Fundamentals of Particle Atomic Layer Deposition    

Al2O3 ALD as a Model ALD System 

Session Notes:


Steve George, Ph.D

Professor of Chemistry University of Colorado at Boulder



0:00:04   Hi, everyone. My name is Steve. George and I first like to start by thanking the organizer's for
0:00:10   inviting me to give a presentation at this particle atomic clear deposition summit. Um, I've worked
0:00:17   on atomic layer deposition on particles for quite a long time, and more recently, we then looking at
0:00:26   atomic clear etching. And so what I'd like to talk about today is particle atomic, clear edging. So
0:00:34   we're gonna be taking things off instead of putting things on. So let me first get up the slides
0:00:42   here, um, share my screen
0:00:49   and get presentation up. Okay? Yeah. So, like I said, I have been working for quite a long time, and
0:00:58   it talked in the field of Atomic layer deposition, but it turns out over the last For five years,
0:01:05   we've started to work on the opposite of atomic layer deposition, which is atomic layer etching. And
0:01:11   we pretty excited about the opportunities that are available for by Tami clear etching with regard
0:01:19   to particles. So what I'm gonna talk about is particle atomic layer deposition how to take it off
0:01:24   instead of how toe How to put it on. Okay, what I'm gonna do through the course of this talk is
0:01:32   cover a couple topics and they're listed here. We'll start out and just I'll define a little bit of
0:01:38   what we do. We mean by thermal atomic Layer a gene, which is the reverse of atomic layer deposition.
0:01:44   So we'll just look at the basics of thermal Ailey so everyone is on the same page as to what that is.
0:01:50   And then we'll take a look at aluminum oxide Atomic Layer deposition, which everyone is familiar
0:01:56   with, as well as then the con. The reverse, which is the aluminum oxide atomic layer etching, and
0:02:03   we'll look at that on particles. And then we'll look at a couple case studies where you'll try to
0:02:09   illustrate what happens with atomic layer, etching and what what are its potential? Ah, the value it
0:02:15   that it might have for particles in general and, for example, will first take a look at the removal
0:02:21   of S I 02 from silicon particles using a silicon dioxide ale E. That's one application of removing
0:02:31   films from surfaces. And then we'll also take a look at an example where we could use a lee to make
0:02:36   particles smaller, and this will be an example where we'll start with zinc oxide particles and we'll
0:02:43   try to reduce their size and we'll see, actually that something else happens. That was somewhat
0:02:48   unexpected. But, uh is related all right. Chemistry is known as conversion can occur, and that is Ah,
0:02:56   that's an interesting process in itself. And then we'll also look at selective removal, and we'll
0:03:01   see how potentially that the thermal Ailey could lead to the removal of particular material from
0:03:08   particles. And this could be. And this could be used then for applications in terms of like particle
0:03:13   cleaning. And then my last example will be how to obtain very ultra thin, continuous film on
0:03:21   particles using a technique called Deposited Etch back. Because we know that with a lady, sometimes
0:03:26   we don't get very good nuclear nation right from the great first start cycle, maybe you have to have
0:03:32   2050 cycles, something like that to get a continuous film. Well, you can do that. But then maybe you
0:03:37   didn't want such a thick film. So then you connect it back so that we call that deposit Etch back,
0:03:42   and I think that will have a lot of applications. Also, four particles Okay, So to get started, then
0:03:50   let's take a look at what is the chemistry involved? Anatomically ECI. And so that's what shown here.
0:03:57   It's basically back to back reactions similar to a L. D. These reactions are also going to be self
0:04:04   limiting. Reactions are, hopefully nearly self limiting, so we'll come in with first reacted and
0:04:10   modify the surface of the substrate. So this is a little different than atomic layer Deposition. The
0:04:16   reactor is gonna come in, and it's gonna modify kind of convert the surface into something else. And
0:04:22   then in the second reaction, this is where the interesting chemistry takes place. A second reaction
0:04:28   will happen where we're bringing a different reacting, and it will. It will interact with this
0:04:33   modified surface and that interaction will then lead to volatile species that will actually leave
0:04:40   and remove the modified surfaces. And that's that's what will give us the chain on. And then once we
0:04:45   once we finished with that, then we can repeat the process here and come back at the beginning and
0:04:50   just cycle back and forth. So a b A B cycling back and forth. But now to remove material as opposed
0:04:58   to such material. Okay, Example that I'm going to show at and give you some data for is aluminum
0:05:06   oxide ale E which, of course, is the reverse of the model system Aluminum oxide ale de. And we'll
0:05:13   see that the way we gonna were gonna do the surface modification is with florid ation. So instead of
0:05:20   the typical water in tm a reaction for aluminum oxide ale de, we're going to run a floor nation
0:05:26   reaction first with HF. So this is gas phase a Jeff's gonna come in and we'll see that it can.
