Functional ALD Coatings for Tailoring Reactions Kinetics

Katherine Hurst – National Renewable Energy Laboratory


Thin films grown by ALD onto catalytic particles can have a significant effect on the kinetics of a reaction. This includes altering mass transport of reactant species, durability, and dispersion of catalysts. Another way is through plasmonic heating from the thin coating. The deposition of a thin layer of plasmonic material can enable photo-mediated hydrogen desorption from coated metal hydrides. Dehydrogenation reactions from traditional metal hydride materials require high temperatures to induce the release of hydrogen. Through localized heating via surface plasmon excitation, the bulk thermal signature for hydrogen release can be significantly reduced. In another example, thin coatings by ALD can produce changes in the reaction selectivity based on tailoring the mass transfer of the reactants to the catalyst surface. The influence of oxide coatings onto photocatalytic catalysts can result in changes in the overall reaction efficiency.


0:00:02 Hello and thank you for tuning in. I'm Katie hurst and I'm excited to tell you about the work that
0:00:08 we are doing at the National Renewable Energy Laboratory. I'll start with some brief comments about
0:00:15 L. D. And how we're using it for catalyst development and then talk specifically about two different
0:00:22 applications. The first one in gas storage and the second one in photo electrolysis where we're able
0:00:29 to change the functionality of the catalyst by using L. D.
0:00:37 When we think of the properties that make a good catalyst. These include high activity, the
0:00:42 appropriate reaction selectivity, good stability as well as being inexpensive in terms of the
0:00:50 material, the materials themselves and the process to make the catalyst.
0:00:59 What determines the catalyst characteristics are really the material properties and these material
0:01:05 properties include morphology, uniformity, surface area and chemistry, particle size density and and
0:01:12 scalability. The old growth technique provides very good control over creating films and particles
0:01:21 with these defined properties. And we can change the material properties by changing the temperature,
0:01:29 pressure, precursor chemistry and exposure times as well as the type of reactor.
0:01:38 So L. D. Is a route to synthesize materials with defined properties, which in turn alter the
0:01:45 functionality of the catalyst performance. And this approach is used for many different applications
0:01:52 of materials.
0:01:56 The first example I would like to talk about is in hydrogen storage and as industries economies move
0:02:03 to use hydrogen as a fuel, we need methods and materials for transportation and storage and we are
0:02:10 working to create materials to do this ideally, the materials would have a high capacity. They would
0:02:18 be easily discharged and charged with hydrogen at near ambient temperatures and pressures and also
0:02:27 be inexpensive. Current technologies including high pressure compression and liquefaction, are
0:02:34 expensive and non scalable. The graph shown here on the right shows some of the materials of
0:02:42 interest plotted with volumetric capacity on the left versus gravimetric capacity on the right. The
0:02:50 color code indicates the amount the temperature that's required to do is orb the hydrogen down in
0:02:58 this regime. We have carbon materials that have a low capacity but also require low temperatures to
0:03:05 dissolve the hydrogen for mental hydrates and chemical hydrates. They have a higher capacity.
0:03:11 However, they also require a higher temperature, sometimes up to temperatures of 500°C.. One of the
0:03:22 approaches that we're looking at at a general is to reduce the overall bulk heat and temperature
0:03:28 required to deserve the hydrogen by incorporating plas Monica nanoparticles with metal hydrates. So
0:03:37 a plasma on is a collective oscillation of electrons that is excited by a specific frequency of
0:03:43 light in a metal or semi metal. And this frequency is a function of the particle size and material
0:03:50 composition. The absorption data and the lower right um indicates that the titanium nitride
0:03:58 nanoparticles used in Our work have a plasma at a wavelength of 680 nm. By exposing the titanium
0:04:09 nitrate particles to light. At this frequency or wavelength, we can excite the localized surface
0:04:15 plasma of these particles
0:04:19 when the titanium nitride nanoparticles are mixed with a metal hydride, similar to this image in the
0:04:26 center. In the lower part of the slide, we have found that hydrogen is released from the metal
0:04:34 hydride at the ambient temperatures without directly heating the material. And so this provides a
0:04:40 really exciting route to avoid energy intensive bulk heating of materials that is costly and slow.
