Published on September 21st, 2020 |
by Chanan Bos
September 21st, 2020 by Chanan Bos
Well, there is only one day left until Tesla Battery Day. For almost a year now, I have been trying to get all the research, the patents, the facts, the rumors into one video, but it will not be ready in time, so this article will be a culmination of almost everything I think I know about Tesla’s next-generation battery. This is not just another article.
Let’s start with the most recent development that everyone is dying to know more about. That would be electrek’s image that supposedly leaks what Tesla’s next-generation cells will look like. According to the writing on the can, it appears to be 54 mm wide instead of 21 or 18, and it has a height somewhere between 900 and 1100 mm, so let’s call these cells 54XXX(X) cells.
If this turns out to be the case, there is one person in the Tesla community that predicted this back in May and I want to give him credit for it. On Twitter, he is known as Sarge Enzyme. What he did was calculate how many watt-hours worth of cells you can cram into a Model 3 battery module if you increase the diameter of the cell one millimeter at a time. The optimum size, according to his calculations, was 53 mm, which is incredibly close to what we see in that leak. However, considering the fact that Tesla might get rid of modules altogether, it would have to be quite a coincidence if that lets Tesla put in exactly one extra row of 54 mm batteries or makes the difference between 54 and 53 millimeters. But even if it was just a lucky guess, it was a good one. What is important to consider is that Tesla’s plans for a next-generation battery are not necessarily something that will be implemented in just one step. In that first step, Tesla might keep the existing module width and upgrade the Model 3/Y to 54XXX(X) cells, but there are plenty of other theories that will be explored in this article.
Unfortunately, I was unable to reach Sarge Enzyme before this story was published to display his work here, but here is a link to a tweet he posted with a link to it.
Battery Master Plan V1 to V3
Now the layman might ask, why didn’t Tesla just go from 18650 cells to 54XXX(X) cells all the way back in 2016 when the Model 3 came out. The answer is that heat kills the million-mile battery. The larger each cell is, the harder it is to cool and it is especially hard for the cooling solution to effectively reach the center of the cell. This explains both how all the way back in 2012, a Tesla Model S was able to accelerate in a way no other production car in the world could. It’s, in part, because they can get a lot of power from 7,104 small cells without straining any individual cell. This is one of the main reasons why they retain most of their capacity for well over a decade whereas a Nissan Leaf from that year on the second-hand market is probably best to avoid.
Small cylindrical cells are a hard engineering and assembly problem, but they offer a lot of benefits. With significantly better cooling and some precise calculations, a 2170 cell became feasible for Tesla in 2016, when the company dipped its first toe into the foray of battery chemistry in the form of a new chemical formula designed together with Panasonic. However, eventually, just like most suppliers, Panasonic did not live up to Tesla’s (or Elon Musk’s) ambitions for yearly improvements measured on the scale of orders of magnitude. So, in what I estimate to have taken place between H2 2018 and H1 2019, Tesla decided it would do what it has done each time its expectations weren’t met — vertically integrate. Though, in this case, that is easier said than done considering that the battery is not just another everyday component. So, let’s start going through Tesla’s journey of patents, development, and acquisitions, starting with one of the simplest, the tabless electrode.
The Tabless Electrode, Part 1
A completed battery cell consists of a can, a top cap assembly, a jelly roll, and the electrode tab connected to it. The jelly roll consists of a positive anode, a negative cathode, a separator between the two that may never touch, and an electrolyte liquid or gel (that will be covered more extensively under the section on Maxwell). Then, finally, the anode and cathode are both just slurries of paste applied to a current collector — a piece of copper foil in the case of the cathode and a piece of aluminum foil in the case of an anode. If you don’t know much about batteries, remember this, as it is the most important part and will be covered again later on in the Maxwell part as well.