0:05:32   Floren ate the surface of the aluminum oxide gives us a couple angstrom thick fluoride layer on the
0:05:39   surface of aluminum oxide. Then, in the second reaction, we're gonna bring in try enough aluminum.
0:05:44   Yes, trying with aluminum, the same trying with aluminum that you would use to grow aluminum oxide
0:05:50   you can use to etch aluminum oxide. And so this is trying with aluminum can come in and undergo a
0:05:57   reaction called Ligon exchange, where the methyl groups on the try meth aluminum can exchange with
0:06:03   the florins in the looming. Um, that's trifle aluminum layer, and that exchange process then can
0:06:09   lead to volatile aluminum species, and we'll see that those are when we looked at this with mass.
0:06:15   Spec. I don't have time to get into too much detail about that. But, uh, dimethyl aluminum fluoride
0:06:21   turns out to be the volatile, the primary volatile species, and so this process then can be repeated.
0:06:27   Floren Ation Ligon exchange back and forth in order to remove aluminum oxide with atomic, their
0:06:33   control. Now the key to a thermal atomic layer. ECI. Whether it's aluminum oxide or any of the other
0:06:41   materials that have been edged over the last 345 years, the main reaction that's that's proved to be
0:06:48   very useful for thermal atomic layer etching is this ligand exchange reaction Ligon exchange. And so
0:06:55   I just show a little cartoon here for ligand exchange. That kind of gives you a little bit of a
0:07:00   picture for what's going on when the try meth aluminum comes in and interacts with the the aluminum
0:07:07   fluoride surface layer, and it's that this has shown course great, with great simplification, just
0:07:12   showing individual molecules. But what happens in Ligon exchange is the flooring Ligon that was on
0:07:18   the aluminum of the aluminum coronated layer. Uh, transfers to the aluminum in the triumph aluminum
0:07:24   and then at the same time, ah, methyl group from the try meth aluminum transfers to the aluminum.
0:07:30   And so that's basically Ah, exchange of Liggins. And what that does then is transfers metal to the
0:07:36   what was the aluminum fluoride layer and a flooring to the Try enough aluminum. Okay, well, now this
0:07:43   dimethyl aluminum fluoride species, it's still volatile, so it just leaves and it turns out that the
0:07:51   aluminum with metal and two florins this is not so volatile. But if a second Ligon exchange happens
0:07:58   to the same aluminum center, then what you do is you basically make time F aluminum fluoride, which
0:08:04   it was the same as the product that you made from the try meth aluminum to start out with. And then
0:08:09   that turns out to be volatile and it will leave. And so that's what the second league in exchange
0:08:14   event on this aluminum center will lead to the etching of the aluminum oxide. And so that's the idea
0:08:21   of Ligon exchange. Transfer the flooring from the metal fluoride to the metal precursor and then
0:08:25   transfer of the ligand from the metal precursor to the metal fluoride is that process of of
0:08:31   transferring Liggins then will allow you to make volatile species that will leave on and allow you
0:08:37   to touch the material. Now it turns out I don't have ah, a lot of time to talk about this. But there
0:08:43   are many metal precursors that are effective for leg in exchange. Now I know why I just highlighted
0:08:49   Try mouth aluminum Because really, I wanted to kind of make a contrast with and she an aluminum
0:08:55   oxide and then ailed de of aluminum oxide. So try meth. Aluminum is a fine precursor for Ligon
0:09:01   exchange, but there are also others 10 ak ak which I'll show an example for in just a few slides.