0:04:48 So oftentimes people talk about gold or silver plasmas and this is similar to the titanium nitrate
0:04:55 plasma. However, titanium nitrate is a more stable material and can retain its nano structure up up
0:05:03 to much higher temperatures and plus it is lighter and less expensive than gold. So we will get to
0:05:10 the L. D. In a minute. But first I'd like to show you the work with the particles. Okay, so in these
0:05:17 experiments were looking at a mixed physical mixture of titanium nitride with ball milled magnesium
0:05:26 hydrated. So that's just in order to make it smaller particles. And then we excited with a 625
0:05:34 nanometer led and that gives off hydrogen and then the titanium nitride and magnesium is left behind.
0:05:43 This tp this light induced distortion spectra here shows the hydrogen signal comes on instantly as
0:05:52 the light is turned on and then it starts to decay. And then when we agitate it, we're exposing a
0:05:58 new material and that causes these spikes here and then again when we turn the light off, you can
0:06:06 see that the signal goes down. We also did the blank experiments and demonstrated that the hydrogen
0:06:13 was not released by just showing just showing the magnesium hydride and the titanium nitride
0:06:20 separately to the light. So it's a synergistic effect that is that occurs. This is a tm image
0:06:28 showing how the titanium nitride nail particles are dispersed. And you can see they're hearing these
0:06:35 black dots here and there throughout. It's very it's pretty well dispersed. And then these are these
0:06:41 big glob, gray glob is the magnesium hydrate. And now the amount of hydrogen that has is coming off
0:06:51 of this material is less than one weight percent. And so in order to be useful, we need a lot higher
0:06:59 um weight percent coming off. For example, the amount of hydrogen that is on this magnesium hydrate
0:07:05 itself is 7.68%. So we're only getting a small amount off um And in order to increase that, we need
0:07:14 to understand well, what is the real mechanism that's occurring here? And how can we increase the
0:07:19 interaction between the titanium nitride and the magnesium hydrate?
0:07:27 This is where L'd comes in. So a child can be used to help us investigate what the mechanism is. So,
0:07:34 in general, there are two main ways that plasmas can control the reaction first is through plasma
0:07:41 ionic heating. So the photons are absorbed and excite the electrons and then they decay down and
0:07:48 emitting phone ins and this releases. This causes localized heating. Alternatively, the Excited
0:07:57 electrons can be transferred to participate in the reaction. So they can transfer from one from,
0:08:03 from the titanium nitrate into the metal hydrate. Um Now there was a lot of different mechanisms
0:08:11 that actually occur when you have electron transfer, whether it's hot electron transfer or direct or
0:08:17 indirect. But the main idea here is that the excited electron is leaving the titanium nitrate. And
0:08:25 so we're going to use, we used an aluminum oxide coating that's surrounding this titanium, not
0:08:33 titanium nitrate to block any electron transfer. And so we've we've grown four and 8 nm of alumina
0:08:45 onto these, directly onto the titanium nitride particles in order to block that. And then we
0:08:54 measured again the hydrogen signal as a function of time. And when we expose it to the light and you
0:09:01 can see there's a quick turn on again for each of them. And the spectrum looked very, very similar.
0:09:09 And this is showing that um the Ox fied the oxide is really not affecting the capacity or the
0:09:19 kinetics of the hydrogen deserving. So if the electron transfer was affected by the coding, we would
0:09:26 expect to see this peak much smaller or a different shape if there was some kind of kinetic
0:09:33 alteration from this this coding. And so this kind of leads us to believe that the dominating
0:09:39 mechanism for this process is plasma ionic heating secondly, to increase the amount of hydrogen diz
0:09:47 orbed, we want to increase the interaction of the thai nitrate in the middle hydrate. So in order to
0:09:54 do this, we grew an A. An L. D. Film of thai nitrate around the magnesium hydride particles. And
0:10:02 this thin film geometry at the nanoscale can create a similar plasma tonic effect as as to that in
0:10:09 the titanium nitrate nanoparticle um in this configuration, when the time nitride material is
0:10:16 exposed to the light, the plasma sonic excitation will cause the metal hydride to diz orb hydrogen
0:10:23 and then transport through the titanium nitride film. In fact, titanium nitrate films have been
0:10:30 demonstrated to have good hydrogen purification behavior similar to that of palladium membranes for
0:10:37 the L. D. We Used tickle and hydrazine at 200 c. in our Vinick reactor. And we and this reactor
0:10:46 allows for airless transfer. The TM on the upper right shows that we were able to deposit titanium
0:10:54 nitride uniformly on the magnesium hydride particles and the high resolution image shows that the
0:11:01 thickness Was around 19 nm. This material was tested for hydrogen description using the six 25
0:11:10 nmeter led. However, unfortunately there was a lot of water that deserved with the hydrogen and we
0:11:18 believe that there could have been some contamination from the hydrogen from the hydrazine and so we
0:11:23 tried to dry it and then redid the deposition. However, once again, water was still evolved with,
0:11:31 with the hydrogen
0:11:35 analysis of the X ray absorption. Spectroscopy for the nitrogen K edge showed the differences in
0:11:42 nitrogen bonding between the titanium nitride nano materials. The black curve shows the titanium
0:11:48 nitride as um the nanoparticles. While the titanium nitride with water. What the water contaminated
0:11:57 hydrazine is labeled the wet curve in blue. And then our attempt to dry the hydrazine is shown in in
0:12:07 red and we can see that even when after our attempt to dry the hydrazine, the hybridization of the
0:12:14 nitrogen atoms is still different than that of the than commercial grade nano thai nitride particles.
