Particle Atomic Layer Deposition:  

Particle Functionalization by ALD: Early Years & Barriers to Commercial Interest 

Session Notes:


Professor Alan Weimer, Ph.D., P.E

Professor of Chemical and Biological Engineering at the University of Colorado 

0:00:10   particle ldiot. See you and also the barriers to commercial interest. So just some background on me.
0:00:21   I'm showing on the screen faculty member because I spent 16 years in industry with the Dow Chemical
0:00:26   Company. Uh, where I quote here today, but I know that there was a real powder technology training.
0:00:34   All of my professional career has been with particles either making particles or or functional izing
0:00:41   particles modifying their surfaces or taking particles and fabricating components. And here what you
0:00:50   see is a photograph of me many years ago standing beside a commercial reactor being constructed.
0:01:02   This is a graphite transport to and basically what happens in here. The system runs it about 2000
0:01:08   Celsius, and you have mixtures of carbon, black and metal oxides that react together by Krever
0:01:15   Thermal reduction that form ultra fight sub micron size known oxide ceramic particles was our
0:01:22   objective. So this is really this is the only thing I ever did. A dad actually made money. Otherwise,
0:01:29   for about 90% of my time, I wasted all their money for this. Everything kind of came together. The
0:01:34   business, the technology markets were their a great group of people and was able to take us from a
0:01:42   century of laboratory curiosity to a commercial facility in about about eight years. As you can see,
0:01:49   I spent 16 years at if you receive Colorado. So this is a photograph of the first commercial plan.
0:01:59   It produced 600 metric tons a year of tungsten carbide powder, something like run power. It was used
0:02:05   to make micro grills. The key to this process was that this particular powder was smaller sized than
0:02:14   anything else that could be made at that time. Other materials were made by hydrogen reduction ended
0:02:20   up with larger grain sizes in the fire of the green, the sharper the bits on the micro drills, the
0:02:27   harder the parts and the final, the particle size them Or could ball you could add for a comfortable
0:02:32   carbide COBOL hard metal end up with with fabricator devices that were both hard and and tough. This
0:02:41   process was sold to Sandvik, largest producer of cutting schools in the world back in the late
0:02:46   nineties. They then scale this process up about three times the size and build a plant in England
0:02:56   core Ahmad. And at that time, when this was sold, the majority of micro girls used throughout the
0:03:03   world were produced from this process called Rapid Carver Thermal reduction. I'm not sure what is
0:03:09   happening today. It was very secretive with that particular So in 1996 I then came to see you and I
0:03:18   have have been here ever since. On one of the urine is that you work in is particle Atomic layer
0:03:25   Deposition. Here is a photograph of the four co founders of Alien and Solutions in 2001. Here you
0:03:34   see Steve George on the right. When I first came to see you, I heard Steve give it outstanding
0:03:40   presentation of the LD chemistry. And I was just amazed at the chemistry. And I asked him, Has
0:03:46   anybody ever tried to Kota particle and responsible? I don't know. I don't put a film on a particle.
0:03:52   I don't think so. And so we teamed up, and that was our objective. Karen Weekly was a postdoc in my
0:03:59   lab. She agreed to spend time turning the company off the ground. And this guy here, Mike Masterson,
0:04:07   was our angel investor. He was a CEO of A L Banana for a long time. Many, many years, and he was
0:04:14   actually my office mate in grad school at the University of Colorado. So I got my PhD and see you.
0:04:22   And then I guess that they figured I was decontaminated by Dow Chemical. And so I could come back in
0:04:30   January of this year. I'll be Nana Solutions and Forge Nano merged to create the bigger and better
0:04:37   for Jun'an Oh, which is really the ultimate manufacturer of code of powder coated particles by
0:04:46   Atomic layer deposition. The other thing for sure that you have to know and anybody NLD knows this,
0:04:54   uh, Dr Santa Ola received the 2018 Millennium Technology Prize. That doesn't happen unless the
0:05:03   technology is absolutely outstanding. He is economy acknowledged inventor of L D. This U S path
0:05:10   method for producing compound Finn films and Rica submitted his nomination. Um, he received an award.
0:05:19   So the point I want to make it a L. D is definitely some pretty outstanding well recognized
0:05:25   technology particular in the area flats and semiconductors and solar panels and electronic devices.