0:09:07   Also, dimethyl aluminum chloride is a good precursor for ligand exchange, as is T I C 04 tickle. Um,
0:09:15   this is a good precursor from for lead in exchange, as is s I sell for as his BCL three. And what
0:09:22   you're seeing here is that they're gonna be different metals and different Liggins on the metal that
0:09:27   can be effective for this process of Ligon exchanges. A fascinating area. Lots of new chemistry
0:09:33   occurring in this league in exchange which I won't have time to get into, but just to give you a
0:09:38   taste of all the things that could potentially happen with different precursors. So it's not just
0:09:42   centric to try meth aluminum. Okay, so let's look at some results because we really want to get to
0:09:48   particles. I wanted to start out those showing you some results on flats. These air quartz, crystal
0:09:55   micro bounce results on aluminum oxide, a l D. Films grown on the courts Crystal Michael balance.
0:10:02   And then what we do is return around and come in with triumph aluminum at HF. So instead of water
0:10:08   just to replace the water with HF, conditions are almost identical temperatures very similar to
0:10:15   aluminum oxide. A l. D. Also, the Doles does sequence here off to 31 30 and seconds is also very
0:10:23   similar to what we might have seen with atomic layer deposition. But in this case, what we're doing
0:10:28   is the court's crystal micro bounce is telling us that when we run these reactions sequentially, we
0:10:33   get the removal of aluminum oxide and we get a very precise removal of aluminum oxide. In fact, we
0:10:41   get on the order of about 1/2 an angstrom of aluminum oxide removed every cycle. And this? This is
0:10:47   these results showing 100 cycles back and forth of triumph, aluminum and HF, and so you can see a
0:10:54   very precise removal. Also notice that there's kind of a zigzag behavior. This is It's a nice linear
0:11:01   loss of mass. But there is some structure, and if we could weaken, blow that structure up. So we can.
0:11:06   We can go in and just blow up and look at not 100 cycles, but just three cycles. What we see is that
0:11:14   the loss, even though there's a very nice linear loss, it's a very digital process. When we come in
0:11:20   with the flooring, when we come in with the HF to Floren ate the surface, we see a mass gain and
0:11:26   then we come in with try mouth aluminum instead of seeing a mass gain, which you would with aluminum
0:11:31   oxide ale de. You see a very pronounced mass loss. So we're losing mass when the try meth aluminum
0:11:38   does this llegan exchange process with the fluoride surface and then, of course, we can make the
0:11:44   florid surface and we get a mass gain when we make the floor so we could basically go back and forth
0:11:49   losing mass gaining, mass losing mass game. Ask when we alternate back and forth between TM and HF.
0:11:55   So it's very much like a L. D. Except now the masses going down instead of going up. Okay, We can
0:12:03   also use other techniques to verify that we really are etching. And also, this is where we start to
0:12:10   introduce the idea of a Leon particles because we can look at particles. In fact, they're very
0:12:16   useful. Their high surface area is We've used them for a long time to do FDR spectroscopy to study
0:12:22   hailed the reactions and now we can do the same thing. Studying ale e reactions on particles So we
0:12:28   bring in infrared being in, and we pass it through and transmission mode through particles in order.
0:12:34   Now to see either ailed de on the particle or Ailes e of a film on the particle and we could monitor
0:12:41   that with infrared and weaken, uh, control the temperature and look a species with, um aspect, but
0:12:48   will primarily be focusing on just growing the film and then etching the film here. So if we started
0:12:53   out with zirconium oxide particles, which are good, uh, high density particles give this nice tm.
0:13:00   They've been a good substrate for high surface area FDR cities, so if we grow aluminum oxide on
0:13:06   zirconium oxide particles, we can watch the aluminum oxide grow. So after two cycles of aluminum
0:13:12   oxide l d 468 10 cycles, we can see the absorbent in the infrared of the aluminum oxide stretch
0:13:21   aluminum oxygen stretch. We could see that increase progressively with number off aluminum oxide, a
0:13:27   lady cycles. So that and we've seen this before in other studies. Now we're doing it, though, in
0:13:33   order to grow a film on the Sacconi, um, oxide, so that then we can also watch it be removed through
0:13:39   an etching process. Okay, so let's do that. Let's turn around now and do the atomic lier etching of
0:13:48   the aluminum oxide. And for these results, these were actually this was the first system we studied
0:13:53   with a thermal atomic letter etching, and that was using tin ack ack as a metal precursor. So we try
0:14:00   meth, aluminum, a work 10 ak, aka work. Other other metal precursors were work, but these results
0:14:06   are with, uh HF and 10 ak ak. So we Floren ate their aluminum oxide with HF and then we remove the
0:14:13   fluoride layer with a to make it volatile with the 10 ack ack. And so what we can see here is that
0:14:20   here we've grown the aluminum oxide. This is again the aluminum oxygen's observance from the
0:14:24   aluminum oxygen stretching vibration. But then, after four cycles, six cycles, eight cycles, 10
0:14:29   cycles, we can progressively remove the aluminum oxide and we can watch the absorb its go down. This
0:14:36   is a sanity check were definitely actually in the film because it we could weaken, See it clearly
0:14:41   being removed in the infrared, just like we could see the mass being removed using the courts.
