undoped and essentially for lightly doped as well samples these are aimd simulations by das demirage and this is a dft configuration here where you have one water absorbed group on the titania and a neighboring one that's hydroxylated we essentially did experiments on this and looked at the theory as well and this is the number density of the hydroxylated groups this is the number density of the water groups as a function of relative humidity or the water that you put in the chamber and the the calculations and experiments agree pretty well in terms of saying that strontium titanate is partial dissociation in other words every other titania is water absorbed and the other one is hydroxylated so if we want to preserve this surface in our experiments we actually have to do a scanning mode so we have to take our sample and literally scan it while we're doing the time resolved experiment so this is what our sample looks like at the very end in an scm and what we did is we scanned it for faster and faster scan so that's the green to the blue and you can see that the kinetics converge this is an optical probe versus the delay we can also take and go to a new fresh sample spot for a new delay position you can see that traces the converged curve so this is analogous to in a cat in a homogeneous catalyst just flowing the solution and taking the data as the catalyst goes by and what it does for us is essentially says that the kinetics we measure is coming from a point on the surface that we can define in the dark so the reason we have to do this is because as we have heard water oxidation restructures the catalyst and we can look at this x-situ with both t-em and s-em and the way we do this i'm not going to go into the details of this but we take the t-em thickness that's been changed after the water oxidation in the s-em area and the current evolved during the water oxidation which we tried to keep with the scanning method about a standard deviation of 0.5 percent is the current so we're always having the same current of all and then what we can do with that is look at the volume change to assume that that entire volume change just dissolved into the liquid and understanding two holes per oxygen we can put a limit of how much of the current went to restructuring the surface rather than water going to O2 and that is a limit of about five percent and so five percent of it is going there but that means a lot for the surface and it means a lot for your optical reflectivity so that's why we do the scanning mode but that that's what's going on in the bulk we can also look at this in elemental analysis and we see the main thing that's being dissolved is strontium into the solution and we can correlate that with ICPE beta okay so with that I'm going to be telling you a couple of stories the first two are a deep dive into the hole transfer reaction at the strontium titanate aqueous surface and the following two are the fate of these TiOH star intermediates and I will be talking about screening but that'll be at the end okay and it'll be involved with the fate of these intermediates so first about forming this TiOH star from which we're going to use ultrafast spectroscopy and the first thing I want to talk about is the structure of these intermediates and the way to get this structure one of the good ways to get this structure is through vibrational spectroscopy so this is a mid IR probe beam and this is an electrochemical cell and inside this electrochemical cell there's a little diamond that diamond is an attenuated total reflection crystal which takes this propagating mid IR beam and creates an evanescent one so one that's decaying in space and what that allows us to do is be sensitive to molecular modes and not to bulk modes of the crystal especially when we have that crystal sitting at a distance from the ATR crystal such that we allow the wave to decay fully that means in the background when we're doing the ground state we're not sensitive to the optical phonons and the bulk all right so with this what we were able to do is identify one geometry of these intermediates and that's the oxyl with the terminal oxygen and a proton transferred to a nearby site that vibration associated with this intermediate looks like this this is the change in absorption we see we only get it when we have whole excitation to the surface we can quench it with methanol which means it's related to oxygen and we do this detailed analysis experimentally using a phano resonance that describes these changes in line shape here so briefly a phano resonance is a discrete mode that's coupled to the continuum and so we're in the middle of a solid liquid interface so we have a lot of continuum and we can couple actually to the solid side as well as the electrolyte side so on this side you're looking at changing the doping density of the strontium titanate and that changes the electronic excitations within this window I should say this is in the window of titanium oxygen stretches but you can also have these mid IR electronic excitations and those are these plasma excitations they couple to that mode and they give you this asymmetric line shape on the other hand you can also couple to the electrolyte so we can get differences by going from air to electrolyte and also by changing the duration of the electrolyte going to hydrogen deuterium and so what's coupling here in this wave number region is vibrational modes of the water so this tells us that it's an interfacial mode so experimentally so we are sensitive to the interface there I will say that they are very nice phano line shapes in terms of what you can see out there I will give you the the details are here I will also point out that if you look at you phanos paper in section four it's a perturbative theory so it means that all the couplings add so that's why you can get the same phano resonance when you have to continue that's coupling to and I think this is a good example of that all right now I want to that's the structure of one of the intermediates that we can get and I'm now go to the formation time of these ti rate star and I'm going to look at that first through up oh I am sorry I am sorry so I jumped so actually to get this vibration to be associated with this axial there was quite a large amount of theory that was done and that was done in collaboration with Dasque Maraju and David Prendergrass and what they did was a strontium titanate slab where they calculated infrared absorption in the bulk in the in the ground state and with the axial configuration this is the optical phonon that we're not sensitive to in the spectroscopy and what they found in particular in the axial configuration is that there's a separated M1 mode here that's due to vibrations of the oxygen underneath that axial and you can think of this a bit in an inorganic chemistry way because what's happening here is when you trap the hole you lengthen the titanium oxygen bond this contracts the unit cell underneath and essentially separates this motion from the rest so already just with the structure