 So thanks for for the invitation and on the purpose that I'm the only one separating this talk from Lance. I'm going to talk about understanding water interacting with different surfaces. Originally I was going to talk about photo catalysis and electrochemical applications but I focus mostly in the electrochemical application setting. So in terms of what we are interested on is really being able to simulate using first principles an electrochemical cell and that we are still very far away from doing it for many reasons and starting from the fact that we need to do similar to in thermal transport long simulations of initial molecular dynamics simulations and large systems. But there are other problems that separate us from getting there and they are related to the understanding of the systems that are part of this electrochemical cell and in particular the key element is liquid water which is always part of these cells and the simulation of the metal interacting with liquid water. So ideally what we want to know is the correspondence between the macroscopic voltage that we applied to an electrochemical cell and what is the actual interfacial charge that occurs at the surface of the metal interacting with water or with an electrolyte which is liquid water containing different ions. And beyond there of course is what is the actual structure of the electrochemical double layer that occurs at that interface and how and applying a voltage to the system modifies such a structure. And so what we aim to simulate is some sort of interface with water and a solid in this case is a metal and we need to have a very accurate representation of the two systems. We need to bring them in contact and the questions that we want to know to start with are what is the structure of the full struggle of all the molecules so the liquid itself at the interface whether it occurs or not a charge transfer between the two systems whether that is modified applying a voltage. And so we are going to start from the most simple system that we can imagine which is just one single water molecule at the surface of the metal. In this case this particular simulation is for palladium and there have been many many studies about what happens what's the orientation of that water at the surface of the metal depending on what is the metal that we have but the bottom line what characterizes interaction is the fact that of course what the metal is going to screen the charges of the water is going to create an image charge at the interface and here I'm showing such image charge for a single water molecule so in blue what you will see is a depletion of electron so a locally positive region and in red is an accumulation of electron or a locally negative region. And so indeed what happens with water absorbs some such a metal is that you have something that is like a mirror and water interacts with the metal by forming pseudo hydrogen bonds indeed you will have a hydrogen bond of the oxygen with the with the local charge underneath and another and of course it's sort of pseudo in that there's no real hydrogen in here but the depletion of electrons represents the the local charge of a hydrogen and it's actually the same on the other side however I mean according to this what what we will expect if everything is electrostatic is that indeed if you just have one single water molecule the most stable configuration will be that where the water molecule is placed with the dipole moment perpendicular to the surface so this electrostatic is the most favorable interaction for a dipole with its corresponding image charge and what you're seeing here indeed is if we just have a classical description just just point charges and describing the water molecule point charges positive and the hydrogen and negative at the oxygen interacting with a perfect metal that creates the corresponding image charges the most favorable configuration is this one in here which with the dipole pointing up the second metastable configuration is this one in here and that's the transition state when the molecule is placed horizontally and we include semi-classically meaning that we allow for polarization both in the in the metal and the water molecule but without any chance transfer the picture is slightly modified we see that we acquire some other semi stable state at this structure which represents the position of the lumpers of the water molecule but you still find that the corresponding up and down dipoles are the most stable however it happens to be that once you include everything including the full charge redistribution and inter and charge transfer between metal and water what you see is that this horizontal position is the most stable and the reason for that is that the pilot repulsion between the electronic cloud of the oxygen and the image charge in the metal is basically dominating the interaction and favors this and this configuration and so if one goes and we want to do now simulation of large systems plenty of water molecules and metal there is the the question of what how should we do it using fully classical semi-classical all fully ab initio simulations and so some of the largest simulations that have been done were done in the group of David charler and what they did is basically using a semi-empirical force field with where water molecules were described rigidly so they are in a way purely classical simulation there's no flexibility on the molecules and everything is point charges interacting with that with the metal and with a force field that actually includes polarization or the creation of image charges in the metal and this is the the model use and is it was a model which is in a way fitted to reproduce ab initio simulations in particular the