0:14:45   Crystal micro bells. There's also ack ack species on the surface and they're they're known to hold
0:14:51   on pretty tight onto the aluminum oxide surface. And there, there and there, there, there,
0:14:57   throughout the etching process. Okay, what about Can we see the films? Can we see the films being
0:15:04   also with with tm studies and we can So here we're we What we've done in these extent studies is we
0:15:12   first grow grew aluminum oxide on 10 I'm sorry on tungsten particles s So this is a tungsten
0:15:19   substrate. We grew with aluminum oxide ale de to get a film thickness of about 16 nanometers
0:15:25   according to the TM. And then we subjected these particles to 50 cycles of aluminum oxide ale e
0:15:34   using HF and try meth aluminum. So now we've gone back to the tm a as the metal precursor. And sure
0:15:40   enough, when we look, we take the particles out. We look at it, the tm we put him back in, we give
0:15:46   it 50 cycles of the daily chemistry and sure enough, we go from 16 to 13 nanometers thickness after
0:15:53   those 50 cycles. So we've reduced the film and also notice we've reduced it and stays pretty conform.
0:15:59   All eso it's it's looking like a lady in the sense that the etching is aken formal removal process
0:16:05   where we can keep going Now we can up in a biltmore. We can we can put the particles back in after
0:16:09   doing tm and then we can do another 100 cycles of Ailey and we can go from 13 down a meters down to
0:16:16   about five nanometers and again we're having a pretty conform. All film on the tungsten particles
0:16:23   that are that are used for these TM studies. Okay, Can we go further? Can be completely remove the
0:16:29   film. Well, sure enough, we can if we put take the particles out after the after put put the
0:16:34   particles back in rather after doing the tm. And then we could do another 150 cycles. Um, I mean, a
0:16:41   total of 150 cycles, that is. And then when we have 150 cycles done. Look, we've got no We've got no
0:16:48   film at all on the tungsten particles, so we've completely removed the aluminum oxide. So this is
0:16:54   this is quite nice because we definitely grow the aluminum oxide, and then we can also remove it
0:16:58   using the aluminum oxide. A leak. Okay, so now I'd like to turn and look at a couple of applications.
0:17:08   Other things. What can we do with particle Ailey? Well, one thing we can do is we can clean surfaces
0:17:14   so we can remove films from the particles, just like we saw with the aluminum oxide and tungsten. So,
0:17:19   no, I'll show another example of that removing s i 02 from silicate particles. And then we could
0:17:25   also turns out selectively remove materials from particles. And I'll show you some results for
0:17:30   selectivity and thermal Ailey. And we'll also see that we can reduce particle size with atomic lay
0:17:36   retching as well as this example that I mentioned at the beginning weaken due to deposition and etch
0:17:42   back to obtain a very continuous, ultra thin film, uh, on particles. Okay, so first of all, just in
0:17:50   general, what can we do with atomic layer etching for particles? Well, one thing we can do is we
0:17:54   couldn't remove coatings or layers that we don't might not want necessarily on the particle. So, for
0:18:00   example, aluminum particles luminous, very reactive. And it's well known to have ah, fairly thick
0:18:06   native oxide of aluminum oxide on the Illumina. So one thing we could do is we could turn around and
0:18:12   take off the aluminum oxide from the aluminum. And so that's that's just a illustrative example.
0:18:18   What we could do Well, I'll show you results for is removing an S i o to film on a silicon particle.
0:18:25   So we're going to start out with silicon particles. They're shown here in the TM and you can see the
0:18:31   oxide, The S i 02 layer. Ah, that's that's coming around the surface of this, the silicon particle.
