When 3G cellular phone networks came out they allowed a data path for each phone of about 1.8Mbps, as a maximum. But by tweaking the 3G and 4G standards, and allocating more and more of the network capacity to data, before 5G was launched, the telecoms industry was happily talking about several 100 Mbps per phone, with almost no change to the underlying base station technology. The standards bodies simply allowed for a protocol called high speed packet access (HSPA) where it better accommodated internet packets in radio, and where people expected to have less conversations on phones.

Expect the car battery market to go down the same route, with the key measurement here being watt hours per kilogram of weight (Wh/kg). Back in September 2020 Tesla had what is now referred to as Battery Day.

All that really happened on Battery Day was that Elon Musk and his team actually bothered to explain the design issues around lithium ion batteries and the tweaks that could be deployed to get around some of them. This is THE big difference with Tesla, by explaining its thinking and the thinking of other battery companies like Panasonic, LG, Samsung and CATL, and logically going through its architecture of choice, it was reasonable to assume that Tesla was years ahead of the pack and was leading in all things battery design.

This is not really the truth, but because it is the biggest customer in town (certainly the US anyway) it is entitled to talk about the technology it plans to use. So if anyone got the impression that this technology would be entirely “for” Tesla, because it owned the intellectual property, they have another think coming as other parallel advances are leaking out onto the market as are the Tesla designs.

So 18 months on from that Battery Day, two advances which came out this week show how the battery market is slowly growing the capabilities of batteries incrementally, in much the same way as the telecoms market grew data rates. It is using both evolution and revolution – with the focus on evolution first and foremost, and the opportunity to shift to a safer battery chemistry as its revolution, down the road.

The first announcement we noted was the arrival of the Qilin battery from CATL, and people in the know have been talking about its development since about Tesla’s Battery Day, mostly because it is its 3rd generation of packaging improvements.

The Qilin battery is mostly about a packing technology (cell to pack or CTP) which is its methods of how to put cells together into the tightest spaces. We saw similar breakthroughs in chip design in the 1980s as IBM internal techniques spread throughout the merchant chip world, and these are similar problems, how to dissipate heat, how to do it faster and more reliably, as the chips or the cells get closer together. In the same way the more processing power brings more heat, so too does more compressed battery power.

At one point IBM put liquid helium almost onto the surface of its latest mainframe chip to cool it down, and threw in a massive metal heat sink above it.

So one announcement was from CATL which is not alone in this packaging space, but it is on its 3rd discrete generation of packing design and it says it has outstripped the 4680 Tesla battery design in terms of watt hours per kg by this route alone, which is hugely impressive. The video on the CATL website talks about volume utilization efficiency, but as this is not defined, we won’t repeat the claims here – suffice it to say that it reckons it can beat 255Wh/kg, which on battery day Tesla said was where the 4680 would put it.

The 4680 has only just gone into production, at Tesla itself and Panasonic, in low volumes, so this does not necessarily mean that CATL is 18 months behind Tesla. It plans to build factories working to this new design by 2023.

There is a strange truth that you always have to remember on progress – if a company, and the best example of this is Apple CPU phone design – designs a component for itself, it does not have to design it for ALL of the other instances on the market – because only Apple is going to use that chip. And Tesla finds itself in the same position. It has “nailed down” the specific improvements that its next gen battery gives it, and wants those additions and nothing else. It is able to design its batteries around its car designs. CATL will have to offer a component to almost every battery maker on the planet for every other car design – some it can lead by the nose, others will want a big say in how things go – and it must satisfy all of them.

For years, Apple fan-boys have been impressed with how advanced Apple application processors are in iPhones, but at the same time few would suggest that Apple is more advanced than say Qualcomm in wireless chip design – so the market sequence is that Apple designs a new feature, puts it into its chips in 9 months and its produces a product with it in, in 18 months. Qualcomm starts at the same time or earlier, but puts the same features it into its general purpose merchant chips asking all customers for feedback, and delivers this after 18 months, and other phone makers rely on them in a further 9 months down the road, coming to the market well after Apple.