0:12:25 In our future work, we're excited to develop tie nitride films using ammonia instead of hydrazine.
0:12:32 And this was just installed into our lab and there are a number of papers in the literature that
0:12:38 show that um tonight tried L. D. With tickle or td matt paired with ammonia has is able to be grown
0:12:46 and and demonstrated in the literature. Um one tricky thing is we need to keep the deposition
0:12:53 Temperature lower than 200 in order to keep the hydrogen in the middle hydrated. And so that would
0:12:59 be a little bit of a challenge. But in this example of a hydrogen storage material, L. D. Has
0:13:07 demonstrated the change to change the functionality of the catalyst by first blocking the reaction
0:13:13 pathway and then second by increasing the surface interaction of the plasma sonic material and the
0:13:20 metal hydrate.
0:13:24 Okay, so now I'd like to talk about another example where L. D. Films can act to change the mass
0:13:31 transport of species to the catalyst surface. This project is working with the esposito group at
0:13:38 Columbia University as part of a larger project led by Shane ardo at UC Irvine. So this image here
0:13:47 shows how we might think of mass transport of reactions to in an electrochemical reaction such as
0:13:56 the hydrogen evolution reaction or h er might occur during photo electrolysis. So the reactant are
0:14:04 here in this bulk electrolyte. They transport to the diffusion layer through the oxide layer and
0:14:12 interact with the catalyst layer and then the products that need to defuse outward. Oftentimes this,
0:14:19 this oxide layer is used to enhance the stability of the catalyst layer.
0:14:28 Okay, so now think of two chemical reactions occurring in the same electrolyte and we want the
0:14:34 coding to enable the hydrogen evolution reaction but not the second reaction. This is our challenge
0:14:41 now in this project to determine the material properties of this oxide film. To engineer this
0:14:50 functionality
0:14:54 and such a challenge with two reactions in the same electrolyte is a situation for a Z scheme photo
0:15:00 catalytic water spilling reaction. So two particles in the same solution are doing different
0:15:08 reactions. And the key to the transport is the mediator species here, if we focus on the catalyst
0:15:17 particle for the hydrogen evolution reaction, we don't want the occident mediator to be reduced with
0:15:25 this particle
0:15:28 here, we can simplify The two D particle situation by focusing on the HDR reaction using a thin film
0:15:36 with a flat platinum electrode and a T. I. O. To coding. So this flat plate model provides an easier
0:15:45 growth surface to create a continuous T. I. 0. 2 or T. I. O. X. Over layer with uniform thickness,
0:15:52 it provides a well defined electric catalyst surface area, which is easier to characterize and
0:15:57 determine the oxide film properties. And it's also amendable to modeling the 1D transport problems
0:16:06 that may arise. So this is the system that we will use to optimize and understand the optimal
0:16:13 conditions for growing and oxide coating. We have grown these films by L. D. Using tickle and water
0:16:21 for a range of different temperatures. The thicknesses of the film here are measured by a lip
0:16:29 symmetry um using a silicon witness sample. And here are the synthesis details. Here we're using a
0:16:39 custom made reactor and our growth rates seem to be reasonable as a function of temperature. The
0:16:49 density of the film will play a large role in determining the mass transport characteristics of the
0:16:54 film. This graph on the left here shows the X ray reflectivity of The T. 0. 2 films that are grown
0:17:04 as witness samples onto silicon substrates. And this is done with an understanding that the density
0:17:11 may vary somewhat when the films are grown directly onto the platinum electrode surface. We're
0:17:17 currently working on determining the density of these films. Um using X. R. R. So in the graph on
0:17:24 the right, um the red data here shows the density determined by X. R. R. For the different films
0:17:32 that are all Approximately five nm in thickness but grown at different temperatures. And then also
0:17:41 plotted in blue. Our values from the literature, from Percy at all, who also use tickle and water.