If it helps you, visualize a battery toothpaste spread over aluminum foil and copper foil on one side and a small tab connected to the foil on the other side. Most of the energy is in this slurry, and the electrons indirectly travel between the cathode and the anode via the electrolyte during charge and discharge, and then current travels both in and out of the battery via the small metal tab connected to the backside of a current collector, one at the top of the anode and one at the bottom of the cathode. Meaning that all that electricity is stuck in a traffic jam on the way to the tiny current collector, causing friction and in turn heat— you know, that big old problem for batteries. (Side note: it’s actually resistance, not friction, but friction is a good analogy.) LG Chem solved the problem by not rolling up the jelly roll burrito. Instead, it cut the jelly roll into pieces, attaching a tab to each slice and then stacking it together as if it was an enormous multi-layer hamburger that then got squished under an industrial metal press (I think I might have to resume writing this article after having lunch).
Tesla instead decided, why not make the entire length of the jellyroll the tab? Technically, it wouldn’t even be a tab any more. Hence, it’s tabless. This way, the current can travel straight up and down, significantly reducing the heat produced, meaning you can either have less cooling or much larger batteries. Another benefit is that the batteries should last longer, which brings Tesla closer to the million-mile battery. There is one more very important aspect to the tabless electrode, but first we need to cover Maxwell to explain how Tesla achieved this.
It would probably be fair to say that when Tesla bought Maxwell is when everyone really started to chat around the watercooler about Tesla making its very own batteries, and this idea has been quite correctly hyped ever since. The importance of Maxwell drycells cannot be overstated. A lot of people have heard of solid-state batteries. Though, not everyone realizes that solid-state batteries are basically just lithium-ion batteries of some type in which the electrolyte is no longer a liquid or gel between the separators and electrodes, but rather a solid slice of material.
Why is this important? It all comes down to the pesky plague of dendrites that always get in the way. Normally, the cathode and anode are applied to the current collector as a wet slurry paste. After charging and discharging, spikes can come out of the slurry and go through the electrolyte, piercing the separator and short-circuiting the battery. If the electrolyte was solid, those dendrites wouldn’t be able to pierce it as easily. However, the technology still isn’t feasible, both because of the inhibitive price as well as the fact that the technology has conductivity issues — the electrons don’t pass as well through it as through a liquid or gel electrolyte.
Drycell technology is literally the very opposite of that. Rather than having the electrolyte solid, it’s the anode and cathode that are no longer a slurry, instead a powder that is sprayed on and calendared so hard that it can become its own self-supporting foil. It doesn’t depend on the current collector to keep its shape. In case you didn’t know, calendaring is a part of the battery manufacturing process in which two big spinning rollers compress the electrode in a similar manner to how one of those big machines rolls over and flattens new asphalt on a road.
With a wet slurry on a current collector, you need to be careful not to calendar it too hard so that you don’t rip the copper or aluminum foil. With drycell technology, once you calendar it once, it is strong enough to support itself without a foil or film underneath, and the self-sustaining calendared powder can be calendared a second time on its own but much harder, compressing the same amount of material to take up less space or have more material in there so that it takes up just as much space as a regular anode/cathode would — hence, increasing the battery density and capacity.
Another benefit is that because it’s all dry, it doesn’t have to go through a drying process. For a regular lithium-ion battery, the time needed to dry the electrode is often the bottleneck of production, and sometimes there are even two drying steps in the manufacturing process. Drycell tech is already dry, so it doesn’t need to dry. Even though there might be a second calendaring step, it would still save a lot of time and space. By using drycell, you get more battery, but you also solve the dendrite problem. While you probably don’t solve the dendrite problem as well as a solid-state battery would, the improvement is still significant.
The Tabless Electrode, Part 2
So there is one other aspect to the tabless electrode patent that I have purposely put after the Maxwell section. It was sneakily hidden in the patent. Here is the specific paragraph:
“Further, with continued reference to FIG. 1 and also as shown in FIG. 2, a second substrate 106 is posed over (e.g. stacked on top of) the inner separator 104. The second substrate 106 has a second coating 120 disposed on a side of the second substrate 106. In some embodiments, the second coating 120 may be disposed on both sides of the second substrate 106.”
