PALD Applications of Interest 

Non-uniform Growth of Sub-2nm ALD Al2O3 Films on NMC Battery Materials  

Session Notes:

PRESENTER

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

Professor of Chemical and Biological Engineering at the University of Colorado

 

0:00:01   Hi, I'm Al Weimer at the University of Colorado and we're gonna talk today about some studies
0:00:09   looking at the initial growth of aluminum oxide by a L. D. For only a few cycles, typically
0:00:16   resulting in films that are sub tune in your we have done particle LD for many, many years. But this
0:00:24   is actually the first investigation that we've ever period of one lifting my battery materials.
0:00:31   These results, I think you'll find to be pretty interesting. So our interest here is the fact that
0:00:40   it was shown back in 2010. The Steve George Group, in collaboration with in Rail and LD Nana
0:00:48   Solutions demonstrated that only a few cycles of ale de could substantially improved the battery
0:00:54   performance of lithium ion Kathryn's. And here in this particular case, lithium coupled oxide. What
0:01:03   you notice is that the initial capacity of lithium coupled oxide the beer material, is very similar
0:01:11   toe that with the two cycles and even the six cycles of ale de however, it's about E 9% capacity
0:01:20   retention for two cycles availed decoding this particular materials were as the beer lithium cobalt
0:01:27   oxide only ends up with about 45% capacity retention also noticed that the initial capacity
0:01:35   decreases for six and 10 L D cycles and that for 10 ale de cycles essentially have an insulating
0:01:44   alumina layer on the surface of those particles. So your conclusion here two cycles appears to be
0:01:51   best. Six ale de cycles tends to improve the performance, and 10 LD cycles is insulated. More recent
0:02:01   studies continuing this work looking at a chronic layer deposition. These were presented in France
0:02:08   API ABC meeting in January of 2019 and what you see is at the LD coded in C M 6 to 2. Materials, in
0:02:20   fact, are superior to those which are uncoated or coated with dry or or wet methods. And some of the
0:02:28   conventionally LD assertions are the ale destabilizes Catholic defects and inhibits the FC I layer.
0:02:36   Protecting transition metals from dissolution improves capacity. Lifetime allows higher walk density
0:02:44   works on a number of different captain materials, L c E O N C, A, A and M C emano generally
0:02:51   lessening six sale de cycles are preferred for commercial interest in the basis of cost performance.
0:02:57   Very low cost the curia only a few LD cycles, especially if the volume of the substrate being code
0:03:05   is extremely high, as it will be 44 can materials now. The film thicknesses we're looking at are on
0:03:12   the order of angst. So the question that really intrigued me, maybe it's called intellectual
0:03:20   curiosity was. Why does this work? How is it possible? Only a few ale de cycles can provide such of
0:03:28   it improved performance. And if you look at all the published papers up to this point, they
0:03:36   essentially conclude that the ale do. Films are uniform on very thin Oprah thin and that they
0:03:45   protect the transition metals, but they're fit enough to facilitate lithium diffusion. InterNAP also
0:03:53   typically sold jail liquid processing. That's non uniform kind of hidden miss. There could be
0:03:59   uncoated areas there, and basically you would not have the ability with those types of films to
0:04:06   prevent electrolytic side reactions. So if we go back and look at what has been previously in terms
0:04:16   of low cycle ale, de uh, Rica really explains it quite well in her paper, essentially, that the
0:04:24   first day of the cycle occurs on the surface of the original substrate. After that, you have growth
0:04:31   on the original substrate as well as growing on the A. L D grown surface and eventually, after a
0:04:37   certain number of cycles, depending upon the growth rate. You will then end up with any LD film and
0:04:44   is growing on itself. Making the film thicker. Greg Parsons has developed a recent model looking at
0:04:51   this nuclear nation on substrate surfaces, growth time. And what you see is, in fact, that you're
0:04:58   growing a film, and after a certain number of cycles, you'll end up coating the entire surface. So
0:05:04   this suggests that the early cycles of ale de or semi continuous and provide for non uniform film
0:05:15   and, in fact, that they look at some of the early studies that John Ferguson did coating for a night
0:05:19   ride with aluminum oxide. He showed that it required about eight a L D cycles in order to initiate
0:05:26   growth on the entire foreign make right surface. So it's not unheard of that you're gonna look at
0:05:32   nuclear nations exposed surfaces. So our hypothesis for this work was that for typically less than
0:05:40   maybe six alumina cycles, which is in the area for cost performance, benefit for industrial
0:05:47   application, we are hypothesizing that film coating is non uniform and non, uniformly thick, as