0:17:48 And you can see that there is a good agreement and the overlapping temperatures. And this provides a
0:17:54 range of synthesis temperatures where we can control the density of these films. And then by
0:18:02 additionally changing the thickness of these films, it can add more variable, more variation to
0:18:10 create an even larger range of densities for the oxide films in order to affect the transport of the
0:18:16 reacting species. Next, we want to test the performance of our coatings and it's important to
0:18:23 determine that the coding has not eliminated the transport to the electrode and the coated electrode
0:18:30 is still active for hydrogen the hydrogen evolution reaction. And I would like to acknowledge that
0:18:37 all of the electrochemical characterization showing this talk was done by robert Stinson at the at
0:18:43 Columbia University. On the left here, we see the current density versus potential for different
0:18:51 oxide thickness is um and we have the bare electrode and then various various thicknesses. And
0:18:58 here's the HDR region. You see that The bearer platinum electrode as well as the thinnest layer at
0:19:06 one five nm has the highest activity. However, when we skate, when we expand that this same data
0:19:13 here on an expanded on the right, um you can see that it it still is active and we do have current
0:19:21 density and this current density actually scales with the thickness of the oxide film. So now we've
0:19:30 established that the buried platinum interfaces electro chemically active. We can now include the
0:19:37 mediator species to the electrolyte to see if the films are selective
0:19:44 In these next experiments, iron is added to the electrolyte and there is a desired hydrogen
0:19:50 evolution reaction and then an undesired iron three plus reduction that can occur and we compare it
0:19:59 to the bear platinum electrode versus the coded the T. I. 02 coated platinum electrode. This shows
0:20:09 the current density versus the potential. And we can see here this is the Bear platinum electrode as
0:20:15 well as the thinnest oxide coating. And we can see in this regime this is evident of The Iron three
0:20:23 reduction, but for the thicker oxide coatings, There is no iron three reduction on the right side.
0:20:34 Here is an expanded view of the data showing this, this large difference in the behavior of
0:20:42 different codings. And this is actually quite remarkable when one considers that the undesired
0:20:49 reaction has a large thermodynamic advantage over the desired hydrogen evolution reaction
0:20:59 for negative potentials where both the H. R. And the Iron three reduction are occurring. We can use
0:21:06 these linear sweep altimetry curves measured to estimate the reaction selectivity of these
0:21:13 electrodes toward the desired er for each thickness of the oxide coating, the bare electrode here
0:21:21 shown in purple along with the thinnest 15 nanometer coating shows these selectivity here. Yet For
0:21:29 films greater than two nm, the selectivity is much greater and it's estimated to have a three and
0:21:38 then a nine times increase compared to the bear platinum electra. Um And so we are still working to
0:21:46 correlate the material properties such as density or porosity. To understand how the transport
0:21:52 properties of these films can be optimized and eventually the sort will be applied to particles. For
0:21:59 this Z. Scheme PC application we will grow our oxide films directly onto the catalyst particles. So
0:22:08 this coding work is really part of emerged larger effort led by Shane ardo with many great
0:22:13 collaborators that involves a catalyst development measuring electron charge transfer and particles
0:22:20 and reactor modeling In summary um L. D. Is a well suited growth technique to control thin film um
0:22:30 and material properties that in turn determine the catalyst performance. In the gas storage example,
0:22:37 an l dioxide coding was used to uniformly block possible electron transfer from a plasma sonic
0:22:43 material. And L. D. Was used to increase the interaction of the catalyst with the metal hydrate and
0:22:51 then finally through good control of material properties that L. D. Offers specifically the density
0:22:58 and the film thickness. The selectivity of reactions can be controlled.
0:23:05 And finally, I would like to acknowledge my colleagues who contributed to this work from an real and
0:23:13 also from Columbia University, as well as my collaborators in both High Mark and hydrogen, as well
0:23:21 as as funding from E. R. E. So thank you for listening to my talk.