0:18:39   And so the question is can we use actually now s I 02 to remove This s I O to thickness. That's
0:18:47   coating the silicon particle. It turns out we can, although here we have to use a different
0:18:52   chemistry. It's somewhat related to the chemistry that we've just looked at for aluminum oxide. But
0:18:57   it turns out s i o to itself does not edge easily. What we have to do with S I 02 to get into edge
0:19:06   is we first have to do what's called a conversion reaction. So we have to take the S 02 and goto
0:19:12   actually higher pressures of try enough aluminum to drive this reaction where we convert the surface
0:19:19   of the S i 02 to aluminum oxide. So essentially we're turning the S I 02 into a material that we
0:19:27   know we can actually cause we've just seen we can actually the aluminum oxide. So we do that we do
0:19:31   this conversion reaction at much higher pressures of try meth aluminum. Then we used earlier in the
0:19:37   slides that I just showed and we turn the surface of the eso to into aluminum oxide and then what do
0:19:44   we do? Well, we proceed to Florida at that aluminum oxide service to make the aluminum fluoride like
0:19:49   we saw before. And then we'll etch the aluminum fluoride with the ligand exchange reaction with the
0:19:56   try meth aluminum to get the volatile dimethyl aluminum fluoride species and that will complete one
0:20:02   cycle. And then when we're done, we still have try meth aluminum there after the aluminum fluoride
0:20:07   layer has been removed. And then it will in turn, convert underlying s i 02 to make more aluminum
0:20:13   oxide. So we basically do this, this process of reactions, it's really just to reacting exposures.
0:20:20   But the triumph aluminum essentially does double duty. The try meth aluminum does the Ligon exchange
0:20:26   on the aluminum fluoride. But then it also proceeds to convert the S I 02 And so here's a cartoon
0:20:34   for this process, where trying that aluminum is going to be doing the Ligon exchange to remove the
0:20:39   aluminum fluoride layer. And then it also will convert the underlying s i 02 to amore of aluminum
0:20:47   oxide or aluminum oxide silicon eight layer And then the HF will come in. And then Floren eight that
0:20:53   layer. And then we'll basically do the etching by going back and forth between T Emma and HF at
0:20:58   these higher pressures where the tm A is known to convert the S i 02 to aluminum box. So that's
0:21:05   that's the idea. Now, how do we know we're actually removing the S I 02? Well, we can follow this
0:21:12   also in the infrared, so thes air studies than on those silicon particles that had the S i 02
0:21:18   coating on them and it resounded 300 See very similar temperature that we used for the looming, um,
0:21:24   oxide before. But with these higher TM pressures now we're in the tours of pressure, whereas before
0:21:30   we were maybe tense of tour. What we see is that we can get and the removal of the essay on to cause
0:21:38   we can watch the s I o stretching vibration be reduced versus number of cycles and that these number
0:21:44   of cycles were conducted at different pressures. And so one thing you notice that as you go to
0:21:47   higher pressures you have a higher trait, and that's because there's more conversion. Uh, that
0:21:54   occurs at the higher pressures. So we go from 4 to 1 toe. Have tour in the in the X ray. It actually
0:21:59   drops quite a bit. We get down 2.1 tour. We can't even remove the eso to. So the conversion reaction
0:22:05   really requires higher pressures of Try meth Aluminum. Okay, so that's one example Another Another
0:22:12   thing I wanted to mention and as a possible application for this particle, Ailey is to be able to
0:22:19   come in with an etching process that selectively removes some material from the particle. Now, this
0:22:25   could be you could view this then as particle cleaning, where you have something on the surface of
0:22:30   the particle that you want to remove and you're going to come in and just take it off and maybe
0:22:34   leave the underlying surface of the particle intact. And so it turns out that this is a very
0:22:40   interesting challenge and chemistry to use thermal Ailey four selectivity. So imagine that we have
0:22:46   ah, surface that has lots of different materials on it. For example, the goal of the thermal atomic
0:22:51   layer etching in selective a thermal alias. Can we come in and just remove, say that yellow squares
0:22:58   and leave everything else behind? And so it turns out that it we can do that on the selectivity,
0:23:03   turns out to be determined by the stability and the volatility of these reaction products that we
0:23:09   form in ligand exchange. So if we can form a product that's stable and turns out also is volatile,
0:23:15   then we can remove it. And then maybe next door on another neighboring material way can. We can't do
0:23:21   that because we don't have the same stability or volatilities in the reaction product. And just to
0:23:26   give you an example of some of those just to show that there can be selectivity, he's now I'm going
0:23:32   back to flat. So this is not on particles, but on flats. You could basically grow different films on
0:23:38   silicon wafers and then we can look at different chemistries to see OK, try meth, aluminum and HF.
