All-Organic Low-Temperature Molecular Layer Deposition on Particle Substrates 

Session Notes:


Tyler J Myers, BS

4th year PhD student at the University of Colorado Boulder   



0:00:01   everybody thank you for tuning in today and thank you for June and for hosting this first ever p a.
0:00:06   L. D Summit and indulging us all over hang out at home and missing some of those typically well
0:00:11   attended meetings. It's great to see how many people are registered for this summit. This is
0:00:17   wonderful. I'm excited to have the opportunity today to speak with you about molecular deposition on
0:00:22   particles. So we're taking a little bit of a different direction from many of the ale. The talks
0:00:26   that are included today I'm only can be quite challenging due to precursor Chiotis using typical
0:00:32   well vapor pressure, organic precursors becomes even more challenging. And we start to implement
0:00:36   this into a high surface area system like with particles. And we even have something to do this at
0:00:41   low temperatures as well so that it could be implemented into some industry like pharmaceuticals,
0:00:47   where you may be too cosom, thermally sensitive powder. So taking all that into consideration, we're
0:00:52   gonna be discussing considerations for reactor design today to do just this type of process
0:00:58   discussed the characterization of an MLT system that we decided in a matter particles and then also
0:01:03   a brief in yet on using Amel Dakota Street drug delivery applications. So what is not your layer
0:01:11   deposition? If you are familiar with a healthy, this should be pretty intuitive. It's essentially an
0:01:16   analogous process. Toe L. D. Instead of depositing atomic layers were gonna be positive molecular
0:01:22   fragments in this case. So it's still the same process where we use sequential self limiting surface
0:01:27   reactions to create these materials on on the substrate. Good stuff we're doing on Nana media level
0:01:34   control instead of ancient level control. In this case, so referencing this figure here essentially
0:01:39   the same process start out with the surface bringing precursor A because a reacts with all of the
0:01:45   surface sites that are there. And still they are all saturated and precursor A can no longer reactor
0:01:50   the surface. That's a self limiting nature that we all know what above bringing figures would be
0:01:55   that reacts with all surface size that were then generated from dozing precursor a reaction. All of
0:02:00   those And now we have regenerated that original surface we have deposited wanna molecular layer of a
0:02:05   material.
0:02:08   So I mentioned that there our considerations to think about when you are doing a multi. So we
0:02:14   designed a reactor just for doing low temperature, whole organic and multi. I'm particles took a lot
0:02:20   of things into and dip layered to make this design work for us. So the first consideration that we
0:02:26   looked at was condensation of vapour pressure again and precursors. It's very easy for these a
0:02:32   condense and say cold spots in the reactor. In the past, we have had lots of issues with these
0:02:37   organic producers convincing years of cold spots or areas of low conducted its. So the first facet
0:02:45   of the reactor is the entire reactor, and all of its components are how is then a nice with thermal
0:02:50   enclosure. So this is essentially just a fiberglass of it. Holds all of the reactor components,
0:02:55   gives all the same dump here to eliminate the possibility of conversation. Do cold spots. In
0:03:02   addition to this, there are holes in the base plate so that we can plumb in inorganic precursors if
0:03:07   we really needed to, or crackers have much higher vapour pressures.
0:03:16   Second consideration is well precursor conducted since the chamber. We do wanna be utilizing these
0:03:21   precursors as efficiently as possible, so in order to achieve that we have very straight direct
0:03:27   precursor dozing lines right into the chamber. And the chamber is relatively low volume. What you
0:03:34   will you see in the some of the schematics that are that are included a way to run. Now, this
0:03:39   chamber, when I say relatively will buy you, means that we do. We have a little bit of space open
0:03:44   for scale up if if that is something that is necessary for any process that we are doing In addition,
0:03:50   these lines don't have any elbows in them. So we don't have any areas where we're going to limit
0:03:54   that conduct. INTs and each of the precursor lands do have their own perfect and evacuation
0:03:58   mechanisms between the pneumatic valves to help combat this as well.
0:04:05   And the final considerations that we are doing this on particles. So we do need some way to agitate
0:04:10   his particles somewhere to disrupt it and continue to expose the surfaces while the precursors are
0:04:16   dosed into the chamber. Now, the way that we do this, this is a rotary reactor. So what we have is a
0:04:22   magnetic feed through that is in vacuum. Which attach is to a cylinder and the chamber on the
0:04:28   cylinder houses. All of the particles that we're using for the experiments and the future is rotated
0:04:34   by a set of magnets. So as that feeder rotates and rotates canister that is holding all of the
0:04:39   powders, essentially disrupting the powders and agitating them so that we can have, we can have the
0:04:46   surface is continually exposed to the precursors.