0:05:33   And we're going to talk about today is the application of avail Devi's thin films to so one of the
0:05:41   things that I want to make sure identify correctly is the particle lt definition that I have because
0:05:49   it may be different from in general. So here on the left you see particles that were coated with
0:05:58   metal ad Adams in this case, platinum. You cooked the internal surfaces of those particles as well.
0:06:05   I've done the right. You see particles and have a thin film on them. Um, the original work with with
0:06:14   putting nano metal decorations on particles for catalyst was that in Finland, here are a number of
0:06:20   the to the early papers. There are a number of other ones as well, and the point I want to make is
0:06:26   that that is not the focus of this talk were really the focus of my workers. My definition of a
0:06:32   particle ale de what I'm interested in is being able to put these films on particles and without
0:06:41   including the particles together Nana glom, aerated films, the definition here, the focus primary
0:06:48   particles could with non particular ale de films. The particles are not agglomeration, meaning
0:06:54   they're not glued together. You don't see cripple junctions here that particles are not being good
0:06:58   together. You also don't see poorest films, which would result if, in fact, you have fine particles
0:07:05   such as you get by chemical vapour deposition. They would end up being scavenged on the surface of
0:07:11   that particular particle. And in order to confirm and verify that you do indeed have primary
0:07:18   particles coated, you have to do more than just t e. And we definitely need high resolution tm. But
0:07:25   you also need ball characterization of these particular materials. For example, surface area has a
0:07:32   surface area changed. Obviously, if you were to for next among these particles of surface area would
0:07:37   be less than what you had calculating if they're if they're all covered individually. You can also
0:07:44   tell by particle size distribution analysis before and after. If that distribution widens out, if
0:07:51   you basically have formed a Guam rinse and then other methods for the chemistry surface. Functional
0:07:57   groups looking at FDR in XPS all of these together before and after LD In order to confirm that in
0:08:05   fact, you have coded primary particles and they're all coated. So if you look at CVD, the precursor
0:08:12   veil, be if you want to form aluminum oxide by chemical vapour. Deposition. You react. Try meth
0:08:19   aluminum tm A with water vapor. And the challenge here is that you inherently will form particles in
0:08:28   the gas phase on. Those particles will then be scavenged on the surface and you'll end up with a
0:08:34   film that is porous. And so here are some examples of that I like this paper published in 2006
0:08:42   because it was done in the food I spent reactor here on the left hand side, you see liquid face CVD
0:08:48   very thick films 1500 enemy. As you can see the poorest feeling the particles were basically
0:08:54   scavenged on the service. You would have examination from the substrate particle, and here we show
0:09:01   gasping CVD. This is a 200 nanometer thick film and you can also see the pores on that particular
0:09:09   particle surface in that particular film again, the result of particles that are formed in the gas
0:09:15   phase better than scavenged on the surface of the film, I would note that you can get CVD to be a
0:09:21   lot better than this. This happens to be the the Micrografx from showing based upon the food I bet
0:09:27   great actor process, which is similar to what we do for for a Ah, I'm sure Steve has already run
0:09:34   through the fundamentals of the chemistry. But just as a review here as relates to what I will be
0:09:40   saying, you have a surfaced with typically oxide or metal or could be something else, and you're
0:09:49   typically gonna have hydroxy groups in that surface. If it's a nitrate, they could be sort of like
0:09:54   amino type groups and then to carry out the first a reaction. You have dysfunctional hydroxy groups.
0:10:02   You bring in T m A and you convert those hydroxy groups that metal groups you convert him all. And
0:10:12   once they are all reacted, there's no further reaction occurs. It's self limiting surface chemistry.
0:10:18   You then have a new surface with methyl groups. You bring in the water vapor. You convert those
0:10:24   metal Bruce back the hydroxy groups, and in that process, he put that in one sub 71 a layer of
0:10:31   aluminum oxide. And then you basically repeat these steps curating out the baby reaction sequence in
0:10:37   order to grow the film.