we can start thinking about how do we map to this delta G1 axis of the first oxidative intermediate and since this bond is lengthening it essentially puts it on the weak binding side whereas Heinz Frey and Thomas Hammond who have respectively studied cobalt oxide and iron oxide found upon hole trapping an oxo which would be a shortening of the bond which would horristically place it on the strong binding side. Okay so now I want to go to the formation of the TOA start in a radius and the way we're going to do that is first optically. So you have a valence band you have the holes that you're creating here and when they trap on the surface you're essentially taking a valence band state and you're creating a mid gap state such that that reaction is spontaneous for the hole. Now when you do this you're going to get new optical transitions one will be an excited state absorption when you're promoting an electron another will be an excited state emission when you're from the valence band another would be an excited state emission when you're using an electron from the conduction band. Now these are at different energies than this free energy because they're electronic but more importantly they're vertical optical transitions that have to do with where the mid gap state is in the reaction coordinate as well as the valence band and the conduction band. I will just say that this transition at the very end of the potential energy surface this means you're on the excited state potential energy surface so you have an electron hole pair remaining here you're on the ground state surface. So this has been looked at by a number of theorists and and experimentalists and because these are optical transitions they're fairly broad so you can be looking at the axial but you can also be looking at the bridge intermediate which is hole trapping on the lattice oxygen. Generically they're called O- also hole polar on and the only thing I want to say about this is that essentially both experiment theory see UV-viz transitions associated with these mid gap states and hole trapping. So we looked at that in strontium tightening under water oxidation and so this is an example of our optical maps we're going from picosecond to nanosecond time scales here from 2EV to 3.2EV this is around the band gap of strontium tightening and you can see in the middle of the gap that you have these absorptive transitions as well as emissive transitions. This is in closed circuit when we have current and this is open circuit when we go. So I want to concentrate now on this emissive transition because you can see that there is a clear growth so that clear growth means that it could be related to the time scale of hole trapping. So we do get a very well-defined time scale so 1.3 picoseconds of that growth. We can then identify the nature of it since these are broad optical transitions using our vibrational probe and we find that the vibration of the axial also occurs with that same 1.3 picosecond time scale. So this emission we find nicely counts the TiOH star intermediates whereas the absorption occurs within the ultrafast light poles so within the generation of electron hole pair carriers and then decays which suggests it has a large portion of the population associated with the valence band holes. Now you might glean that from a simple diagram like this this is just a suggestion as to why this is happening but when you're taking a state from the valence band and you're using it to create the mid gap state this transition is going to have be complicated by essentially the valence the the complicated by the population of the valence band states you have left to create it whereas the electron and the conduction band here is not at all related to whole arm group formation so it can be a sort of independent count of these mid gap states. So I have one more way of defining this as the time constant of the whole trapping reaction and that's some recent work we did in the lab so these are optical transitions these are mid IR transitions we also looked at coherent phonons so when you create this TROH star you're distorting the structure at the surface for each one and you create many of them you could create a continuum interfacial strain and if you do that at an interface what's going to happen is you're going to get an acoustic propagating pulse into the bulk of the crystal and then our optical probe will scatter off of that acoustic strain pulse and across the interface and have coherent oscillations in the data and those oscillations will have a frequency that depends on the optical wavelength because this propagating wave is propagating you're going to get a dispersion relation so this is the frequency of each of these oscillations versus the optical wavelength and the model essentially includes the acoustic velocity in strontium tightening these are the oscillations we see in the data we can remove them by that model by knowing the dispersion relation and what you have as data is essentially the phase of these oscillations as well as the amplitude and what Ilya Vinogradov did in the lab is he took a model that was done before but expanded it to kind of the broadband spectroscopic range and that model includes the electromagnetic interaction so how these interfere with each other as well as how that's propagating acoustic strain is related to the interfacial strain so this is the propagating wave and this is the interfacial strain and so what's input into the model is the spatial decay of that interfacial strain which is exponential and a formation time which is an exponential rise and so this is that strain wave this is what's related to the spatial decay this is relates the formation time and this is just a convolution with the time resolution of the experiment and with that you can start understanding the phase and amplitude of those weights and particularly from the phase we can get a good amount of information so this is the phase in the closed circuit experiment and what's fit here is that we get a formation time for the interfacial strain also of 1.3 picoseconds which helps assign it to the TiOH star that are being created now it's also an independent measure because it's a phase we can also do that with different experimental conditions that phase also tells you that the essential the nature of the strain at the surface which is a C elongation versus the A and that C is in the direction of the electrode electrolyte interfaces so you can think of overall an expansion we can also do this experiment in open circuit where we're not creating a lot of these polar arms and what we see there is we can only see these oscillations at very high fluids and that phase also the phase is very different that all gives us that this is really carriers that are coupling to the lattice where this is polar arm formation all right we can now so we can do do something with that or I think I could do did that really well with this model and I refer to you to this paper we can get that formation time to the air that's shown there but the spatial