0:04:52   So taking old cast some schematics. Here on the left side, you'll see a from a bird's eye view top
0:04:57   view of the reactor, and on the right, there is a cross section on the left schematic. You can see
0:05:02   the very direct precursor lines. They go straight into the chamber, and they are straight lines next
0:05:07   to the bear times of the left. There is a port for a considerable thermal couple probes, so that's
0:05:12   gonna sit right near the canister so that we can check the temperature. Be very accurate. Make sure
0:05:17   that we are at actualized with thermal temperature compared to the other. You will see gay valve to
0:05:23   isolate from the pump and those kind of cylinders in the back and century. On the top of this
0:05:30   schematic. Those are the magnets that are gonna turn. Turn the few through that is inside the vacuum,
0:05:36   and if you look over on the cross section, you can actually see the feed through. This is the
0:05:41   straight kind of shaft that goes into the chamber and attaches to that canister inside that chamber.
0:05:46   That's where all of the powder is going to be. House. You may also also see a small VCR port just
0:05:53   above where the shaft of the feet through is. This is gonna be for a conducting limited evacuation
0:05:59   line so we don't disrupt the powder essentially, during the initial vacuum come down. You may also
0:06:07   notice that the reactor is on an angle, so it's tilted up at a seven degree angle. This is kind of
0:06:13   the magic angle that was experimentally figured out. In this case, we're using an open ended
0:06:18   canister to house the powders. Many of our rotary system before used a porous canister. Since we are
0:06:24   using sticky will, vapor pressure, organic precursors, we decided the best to go with the open ended
0:06:30   so we don't clock up those pores. So the seven degree angle helps keep powder inside the canister
0:06:35   without having all just sit in the back and not continue to be agitated throughout the process
0:06:43   for additional agitation. We did design this powder scraper. We were having some issues with powder
0:06:49   sticking to the wall of cancer. Whether do the static electricity or that was due to the M. O. D.
0:06:54   Chemistry and gluing It's the wall. So this product scraper attaches to the front conflict of the
0:07:00   reactor and sits right inside the canister, the tip right at the top, so very close to the top. So
0:07:06   as the cancer is rotating around, the scraper is going to catch any of the powder that was stuck to
0:07:11   the wall and not get back off. So just a additional conditional mechanism of agitation. And here
0:07:18   we're showing you the actual s of their own closure on the left is the oven closed? So this is what
0:07:23   it's gonna look like when the reaction I was actually happening on the right is the reactor inside
0:07:29   the enclosure with the enclosure open, just to show you how the reactor is sits inside of the other.
0:07:37   So now let's move on to the actual chemistry that we decided to demonstrate here, and the system
0:07:42   that we decided to to go after is an organic polly and MLD system. So here we're gonna be using a
0:07:48   deep local ride and ethylene di amine to create the so called nylon six to polymer. Now, this may
0:07:55   look familiar to you. If you're familiar with MLD, the nylon 66 power for the typical nylon that
0:08:01   people think of has been deposited in the past. We figured because of that, this would be a good
0:08:06   system to try and demonstrate and establish a baseline for particles. We know the pallium it's gonna
0:08:11   be done. We know they have favorable thermodynamics, and these two producers have relatively high
0:08:17   vapor pressure in those. Well, so I was just a good but we thought easy system to try and implement
0:08:23   into particles substrates.
0:08:28   So since we are dealing with molecular fragments here and not atomic layers, the growth mechanism is
0:08:32   gonna proceed just a little bit differently. So on the left here, I'm showing you ask 67 degrees
0:08:37   Celsius. What? The growth of the poly and it looks like so over a number of some of these cycles, we
0:08:43   get about a four ancient per cycle growth rate for the center of the system. Now you're wondering.