0:10:40   So we eliminate the scavenging. We grow films on primary particles of tackling and together Here is
0:10:50   a very nice Micrografx. John Ferguson, his work published to pass it to you. Here he's showing Born
0:10:57   Night cried fields on the surface of zirconium oxide particles. You can see, of course, the atomic
0:11:03   structure of zirconia. Zirconium, oxygen zirconium option is black and white lines. What is
0:11:08   particularly impressive is the fact that every single primary nanoparticle eyes could have with the
0:11:15   25 Angstrom Fillol born. I cried without gluing of particles together, and that is a very, very
0:11:21   significant finding. In fact, you could cut primary particles about going together as a particle
0:11:27   person my entire life. When I saw this initially in some of our Micrografx size, totally amazed that
0:11:33   you could actually pull this pool is. And of course, that was shown by AM Dylan and others carrying
0:11:41   out multiple a bi cycle theory of 3000 B cycles, you can achieve very, very linear in growth rate
0:11:50   viewer of 1.29 anxious per cycle. Here you're looking at a surface roughest of only four. Angst
0:11:56   comes for a film thickness. It is 3860 Angstrom. Pretty, pretty amazing what you could do with with
0:12:03   a L. D. So the idea is, can we do this on a particle and really work focused on a number of
0:12:10   different materials? We looked at Warren. I tried the application. Here was these boring maker as a
0:12:16   thermal filler. It's also insulate er elect chronically, it doesn't conduct. Electricity is used as
0:12:23   a pillar to increase normal conductivity for electronic packages. The problem is that the born night
0:12:30   right there is that there's a limit in its loading into a resin that is used to essentially glue a
0:12:40   chip onto a substrate. Um, and if the viscosity goes up, there's a drop that comes out of a syringe,
0:12:48   and that drop it doesn't flow over the chip to the substrate. And so it's really critical to have a
0:12:54   viscosity that meets the conditions that is gonna be floatable is going to contain enough of the
0:12:59   filler in this case Born. I try to be able to also improved thermal conductivity of that of that
0:13:05   plastic package. We also had a lot of interest in iron particles trying to put barriers that on iron
0:13:12   to prevent oxidation. That application was for still stealth aircraft. Basically, could you prevent
0:13:21   the oxidation of iron without changing the particle size. We had interest in being able to
0:13:26   demonstrate that we could coat polymer particles without gluing them together. So I have some work
0:13:32   to show you here on polyethylene early work. We also did some like earlier deposition films. These
0:13:40   could then be oxidized because the MLB it produces an alley, Kona has an organic component and
0:13:47   organic component. You could remove the organic component and produce a very high surface area film
0:13:53   on a surface. And we also wanted to demonstrate that we could coat the internal pores within hype
0:14:02   ferocity, maybe 85% poorest polymer particles. So if we look at the board and they cried first again,
0:14:12   it has really high thermal conductivity there, used to enhance heat conduction boring Laker is
0:14:18   fairly inert. What I show here in this graph I za silica thickness and angst rooms and how that
0:14:25   relates, theoretically, to the thermal conductivity of a reborn night cry. Players have silica
0:14:31   coating both sides of it, and the idea here is to increase the born nitrite particle concentration.
0:14:39   The company that funded this work was selling Bora nitrite, and they wanted able to increase the
0:14:45   amount of born nightmare that could be letter that they could basically sell MAWR material to the
0:14:50   user. The problem. Waas that the night Cried did not interact well with the coupling agent for the
0:14:59   poxy. And could we could we make that for a night look like an oxide? But do that? We had to make
0:15:07   sure that that film was extremely thin because of thicker. You make the film the lower the thermal
0:15:14   conductivity defective thermal conductivity in that particular system. You also have percolation
0:15:20   issues, but the focus at that time to make the nitrite surface look like an oxide bond with the
0:15:30   coupling agents to reduce the viscosity and maintain thermal conductivity was to have film with her
0:15:37   lesson about one Nanami, or thickness, because the thermal conductivity decreased by about 50% for a
0:15:45   one nanometer film. So the focus of early research back in the early two thousands was for Oprah
0:15:52   Thin films. Born nitrate was a key material because of its thermal conductivity and the fact that
0:15:59   was electric. So there's a challenge in making these offer thin films and Riga her paper, this paper
0:16:09   incited probably now about 1500 times and what she stated there was. In fact, the original first
0:16:19   cycle is going to coat the surface of the substrate. It's not gonna coat it all. It's gonna take a
0:16:27   number of avail de cycles where you are coating basically the substrate and the coating material
0:16:33   until you get to a point based upon the growth rate and the chemistry that you will finally be
0:16:40   growing only on the surface of the NLD. Grow material and great persons recently developed a model
0:16:47   which basically shows this where in fact you laid out your nuclear. You gradually grow this film and
0:16:55   so depending upon the surface chemistries that are involved here, it would take a number of cycles