Let’s decode that corporate lingo. As you now know, with drycell, the current collector during the manufacturing process doesn’t need to hold wet slurry to keep it from falling apart, so, technically speaking, there is no reason why you can’t have the anode (and cathode) on both sides of the current collector without smearing a paste on all the rollers trying to move the electrode along on the line. It is why in the image above you see that the conductive portion of the tab that covers the whole jelly roll has an electrically insulative material on top of the part that touches the current collector. Because both sides of the current collector are used, you can divide the number of current collectors needed in half.
I have heard at least one estimate that this could reduce the jelly roll size by 30% — or, inversely, allow for a 30% increase in energy density. (Side note: there is now some experimental technology that could do this with a wet slurry without adding extra drying steps, but with drycell this is still simpler.)
Maxwell Enables New Chemical Formulas
Most interestingly, it seems that drycell technology is also a lot more stable. Cobalt, one of the biggest thorns in the side of lithium-ion batteries, is a stabilizing agent. A battery can easily be made without cobalt, but at the higher energy densities we so desire for our electric vehicles, it just wouldn’t be very stable without it. With drycell, as well as other improvements to the chemical formula, the need for cobalt can be avoided, and thus entirely eliminated from the product.
Right now, Tesla uses NMC (nickel-manganese-cobalt) batteries for its energy storage products and NCA (nickel-cobalt-aluminum) batteries in all of its electric vehicles. A lot of people wonder what chemical formula Tesla’s next-generation batteries will use. The answer was basically hinted at in the previous paragraph. It’s actually a lot simpler than people think. The formulas will be NM and/or NA. As a matter of fact, this would not be unprecedented. A company named Sillion already sells cobalt-free batteries that they call NMx batteries. Though, the ones Tesla will make will undoubtedly have much higher energy density.
LFP in EVs — Will It Last?
There are some people and countries (especially China) that are heavily betting on LFP batteries that in the next few years will evolve to become LFMP batteries. However, the LFP batteries that are about to go into the Chinese-made Model 3s have just barely, due to some new tricks like cell-to-pack technology that eliminated the need for modules, exceeded the energy density required to meet the range requirements for the Model 3 Standard Range Plus (SR+) and are not likely to play a large, lasting role in Tesla’s future battery plans. Maxwell tech can also play nice with LFP — cooling systems could be simplified further, and if space isn’t an issue, it could be valuable for grid storage. The deal Tesla has with CATL for those SR+ LFP cells is just a marriage of convenience until Tesla manages to produce as many batteries as it truly desires.
The Other Aspect of the Chemical Formula
In some ways, the chemical formula of a battery is almost like programming. It is something that can be updated and doesn’t require new hardware to be installed in the factory (most of the time). Ever since Tesla developed the original formula for the 2170 cells with Panasonic in 2016, Tesla has been improving on the chemical formula every year. In fact, the LG Chem cells used for the Model 3 Long Range (LR) made in Shanghai are slightly better than those made by Panasonic in Gigafactory 1 (GF1) even though LG’s cells used to be worse than Panasonic’s. This is all because Tesla had a new chemical formula for 2020 that LG Chem implemented, while Panasonic was too busy ramping production to switch over to the new chemistry (information accurate as of April 2020).
While we don’t know how many versions of the chemical formula Tesla has developed over the past few years (likely one per year), we do know that the newer ones are referred to internally by Tesla as L cells and the previous ones were D cells. In general, this is just an interesting tidbit of information, but the point is that one way Tesla will keep improving its batteries year after year is by making improvements to the chemical formula.
The Fate of Tesla’s Battery Partners
From the standpoint of Tesla’s mission, we know that Tesla doesn’t like to place patent booby traps for the companies following in its footsteps and doesn’t usually take legal actions unless trade secrets are stolen or precious talent is poached. It could be argued that if Tesla was to supply Maxwell, tables electrode, Roadrunner technology to its suppliers, like Panasonic and LG Chem, by selling or leasing the custom factory machinery Tesla designed, Tesla could ramp up battery production more quickly.