0:08:48   OK, that seems a little low on the right figure here. This is one ideal model air. Would it look
0:08:55   like if we were depositing a polymer? We would expect to just go straight up, kind of like a forest
0:08:59   of Palmer. And in that case, the ideal model later would be around 18 extremes. So if we're
0:09:05   depositing straight one on top of each other and continue to grow straight up, we should, in theory,
0:09:10   get 18 ancient per cycle of growth. Three. However, we don't see that
0:09:17   because the polymer is going straight up initially. Eventually, it's going to collapse under its own
0:09:22   weight. We're gonna have bond rotation, and we're going to end up with something that looks like
0:09:27   what we see on the bottom right of the slide here. This man ideal amount aware, so resembles more of
0:09:32   a plate of spaghetti than it does a forest of of polymer. And that's what's going to end up,
0:09:38   resulting in this some on a layer growth.
0:09:43   Even so, we can look at different temperatures here, and even with that interesting growth mechanism,
0:09:49   we do still get when you're growth rate over a range of temperatures. So here we're showing two
0:09:53   different curves on. This was done on Sacconi, um, dioxide witness wafers and then submitted for ex
0:09:58   situ analysis with extra reflectivity and spectroscopic ellipse summitry I'm over showing. Here are
0:10:04   a few different growth rates and different temperatures. The green curve is the same that I showed
0:10:10   in the last lead. So that's that for ancient percent growth, three as 67 degrees Celsius and I'm
0:10:15   also showing to other temperatures here as well. This is a temperature above and below that 67
0:10:21   degrees Celsius. So above 50 67 we have occurred in the red, which is 90 degree Celsius. We get
0:10:26   about 1.3 ancient per second growth rate and below 67. We use 57 degrees Celsius and we get about
0:10:33   1.6 ancient percent growth rate. That's his blue curve here. Now it's interesting to see that both
0:10:38   above and below that 67 degrees Celsius we see a decrease growth rate.
0:10:46   Looking at the data just a little bit differently, we can see some very interesting temperature
0:10:50   dependence on the throw three here so again that for ancient per second growth rate is seen at 67
0:10:55   degrees Celsius, which is a peak in this curve here. But if you go above or below that temperature,
0:11:01   you get a decrease growth rate. Now. This isn't so surprising for the higher temperatures. Many m of
0:11:07   these systems in the past of demonstrated that higher temperatures will result in a lower growth
0:11:13   rate, and this is due to the short residents. Time of the precursors on the surface is not enough
0:11:19   time for the reaction to proceed before Diz Orbs that many haven't with that the the weather
0:11:25   temperatures. And in this case we do see a D whose birth rate that we're attributing Teoh not quite
0:11:31   reaching the thermal activation barrier needed for the reaction. Proceed, We did attempt to do some
0:11:38   experimental both if you degrees Celsius. We had a negligible growth. Didn't really see anything
0:11:44   happen here. Another interesting thing that look as that the decrease seems to go more rapidly on
0:11:50   the lower side of the temperatures and it does on the fest are on the higher side. Excuse me,
0:11:57   suggesting that maybe the thermal activation barrier plays more of a role in this from the system
0:12:03   that a residence time does that case, which is very interesting to see these trade offs those we
0:12:08   start to see maybe an MLT window opening up here and you get these trade offs here that you are
0:12:15   using low vapor pressure precursors. So maybe a higher temperature would help you out in terms of
0:12:21   getting vapor pressure in terms of better purging. But you're going to sacrifice your growth rate
0:12:25   for that higher temperature and those better vapor pressures.
0:12:30   Now, of course, we have good growth we have when you're growth. You wanted to make sure that we're
0:12:34   actually depositing the correct composition and the right polymer that would be hoping to hear.