The battery market is being used by Tesla in much the same way. It gathers battery makers around and says “Who wants to have a big piece of our business,” and then asks “What features have you got up your sleeves?” and then it chooses some it likes, and sets a design in concrete, and puts it out for manufacture.

We know that the first 4680 manufacturer is Panasonic, but we also know from an announcement picked up this week by Korean media (the Elec) that Samsung has a 4680 in testing, and says it may cut the design in half and offer it as a flatter 4640, with some manufacturing upsides (like thinner cathodes being easier to roll up reliably with the anode and separator foil, giving better distribution of the electrolyte also.

These shorter variants are better for flatter designs but can be stacked on top of one another, and the publication speculates who will go for these – BMW, Volvo or Stellantis are the likely candidates – well at the right price they are always going to put this advanced technology (Wh/kg) into their cars, and this could be at about the same time as Tesla (in volume), or at least not long behind it – guaranteeing the key ingredient of range.

The CATL approach is another way forward and the packing technique can be used with either NMC or NCA or LFP, creating different energy densities with the 255 figure for the first two and LFP closer to 160 Wh/kg. Given that LFP is cheaper to make, lasts longer, offers more charge cycles, is safer and less likely to result in thermal runaway, in markets where range anxiety is not a major issue (i.e. anywhere outside the US) this will likely be the dominant technology sooner rather than later. CATL supplies all of Tesla’s batteries for its Shanghai factory for these reasons.

But the more advanced NMC version of the Qilin battery pack can offer a range of 1,000 kilometers, which may well suit the US market. The packaging technology includes putting liquid cooling between the cells, rather than underneath them, multiplying total cooling area by 4 times, and this can be placed under the control of a BMS, so that it can “manage” overheating events. It boasts 13% more power than the 4680, and will go into manufacture next year. The system will charge to 80% in 10 minutes.

So here we have a dichotomy between tubular and prismatic design, which offer different design advantages. If you think of CATL as the Qualcomm in the equation, this design could seep into multiple car brands, and Tesla can continue to diverge with the market, supporting the 4680 tubular design. There are other issues like structural integrity improvements so that the battery can be part and parcel of the car design in the CATL packing announcement. The proof of which approach is better will likely be delivered in which approach gets to high yields first – and given that Tesla has been talking about “money furnaces” this week, there is a strong whiff that the 4680 is a nightmare to get right.

CATL talks about an internal crossbeam, a liquid-cooling plate and a thermal pad from its previous generation, being all integrated into an elastic interlayer, with built-in micron bridges to accommodate changes inside the cell, improving the battery reliability throughout its life cycle, with more stable load bearing perpendicular to the driving direction, improving shock and vibration resistance for the battery.  See the video here. The CTP 3.0 battery pack even has a protective vent for thermal runaway risks.

Interestingly the LFP variant would likely end up with over 60% of the 1000 km range, so well over 600 km, which pretty much beats anything on the market right now, with the exception of the Lucid Air – certainly better than any Chinese built car – so the uptake of this will be considerable.

So this schism is likely to continue – with Tesla forcing most battery makers to go down the tubular route as well, and rival car-makers wanting to cover both designs, with CATL going its own way on packing – until the best of both worlds get overtaken post 2024 with solid state lithium ion which may hit real volumes until a couple of years later – by then at well over 500 Wh/kg.

Rethink analysis of future zero emissions transport: Harry Morgan

Heavy duty transport transition WILL rely more on hydrogen than batteries

Talking with start-ups: SunOyster Systems

Global Wind Power Forecast - entering boom phase by 2025

Talking with Start Ups: Elestor Hydrogen Bromine battery

Talking with Startups: SunlitSea, why Norway chases floating solar so hard

There are too few global lithium-ion gigafactories for EVs

To contact Rethink please email us at [email protected] Alternatively call us on +44 (0)1179 257019