One thing that has been made clear is that Panasonic isn’t going to leave Giga Nevada anytime soon, even if it just produces the same old 2170 cells. Tesla could build around Panasonic, or build with Panasonic. There is the possibility that Tesla will do this, as it could be in the best interest of the mission. Tesla would not only be Panasonic’s client, but also a supplier. Tesla might do this even if it isn’t necessarily in the best interests of company value and growth and could even prevent Tesla from cornering the market.
Sometimes it feels like Tesla has split personality disorder. On the one hand, there is Elon Musk with open arms willing to help, do what is best for the mission, and listen to customers. On the other hand, you have the traditional business side of Tesla that doesn’t respond to journalists, is willing to cut corners, and is often indifferent to unjust actions taken against customers who are unable to reach Elon Musk on Twitter.
Here is another major aspect of the battery that Tesla has plans to implement at some point down the road. Right now, within the battery pack, there are modules that hold the battery cells. Why do we need modules? Well, they do make the task of engineering the product a little simpler. In the original Tesla Roadster and the Tesla Model S, the idea was that broken modules or modules that have lost their capacity can be replaced. When it came to the Model 3 the idea was to make a battery pack that can last the lifetime of the product, one of the first steps toward the million-mile drivetrain and the robotaxi fleet.
However, Tesla still had a battery pack team, a battery module team, and a battery cell team, and the people in the module team did not volunteer to be terminated and/or transferred, so the battery module team did what a battery module team does — designed a better battery module. It is only now that the battery module team has been integrated into the battery pack team that the next pack won’t have modules.
On more than one occasion, Elon Musk has indicated that he wants to move to cell-to-pack technology and eliminate the module. How soon that change will come, however, is a big question. On the one hand, if Tesla added some extra glue in there to compensate for some empty space left by the 2710 cells, it could change over immediately. However, to let the cells truly fill a battery pack, the whole vehicle needs to be adapted to accommodate a battery pack with different dimensions. Then there is again the fact that somehow switching to CATL prismatic cells is easily doable, so I really look forward to a teardown of Tesla’s prismatic LFP-cell-filled battery pack. One way or the other, switching over to a cell-to-pack design could again increase the density of the battery on a pack level.
Tesla’s Custom Factory Machinery
One reason Tesla couldn’t vertically integrate battery production as it has been able to do for a lot of other components is because the battery industry is full of industrial secrets. Even in GF1, Elon Musk can’t just take a stroll through the Panasonic side of the factory unaccompanied or start winding gears on the machinery there. While there are industry-standard components and machinery for sale, they won’t make state-of-the-art products. For that, you need custom factory machinery that is not for sale.
Tesla now owns Hibar, Maxwell, and Grohmann Automation and is now fully capable of designing custom battery manufacturing machinery that will hopefully run circles around the technology of other battery makers.
One very important aspect on the road to 2 TWh is something we have seen before. Just as with the giant injection molding presses Tesla is installing to make parts of the Model Y in one go rather than assembled from hundreds of pieces, this is also something that we can expect to find in Tesla’s battery efforts. Tesla will undoubtedly make larger machines than have ever been made before, and one specific prediction I am guessing will happen is that the machines will be wider so that more square meters per second of electrode can go through the machines at once. These rolls are always cut into slices — the wider the machines, the more slices one can be cut into.
In Tesla’s battery production, I truly wonder where the bottleneck will be. My guess is the cell formation cycles at the end of the production process where the cells just sit still in a warehouse anywhere from 20 hours to days before the bad apples are taken out and the rest are sorted into groups with similar characteristics.
It would be awesome if battery cell production and vehicle production were under the same roof and the formation cycles could take place directly in the battery pack as it goes through vehicle assembly. That could save 20-30% in both time and cost. Though, it is unlikely to be possible unless Tesla has done some very creative revolutionary engineering. Thus, it is likely just wishful thinking.
The Sillion Acquisition
As part of my research in the past year, I have been digging around the patents of Tesla as well as the companies it has acquired. One of the lesser known ones is Sillion. The most interesting part of this acquisition is its non-flammable ionic liquid electrolyte. The description in some of Sillion’s patents are somewhat reminiscent of how the BYD Blade battery is supposed to be fireproof by ensuring that the chemicals in the battery require more energy to combust than is generated by a thermal runaway event. Translation: fire isn’t hot enough to combust the chemicals. With this, battery fires could be a thing of the past.