decay is something that is harder to get at and what we could do is only put a lower limit and that lower limit means that there's an upper magnitude to the strain of about a percent so what I think that this does is I'm just going to summarize the formation of these tioh star intermediates is we can see them through the emission using electron volt energies we can see the formation also using hundreds of milliolectron volt energies and we can also go to the gigahertz which is fractions of a milliov and see them there so that means that we can look at them from the electronic transitions we can look at them from the normal modes they create as well as from the environment this is kind of the continuum strain you would expect from a collective all of these are states that you would expect to go into defining a free energy of the product tioh star so we assume we know what we're starting with which is valence band holes and we're defining the states of the product the fact that we see a close 1.3 picoseconds between all of these states tells me that we can define a common temperature for that free energy it might be some elevated temperature and most likely is because we're using an ultraviolet light pulse but common enough to the states so i think that's important i think from an experimental point of view this is also seeing this across electromagnetic spectrum is important because we can use this as cards in the experiment here this is better at calculating or seeing the total tioh star population whereas this will track a particular geometry and this will tell me something about the environment so with that we have a time scale for the whole trapping event now i want to that begs the question can we get a reaction free energy difference from our spectroscopy and the way okay so that's going to the next part of the talk which is essentially this tioh star intermediates are creating this picosecond time scale and they're fairly stable and that suggests that we could get an equilibrium constant from it and the way we're going to get the equilibrium constant though is not going to be following exactly the time scale of the back transport but rather shifting reaction conditions so i'm going to shift the reaction condition of ph in the dark that's easy for me to do and easy for the lab to do in ph just change the concentration OH minus through the ph and see how that affects the products that we create so this is the spectra that we get so again these optical naps delay and energy but now as a function of ph and you can see that the emission becomes a lot more as we go to ph 13 so already suggesting that we're creating more to a OH star when we have higher ph or a higher hydroxylated surface but if we're going to get free energies and we're going to get equilibrium constants this reaction better be common to this data which means i want to be tracking a common enough species on the product side and reaction kinetics that are essentially the same so how do you go about answering that question and one of the ways is to see if you can get a common basis set across the data and that's what ilia and a retro did they used a principal component analysis and both of them started with a singular value decomposition and singular value decomposition for each of these optical maps immediately spits out for you that you have two dominant components and then what ilia did is he took those two dominant components for ph 13 and just reconstructed the ph 13 data from them and then he took the two dominant components from ph 7 and used them to reconstruct the ph 13 data this is a simple rotation of the basis okay so that's a nice mathematical way we can do that as a function of a large range of ph and so we're basically taking the time scale the energy scale the ph as a data set and saying that we have a common basis that's a mathematical description of it but we really want this basis set to also reflect the optical spectroscopy and that's what a retro did and he constrained the basis it's not a simple rotation basis anymore but constrained it to have a pure absorptive component within our broadband window and a pure missive component and so this is the common spectra across that entire data set so it tells us that essentially we have t i r h star species that we can count by this emission that is fairly common across the data okay we can then look the kinetics associated with these species and we see that we have that picosecond component and then we also have another 60 picosecond component but the point here is the kinetics is common and we're just changing amplitudes so changing populations so now i'm going to look at this two picosecond component that we've already assigned and that is the one that has the nice ph dependence so this is the amplitude as a function of ph and this is its growth the 60 picosecond component which we haven't assigned yet doesn't contain that ph dependence now all of this analysis suggests that this is still related to 2 a h star that might be essentially from a different origin i want to take a look at this nice ph dependence here which if you look at it is in fact a sigmoid okay a sigmoid harkens of a reaction isotherm so if i were to think about just a simple absorption isotherm like this and this would be the coverage of the hydroxylated species if i tune the ph i'm going more i'm creating more of the product and this rise in the sigmoid would tell me something about its equilibrium constant the fact that you have a flattening here in the sigmoid means that it's a langmuir isotherm so that you have a limited amount of surface sites that can be hydroxylated now this isn't of a simple absorption isotherm it is of a metastable intermediate so this coverage is of tioh star so that what he immediately told us is that well we can construct an effective equilibrium constant here of the two sequential reactions and in this reaction okay the limiting reaction would be what's conserved is the number of holes i've created by my laser pulse which is either in the valence band hole or in the trapped hole so in this reaction i'm exchanging this for this and this is my tuning variable so as to say in this context that the number of holes we can create is only two percent of the surface site so you have many more of these than of these that you can create so with that we combine these two reactions through the common reactant which is the hydroxylated site and we get again a langmuir isotherm but now an effective k hey there's one approximation here but it's quite negligible so that k is now a combined k of the surface p k a and then of the hole trapping reaction and this is how we define our equilibrium constants so what i think this shows is that if you can time resolve intermediate populations you can in their meta stable you can start getting to thermodynamic parameters of the driven catalytic surface so you might ask me well i said a lot about this reaction here what about this reaction well okay maybe you can tune it with ph but is that really happening so what we can do