0:05:55   would be characterized by prior work such as John Ferguson's work and essentially were stabilizing
0:06:01   the transition metal oxides in the presence to the electrolyte without blocking you. If human
0:06:06   equation pathways essentially that the non uniformity promotes lithium diffusion. So how do we
0:06:14   investigate this for films that are typically a new order of anxious? What kind of methods continues
0:06:22   fertilization? Conventional methods such as XPS have a resolution about five nanometers and start
0:06:29   looking at some surface, which is going to be typically we're interested in sub one nanometer films.
0:06:36   Eyes Not Going Teoh Give us very high quality characterization. We know that low energy ionization
0:06:43   spec Cross could be is going to look at the first atomic layer and top. Simms has a resolution of
0:06:50   about one nanometer, so we're going. Teoh. Use those two characterization methods in order to
0:06:57   investigate the resolution of surfaces. Lesson one nanometer for up to about six ale de cycles
0:07:06   coating on on heroes so low energy ionization spectroscopy looking at the first atomic layer, its
0:07:15   quantitative. But I can't measure lithium. Top Sims, on the other hand, has a resolution of one nana
0:07:21   meter. It's qualitative but a can measure lithium, So the objective here is to use these two
0:07:28   characterization measurement methods together in order to identify what is occurring during the of
0:07:38   those materials. So in order to fabricate materials for investigation used, a fluid ice bed reactor
0:07:46   loaded it with NMC. 111 repurchased, carried out 2468 10 12 and 15 hailed the cycles and remove
0:07:56   samples along the way. These were not separate batches. In order to avoid batch, the batch
0:08:02   variability. We get elemental analysis and high resolution spectroscopy to understand the deposition
0:08:09   in the nature of the films, fabricated electrodes and tested them and cooling sells for 250 cycles.
0:08:16   And we showed that four alumina LD cycles actually performed the best, which is consistent with the
0:08:23   prior studies that we presented here earlier. E ds mapping shows that there is apparent uniform
0:08:29   aluminum distribution across the entire population of particles with no uncoated particles looking
0:08:37   at even four cycles in that particular right. So what do we see from the lice elemental spectra on
0:08:48   the left? What you see is the growth of the aluminum containing layer. The aluminum oxide is clearly
0:08:56   visible with increasing ale de cycling. On the other hand, on the figure on the right, we can see it
0:09:04   the presence of manganese, cobalt and nickel. What? The surface is completely suppressed by the 10th
0:09:10   L D cycles. Overall, alumina is growing. Believe it's aluminum is growing and we see that is
0:09:17   covering manganese Couple in. If we integrate these spectra on, we normalize the meaning of these
0:09:28   cobalt and nickel signals to the uncoated sample. Looking off at the at the left Ah y axis. Uh, that
0:09:37   normalization is to the uncoated sample in order to determine the fraction of the original manatees
0:09:43   couple of nickel surface Adams that were covered over the course of the deposition. The aluminum
0:09:49   data, on the other hand, were normalized to the sample could've with 15 cycles since this represents
0:09:56   the maximum out of aluminum deposited within this study, and what we see here is that between four
0:10:04   and eight LD cycles, approximately 80% of the manoli's coupled and nickel was covered. However, the
0:10:12   surface contains only about 40% of the ultimate aluminum coverage, and from 10 to 15 l d cycles, the
0:10:22   main of these couple of nickel ore nearly two completely covered. But the aluminum signal is still
0:10:29   increasing, indicating that there's a component of the original surface is still present because
0:10:34   it's not completely covered by the aluminum layer. So the question is, What is that? Well, we
0:10:40   suspect that the missing cup composes lithium, since it can't be detected by this particular method.
0:10:49   In that case, we then went to talk Sims on looked at the integrated intensities of mass selected
0:10:57   irons. Lithium, nickel, manganese, cobalt and aluminum is a function of pixel position, with a
0:11:03   resolution of about one. And what we see here is that the nickel mag unease and cobalt are all
0:11:12   suppressed after 15 ale de cycle. We see that clearly aluminum eyes growing, it's a visible. It's
0:11:22   clearly visible at 4 15 and they all said the city that lithium is still visible on a surface even
0:11:30   after 15 lt cycles. Note here is that the intensity that color is really correlated the elemental
0:11:37   concentrations in the measurement death and the variation. You see there is really a function of of
0:11:45   particle size, crew, mature and Z height. But if he was still present even after 15 cycles, this is
0:11:53   actually the first investigation of aware of, which indicates that lithium is still present on the
0:12:01   coded. He'll decoded rules. So here we look at the absolute signal counts. This is qualitative, not