0:17:01   in order to basically put down that film across the particle service. This suggests that that these
0:17:09   particles and films their semi continuous and you have known uniforms field roof. So if we look at
0:17:16   the work that John Ferguson reporter back in 2000 um, for for born nitrite, what he shows here is it
0:17:25   takes about eight a l D cycles in order to initiate the alumina growth on the entire border nitrite
0:17:32   surface. So here we're looking at about eight cycles in order to start growing aluminum oxide across
0:17:40   surface. Here is Ah Micrografx of the Code of Born. I tried note that the edges and the basil planes
0:17:50   are coated with aluminum oxide, and a reason for that is that tm a behaves like a lewis acid. You
0:17:57   have functional groups on the edges that you can react by atomic later deposition. The basil planes
0:18:02   only have an electron pair this active for Lewis acid. So basically, because of PM a behaving like a
0:18:09   loser, Lewis acid is possible for alumina to basically closed the entire born A crime, however, get
0:18:18   silica is desired as the coating for the outer surface. Silicon tetrachloride will not behave like a
0:18:25   Lewis acid, and in fact, you can see here that you can coat the functional groups on the edges very
0:18:31   nicely, but that the basil planes you end up with patchy coatings. And in fact, it's possible to
0:18:40   have films that are not uniform on the basil planes. But our uniform is, and here we show the
0:18:48   results of this were indeed, by coating the border nitrate particles weekend, we can decrease the
0:18:53   Scots and the It's also possible to make particles here where you put that aluminum oxide first in
0:19:01   order to create a bonding layer and then you can put the silica on on top of that, what about the
0:19:10   iron? So you were looking at at five micron iron particles, And again, the objective is to provide a
0:19:16   a barrier to oxidation to the iron particles Here were showing results of a thermo gravimetric
0:19:23   analysis as temperature has increased up to 265 degrees Celsius for various codings of aluminum
0:19:32   oxide on on the iron particles. And here is the onion could've material. You can see it oxidizes
0:19:39   picks up. Wait. When you do 10 a. L d cycles those particular particles were the film was about 1.3
0:19:47   nanometers. Based upon the growth rate, you concede it oxidizes almost the same as the uncut of
0:19:52   material behaves very similar that, however, when you now go to 25 cycles or 50 or 100 they're all
0:20:00   essentially the same. About a 2.5 nanometer film protects against oxidation is there is a
0:20:06   substantial difference between this real much only 1.3 nana meters, and that's really what's about
0:20:12   2.5 substantially different properties to Nana Miers appears to be about the key film. Thickness
0:20:22   again identified back in early two thousands that the properties of those particular materials were
0:20:28   substantially different based deployment thickness of those films and I think do it particularly the
0:20:34   fact that it takes a number of cycles in order to, um in order to cover that surface and then
0:20:41   effectively provide a very We also were interested in coding polymer particles here you're looking
0:20:50   at at polyethylene 33 micron polyethylene particles with films of aluminum oxide grown at 77 degrees
0:20:58   Celsius. And what you see is that you don't have triple junctions here. Basically, you could do this
0:21:03   without aggregating. The particles are over high density polyethylene, the uncoated polymer, as well
0:21:12   as after 75 tm a water cycles. They grow aluminum oxide again, not showing any on. So the question
0:21:21   was for us was, How is this actually able to be accomplished? Because there are no functional groups
0:21:28   on the surface of the high density polyethylene. But there are pours present, and here you can see a
0:21:34   50 cm. You can see the aluminum oxide on the surface and you can also see some aluminum oxide. It
0:21:41   looks like it's inside. The proposed ailed film growth mechanism John Ferguson came up with was that
0:21:50   essentially you have TME and water going into the pores, diffusing into the pores. And eventually
0:21:57   aluminum oxide grows a grows within the poor and outside of the poor, and that eventually you
0:22:02   basically grow, um, aluminum oxide on top of aluminum oxide. So you essentially have a physical type
0:22:10   of mechanism for bonding. Theologian Mina to those Palmer surfaces that don't have particles, I
0:22:18   couldn't look identical 85% ferocity and do a cross section on these particles coated. And this is
0:22:26   what the inside looks like. If we do E. T s math and you can see the aluminum deposit within the
0:22:33   pores of that particular, if we look more closely on the left hand side here in this particular
0:22:40   image, the aluminum oxide is bright on the right hand side. The aluminum oxide is dark. Here, you
0:22:46   can see a poor, Here's the polymer, and here's the aluminum oxide feel on the pope. It's also
0:22:54   possible to do molecular earlier deposition on on substrates and here we're looking at spherical
0:23:04   silica particles made by the Stober process. And in this process we are basically being able to
0:23:14   create aluminum Alcock side film with value cone in this particular system using F ling glycol in
0:23:22   place of water. So tm ethylene glycol to grow in this particular film. And here you can see about a
0:23:30   seven Nana meter film on the outside of this particular case, a t i 02 particle.
