Silicon anode scale up process overview and key hurdles to production at scale

Isaac Lund – VP of Business Development GDI

0:00:01   Hello. My name is Isaac Lund the VP of business development for graph Phoenix Development
0:00:07   Incorporated or G D I. To make things simple. Uh, the title of my talk is Silicon Eno Process
0:00:15   overview and key hurdles to production off silicon and at scale.
0:00:23   Mhm, mhm. As a quick outline, I'm first gonna introduce myself and give a little bit of my
0:00:30   background. Then I'll survey of a different silicon, an ode solutions out there on the market. I'll
0:00:35   discuss a couple of issues that limit the percentage of silicon and notes that are incorporated into
0:00:43   cells and electrodes currently. And then finally, I'll introduce G. D I and show the current state
0:00:51   of the art in terms of our own development and testing. So as a background to myself, um, I started
0:00:59   to work on silicon, and it's about 10 years ago. This is an outgrowth of the research I was doing
0:01:06   during my doctoral, uh, dissertation. So I was researching silicon nanowires for on chip
0:01:14   Interconnects. It called the Nano Scale Science and Engineering at SUNY Albany. I discovered an
0:01:21   interesting process to bake Basically a heterogeneous core shell, nickel suicide and wire. Uh, spun
0:01:28   that out into a startup company called Best Technologies. We're really looking at, you know, the
0:01:35   material characteristics and the electrochemical stability. So from there, I moved over to BMW,
0:01:42   where I was looking at the integration of silicon animals into a large format cells and kind of the
0:01:48   integration into, uh, end user applications. And in this case, it was, you know, automotive
0:01:55   applications. And so, looking at mitigating all of the problems that Oh, yeah, enable, uh, the
0:02:05   integration of silicon and nose into larger format cells. And then finally, I moved over to G D I
0:02:12   where I am currently now, which is taking a kind of re look at the strategy of making you know, uh,
0:02:20   lithium ion batteries by focusing on scale processes and then trying to figure out the stability
0:02:28   issues. Yeah, So in general, there's four different ways of manufacturing Silicon based, and it's
0:02:37   three of them are particle level manufacturing. That then have to be, uh, mixed with either
0:02:46   gratified or other conductive carbons, right? And binder systems, uh, coded onto a current collector
0:02:53   and then tried and integrated eso. The first production method is you just take a S i o X particles
0:03:01   that are out there. You know, uh, in industry, you mix them with a bunch of graphite and then you
0:03:10   integrate them into cells. Um, it's limited in terms of the percentage that you can integrate s i o
0:03:18   X particles to sub 10%. Unless you integrate a pre litigation step. This is due to the first cycle
0:03:27   efficiency of S I O X being rather low. So in the seventies percent. So the next strategy is to take
0:03:38   metallic silicon particles from you can get these from a bunch of different suppliers, right? Some
0:03:45   people will say that it's just, you know, the waste material from solar cell cutting and
0:03:51   manufacturing. But, you know, you could look at a bunch of different raw material sources for this.
0:03:58   So you take the silicon nanoparticles. Um, you coat the surface with some sort of blocking layer
0:04:05   thio help with kind of the surface stability, right? This would normally be a carbon based coating,
0:04:13   and, you know, uh, at this application, I'm showing a graphene based coding, and then you basically
0:04:21   do the same thing where you, uh, mix it with a bunch of other carbon materials, and then you coat it
0:04:27   onto a current collector and you dry it. So, uh, right now out there on the market, you confined
0:04:37   cell set, you know, include sub 10% off, uh, silicon metal base material, right, um, into a nana.
0:04:49   And then, uh, there's this proposed, uh, mechanism for creating silicon particles, uh, in silicon
0:04:59   carbon composite particles where you take, uh, activated carbon or high surface area carbons as your
0:05:08   raw material. You infuse it with silicon through gas face reactions from a silicon precursor. Uh,
0:05:17   most of the time, it's styling. That's your silicon precursor, right? That it basically decorates
0:05:25   the carbon with a bunch of silicon. And then you take that you mix it right with a bunch of other
0:05:32   carbons and binders, right? You coat it and then you dry it. Hey, and what, uh, g I is proposing and
0:05:41   what others have proposed us Well, is doing the directional ized deposition of silken directly onto
0:05:50   current collectors utilizing plasma enhanced chemical vapor deposition from, you know, uh, basically
0:05:57   the same gas precursors as the last option. Um, and this is a process that is commonly utilized in,
0:06:06   uh, thin film solar processes and then also semiconductor processes. So it's basically leveraging
0:06:14   the A tool set logistical supply chain from those industries. So as stated before these air particle
0:06:23   based solutions, you still need the mix coat calendar center slit and dry them right. Whereas for a
0:06:31   direct deposition of ah, silicon based in ozone to a current collector, you basically just load up
0:06:38   the current collector. You know the role of it. It goes through a machine, that deposition is done,
0:06:44   and then you roll it back up and then you, you know, split it and integrated. So it's compressing
0:06:50   five steps into a single step.