0:23:44   Can you remove zirconium oxide? Well, no, you can t doesn't that chemistries does not work for
0:23:50   zirconium oxide, but it does edge half a mock slide a little bit HF in in tm a and it etches try
0:23:56   meth. Aluminum. Really? Well, because we know that we just saw that earlier silicon nitride silicon
0:24:02   dioxide. It also doesn't, because now we're using pressures that are more like 1/10 of the tour for
0:24:07   tm A As opposed to the tour pressures that we saw for the S I 02 And we're getting actually, um,
0:24:14   products out with try meth aluminum that are volatile because we're transferring methyl groups onto
0:24:21   the aluminum fluoride. We're transferring methyl groups onto the half ian fluoride that are part of
0:24:26   the reform during the hft m A reactions. And it turns out that they have, ah volatility and are
0:24:31   fairly stable. Okay, so that's trying not aluminum in HF. We didn't edge zirconium oxide. Now the
0:24:39   question is, can we with a different metal precursor, see different selectivity? And it turns out we
0:24:45   really can't because we add a chlorine to try meth aluminum that gives us time off aluminum chloride
0:24:51   that it turns out that we can actually coney um Oxide, In fact, we could etch it really well, once
0:24:57   we have this chlorine, let me go back for a second. We try math, aluminum. We didn't natural at all,
0:25:03   which dimethyl aluminum chloride. We etch the Cicconi mind according oxide. Very well. So it turns
0:25:09   out that depending on what Liggins you can transfer in ligand exchange, you can either make species
0:25:14   that are are stable and volatile or not. It turns out zirconium chlorides turned out to be very
0:25:19   volatile and stable. The zirconium metal compounds are not not so stable. And so we were able to
0:25:26   etch with diamond aluminum chloride. And we can't actually Coney box hide with TM. Another example
0:25:33   of selectivity is looking at the Ailey with HF and tickle and so forth. Now tickle, it turns out,
0:25:41   can match both zirconium oxide and half game oxide, but it can etch aluminum oxide. So here we see
0:25:47   that there is no etching for the aluminum oxide, although we can access the Cicconi, um, and half me,
0:25:52   um, oxide species. And now it turns out the aluminum oxide doesn't match. Not because aluminum cool
0:25:58   rides are not volatile, but because it turns out this chemistry of converting the aluminum fluoride
0:26:04   to aluminum chloride isn't favorable thermal thermal chemically than that. So that's why we don't
0:26:09   get any touching of the aluminum oxide. But these exact. What these examples show, though, is that
0:26:14   we can get selectivity and that selectivity is very dependent on the metal precursor. Just as
0:26:20   another example of selectivity, we can also look This is going back to courts Crystal micro balance
0:26:26   studies so we can grow aluminum oxide. Sir, Here we're growing aluminum oxide by LG and watching the
0:26:31   mass change. And then we grow half iam oxide by LG and watch the mouse change. But then we stop and
0:26:38   then we each have any of oxide by atomic layer ECI. And so is it. Then that could be done very
0:26:44   nicely with HF and the DSL four. And so we that's the half new. So we etch Etch Etch Edge. But then
0:26:50   it turns out when we get to the aluminum oxide interface, which is about here, the itching stops
0:26:57   because that chemistry of tea I sell four etching aluminum oxide is not favorably is not favourable.