0:23:38   Now when you take that value cone and you can all sign it and here we're looking at in calcite at
0:23:46   four degrees C. In the presence of here that 25 enemy or thick film it shrinks. It's now about a
0:23:53   centimeter sick, but it's extremely porous. And now we have about 1250 meters squared per gram
0:24:00   poorest film fabricated from aluminum Alcock side run film, which is then Cal signed to remove the
0:24:09   carbon. In fact, it's possible to control the size of the pores that are generated in this poorest
0:24:15   film by controlling the Sally Kohn to begin with, and I think essentially the length of the carbon
0:24:22   carbon bonds in that particular. So how is this done at larger scale so the original work was done
0:24:32   with a tungsten grand pressing particles into a tungsten grid passed on, and I are being through
0:24:38   doing FDR analysis surface chemistry. However, in a fluid eyes bed reactor, it's possible to have
0:24:45   many, many particles high surface area inside a crude ice bed reactor. And because of that surface
0:24:53   area, you can follow the surface chemistry with external mass spectrometer. I shouldn't here. Ah,
0:24:59   stir as possible. Use a stir and a small system. It's possible to actually vibrate this entire
0:25:06   reactor chamber or, most likeliest, possible to pulse the gases in order to be able to overcome any
0:25:15   kinds of inner particle forces that may initially form, say, from Vander Waals forces of opera fine
0:25:21   particles within that particular system. I also want to note that you can make giant fluid eyes,
0:25:27   bids FCC and basically all the gasoline producing over cracking down on the hydrocarbon industry is
0:25:34   using giant fluid ice bed reactors that might be 40 you know, 50 feet in diameter. Um, yeah,
0:25:41   absolutely Giant. Most of the polyethylene grown in the world has grown and huge food, I bet
0:25:47   reactors, my gas based politicalization. So food I spend reactors very familiar with. I worked with
0:25:54   them all my life. I believe they are easily scaled up. Uh, for So what are the barriers to
0:26:01   commercial acceptance? Well, two in particular. One was that, um, scientists at the time particle
0:26:10   experts it would have 10 particle conferences were unwilling to believe that you could coat primary
0:26:17   particles by atomic layer deposition without gluing them together. This was a slate as as 2000. For
0:26:26   5 4006 I attended a conference put on by the International Fine Particle Research is to to free Um,
0:26:36   I was an invited speaker in Santa Barbara, California. I gave a talk in early talk on on a L. D. And
0:26:43   was just totally blasted. There was a particle expert there from Europe, well recognized, one of the
0:26:49   best in the world. And he just totally blasted being his conferences, that that's impossible. There
0:26:55   is no way that you can coat individual particles that gluing them together because of Andrew Walls
0:27:01   forces. And so it was clear to me that we had to prove that this was possible because people 78
0:27:08   years after we started doing this work still couldn't believe that you could cut primary particles
0:27:14   by L. D. In this particular manner, I would like to say that I saw this person give a plenary talk
0:27:24   three years later at the party conference in Nuremberg, Germany, and he talked about how great
0:27:30   particle a L. D. Was. And they went up afterwards and I asked him, I said What? You blasted me three
0:27:35   years ago in California. We're talking about this being a fantastic process and he said, Hey, it
0:27:42   took time to sink in. So he became a believer also in what I call particle Ale de. We're gonna talk
0:27:50   more about this and how we can basically prove that you could cook primary particles. The second
0:27:56   period of commercial acceptance, which still present today, Um, for some companies is the fact that
0:28:04   they can't believe that particle oil be is a low cost process with improved cost performance
0:28:09   benefits. So we're gonna talk about two of these aspects as well. So what we did is we took a number
0:28:17   of nanoparticles, Laconia fumed, silica and tight Titania particle size roughly 25 to 40 an animator
0:28:26   surface area around 50 meters square program. We coated those particles by atomic layer deposition
0:28:32   with different thicknesses. We then did high resolution microscopy. I look at the surface is where
0:28:39   they're crippled. Junctions reduce Carter particles primarily coated. However, you can tell just by
0:28:44   looking at a high resolution tm you need the high resolution TM, but it's only looking at a very