is we can change our sample to change the nature of this reaction and so we did that by going from lightly dope strontium tightening everything i've been showing you here is lightly dope strontium tightening niobium replaces the titanium here and that's what gives us the shocky barrier and you can also get the shocky barrier get water oxidation with a highly dope sample get all the restructuring as well everything happens here too but here you have a ph dependence and here you don't and you can that bears fruit in the principal component analysis there's ph dependence here and there's no ph dependence here if you look then at the ambient pressure xps which is telling us what the surface looks like under neutral conditions you see that in the partial in the lightly doped strontium tightening you get a partially hydroxylated surface so this mimics a lot the undoped sample that i showed you earlier yeah so these are the hydroxyl groups we're fitting these are the water absorbed groups this is the carbon species this is the lattice oxygen and this is gas phase water when we go to a highly dope sample we have essentially a factor of two to three increase in the hydroxylation and that suggests a closely fully hydroxylated surface near the neutral conditions so we're we're dealing we are getting to the detailed coverages here but i will say that if you look at the low relative humidity range when you have water the amount of dissociation that you have there is about less than 10 maximum 10 which is this is at least 50 percent and then it grows what with relative humidity at ultra high vacuum these samples look quite alike okay so it's only when you add water that does it does it change so um this gives a picture oops this gives a picture of essentially a lightly doped sample where we have hydroxylated surface sites among a water network so there are two reasons i believe we can get this sigmoid is that we can push the hydroxylation in the basic pH range and the other reasons is we're look using the optical emission to average over all the different intermediates the surface so we can get basically something that's common on the product side that we're looking at so doing this i think we were able to isolate the first reaction step of water oxidation using the surface acidity in the dark should say this is a collaboration with Jan Rosmicell he did teach me a number of things about the oxygen pollution reaction made me more confident about the model with that i want to now place these two reaction steps so now we have a surface pka and a whole trapping reaction on the one delta g one that's the usual one that people use for volcano bots okay so these are these two steps here okay and we want to create a delta g one that we can put on an electrochemical scale for example strut weak binding here to strong binding there so this is non-trivial and it requires a couple of steps one is that we start off with the equilibrium constant that we measure and however that equilibrium constant that we measure is one of surface coverages where we have a laser creating many holes in the valence band when you want to compare to calculation this is a single site so you take a single hole single tioh minus site is calculated and then you get the delta g and you can get bright and equilibrium constant for that but we have a degeneracy of holes on the valence band side so you have to divide basically divide the partition function by that hole density so that's what we do to connect to the calculation to a single particle like delta g right and then what we do is we have two parts here we have the pka and then we have the whole trapping delta g we take from calculation the pka which is a pka of eight and then we take a calculation that was done for specifically for strontium titanate by das we take the highest one that was calculated so minus point four eb we put that in and then we derive back our experimental k and that is five in our experiment we get an equilibrium constant of 150 so what that would do would push the sigmoin from ph 11.7 for k of 150 to ph of 13 for k of five okay so it gives you an idea of the difference with that we can then essentially extract out that minus point four eb and add it to the valence band edge to put it on the electrochemical scale so in this context in this delta g one all the ph dependence is in the valence band edge so that's the surface pka reaction we're saying and that on an rhe scale would give you one number on an nhe scale would change with ph so with that we can put that on these volcano plots and this is then gives us essentially 2.4 electron volts puts it on the weak binding side now this is still approximate and i think that it's partly maybe theory but not from an experimental side i think the major issue is this part here dividing by the whole density because not all of the holes we initiate with the laser pulse are necessarily going to be evolved in this one reaction however even given that i am comfortable with putting it on the weak binding side because essentially this equilibrium constants can change by a lot and this is a reasonable place for it to be um for the pka and and for for what the delta g's are um okay and it also just the idea of seeing an equilibrium constant is important for defining a metastable intermediate of a reaction so um with that i am now going to go to the fate of these taoh star intermediates and that's going to be on the microsecond time scale so i'm going to start with just a proposed mechanism and this is just a proposed mechanism right now um to guide our discussion so the idea is is we create these first oer i mean the intermediates all at early time scales and then there's essentially dark and chemical steps that leads you to the rest and so that would then mean we're going to be looking at the decay of these intermediates towards those dark and chemical steps and so it suggests um because we're looking at a decay and because the dark chemical steps the transition state theory would be important a hallmark of this would be if essentially you create all of these initially and then the time dynamics is fairly uneventful until you get to the transition state and then you see a marked decay okay so we looked at the emission again at um microsecond time scales and we're looking at this here with just a single wavelength probe at 400 nanometers and it's okay it's fairly unembedent full in the nanosecond time scale but we are now going to be looking at that more closely um this was done with a 30 nanosecond laser but now we have a better one that we can really track with a picosecond through the microsecond with the full broadband spectroscopy but this is what we've done in the past we do get a marked decay on the microsecond time scale so why you might see a decay in the emission um going towards the next step in the cycle is because these tioh star intermediates contain charge and so they would be in the middle of the gap but if you form some neutral oxygen oxygen bond they would no longer be in the middle of the gap so we now