0:27:03   Thermo chemically and so the aluminum oxide act as an f stop. So again, it kind of shows in
0:27:08   illustrates selectivity. Okay, so a couple more examples, um, particle size reduction. One thing
0:27:16   that atomic there actually will be useful for with regard to particles is making them smaller. So
0:27:22   imagine we had had selling I particles and we wanted to make them smaller. We could do so many
0:27:28   cycles of the daily process to make those particles smaller. And one reason to do that for these
0:27:34   materials that semiconductor materials, especially, is that we could tune the light emission then
0:27:40   from the quantum dots from these your light emission from these particles and so cats sellin eyes
0:27:45   Very nice example larger, larger sizes, arm or in the red. And then as we go back down in size, the
0:27:53   light shifts to the blue. Okay, now the example. In fact, we tried this. In fact, Tyler Myers, who's
0:28:01   also giving a talk at the P led summit, tried this and we thought, OK, we'll start with zinc oxide
0:28:07   particles. Maybe they're start out around 10 men nanometers and then after a variety of cycles of HF
0:28:14   and in trying with aluminum on, this is chemistry that would we demonstrated earlier in the paper
0:28:20   that shown we thought, well, we could just make these particles smaller by going back and forth with
0:28:25   our exposures of HF and, uh, and trying to aluminum Well, it didn't turn out that way. It turns out
0:28:32   that zinc oxide also can undergo this process called conversion, similar to what we saw with the S I
0:28:41   O to converting to aluminum oxide. And so, it turns out, is even better on zinc oxide. Zinc oxide
0:28:48   isn't as stable and oxide as the S I 02 And so what happened was zinc oxide. The surface of the zinc
0:28:55   oxide, in fact converted to aluminum oxide. And it actually converted to a pretty thick aluminum
0:29:01   oxide layer about one nanometer thick. And so that first dose of the TM A on the zinc oxide
0:29:08   particles did a lot more than just Ligon exchange. It also converted a lot of the zinc oxide to
0:29:15   aluminum oxide. And so Tyler was able to determine that that it did this based upon flats and doing
0:29:23   X ray reflectivity to measure the thickness of the aluminum oxide. But then he did also look at
0:29:29   particles, and so he put those 10 nanometer zinc zinc oxide nanoparticles in and then it turned out
0:29:36   with single exposures of T m A. He was able to see really quite large mass changes in the in the
0:29:42   collection of the particles that increased, increased and then levelled out as a function of
0:29:48   exposure to try enough aluminum. And so what? This is this is the mass loss upon conversion of that
0:29:56   one nanometer of zinc oxide on the surface of the zinc zinc oxide particle to aluminum oxide. And so
0:30:02   he was able to show that very nicely just by weighing the particles that were in his particle rotary
0:30:09   reactor. And so, in fact, what happened? We didn't make the particles smaller, but we also made like
0:30:15   a core shell particle where we started out with zinc oxide. And then, as a result of the try meth
0:30:20   aluminum exposure, we've actually put a nice aluminum oxide layer of about one nanometer. That was
0:30:26   the mass loss was consistent with the one nanometer conversion of zinc oxide toe aluminum oxide.
0:30:34   Okay, so that's an example. And then my last example here is going to be for the technique of
0:30:42   deposition and etch back, and this is a very nice technique. And I think this also shows how well
0:30:48   the atomic clear deposition and atomic clear actually in can work together because we know that when
0:30:54   we do atomic layer deposition. The film's always don't nuclear eight so rapidly. In fact, sometimes
0:31:00   it may take tens of cycles, sometimes 50 to 60 cycles. If we're trying to do it, for example, atomic
0:31:07   clear deposition of a metal on an oxide you could take it could take 5100 cycles. Toe actually get a
0:31:13   fairly smooth film because what happens is the the Ale de nucleus and little islands. And then those
0:31:20   islands grow. And then eventually, only after many, many cycles do the island's grow together. And
0:31:25   then finally, you'll get a fairly continuous film after many, many a lady cycles. Now that's fine.
0:31:32   That's known. But let's say we didn't want a film that that that was that thick. Let's say we really
0:31:37   wanted a film that was a lot thinner, but we could never get it directly using a lady because of the
0:31:43   nuclear ation difficulty. So what you could do then is to grow a film with a lady, get it thick
0:31:50   enough so it's nice and continuous and fairly smooth. And then you could turn around with atomic
0:31:56   layer etching and remove the Phil and etch it down until it now is at an ultra thin, continuous film
0:32:04   that maybe is what you really want it but couldn't get directly with atomic layer deposition. And so
0:32:09   this this process of a lady and a Lee is a way to get then ultra thin Struth films on to circumvent
0:32:17   the problems that you would have with the L. D nuclear nation. And so this is a very nice way t get
0:32:24   at these really ultra thin on smooth films. And by the way, I should also add that the thermal daily
0:32:30   process actually also acts to smooth the film itself. So you might wonder what happens when you etch
0:32:37   what we've. What we've found in recent studies is that, as you thermally etch, you can also sue than
0:32:44   the surface. So so the surfaces get smoother as you do the thermal. Ailey, on example of this
0:32:51   process of depositing a thicker and then etching back, is for this. Now, this is a semiconductor
0:32:59   application where sometimes you want to deposit crystal and half the um oxide. But it turns out that
0:33:06   when you deposited it's a Morphosis and then to crystallize it turns out it's crystallization.