0:28:52   small smidgen of the actual sample. So you also need bulk analyses. You need surface area and
0:28:58   particle size distribution of those particular particles were being coated. And then methods also
0:29:06   like FP, our and XPS to be able to also evaluate the bulk chemistry of. So these air the code, the
0:29:15   particles that we started with noticed after, made inflaming, accurate. It's if somebody's particles
0:29:20   are automatically aggregated, you're not gonna break up the aggregates. But the idea is not the coat
0:29:27   the particle aggregated. And here are some of Lisa Kim's work with with fumed silica, different
0:29:37   degrees of magnification noticed that you don't see cripple junctions here. Basically, you have
0:29:43   coatings and all these particles, but you're not gluing the particles together. And in fact, if we
0:29:48   do a higher resolution, look at this. Every particle looks the same. A 5.2 nanometer film on every
0:29:55   one of those particles and people just amazed that in fact, you can. You can pull this off here, we
0:30:03   seize, or Cody and nanoparticles could it with silica. And here we see zirconia nanoparticles could
0:30:11   with titanium dioxide. So basically different systems before I should were nitride films a lot of
0:30:18   different variation to the substrate in the particle surface. It's also possible to design and
0:30:26   fabricate particles with multiple films. Here you're looking at a particle that has three different
0:30:32   coatings on it.
0:30:35   So to look more at the bulk, here's a result of alumina coated silica P i R. Here you can see the
0:30:42   absorbent spectra. Here's the aluminum oxide, um, absorbing spectra and you can see the ball alumina
0:30:51   feature. Uh, the uncoated silica does not have the bulk alumina feature. However, after 20 and 50
0:30:59   cycles, you can see the bulk alumina growing on. You were looking at XPS, and what you noticed is
0:31:07   that we have attenuated the silicon and silica dioxide, pinks, aluminum peaks or presence of
0:31:14   basically we're coding now. Have we have we glue them together? No, the data would say we have not.
0:31:22   Here's a particle size distribution before and after. In fact, if we had Cody's particles ust a
0:31:29   broadening of this distribution, we don't see that indicating is based upon a particle size
0:31:34   distribution. We have not Cody's particles that could articles and surface. Syria tells us the same
0:31:40   thing. Here you're looking at a theoretical 25 5100 Nana Meter film thickness on films and here,
0:31:53   particle size. I'm sorry, film thicknesses and the surface area is predicted, and then you see the
0:31:59   experimental results, and basically the predicted surface area is in line. The experiment was in
0:32:07   line with the predicted surface area, So surface area particle size distribution in line indicating
0:32:15   no next high resolution indicating no next fpr XPS indicating, in fact, the bulk surface are all
0:32:23   covered. So how could you do that? How can you pull that off? That was really a question for us. We
0:32:31   looked more in detail like a food ization behavior. Fine particles on the left. Here you see 150
0:32:39   micron diameter nickel particles a lewd eyes. Individually, they're obviously can be coded
0:32:45   separately on the right. These are seven or 13 micron diameter, boring nitrate particles. There's a
0:32:51   341 Mike Run bar noticed at these particles fluid eyes as dynamic aggregates. So what happens in the
0:32:59   food eyes? Bed is the particles air effectively larger and they stay in the bed even though you may
0:33:05   have nano particles, Um, show in seven or 13 micron for a night right here I'm showing nano
0:33:12   particles. EULEX 50 fumed silica and they will basically fluid eyes as aggregates. Here in the head
0:33:18   space, you have an aggregate of silica fume silica fluid izing of those aggregates will shed
0:33:27   particles and then other agg ress will pick them up. And so we have named this dynamic aggregation
0:33:34   where ultra fine nanoparticles will fluid eyes is effectively larger particles on the order of maybe
0:33:40   500 microns. They stay in the fluid ice bed, they don't get carried out, and in that process is they
0:33:46   move. Besides other aggregates, they shed particles from one aggregate go to Teoh to to the other
0:33:52   aggregate. But in fact, all of those particles are code Exactly the same for those here you're
0:33:58   seeing at five point thine enemy or film of aluminum oxide on the surface of the fumes.