um take this and we fit it with bi exponential decays and we get distinct time constants so marked microsecond times constant we get one of six microseconds something okay one of um eight microseconds one of 60 microseconds and we have the fact that we're fitting this with these um this kind of exponential these bi exponential decays means we have two parallel decay routes so the tioh star population starts out here and then it bifurcates into two populations that bifurcation is ph dependent so at higher ph the faster pathway is favored and at lower ph the slower pathway is favored we also deuterate the solution so we go from hydrogen to deuterium and we get a very nice and marked kinetic isotope effect of 1.75 on the slower route that you can see directly in the kinetics when you compare the ph the ph solution to the pd solution here so blue to red so that suggests that we have some on the slower route some marked oh breaking event that is screened so it's a low k k k i e all right this bifurcation is also there so we change the salt concentration in the solution and keep the ph constant and you can see the changes in the bi exponential decays that we analyze here so these are the amplitudes of those populations as a function of the salt concentration this is the slower pathway that's what's favored at low salt concentrations and you can see that the faster pathway grows in with direct exchange to the slower pathway so we haven't normalized at all these amplitudes okay and it does so up to a certain saturation point okay so this direct exchange is what suggests to me that the screening layer is involved so here if you basically increase the salt concentration you can screen a constant charge at the surface with a single layer of that salt whereas when you have less of that salt concentration in solution you would have a thicker screening layer okay so the reason i haven't mentioned the screening until now is because in the whole trapping reaction we don't see a dependence on the salt concentration and why we might not see that is because we have a constant voltage and we have a constant fluency excitation so a constant amount of charge here that we're putting on the electrode and so the potential difference between that and the counter ions would be the same with increased and maybe the potential difference is what's relevant for the whole trapping reaction whereas in this microsecond timescale what we are changing is the electric field so when we go to higher concentrations we're creating a screening layer in a thinner distance this is just the debiling which would imply that the electric field increases so the suggestion here is that this bifurcation is related to strengthened electric field allowing essentially the faster pathway to come into play okay so that's just a suggestion based on this bifurcation and where this effect saturates so i've been talking about these pathways and of course these pathways should relate to activation barrier heights so then we should be able to modulate those activation barrier heights so we did that first in the conventional way using temperature and this is the rhenius law you have the log of those kinetics versus inverse temperature and you get slopes for each of them in the range of 0.4 to 0.5 EV for the activation barrier height you get prefactors in the terahertz range which by transition state theory would suggest that there's vibrational motion that helps you get to the transition state so that's a conventional way to identify activation barriers but there's another way we can do that which is using the screening dependence and so what we find is that you have a log of the kinetics and those rate constants are also dependent on the screening and here we now cast that screening as an ionic strength and those rates increase with ionic strength okay so how screening might be involved is that these counter ions can screen a charged transition state helping if these are like intermediates they can help and get through that transition state to the next step now what I have shown here which I learned earlier on is that the Bronsted-Beriem equation I learned it predated actually transition state theory and what it said it was is usually used for homogeneous solutions is that is essentially describing this screening of a charged transition state where z is the charge in the transition state and the square root describes that it's a screening effect and i describe is what we modulate in the experiment and what we get is approximately a square root behavior but what we also get is a charge that's reasonable and that it's less than the full proton transfer and the charge is higher for the pathway that's favored at higher ionic strengths okay so what we then did is this is basically kinetics of the decay of this TIH star population we simply mapped it onto two different possible mechanisms for these activation barriers so there's a lot more work to do here and to elucidate these mechanisms we just mapped it onto them based on the reaction condition dependence and so we mapped the faster pathway to biradical recombination to form the O-obon and that's because it happens at higher ionic strengths the nucleophilic attack pathway we mapped onto the slower pathway essentially because it occurs with some HDKIE and it is slower so it would involve some lattice intermediate all right um there's a lot more to do as i said we are uh planning to do right now with those maps that i showed you those optical maps to do vibrational spectroscopy on them using kind of second stimulated resonance Raman spectroscopy and that's where we hope to try to resolve the oxygen-oxygen bond on the microsecond time scale and maybe something about these transition states so that's experiment long time and coming and part of it was to get really nice optical spectra was point one so i want to just introduce two other experiments we've done not on strontium titanate but onto other ones and it's just one slide actually but it has to do with the screening layer as well so the fact that these TiRH star population bifurcates like this and that there might be dark chemical steps that guide the bond formation uh suggests that surface mobility is really important and that's something where it's really hard to get to experimentally and it's hard to get to experimentally on transition metal oxide surfaces especially but early on we looked at this with just a simple simple conventional semiconductor gallium nitride and what we did is we um created a grading of charge carriers in the semiconductor and we could change that grading spacing and then we can diffract off of that grading using our spectroscopy and this is the diffracted pulse to get the rate at which the decay the grading is decaying so the great leaving and that would tell us about the diffusion or the lateral diffusion along the surface so this is done in an undoped gallium nitride semiconductor and this is the rate at which that grading decays versus the cue spacing of that grading which we can control and from the slopes what