0:33:11   Temperature is very dependent on its thickness. If you have a thin film, you have to go to a much
0:33:16   higher temperature to crystallize it. Then if it's out of lower, unless you unless you have ah,
0:33:22   thicker film. So if you have a thicker film, you can you can you can crystallize it at a lower
0:33:26   temperature. So what you can do then is you can deposit an amorphous film, and then you can Neil
0:33:34   that film up to crystallize it at a lower temperature to get it nice and crystalline, and then you
0:33:39   can edge it back in order to get your ultra thin crystalline film. And this was recent paper that
0:33:45   just came out that illustrated how we how you could do that to get really ultra thin, crystalline
0:33:50   films off half am oxide that you couldn't have gotten directly because you couldn't have heat. You
0:33:55   couldn't have heeded the some sample of high enough to get it to crystallize. When it's thins, we
0:34:00   have to crystallize it when it's thicker. And then, uh, once its crystalline entered back to get
0:34:05   your ultra thin, crystalline Phil. Okay, so that's the last example for me. So now I basically just
0:34:12   like to conclude on and kind of review what we've just been talking about. So hopefully when I've
0:34:18   been able to show in this talk today is that weaken use thermal Ailey and that basically a LD
0:34:25   thermal Ailey functions as a lady in reverse. So it really has all the properties ale de. It's just
0:34:32   we're going backwards and we're taking material way so the aliens away to take it off a lady is the
0:34:37   way to put it on, and but really, the way they work the way they work with sequential self limiting
0:34:42   reaction is very similar. Also, we saw that one pathway for thermal alias, Floren Ation and Ligon
0:34:49   Exchange, and that's been one of the dominant pathways that's developed as a way to etch oxide and
0:34:55   nitrite materials. Now there are other pathways to but floor nation. Ligon Exchange is probably the
0:35:00   main one to date for oxides and nitrites, and then we saw that we could we could use silicon dioxide
0:35:07   daily to remove S i 02 from silicon particles so we can see that there are applications now where we
0:35:14   can use a lee on particles to do things useful, like remove codings that might be on the silicon
0:35:20   particles. Also, we saw aluminum oxide could be removed from the tungsten particles. And then we
0:35:25   also saw some really interesting chemistry. Where try meth. Aluminum was able to essentially convert
0:35:32   zinc oxide to aluminum oxide. We're trying to make the zinc oxide particles smaller, but we found
0:35:37   this conversion reaction was was really at play and get a conversion process which also could be
0:35:43   used to catch but, uh, interesting in itself just to look at the conversion. And then we also looked
0:35:49   at possibilities of thermal alien for selective removal of Mitchell from particles. This is really a
0:35:55   cleaning example. And then also we looked at the possibility for using thermal daily to form ultra
0:36:02   thin and continuous films on particles using this deposit and etch back technique. Okay, so that's
0:36:09   that's the talk. Let me stop share and just conclude Here, start my video again. Um, yeah. So hope.
0:36:17   Hope this has been useful. I know this is a little different in the sense that it's atomic, clear
0:36:23   etching and not atomic layer deposition. But hopefully I've been able to show that the atomic layer
0:36:28   etching is like a lady in reverse, and we're hopeful that it will have a lot of applications on
0:36:35   particles. It's certainly having a lot of applications in the semiconductor world now, and we're
0:36:40   hoping to extend that into the world of particles. Okay, so thank you very much. And, uh, I will. I
0:36:48   guess we're gonna be answering questions by chat. Uh, throughout this meeting, I'm pre recording
0:36:52   this. Of course. So I can't say we're gonna ask her questions now that hopefully will answer
0:36:58   questions later. Okay. So hope you enjoyed that. And thanks the night.