0:12:39   We're looking at an XPS survey scan of a zirconium dioxide witness. Wait for that. We deposited and
0:12:45   I won 6 to 1, referencing the polymer down in the right hand corner. Here. What would be expected to
0:12:52   see in this XPS is oxygen, carbon and nitrogen. And that is exactly what we see in the SPS. Very
0:12:58   large signals from those three elements are interestingly, we do see a small P for Korean as well,
0:13:06   popping over to the table and working out the expected at Howard Percentages there relatively close
0:13:11   to view, looking at 16.7% for both oxygen and nitrogen, 66.7% for carbon. And of course, if this
0:13:19   proceeded in a perfect fashion, we wouldn't have any Korean in the film at all for a little bit on
0:13:24   the low one for the oxygen and nitrogen and a little bit on the high end for the carbon. We are
0:13:29   attributing that increased amount of carbon to add petitions. Carbon. This wasn't ex situ analysis,
0:13:35   so the sample did have to be transferred from the MLB reactor over to the XPS. There are a couple
0:13:44   reasons why there might be Korean in the film. First of all, we are operating in a static dose
0:13:50   regime here. So the precursors are sitting in a chair for quite a long time, and it has been shown
0:13:56   that there can be diffusion of precursors into these polymer films. So what likely happens is we
0:14:02   don't send a typical chloride into the into the chamber, sits in the chamber for some predetermined
0:14:08   amount of time, cut off from the vacuum, and we have some dual core I'm diffusing into the MLD film
0:14:15   now sufficient purging may end up, may end up taking out all of the diffused precursor, but that
0:14:24   would be pretty difficult to get all of it out in the end. Another reason we might have Korean in
0:14:29   the film is we are operating and stuff on away or growth here. It's possible that some of the
0:14:35   polymer chains are not totally reacting 100% so it might have some chains in the film that are
0:14:41   terminated after in difficult chloride does, which would mean the end of the polymer chain is a
0:14:46   Korean it corn. From that and then also with the nylon 66 works show that there could be possible
0:14:52   core insults that are incorporating to the film as well. So a lot of reasons why there might be some
0:14:57   Korean in the film. Thankfully, it is relatively low in this case, only at about 1.3 percent. So the
0:15:04   XPS was the end of the characterization witness wafers and our moving to characterization on powder.
0:15:09   How the first we did here was FBI are at the end of the end. Zirconium dioxide, you know, powder,
0:15:14   the Emily was quoted on the powder and the president to Thompson grid and put it into through the IR
0:15:19   instrument we're looking for is a few different functional groups. Here I have those color
0:15:24   coordinated from the reference polymer on the left and then where they correspond to in the actual
0:15:29   FT Iris scans. So it would be looking for are the carbon hydrogen, the Amit and then the nitrogen
0:15:35   hydrogen as well Taking a look at the highway of number region Here we do see the nitrogen hydrogen
0:15:41   stretch coming in at just under 3300. So at 30 to 90 wave numbers now we do see the carbon hydrogen
0:15:48   stretch coming and just below 3000 way of numbers as well. So good to see those moving down to the
0:15:54   low way of number of region. We do see a number of modes that correspond to the to the Amit
0:15:59   functional group. So the amad carbon double bond action stretch and then the carbon oxygen, nitrogen,
0:16:07   hydrogen bending road as well as a sanity check. We didn't make sure that these were the same
0:16:11   functional groups and same wave numbers that we saw in the nylon 66 work, and they are so way were
0:16:19   pretty happy to see that those were those were consistent with each other. Find a little bit of
0:16:25   characterization of the film in Terms of Composition is a TDs map here of the MLT film coated on and
0:16:32   organic substrate. This organic strip straight was first coated with a lady aluminum oxide, and we
0:16:39   do see very clear layers of the polymer of the aluminum oxide and of the substrate from that
0:16:45   aluminum where is very nice and showing the separation between the two and you can see where the
0:16:49   polyamory film is due to the very clear nitrogen and Corrine signals in this case as well. Again, we
0:16:57   do see, according to the film. But the atomic percentages that were backed out with with this
0:17:01   experiment showed only 0.8% according so, you know, lower than the XPS. But it's nice to see that
0:17:07   the composition as deposited on a powder, is the same composition and consistent with the XPS survey
0:17:12   scan. So now, moving on to actual images of the Kennedy family wanted to make sure that the growth
0:17:19   on the on the particles was consistent as well with what we saw in the witness waivers. So here I'm
0:17:26   showing the TM of the M L D on a zirconium dioxide nanoparticles. Glamorous. So this was 60 cycles
0:17:32   of MLD performed at 67 degrees Celsius. Now, if we take that original growth rate of four ancients
0:17:38   per cycle that we saw a 67 degree Celsius on the witness wafer, we would be expecting around 24
0:17:44   Manny meters of M o d film in this case. And that is exactly what we see here. Maybe a little bit
0:17:50   more on the edges, but the outer edges so very confortable in smooth film where they thickness,
0:17:56   that's consistent. Without four action per second birthday, we would hope to see a 67 degrees
0:17:59   Celsius. Now we do see a thicker film and some areas of low conduct ins. As you can see where kind
0:18:07   of the glamorous are coming together, there's a much thicker film there That's unsurprising, I think,
0:18:14   and sufficient purging high temperatures could likely combat that at the end and a sacrifice for
0:18:20   growth rate and first cycle time as well. But this would suggest that maybe and all are gonna gamble.