you get is the diffusivity and it's the charge diffusivity it can be of electrons it could be of holes or it can be of electrons and holes and so in undoped we basically mimicked what the what was seen before um 400 gallium nitride also using this transient grading spectroscopy and then what we did is we simply put it in a semiconductor surface and uh so then we took instead of an undoped gallium nitride we took an undoped gallium nitride surface and we looked at it in air and we got the same slope as we would get for the um for the undoped one but when we put it in um solution we got a much higher slope which means a higher diffusivity okay and so uh that suggested that essentially when we have charge carriers localized at the interface with an electrolyte uh that electrolyte helps that charge it increases its diffusivity or its mobility and you can see that by all the checks we made here on the different types of samples just undoped air undoped electrolyte undoped then gallium night or doped gallium nitride air and with two electrolytes and here is only when we get that factor to increase so we need the charge and the electrolyte to increase um the charge diffusivity and this is a holes at the interface so Tony I think we lost you we cannot hear you anymore hear us please disconnect and reconnect let's see something like this had to happen at a certain point that is uh what I was afraid of yes but in your case it worked perhaps someone can text her a message or some other way I just yeah I wrote an email but if the internet collection went down yeah but somehow I doubt that the internet connection is completely down to bolder does someone of you have her presentation no also even if you had the presentation you wouldn't be able to explain what is true it can be done by phone it already happened okay if one of you oh no okay you can give her a few minutes tomorrow no yeah yeah sure that is also true yes yeah tomorrow we have a shorter day so we'll be totally well we can give her a couple of minutes to see if she comes back if not yes it was almost yeah this is Francesca I texted her I can try to call her and see if she okay did you hear the last part of my talk or was it gone depends on how long did you keep talking exactly yes the class was something like five minutes ago we had the cut let's say but I didn't pay attention to the slide number was it oh god but but we can confirm which one you you said five minutes so I would say yes this one did you see yeah yeah and this one did you see yes okay so it's just the last one okay good okay add my my conclusion slide okay all right but it might have been worse for the questions I don't know do you want no questions you close the sharing again oh do you want the sharing okay okay let's do the sharing no please go ahead and finish your your presentation even if we're on 10 minutes yeah yeah yeah go ahead and finish okay I'm just I don't know what I just did presentations summer I will get this okay excuse me stopped at least the last yeah okay share do you see it no you see my oh no I have to share it and mention do you see this we see now you see the other one the presentation mode okay now you see it right yep okay I'm gonna go to my last one so this is the what you didn't see so basically charge diffusivity mobility is really important at these interfaces it could be very important for dark chemical steps but it's really hard for us to see it in transition metal oxides early on we did something on a simple semiconductor where we created a grading of carriers inside the semiconductor and we diffracted off of it and that grading essentially as it disappears is going to tell you about the diffusivity of charge along the interface so this is the decay of that grading that rate versus the q spacings and from the slope basically you're changing how close these peaks are to each other but from the slope you can get that diffusivity constant and so this is an undone gallium nitride this is an air and an electrolyte so we get the same thing and both and that was done before and we could say okay we had the same thing that others had and then we went to end up gallium nitride surface and there you would get mid gap states so you would get charged into you know things that are trapped at the interface and when we did it in air we essentially get the same thing for the whole diffusivity that we would get in the undone but when we put it in the electrolyte we got a higher diffusivity and so that suggested that essentially a picture like this when you trap the charge the electrolyte helps it move along the surface and so here's this undone gallium nitride and air and electrolyte we get the same whole diffusivity and end dope gallium nitride and air or we get the same whole diffusivity but in electrolyte we get a factor of two higher then the other part was that we were so this is all about the screening layer essentially and um we also looked at that in a battery electrolyte not the diffusivity but here we're just counting charge um the positive charge in the lithium ions and negative charge in the perforate and we're using this atr spectroscopy which found pretty powerful to get at that and we looked at the number density through our absorption cross section versus potential this is the nominal pzc which we define as the open circuit voltage go into the negative side or positive side and including lithium ions here cumulant perchlorate ions here and um and then we did an agreement with oligodon on this oligodon into the nd simulations what you do require for this is putting a factor for the surface enhancement that we observe on these gold electrodes but that factor is on every potential the same and then just this is the summary slide if people want to see this is essentially what I talked to you about is forming this to OH star intermediates through their electronic states and emission through their normal modes with vibrations and something about the continuum strain within coherent phonons that all gave us a whole trapping energy of 1.3 picoseconds that allowed you asked to ask the question well what is the reaction free energy difference and we isolated the equilibrium constant of this by changing but by the century the surface acidity in the dark and we got an equilibrium constant of 150 which placed this on the weak binding side of the volcano and then we looked at the fate of these in marine intermediates at six at microsecond time scales we see two transition state pathways which we loosely assigned to these two mechanisms and then hopefully I giving you an understanding that this is we it's it is important to isolate individual reaction steps within a continuous cycle and I think fundamental logical theory is a good bridge to try to help you know theoretical descriptors be related to time result kinetics I think if we see this more and more we're going to see that there's going to be a lot of opportunities to select for different pathways by for example surface modification I showed you one with the salt concentration and the EDL