0:06:54   So for the different, uh, to go Maurin depth on the different solutions for S I o X based particles,
0:07:03   uh, normally requires a pre litigation step due to its low, uh, first cycle efficiency. And you also
0:07:10   have to provide a conductive network, uh, to be able to enable cycling, because s I o X is insulated,
0:07:19   right? Also s I o X particles air usually not monolithic. It's normally a combination of silicon s i
0:07:28   02 other conductive phases, right? And so these can actually be very complicated particles and
0:07:35   expensive particles to manufacture right and to integrate into a high percentage of S I O X. And to
0:07:43   your a note, you basically have thio integrate, integrate institute pre litigation. So for another
0:07:53   process, if you take a metallic silicon rate on you, coat it right. You have to make sure that the
0:08:02   surface is completely coated, right to inhibit basically any electrolyte from seeing the surface and
0:08:09   creating some sort of parasitic degradation, you know, unstable s EI formation. But you also have to
0:08:19   allow electrolyte to basically wet the silicon particles. Well, right, so there's some sort off
0:08:29   enter composing processes going on there, right? The other thing to remember is that you know,
0:08:37   silicon particles, right metallic silicon are not all of the same quality, right. You'll have, uh,
0:08:45   different levels in different types of contamination into the silicon particle, based upon what you
0:08:53   know, quality you're purchasing. And so, you know, you can look at things that are relatively cheap
0:09:02   and basically have ah lot off contamination to things that are less contaminated and more expensive.
0:09:11   So looking at the two processes that are utilizing, uh, silicon from a gas based precursor, the main
0:09:23   difference between these two processes is actually the circulating utilization or the gashes
0:09:30   Precursor utilization, which would ultimately, you know, drive up the cost of thes, uh, silicon
0:09:39   based solutions, right? And so for infusing high surface area carbons with silicon that is normally
0:09:48   done through a thermal CVD process Silent utilization There is normally the 20 to 30% range, right.
0:09:55   So if you have your costs, the silent is $10 for a kilogram. You know, you divide that by point to
0:10:03   write the cost of, you know, your energy based material would be at least 50. So, you know, and then
0:10:12   for plasma based DVD, you can basically take this silent utilization depth up from the 20% that you
0:10:21   get in thermal, right all the way up to 70 80% so heavily reducing basically lost. And the cost of
0:10:31   the silent gosh is precursor. So the difference in the two processes really has to do with they
0:10:39   gashes molecules, right? And so, under thermal CVD step, they're basically randomly oriented, right?
0:10:47   And so this occurs due to, uh, even toe have high surface coverage. Right? So you can't have a
0:10:55   directional ized or shadowed deposition process, whereas for plasma CVD, it could be directional
0:11:03   ized, right? And so you can get a high utilization of in incoming particles, right to incorporated
0:11:11   particles into your silicon. And so some of the problems that I see that have been limiting the
0:11:22   amount of silicon, uh, you know, that could be integrated into batteries and to basically a
0:11:29   composite electrodes where you mix, you know, silicon with certain amounts of graphite, right? And
0:11:35   then coat it, you know, uh, cycle it. So one of the main challenges right is that the graphite
0:11:45   litigation potential is below the silicon litigation potential. And this means that in order to Elif
0:11:52   the eight or get any capacity from your graphite, you're basically having toe have silicon go
0:11:59   through its full litigation in faith expansion. Right. So this is going from, you know, just pure
0:12:07   silicon all the way up. Thio have I 20 one s I fire for lie 4.4 s. I right. And this involves all of
0:12:17   the volume expansion that is associated with that as well as creating a silicon alloy material that
0:12:27   is the most reductive. And so it makes passive ating thesis surface problematic. Right? So maybe we
0:12:36   need to transfer from a silicon carbon composite to basically an old silicon alloy, right?