absorption of water on platinum with a monolayer both geometry and energy but never a full bulk system and so what was observed in the simulation of this system is that indeed depending on what is the determination of the metal so either platinum one zero zero platinum one one one interfaces the the structure of the water layer at sort of the interface is very different and in particular what they described is that in one zero zero interfaces what one obtains is some sort of a hydrophobic water layer what water gets so well bounded to the metal forming bonds with the metal within the water molecules in the metal such as it doesn't favor the formation of further hydrogen bonds with the with the upper water layers and so what you guys is a highly hydrophobic layer and not because of the not on the metal itself but on water itself and and if we actually go to the one one one structure what happens now is that we have much more freedom for the water molecules to absorb either horizontally vertically with hydrogens pointing up or down the layer and now we open the path to many more hydro and bonding structures with the water and that automatically makes this this layer very hydrophilic but of course here is the question of how much is this correct if from what I said before and the fact that there is not actual chance transfer interaction described in the system and and no no effect of Pauli repulsion of or any other quantum mechanical effect that can occur on all and also what is the the effect of further polar isasol in the liquid water and so the first thing that we did is precisely study a system which is much smaller than the one done by the group the Berkeley group but in a way much more expensive from the point of view of simulations because we we have to simulate the metal and the liquid everything have initial and so there's a I mean this is the summary of a lot of work and of course a lot of these simulations will depend on what is the initial structure that you have for the for the water so first of all we start doing equilibrations using classical force fields and then once we don't different equilibrations we move we switch to quantum mechanics and within quantum mechanics we also have the the question of what how do describe the liquid water which is we all know and one of the the systems that is most difficult to describe at the standard with the standard DFT approaches and so I'm gonna I'm gonna present which are the results and hopefully that will give us an idea of why why these systems are complex and in part also associated to the fact the liquid water itself and is close to a somehow a critical behavior which give rises to structures which largely depend on what are the conditions that occur and the interface between the the water and the metal so if we look at the results of these simulations and and we look at what is the order in of the first layer interacting with the metal what we see is that we actually can have two different types of order one which we call highly order where you also here basically in in red is the density of oxygen and in blue is the density of the hydrogen and this double peak is characteristic of this sort of double layer and you see these two peaks which corresponds to hydrogens which are pointing towards the metal or pointing outside of the metal and then you can also have the structures where indeed you have a much more favorable configuration for molecules with most of the hydrogens pointing towards the metal and so that probably on itself says little about what's going on but it's easier we go now and look also what happens in the in the in the plane and so this structure with the up and down type of positions is a really very order structure so this thing here is the is the unit cell and everything else is pretty boundary conditions and what you see that of course is that indeed water is absorbing forming what is so called a two-dimensional ice type of layer and that structure is very stable and very order and very what we call low density type because indeed is is relatively close to to a very hydrophilic and hydrophobic layer if you go to the other side where we had a a favor of larger hydrogens towards the metal and what you see is that there's much more disorder there's much more mobility in the two systems and and what I'm showing here in comparing is what happens whether you simulate using a standard gga approach or when we use a Vanderbilt's interactions and it seems like indeed the Vanderbilt's interactions modify a lot the structure in the system however that is not really the case what is occurring in this in the system if we we look at the full detail of the simulations is that so this was the the sort of one type of layer order layer two-dimensional ice layer and this is much more disorder more high density type these are the results we do just a pb simulation so without Vanderbilt's and with Vanderbilt's and what of course is that both of them are actually giving the same type of structures but the Vanderbilt's one has a very fluke 20 type of domains so we as what we are seeing is that that we don't have a unique type of structure forming at the interface you have domains which intercombelled from one another and the only difference between the Vanderbilt's and non-vanderbilt's is the fact that this is very stag and the diffusivity is much smaller and we don't have timing scales long enough to explore the motion between the two domains and so if we look at a different also this of course with palladium if we look at gold and in this case the chance transfer between water and gold is much smaller than with palladium so the description of this model should be much closer to to the classical description so it's very