0:18:25   These system isn't the best for doing Cody's in high aspect ratios, which I don't think anybody
0:18:31   would be would be telling you to do that at this time. Anyway. We also find that there's consistency
0:18:39   in the control over a number of cycles. In this case, I'm showing 10 psychos, 30 cycles and 60
0:18:45   cycles again, all the positive at a 67 degrees Celsius. So taking that same for ancient procreate,
0:18:51   you would expect we'd be looking for foreign anti meters for 10 cycles 12 90 million films for 30
0:18:57   cycles and then 24 90 meter films for the 60 cycles. That is exactly what we see here on the very
0:19:02   first figure figure. A. We do see about four nanometers around the outside of these nanoparticles.
0:19:08   Very nice and weaken formal film. It looks really nice moving over to day 30 cycles. Again, we do
0:19:14   seek about 12 millimeters, which is what we would expect, and then seeing the 60 cycles is the same
0:19:20   image that I showed in the previous slide. Here. We do get that 24 manu meter per cycle growth three
0:19:25   here. So not only can we do this Emily system on powders, but we do have, and though the control
0:19:30   over the thicknesses and get to quite substantial thicknesses of this Palmer as well.
0:19:37   So, in an attempt to get around some of the agglomeration issues that we're having with the
0:19:41   nanoparticles, we decided to attempt this on some larger powder. So here we're doing the empathy on
0:19:46   a cellulose polymer. Micro powder thes are quite large particles on the hundreds of micron size. We
0:19:53   did the same 60 cycles as 67 degrees Celsius, hoping to see a 24 90 meter film with that for action
0:20:00   percent growth. Three and again that is will receive. Here is a 24 90 meter film around the outer
0:20:06   edges of these particles conform all coating that agrees with the witness way for measurements. And
0:20:11   what's really exciting is that we're showing consider CIA both across size and composition of
0:20:17   substrate that we're trying to do this and all the system on. So we're not showing any were nuclear
0:20:22   a Shinto ways or something like that, from going from an inorganic to an organic powder final
0:20:28   particle substrate that we decided to test this system on is a Matt Foreman powder, and this is an
0:20:34   anticipation of moving towards some industry application, particularly the pharmaceutical industry.
0:20:39   So if you're familiar, metformin is an aural diabetes medication. And we essentially thought that if
0:20:44   we could coat this system onto an actual active pharmaceutical ingredient, this could pave the way
0:20:50   for MLD actually being used for for something like a drug delivery coating. So we did again the same
0:20:57   60 cycle of amoled, the A 67 degree Celsius. And I'm sure you caught on. By now, we would be looking
0:21:02   for a 24 Nana meter film on these powders. And again, this is what we see you. We see about 24 90
0:21:10   meter film fairly smoothly conform on the outside of these particles. So not only do we have the
0:21:16   ability to coat inorganic particles polymer particles, but we can also call molecular solids with
0:21:23   this MLD system as well. And I think this is very exciting to not only be able to do the M of these
0:21:29   systems are a pharmaceutical powder, but to also demonstrate that system at a low temperature, which
0:21:35   may let us do these types of coatings on third with sensitive powders as well.
0:21:41   Now, the reason that we would want to use MLT on these types of pharmaceutical powders is for an
0:21:48   application and, say, drug delivery. So there are plenty of Palmer's that have been used for drug
0:21:54   delivery applications for many decades. Now, at this point on, there are a ton of sophisticated,
0:21:59   sophisticated systems that are used for direct delivery as well. There's not enough time to go
0:22:04   through every single type. I will only briefly talk about these two here and really only spend a
0:22:11   good amount of time on the U. F. One the time release. So there two types that I discuss the time
0:22:16   release on the left is to kind of eliminate the having to dose over and over again, a meditation so
0:22:24   typically taken medication. You have an increase in the concentration of the drug in your
0:22:28   bloodstream that passes out, you take another dose and you get this a Silla torrey fashion of
0:22:33   concentration of the drug in your body. Instead, we can have a time release system where the system
0:22:40   keeps a control. Standard amount of concentration in the blood had a particular for a particular
0:22:47   amount of time, and that way you don't have to continue taking medication over and over again at the
0:22:51   touch of systems are typical used for these time release methods are either a monolithic polymer
0:22:57   drug system, meaning that the polymer coating and the drug are homologous in this case. So as the
0:23:04   Palmer is dissolving, it is also dissolving the drug into the bloodstream as well. Now the other way
0:23:10   is to have just a simple polymer coating, and this is going to be a diffusion based mechanism where