So, What Will Tesla’s Next-Generation Battery Be, And Be Capable Of?
So, before we poke some holes in existing information, let’s quickly recap what exactly Tesla’s news cells are and what they are capable of. My prediction is that the new batteries will include Maxwell’s dry electrode technology, which will increase cell energy density significantly to somewhere between 300 and 600 Wh/kg (current energy density at around 260 Wh/kg). I believe Tesla will increase the speed at which batteries are manufactured since the need for drying is eliminated, which also decreases the amount of space needed to make batteries, and of course will eliminate cobalt from the chemical formula.
Then, thanks to Sillion, Tesla will make batteries that are more heat resistant and hopefully non-combustible. Then, thanks to cell-to-pack tech, the cost of the battery is decreased and so is the time needed for assembly, in addition to the fact that more batteries fit into a pack. Finally, there is the tabless electrode, which allows for larger cells that last longer.
Here is the same information but a bit more broken down by acquisition/patent:
Maxwell’s drycell technology:
- Increases battery density to 300–600 Wh/kg (more battery fits in a can)
- Increases manufacturing speed (by eliminating drying)
- Decreases the floor space needed for construction (by eliminating drying)
- Eliminates cobalt
- Enables larger cells to be built without compromising cooling
- Makes electrodes double-sided
- Allows more cells or larger cells to fit in a battery pack (more Wh/L)
- Simplifies manufacturing process (time = money)
- Allows Tesla to make a fire-resistant battery like the BYD Blade
The Cell Size Dilemma
For more than a year now, I was fairly certain that Tesla’s next-generation cell would be bigger than the 2170 cells that it has now. The recent electrek leak is quite incredible to the point of unbelievable, but I have a feeling that is only because we don’t have the whole picture. Ask yourself this: does it make sense to have the same size cells for both the Model 3 as for the Cybertruck and the Tesla Semi? With any battery pack, it’s always about filling the pack to the brim with batteries, not wasting any space at all. From older visualizations, we know the Tesla Semi has multiple battery packs the size of oil barrels (square ones). When it comes to a skateboard battery, a Cybertruck might be able to have a taller battery pack than a compact sedan like the Model 3.
Most factory equipment for cylindrical cells can be adjusted for various cylindrical battery sizes. In that light, does a one-size-cell-fits-all approach have any obvious benefits? Personally, I don’t see it. With all of the improvements in technology, Tesla might be able to double the range of a car, but is that actually useful for the robotaxi revolution? Is increasing the height of the Model 3 battery skateboard to accommodate this new 54XXX(X) cell possible or necessary? Hell, these cells might even be placed sideways, decreasing the height of the battery pack and still providing the same range as a Tesla has today. (Just to be clear, that was a joke.) There is a good possibility that the 54XXX(X) cell is for the Cybertruck and/or the Semi and that Tesla will make the same 2170 or somewhat bigger cells for the S3XY models.
This larger cell would very much make sense for the Tesla Semi. Tesla, with new breakthroughs, may have improved battery thermals, but it’s prudent to not get crazy here. So far, one of the ways Tesla has kept its batteries cool is by having so many cells, so that during use each cell is not strained a lot. A Tesla Semi would probably need as many 54XXX(X) cells as a S3XY car needs 2170 or 18650 cells. Again, not a lot of stain on each cell except for charging, but that is what the drycell and tabless electrode help compensate for.
Throughout the past year, I have had my own theory on the new battery cell size, and it could still come true. First of all, for those who didn’t know, Tesla’s current cells are actually 21700 (21 mm x 700mm) cells, not 2170 cells. It’s just Tesla that calls them that. The issue with the 40 or 54 mm diameter cells is that, unlike 21700 and 18650 cells, they are not industry standard. In other battery research I conducted in 2018/2019, I made a spreadsheet with detailed information on 97 battery manufacturers. One of the things I noted was battery formats, as well as the sizes they produced.