so that's my talk thank you so I will ask the audience if they if they have any questions and if not I have a ton but please go ahead if there is yeah James Darant just a second yeah you can unmute yourself yes I have it's fine thank you Tanya that was very nice I enjoyed um I was obviously one of the things we think a lot about is what the energy the surface um holes or the TIOH you talk the states are on the surface and you can consider the energy of those states versus depending on the edge but probably it's more relevant to consider it versus the energy of localised bulk polo arms I I wonder how and I'm there's quite a lot of work now on electron polo information but I'm not so familiar with whole polo information I I wonder what you think about how much this the relaxation energy of the surface TIOH groups might be compared to the relaxation associated with whole polo information so that that's actually been um calculated in strontium titanate and then I'll say experimentally but the calculation is that um this was done by Genotti for strontium titanate and um he did and he found for the that there is a whole because there's electron polo arms in in strontium titanate and I draw those kind of close to the conduction band when you dope but the whole polo run is minus 0.05 EV versus the valence band edge and this is what Das calculated for for these bridge and oxyles is minus 0.4 so that's the difference in the calculation um experimentally though um one should keep in mind that even if you have like a small difference to or small free energy to create the whole polar on it can create UV vis absorption because these are optical transitions that are reaction coordinate and for example Genotti finds and he actually calculated that for the bulk whole polar on is 2.5 electron volts would be the emission and um so how we relate this to surface poles is um because it's essentially a sensitive to the surface and not the bulk um but and and that they are much more favorable to be created. I also think and I know uh Mikhail Sprague has probably calculated this in TIO2 the bulk versus this is surface. Yes um I am I never trusted these results can I can I show this to you? Sure. I have to uh share my screen is that possible? Oh I will stop sharing. Maybe this is asking too much for me. I'm not sure whether can I? No then I will just tell the results um so this is uh um calculation we did with Jung and what we found very disturbingly that the A the reorganization energy for the OH-2 OH dot is the same as in bulk water and not only that even the levels themselves are in the same place so the surface OH- is very hard to distinguish from a bulk OH- both with respect to the levels and of course the reorganization energy. We know where the hole is because that's where the proton is missing but I never dare to publish it and Jung probably was rather happy that I didn't because it can be a DFT error, right? You understand what I'm saying? The pictures are the same. So you're saying that the OH- to OH radical in the electrolyte is the same as on the surface? That's right, yeah. Yeah I mean what we're looking at is on the surface, right? There's uh that we don't see OH- and OH- radical in the solution based on our structure. No no I understand that but I was just mentioning that uh so my my my explanation was that the tight hydrogen bonding of the guy on the surface is essentially the same as in the water but that was of course mere speculation. Yeah that is surprising to me because I remember your paper with the transient OH. I don't know if you guys remember the the JVCC you had a transient OH and it was very short amount of time and then it became the TOH star that I'm talking about and the free energy difference. Yeah yeah no I work this out a little better now. I can send you the picture by email and you can stare at it. That's not that transient OH but it's it's okay yeah I would love to see it. Yeah so as you understand the whole the un-relaxed hole is below the valence band so obviously it doesn't exist for too long. Okay I'll send it to you. Oh the un-relaxed hole of the TOH minus or the transient TOH is below the valence band so it doesn't live that long yeah yeah that's that's published right that's not something about one volt. Yeah yeah yeah a question a question I have is regarding hole accumulation at the surface so you might have seen Professor Darren's talk the fact that holes can accumulate the surface and then drive a series of chemical reactions chemical steps and it seems to me that your cartoon in which you had that cascade of steps and you mentioned the fact that the first one is light driven and then you have a series of steps that are dark steps yeah so is this consistent with the picture that Professor Darren has proposed so you accumulate the holes at the surface right yeah some chemistry happens right but chemistry is not a sequence of one electron oxygen but it is can be single steps can be multiple steps so this is like a thermo chemistry it's not an electrochemical step is that emerging from your work too? I would say that on TiO2 so we're very much on strontium titanate but on TiO2 the picture that you just said I would say is consistent with this we still have to look at this picosecond to nanosecond range like I said and also assign the 60 picosecond times constant but if I look globally that idea that there's like something on basically we have a similar optical cross section at nanoseconds that hundreds of nanoseconds that we did at two nanoseconds and that's a little bit so but we want to do that in detail I want to say we're going to say but then a March decay afterwards suggests that these electron transfers are happening early and then there's dark chemical steps and yeah that's that's yeah but now you also have a way to to measure the coverage of these intermediates so you could do the very same thing that Darant has done with optical spectroscopy to correlate coverage of whole trapping intermediates and photorelectrochemical performance to see if indeed there is this multi-hole radar terminus step or not so it would be two different ways to different spectroscopies to address yeah the same issue right so one thing that we don't do a lot of but we do some of is a fluence dependence which is essentially what James Duran is using part of the reason we don't do that is because we want to keep the interfacial energetics the same for what we're looking at and then we're spanning the reaction conditions so in in the what I want to do to make a connection here is that in order for us to really make this connection we need both the TAOH star as the well as the bulk holes we need a count of both and I think that's why I put that the TAOH star is a good count of emission through the emission through for all these reasons whereas the absorption has a lot and I showed on one of the slides it's really with the ultrafast light pole so which with carriers coming and then has a lot of the valence band holes but I can't say that it's all the valence band hole population