similar to platinum and indeed what we see is that the results for golds are very close to the results that were described in the in the Berkeley Groups paper and where we have a very hydrophobic layer followed by a very hydrophilic structure and so how why is this relevant where it's relevant because if we want to understand what is the the the change of work function of a metal upon salvation we need to know what is the actual structure of the layer and the layer happens to be very dynamic and very different in nature so we want to calculate such change of work function what we need to do is basically calculate what is the work function of the metal in vacuum and then the change of it which is due to the dipole of the dipolar structure of the of this of the liquid water layer plus the polarization induced by by the metal and that polarization is basically can be decomposed further into the image of the metal interface the polarization of the water and then the net charge transfer between the two systems and so we do that analysis what we see is that indeed um depending on whether we have one type of order or the other type of order we are going to get so this is for palladium in the case of palladium we have the the the first order that I described with the bilayer has an eight dipole which is pointing out from the surface and that will decrease the work function of of the solvator surface and the other type of order has an eight dipole which goes towards the surface and that increases dipole the sorry the the work function and in the case of gold you get an average and this actually has these two are very large so these two dipoles are large and opposite directions in the case of gold you have a small dipole and always out which is a net decrease of the work function so when you look at experiments indeed you can have before I go I go to that just one extra thing which is and this was already shown by the group of growth they were actually only looking at single bilayer so not a full liquid layer and as a function of the metal and what they were saying is that depending on how is the the bilayer which can be with with hydrogen down or with hydrogen sub the net change of work function decreases as a function of the the actual natural work function of the metal so what we are studying is for gold and for a palladium which are quite extreme cases gold and has a very small change of work function and palladium are rather large one and so so if we were to compute what is the net this is the the delta work function so basically what we are calculating is the change on work function and after a salvation of water without salvation and and so what we see is that and this already includes averaging for over the two structures is that is relatively large for a for palladium and relatively small for gold there is a change of signing gold whether we use a pbe or van der Waals but in both cases what happens to be is that the one that compares much better with experiment is the is the van der Waals results which are which are sounding here so this is the solvated experimental wave function and and this is with van der Waals so relatively close and this is this is for gold but what I want to say is that in order to obtain these these numbers and one cannot just simply do the simulation of one one single monolayer or one needs to actually do a long molecular dynamic simulations but they correspond to really average of these two type of domains and and the connection I wanted to make to liquid water is that these two types of domains are actually in a in fact we believe related to the to the underlying water criticality of having a low density and high density types of fluctuations which they are enhanced by the interaction with the metal now I told about electrochemistry and of course what we say that indeed what we want to know is what's the effect of applying a voltage over this and which I'm not doing just yet not even here and so the the next thing we did is actually charging our surface in the standard way of charging surfaces and or charging calculations in in in a Venetian simulations which is just adding a net charge to the system with a compensant compensating underlying background and then relate that to what will be the situation with an applied voltage so so doping with electrons or charging the surface with electrons will correspond to a sorry this is charging the surface removing electrons will correspond to a negative voltage or in charging with electrons to a positive voltage and what we see is that indeed charging the surface is what it tends to do is to even enhance those two different types of structures it makes an either more order layer for in the case of of positively charged and in the case of negatively charged in favor so much more disorder layer now in reality what we want to know is exactly what occurs if we have an applied voltage which we cannot do and so what we did is basically propose to couple the standard transport non-equilibrium rinse rinse function formalism that is used to to study electronic transport in in systems and nano systems so apply exactly the same formalism but where our scattering region now is going to be the actual so-called electrochemical cell which is metal plus water in in between and so we can actually now establish apply a voltage and taking out of equilibrium the the Fermi level of the two sides of of our cell and in the standard non-equilibrium transport formalism so as we say this arrangement and is the same as what we encounter in electronics transport we have a central scattering region couple two electrons and now these two are going to serve as our reservoir and in principle the voltages that we apply and of course we can