0:12:47   The other issue that is not discussed heavily out there in three industry is that, uh, the solid
0:12:57   electrolyte interface or the C I A. The passive Asian interface on you know, carbon is not the same
0:13:03   as the passive Asian interface on graphite. So graphite has a selective passive ation interface
0:13:10   based upon where the lithium inter clips to or basically along the cross section where basically the
0:13:20   Salvation Show is co inter Kalay tid and then decomposed. And it creates this three dimensional
0:13:26   interface that also access kind of a adhesion or pinning point for the rest of the S E. I, uh, you
0:13:34   know, my hypothesis is just also helps it in terms of swell, right. Whereas for a carbon interface,
0:13:42   uh, this co inter Kalay Shin of the Salvation Shell actually doesn't happen. It's blocked, right?
0:13:48   And so you're going to get you're going to get basically a random mosaic of the decomposition
0:13:56   products of the S E. I uniformly distributed all over basically that carbon interface right, or the
0:14:06   blocking interface for the silicon particles. So, in order to increase the passive ation
0:14:16   capabilities of the salt, electoral and interface. It's been shown that a lot of people are
0:14:22   integrating a certain percentage of FCC, uh, into the electrolyte, but not too high of a percentage,
0:14:29   because it has a hi vaporization uh, pressure, Um, in order to basically help stabilize the C I
0:14:40   interface so But once again, this kind of strategy does not necessarily create the same binding and
0:14:49   pinning points that you know the S C I onto graphic does. So, you know, here's changing from a
0:14:59   mosaic interface to basically in L. A F dominated based interface. So as an introduction to G D. I
0:15:10   so g d I is basically a mix of energy storage experts and ex Kodak engineers. So the Kodak engineers
0:15:19   basically put a focus on the You know, if you're gonna build something, make sure you build it with
0:15:25   scalable processes. And so, you know, basically X half of our team is Kodak with over 2030 years of
0:15:35   experience and kind of scale up in manufacturing.
0:15:40   So this has made a, uh, focus on utilizing scalable processes. We started with ultra capacitors in
0:15:50   2014 focused on utilizing role, the role coding processes with Equus or green solvents to be able to
0:15:59   plug into the Kodak network. So scale up codings thio thousands of, uh, meters of material at a time,
0:16:07   right? Integrate them into cells, integrate those into a validation projects in kind of campus
0:16:15   shuttles, right. And then also great into storage boxes for lifting Mayan batteries. We looked at
0:16:22   utilizing basically the same processes on the cathode utilizing water based finders. Thio plug into
0:16:30   currently existing role the role, uh, processing machines and then, uh, on the and look at utilizing
0:16:40   a silicon based and deposited utilizing, uh, solar cell processes because all of the tool sets and
0:16:48   supply chains and you know etcetera are scaled up for that. So going
0:16:57   so going on to our, uh, a current developments, status and vision. So G d i s, uh, currently
0:17:08   leveraging solar cell processes on a sheet. The sheet tool, um, you know, creating single layer
0:17:15   pouch cells and doing all of our development on that we're in the process of validating a, you know,
0:17:23   roll the roll double sided deposition scale tool, right. That would be able to enable us to scale up
0:17:30   our processes. So, you know, in terms of some of the benefits of utilizing high silent utilization.
0:17:38   Right, So the cleaning is inversely related to silent utilization. Most of the power to run CBD
0:17:47   processes is actually dominated by the pumps, and turbulence is the dominant Yeah, force and gas
0:17:54   flow. So that means you can do vertical processing and reduce your overall footprint area. So, um G
0:18:01   d eyes looking at and co locating its processing with silent production there's a lot of
0:18:08   capabilities not only of the U. S. You know, with a theoretical, uh, supply of 142 gigawatt hours of
0:18:16   an old material per year. But, you know, worldwide, it can get up to the terawatt hours, and this is
0:18:22   currently available silent production facilities. So, in terms of full cell results, uh, this is our
0:18:31   results in terms of integrating, uh, in a single layer pouch, full cells, right at high sea rates,
0:18:40   medium see rates and low see rates with different amounts with off the shelf commercial. NMC 62
0:18:51   cathodes Um, we've also been doing some work in terms of proliferation, right? So pre litigation for
0:18:59   us helps us avoid some of the first cycle inventory losses right? it basically puts the cathode into
0:19:08   its optimal working area. And it allows integration of conversion cathodes also for us, right? If
0:19:18   you eliminate some of the Catholic access, you can get all the way up to a 400 watt hours per k g.
0:19:24   And this is in a 18 6, 50 miles. So that's basically it. Thank you for your time. And I hope you
0:19:31   enjoyed this presentation.