0:23:15   the a. P I is diffusing through the film at a particular rate that's determined by the thickness of
0:23:20   that coating. Now the other interesting applications with stimuli responsive films. These are
0:23:28   typically more complex films as well. So we're looking at hydrogel is my cells, maybe some Marasco
0:23:33   polymers and crosslink networks as well, and you can imagine all different kinds of environmental
0:23:39   stimuli that could change the morphology of these of these methods. So something like a pH change
0:23:46   going from, say, stomach to intestines, a temperature change, maybe ultrasonic or electromagnetic
0:23:52   stimulus as well. You can imagine a my cell hitting some difference to be a lie. It's morphology
0:23:59   flipping around and really see the drug that way. But due to the actual current state of MLD, I
0:24:07   think the best way to look at is the time release applications. So these more simple polymer
0:24:14   coatings he's more monolithic polymer drug type type methods and one in particular I would like to
0:24:22   discuss is this reservoir methods. So this is essentially the simple polymer coating that I
0:24:27   mentioned before. So you essentially have a pharmaceutical ingredient and you have an MLB film. That
0:24:33   is, it's coating on the outside of this AP I. So as time goes on the AP, eyes going to diffuse
0:24:41   through the MLT film into the blood stream has some controlled rate that's based on the film
0:24:47   thickness. So as time goes on and concentration of the AP, I inside the film is going to decrease as
0:24:54   it's released into the body, but that this would be controlled by the MLB thickness. So if we're
0:25:03   looking at what the expected release profiles, but it looked like we'd be looking at something that
0:25:07   should look like the figure on the right here. So we would imagine that a thicker MLD film would
0:25:13   result in a slower drug release because we have a larger distance for the A P I to diffuse to
0:25:20   actually get to the bloodstream. So looking at these three red curves, the top curve so the fastest
0:25:25   regulars would be your fin ist MLD film and the bottom curve where you have the slowest virtually so
0:25:33   that would be your thickest motu film. So you increase the thickness of the MLT film, you would have
0:25:38   a much markedly slower, really something drug. Now this is corroborated with some literature already.
0:25:46   So there's a nice our journal article and come communications by the only group in the Netherlands
0:25:51   where they use an ability system to coach some protein powder in this case, and then look at the
0:25:56   control were released of that protein into a solution. And as you can see, just as that, those
0:26:04   figures suggest in the previous slide here that a thicker MLD film will give you a solo release of
0:26:11   whatever family film is coating. So, looking at this figure here, the red dots are going to be a
0:26:17   bear protein in this case, and then you go from two cycles to six cycles to 10 cycles of penalty on
0:26:24   the protein powder. The two psychos of any of these shows the fastest release other than Bebear
0:26:30   protein and then as you increase the thickness, increased the number of cycles of the MLD. You do
0:26:35   get a slow release of the protein into solution. So this does corroborate that there is quite a bit
0:26:42   of potential for using about the to control drug release profiles. We are working on some some
0:26:48   testing of our films as well to get some figures and curbs that look like this. So we're very
0:26:53   excited to to see what those what those released profiles they're going to look like in the end,
0:27:01   just a sum everything up first. I believe that reactor's design is quite important for doing such
0:27:07   English application for something that is is difficult to do as a low temperature MLD on part of
0:27:14   those substrates. So there a lot of considerations taken place when you're working with a process is
0:27:19   very fickle, and making sure equipment can do it is I was going Teoh Onley benefit you in the end.
0:27:26   The second is we're able to deposit films with expected compositions using this low temperature. I
0:27:32   saw thermo approach, so we're getting the right composition of Palmer and we are getting good growth
0:27:37   rate as well. Third weekend do control deposition of an MLT film on particles. So we showed. And not
0:27:45   only can we do out over a very number of cycles, but we can also show control and consistency of the
0:27:51   system over a various number of sizes of particles and composition of particles as well. And then
0:27:57   finally, we believe that this could have applications in the pharmaceutical industry. There is a
0:28:02   literature showing that Ahmadi can do some extended release of a P I. We're hoping that our films
0:28:08   may show something similar when we are done testing them as well. So with that, I'd like to thank
0:28:14   everybody for listening. I hope I was able to get everybody's questions in the Q and A. Please feel
0:28:19   free to email me or Professor George with any other questions that you might have about this project
0:28:25   and enjoy the remain of the summit. Thank you