actually match now to experiment and the other thing that we can do so that's the description of our region of course the scattering region does need to include a quite a large section of the metal because whatever charts transfer that occurs between the water and the metal and needs to be accounted in the scattering region whereas the electrons are supposed to be already in in equilibrium and far away from the from the from the scattering and the interactions with the water and so we can do this and we can even now as long as we ensure that we don't have and that's the actual system that that we can simulate and we can do this and we can also compute the the forces within the the scattering region which are warranted to be correct as long as we have no actual electrical current circulating through the system which in our case is correct because the band gap of of the scattering region of liquid water is larger than eight electron volts and so we can have ionic currents and that those are correctly defined but we as long as we ensure that we don't have electronic currents and we can really simulate this type of setting and so that's what what we really want to do but with this this work was just a proof of concept and to prove that that it works so we started again back with one molecule but now what we are doing is really doing relaxations of the of the structure of this molecule in the presence of an applied voltage so in this case what you're seeing is what is the configuration for a positive voltage so the water molecule is here and as you see that means it's interacting with the negatively charged surface and so one of the things the first thing that that we we identified is that in this the potential energy surface of this water at surf at the surface as a function of the voltage is really far flat and with the voltage meaning that up to voltages is smaller than than one one volt there's very little happening the equilibrium configuration strips shifts very little voltages and it's only when we go to voltages larger than one volt that we start seeing that we are starting changes and in this these are the structures this is the equilibrium is all doing a standard conjugate gradients optimization and so what you see is that it's a very highly asymmetrical simulation with a positive voltage I mean charging with electrodes the surface what happens is that the molecule gets unbound and now of course the so the the molecule basically goes far away from the surface and gets unbound with negative voltage you have a very small change to shift up the hydrogens very little and the the distance between the oxygen and the and the surface decreases slightly but you can see how how asymmetrical is the structure with the voltage and the reason we understand that is actually as I started the talk with and exactly the this poly repulsion between the charts in the water and the metal is strongly bias dependent and so what we did is again do the simulations but now subtract and repeat the simulations with for the for the electrode itself without the water and then the water along with the with the voltage and subtract one volt for each other and so in a way what we are doing is removing what are the effects of the electric field that is coming from the metal and looking only at what are the effects from the from the charts exchange between the metal and the water and so this is with with zero bias and what you see with zero bias basically what you want to look in this picture is the larger the blue region is the larger is the poly repulsion so you have already a very large poly repulsion between water and the metal when you decrease sorry the opposite the less you have it in here the less the less blue region is the the larger the Fermi repulsion and so you see that you decrease the Fermi repulsion when when you apply a positive voltage and you increase it when you apply the negative voltage and and the reason for that is basically because you really change how much is the the amount of chance transfer that occurs between the two as a function of the voltage and so that's where where we are now so this is for both of course which already to start with there is a very small chance transfer so indeed if you look at the numbers in here the iso surfaces are very small so so 10 to the minus three electrons per cubic ounce drum so how much you see the asymmetry between applying a positive or negative voltage in the case of gold is small because it's all to start with there is a small chance transfer between the metal and repeating the same with palladium makes the simulation of this electrochemical cell much more complicated because now the region that we need to describe within our scattering section the region of the metal is much larger because precisely of that charge interaction and this is what we are working on now and hopefully we'll have results for not only for different methods but also for larger for larger cell simulations so I'm going to stop in here and the first thing that that I do want to highlight is that indeed the order of water at biases interfaces it really reflects and the fact that water has this instability of forming a two different type of structures low density and high density and the second is that indeed we can now simulate electrochemical cells using the non-equilibrium response and formalism to describe this system until an external bias and we can do that without having to add charge to the system so really being able to to compare to methodology and so we can now do molecular dynamics as I say as long as we ensure that there is no electric transport through the system which is ensured with systems like this which have large gap and with this I'll finish thank you