because it is a 1d spectroscopy so it's the populations of both and we're using the valence band states for that transition that are the ones that are also creating the whole polarons or the star so the time dynamics is going to be a complicated by the valence band states how they're changing in time as well as the whole polaron states how they're changing in time whereas the electrons are coming from the conduction band are independent the other thing is you're going to have overlapping transitions that's another thing and the reason I just point out this is because it's 2EV we have a peak there people assign this to O- in oxides and so on but I think if we get more data for both of these transitions perhaps we can take out the populations and then for example do a fluence dependence to see because one of the things that also when we're doing the equilibrium constant what I would like to know is how many holes are involved exactly to create this amount of TiOH star and we're gonna we have more holes than we have TiOH star from what it looks like right so and and it could be 50% or something so calibrating that would be good and so I think what we need is a broadband spectroscopy where we have more information on the two populations and then a fluence depends I will say in terms of fluence because I didn't get to in the talk is that you're going to saturate the emission if we do this scanning mode where we're really looking at a fresh sample spot you saturate the emission at a certain fluence which makes sense because you only have too much so much TiOH minus there at the surface that you're using whereas the absorption increases linearly that linear increase though can be of valence band holes you're creating or of total charge at the surface right it linear increases one has to you know and deconvolve a bit so I think that I think yes I think that in short we should be able to eventually make that I also think theoretical input would be good here for the optical admission and absorption and what it's going on perhaps some ideally DFT calculation but I know that's very hard for electronic transitions but even in terms of what I'm talking about like if we have kind of like you know Fermi's golden rule you know what are the you know I can give you the number of poles that we create and things like this and that's something that would be good perhaps as you as you were referring to my work as Emanuele I maybe I should have a little comment on that as far as I understand a lot of the first part of your talk kind of was the formation of the surface holes which then in my talk I focused on how those holes or surface TIOH groups come together to drive or draw stations so in that sense I think what we do is quite complementary the bit where maybe we have some differences on the time scale of decay of those holes at the surface where you see quite a lot faster kinetics than we observe and I am aware within ultra-fast measurements it's quite hard sometimes to avoid recombination losses and I know you've done quite a lot of work to try and show it's not recombination and all this sort of thing and maybe strontium titanate is really unusual at having much faster kinetics than other materials we've started to look at that and we haven't I'm aware the measurement the reason why we do our measurements with these long LED pulses is to try to avoid the issues of the ultra-fast laser pulses and all the high intensities they create and all this sort of thing but we're certainly planning to do some lower light flux measurements on the STO to see if we get similar time scales to you or while our slower time scales so yeah we are you asking and you know we do do use as ultra-fast laser pulses we do get 75% in greater quantum efficiency with it and I know that sounds unusual but it's not that unusual in the context of a single crystal because if you actually use the diode well I mean the diode is but and but it does limit that we it's hard to get to other materials so we use this as a model system and then um in the future try to get it to other materials too with it but I think you were pointing out microsecond time scales versus your millisecond yeah because most of our warthog stage is pretty second to seconds um yes and no it depends on what special region you're talking about like these microsecond kinetics are not inconsistent with oh oh bond formation because I mean oxygen evolution might have you know so oh oh bond formation it's half an EV electron volt scale and terahertz so the the time scale you're looking at also depends on the spatial scale you're looking at you know and where you're going from and so um yeah I I think it's and and we do we do are seeing oxygen in the in the electrolyte we are seeing ultrafast we're seeing start separation so most of that shot yeah most of that is leaving at microsecond time scales but it might be that other things are rate limiting of the reaction like the oxygen evolving um and it's very hard for me to connect to the turnover rate yet and um because it's very site dependent how many sites are involved in that millisecond time scale let people talk about or second notes I know that our for our spatial site we're we're doing one oxygen per the per site per second for for the spatial region that we have which is not inconsistent with with some of the things on oxides and that that's how what we produce and what um what we produce is that and then I forgot what else I was going to say it was fine okay I went out there was a you know I hope you know there's a lot of things of operando and in situ and that kilohertz laser pulses aren't catalysis necessarily and I just I don't know how to answer that because catalysis is inherently transient and that transient steps give you a steady state we pulse that catalysis and we get a steady state we get a steady state on those scans of 0.5 standard degree we can we can do that on a clean enough sample you know do that for for a while so the idea that catalysis isn't inherently something that you can control by pulsed excitation to get a given steady state I don't think is true because you can also continuously light excite and just continuously have holes but that's just one rendition of that steady state because the process itself is inherently transient I mean I suspect if we get into too much detail on the transits of adoscopy that might not be quite everyone we can have more discussion about the best way to do the details that was not just to you James it was just a general what I've been thinking yeah okay so yeah I think we're done for today thanks a lot to all the speakers and all the people that participated in the discussion and we can reconvene tomorrow in the afternoon for the last part of the of the workshop so thanks again professor bandical nice to meet you actually I I uh Elena Magnano was in my lab for a little while and we did some resident xps I know nothing about it but she did it for us and then yeah I just want to tell you nice to meet you too great talk yeah okay see you bye bye bye everyone right