0:34:08   What about the cost What could we do here to take a look at the cost that indicated this is Lou cost
0:34:15   process. So in this case, we ran a large fluid eyes bed six inches in diameter, 12 inches tall,
0:34:24   Curia atomic clear deposition at 350 I degrees at 77 degrees Celsius. One tour pressure and, in fact,
0:34:37   thes five micron iron particles. About 12 kilograms of material. Total surface area, 1200 meters
0:34:45   square. We carried out 22 cycles to put a 2.5 anatomy or film of alumina on the outside of his iron
0:34:53   particles. Basically, we know from the prior TG a work that a 2.5 man Amir film will provide a
0:34:59   substantial oxidation barrier. What always amazes people is that to pull this off on a kilogram of
0:35:07   composite, you have to lay down 1.4 grams of aluminum perfectly. And of course, anybody in a
0:35:14   particle ale The space knows of Dad is a piece of we do with the fluid eyes bid. And because of the
0:35:21   surface area and the reactor, we can follow the surface chemistry. So here, what you're seeing is a
0:35:27   signal from the men from the downstream mass spectrometer. In this particular case, uh, this
0:35:33   obviously the signal is a log arrhythmic scale. The red here is T M A. Um, you're in the noise down
0:35:42   in this. In this location of the signal, you're in the noise. We then start the first dosa tm a dose
0:35:51   aluminum hydroxide, the hydroxide surfaces reacting with the PM, a dose starts here. We're looking
0:36:00   for methane. You see the method for its receding in the mass spectrometer? It isn't until all of the
0:36:09   surface punctual groups are converted methyl groups that all of a sudden we see tm a breakthrough in
0:36:16   the mass spec and the same thing with the second step here we start the water dose. You can see tm a
0:36:25   being form and it isn't until the methane starts to drop that we see breakthrough of the water vapor.
0:36:35   So the indication here is that we can get near 100% effective use of gases on essentially not waste
0:36:44   gases. Yet remember, you can have hundreds of thousands millions of square meters of surface area
0:36:50   food, eyes, bed and the surface of the inside of the bed is minute like it's nothing compared to
0:36:59   surface area on the particles. So if you look at that in terms of a cost analysis, if we want to
0:37:05   make 100,000 kilograms of this composite, these five micron iron particles with A with a film of
0:37:12   aluminum oxide for a 2.5 Nanami or film and the precursor costing, say, $350 a kilogram. It costs
0:37:21   about $50,000 to coat and make 100,000 kilograms of that composite. However, if we have to make a
0:37:28   thicker feeling, the compensate for porous coating, such as you would get in C v D. I pick here the
0:37:34   200 nanometer film because that's what I showed made in a fluid eyes. Bad in one of those early
0:37:39   slides. The cost of that coating is about $4 million. So this difference between $50,000.4 million
0:37:46   dollars is really this cost savings. To be able to put down a perfect, uh, echoing barrier by atomic
0:37:56   layer deposition the path forward Um, what I want to indicate here is the interesting in particle
0:38:06   will be over time. Here I'm showing that between 1999 2000 and 19. We've grown from only a few
0:38:14   citations, a few papers to a roughly 30,000 total citations today. And what you couldn't see is we
0:38:24   have over 5000 citations a year, Basically, exponential growth. Uh, these are our citation numbers
0:38:31   from the rebel science where we we combined the term particle and in quote patients, atomic layer
0:38:38   deposition. So there's a huge amount of interest in particle ale de from the early work back in the
0:38:45   very late nineties. With that, I want to conclude I get Teoh give this nice talk, but the work is
0:38:54   actually being done by my graduate students and postdocs. I have outstanding PhD students, and I
0:39:01   really want to thank them for being three. The doers in all of this work that we've done over over
0:39:10   time, thank you very much, and I'll be happy to take questions. One. This