The two most common cylindrical sizes above 21700 were 26650 and 32700. If Tesla is going to make a larger cell for the S3XY line that is not the 54XXX(X) size, its likely that Tesla will switch to one of those — probably 32700. Think of it this way: the 54XXX(X) cell is a lot taller than a 650 mm or 700 mm cell. Does Tesla really want to redesign all of its models to accommodate a higher battery pack? No, not really. So, basically, sticking with the 650/700 height is a good bet. By removing the modules, you have more space to the sides but not a lot more space above and below the battery.
The Terafactory Conundrum
Tesla has quite correctly said that to truly achieve its mission, it needs to get to 2 TWh of battery production to secure a fair fraction of the future EV market. However, getting there is going to be rather difficult. In this paragraph, I am going to exaggerate how much various technological improvements can help Tesla get a factory to 1 TWh to illustrate a point.
A terawatt-hour is 1000 gigawatt-hours. That is 10 times more than the 100 GWh a year that Panasonic was supposed to be able to make with its 260Wh/kg 2170 cells. If a battery terafactory is the same size as Giga Nevada was planned to be, and you have 50% more floor space because 2 drying steps in the cell production process were eliminated as well as the need for battery modules, then you suddenly get 200 GWh — nearly double the density to the 500 Wh/kg Maxwell promised. Plus, you get 384 GWh, 90% bigger 541100 cells, 730 GWh, 40% density increase because of double-sided tables electrodes, 1,022 GWh, and hallelujah, we made it to a terawatt-hour. But, basically, we needed a battery 4 times as dense, twice as much floor space, and cells that are 90% bigger in order to have a terafactory.
Does the world need terafactories? Indeed, it does. The world makes more than 90 million cars per year. 1 TWh is about 16 million Model 3 Standard Range Pluses. That is 17% of the world’s automotive passenger demand, and we haven’t even mentioned home and grid storage or industrial needs.
Though the idea was that a terafactory would enable Tesla to make many millions of electric vehicles with 60–100 kWh batteries. With 54XXX(X) cells, you either abandon the skateboard design or you have a vehicle with a battery pack much larger than 100 kWh but not nearly as many millions of vehicles as the world will be in dire need of. Yet another downside of 54XXX(X).
On Tesla Battery Day, I expect the California company to announce a 300–600 Wh/kg density battery in 3 different cell sizes, 54XXX(X) for Semi, 32700 for future passenger vehicles, and 2170 for existing products — and maybe even for battery partners. Furthermore, I expect Tesla to say that it wants to reach 2 TWh, that they will need 3–6 battery factories for that, and that in the future, each factory could potentially reach a terawatt-hour. I expect that initially in Tesla’s first phase, its next-generation battery factory will produce 300 GWh each. The batteries will last a million miles and be fireproof, and extend the range of vehicles somewhat (maybe 30%), but will not double it because that is not a priority and is not really necessary.
Bonus: Tesla Next-Next-Generation Batteries
Just for fun, what can we expect to come after this? One thing is for sure — Tesla has decided to support drycell technology with a roadmap to 500 Wh/kg batteries instead of solid-state batteries and is very unlikely to switch. On a funny side note for anyone wondering, drycell solid-state batteries where the electrode and electrolyte are solid are not likely to be practical. It’s like two crackers with no butter — they won’t hold together.
As for future chemistries, right now that is a big unknown. There are a lot of interesting competitors in that race. Lithium-air is a good possibility, with a theoretical density in the kWh/kg range rather than the Wh/kg range. Lithium-metal is a possibility, but a lot of things would have to go right in the development of that chemistry. Sodium-ion is an interesting option. What is especially important is that, if Tesla is going to make an electric jet, it has been pointed out that a high battery-to-mass ratio would be necessary — like filling up strange shapes and corners with flexible batteries, something that is not necessarily dependent on a particular chemistry type but is an important advancement that needs to be made and iterated on in the meantime. However, all of this is still a decade away, so I wouldn’t worry about it too much.
Have a tip for CleanTechnica, want to advertise, or want to suggest a guest for our CleanTech Talk podcast? Contact us here.