0:12:12   quantitative looking at originally at the residual signal relative to the uncut example for the main
0:12:19   composed with your nickel, manganese and cobalt after 4 15 cycles. And what we see is that the
0:12:26   growth of the aluminum signal begins to level off between Ford. 15 cycles at the ratio of aluminum
0:12:34   lithium is increasing with an increase in. We also see that the cobalt, nickel and manganese signals
0:12:43   are highly suppressed between cycles four and fit. But we see that there is still significant
0:12:50   lifting and present even after 15
0:12:56   and we believe that that has to be in amorphous form of lithium aluminum oxide, likely lithium still
0:13:06   present at the surface. In that form, it could be that their that their existing lithium surface
0:13:11   sites are not initially coated. Where could also be that the surface sites generated by the
0:13:18   potential lithium migration through the substrate in the existing bill. So basically some validation
0:13:25   that Livia may actually migrated through that film,
0:13:31   and here we're looking at clearly seeing that the L. D is preferentially coating of transition metal
0:13:38   surface sites. Nickel. Matt Megan, Easton, Qiu Bo over lithium surface likes of preferential coating
0:13:46   of those transition metal oxides versus lithium. So in order to try and get a better understanding
0:13:55   of what is happening system we went back and we looked at 0 15 cycles ale de of coding. Pure lithium
0:14:05   hydroxide particles and coating with pure lithium per minute particles and also compared that to the
0:14:12   NMC 111 materials that were coated. And what we see is that there's a substantial difference in the
0:14:20   aluminum growth on lithium hydroxide versus lithium carbonate. Lithium carbonate tends to go back
0:14:30   through the 00 point, which is what you wouldn't expect. We're atomic layer deposition with uniforms.
0:14:37   Film growth, however, noticed that for lifting of hydroxide there's a substantial demarcation and it
0:14:45   would appear that there had that there may be some known ale de still a chemical reaction which
0:14:51   initially occurs, and then after that you have led growth. And so if you look at the blue life
0:14:57   lithium hydroxide possibly, uh, the aluminum, the aluminum oxide is reacting, say, with lithium
0:15:09   hydroxide and maybe forming lithium aluminum oxide. In fact, if we thermodynamic lee good good
0:15:17   effects age. The Thurman medically stable product of aluminum oxide with lithium hydroxide is
0:15:24   lithium aluminum oxide and may also be that there's a demarcation, although we can't see as easily
0:15:33   with him with him. See material. So the take away for the ask produced LD Coated Can materials again
0:15:42   on the left General theory is that these films are typically for less than two nana meters sick.
0:15:47   Their uniform at the Earth been enough to allow lithium to diffuse through the film. What this work
0:15:53   is showing with novel surface analysis to combine characterization of license tof Sims is that that
0:15:59   coding is non uniform and non uniformly thick, particularly for lesson about six ale de cycles, and
0:16:08   that surface is probably a combination of aluminum oxide and lithium aluminum oxide.
0:16:16   So, in summary, low psycho whale, the aluminum films improve its ability of intimacy cathodes. No,
0:16:23   Phil was less than about six cycles generated on intimacies have non uniform chemistry and thickness,
0:16:30   as we would expect from from prior studies that we talked about John Ferguson work back in in 2000.
0:16:38   We also have noted here, a combination of lice and top Sims allows us to investigate these films.
0:16:48   They're less than than one nanometer thick and essentially conclude that illuminate departures to a
0:16:55   higher degree on transition metal sites. And it does on lithium, most likely the loose cycle alumina.
0:17:04   All the films improved the cycling stability of battery cathodes through this preferential growth,
0:17:09   which is unexpected. That stabilizes the principia metal oxides of the presence of the electrolyte
0:17:15   without blocking the lithium interpolation pathways, typically through what is most likely amorphous
0:17:23   lithium aluminum oxide. I'd also like to mention that this work is published, you know,
0:17:31   Bucs Supply, Dana Materials Journal in October of 2000 and 19. And if this paper is open access here
0:17:41   is, uh, a photograph of a commercial special ale. The reactor that can process three metric tons of
0:17:54   day can materials on this was actually the first commercial reactor was delivered to a battery
0:18:02   materials factory in January of 2019. So if you're interested in commercial processing for large
0:18:11   ball quantities continuously of these thanks of materials, please contact Ford. I like to
0:18:19   acknowledge my students they have actually fantastic students. Um, I can give this talk here, but
0:18:26   they are actually the ones that are doing all the work. And I really appreciate everything that they
0:18:33   have been relative to these.