Cell manufacturing - why so many fail along the way

As currently I am holding position of CMO in the batteries manufacturing company, don't expect any data, graphs or even funny pictures. This article is only my personal opinion, directed to my friends all across the batteries world. They will understand...

Historic context

Before getting to the main point I’d like to give brief historic context.

In year 2008 Tesla presented its Roadster. Toyota launched Prius one year later and Nissan presented Leaf in 2010.

To be fair – in year 2008 there were also Beijing Olympic games, which really accelerated e-mobility in China, but this was only slightly noticed in the western countries.

Back then HEV was treated as a good solution for fuel economy improvement for taxi drivers, and BEV was seen as a toy for wealthy geeks from West Coast California.

The real breakthrough happened in September 2015 when famous Dieselgate was announced to the public. Just one year later Tesla announced release of its Model 3, which – unlike Model Y was much more affordable.

Thus year 2017 was a time when entire e-mobility concept gained real momentum.

Announcements of new models started to pop up like crazy: Jaguar I-Pace, Audi E-TRON, Mercedes EQC, Porsche Taycan, VW ID.3…

But the promises from already late OEMs soon hit a wall of reality: to produce EV you need a lot of batteries, and supply of cells could not cover the enormous demand.

European OEMs hastily signed contracts with mostly Korean manufacturers: LG, Samsung and SK Innovation to bring production to the Continent, and soon realized how big dependency it throws them into.

So if year 2018 was a time of panic buying, the 2019 started to show cell manufacturing domestication trend. So Northolt has been announced, and then avalanche of others: Northvolt Zwei, Britishvolt, Italianvolt, Norvegianvolt, Morrow and finally VW with its ambitious goal of opening six cell plants until 2030.

The period between 2019-2022 was a time of great optimism and bold statements, but in 2023 we start to realize that the claims are not followed by execution. The best example is of course latest case of Britishvolt, but let’s not be too harsh on them as all the cell projects are running behind schedule, including Northvolt and Tesla itself!

Of course the companies must say something to investors, so we can hear about raw materials shortage, covid, chips and other electronics disruption etc.

Well, it takes only couple minutes of research to find out that the curves of declared vs actual capacities of the battery manufacturing, started to split even before covid.

In my opinion the reality is much more grim, as it is a systemic problem associated with hi-tech, hi-volume manufacturing.

Let me explain...

Battery manufacturing is boring as hell - and the investors expect fireworks

Linkedin is filled with sensational graphs showing how big the battery market is and how fast it will grow. Within only few years all the vehicle industry focus shifted from the vehicle itself, to its source of propulsion - cells.

We do not discuss about new body, aggressive look, more horsepower - instead everyone is awaiting next battery day from Tesla, next batteries swamping station from Neo, or implementation of blade cells from BYD.

This hasn't passed unnoticed to the people of business, who sense enormous opportunities to get into this market.

So for example Northvolt was able to raise $5.49B?over 11 rounds - huge sucess. But if you are looking for money from investors, you must show them some vision, have good pitch, make flashy and noisy BOOM!

This is not that difficult with the round one, where you show future market share, and stylish renders of the plant, but round 1 is just beginning of the journey.

So let us pretend together that we are building the cell plant.

After fiery speech and excel projections we have got first financing! Yeah baby jeah! Time to get to work.

Qualified personnel shortage

Very quickly we find out that there is not enough batteries specialists in the market, so you we to compete to get them. We take whoever wants to join us which results with assembling multinational team. This poses Babel Tower challenge, where different styles of work need to somehow produce unified outcome - oh, now we start to understand why the largest corporations build this tolerance and inclusivity culture...

We also need to recruit many engineers a long time prior to production, which also raises cost. This is caused by the fact that when the production will finally ramp up, its impossible to simply man the stations overnight.

So we need to hire a lot of people - our expenses are rising and we haven't even started any meaningful work.

Manufacturing equipment suppliers

Alright, after initial chaos we start our work with selection of the chemistry and form factor. Obviously we don't have any knowledge, so - lets just copy (correction: take inspiration from) CATL, LG or Tesla. It will be easy - we will just go to the equipment suppliers and say: give me the same stuff Tesla has. After all, if you want to produce screws or nails, you procure equipment and that's it - you are ready to go!

But here is the second unpleasant surprise: in the battery business, the equipment manufacturers have limited knowledge about the call manufacturing process! This is caused by the fact that their typical partners are the companies who very well know what they want from their machines. The equipment vendor has limited access to the process know-how post SOP and is not even seeking to gain this knowledge, as this could lead to accusations of technology transfer from their premium partners - huge cell companies like LG or Tesla.

What is more, the cell manufacturing process is various for different stages of production: Mixing is a typical chemical process, while slot die coating reassembles paper industry (roll to roll transfer), assembly is a typical automated mechanical process with cutting and welding, while cell formation is a pure electrochemical process, where all the operations cannot be seen otherwise than through the cell's electrical parameters. That is the reason why usually equipment manufacturers focus on the only one process: mixing, coating, winding and assembly, or formation. So if we were counting that we will just pay huge money to the equipment supplier and he will set up production for us, then we were wrong.

Some of you may say - well this is ridiculous! No-one is going to the equipment suppliers with query to set up entire process. Well - ask any single equipment supplier like Wuxi, Manz, Eirich, Zeppelin, Koem etc. They will have pretty good stories to tell you...

Alright, back to our journey - We don't have cell design, and suppliers are not going to partner with us without it (they are evaluating their partners as they don't want to waste resources on the lost cause).

Cell design - blocker for any rookies

So we need to look for some know-how providers. There are such entities and we can divide them into two parts:

1. Private-owned business
2. Academia

I will start from the second one. There are very good academic entities like Fraunhofer Institute, where people have understanding of batteries and are implementing a lot of innovations. However, the issue is that academia functions mostly due to the government-based funding, which put a lot of constrains into its business activity. How do you feel to fund research which will have to be at least partially open to public? What is more, the academia is working at its own pace and its own methodology. It gets somewhat philosophical, but although tech business likes to say its based on science, in reality business and science are the two separate worlds.

Science is based on axioms and theories, then it forms hypothesis and as long as hypothesis is not proven, it cannot become new theory. In other words the pace of academia is linear and iterative - unless we are not 100% certain about something, we do not incorporate it as such.

On the other hand business is all about predicting based on the limited information we have. Thus business doesn't wait until the trend is obvious and proven - it cuts corners, it fill the gaps and make calculated risk guess. If it manages to the bull's eye then it earns big money, if it misses - dies.

So the discussion between business and academia is neither easy neither pleasant for both parties. As those two logic systems are not fully compatible.

Therefore very rarely the project of development of the cell is given to the Academia - usually the cooperation has much narrower scope, like special additives to electrolyte, separator coatings etc.

So we must consider option No. 1 - business aimed at providing know-how. Here again we encounter challenges, as those companies usually have of the shelf technology for older formats like NMC532 18650 cylindrical. So if we want something like High Nickel 4680, or Sodium stacked pouch blade cells, they will struggle to deliver it. So we are facing very big decision now: whether to pick old, proven technology, or try to develop (and I mean develop) state of the Art.

External consultants support

But, this dilemma must be put aside for a moment, as we have just realized that our round 1 financing is dwindling as a staggering pace and we are already behind the schedule given to our investors.

So we look for some consultants to help us out, very quickly we realize that independent consultants don't have a full picture of such a projects, as those who have are very quickly taken out from the market (guess how I have ended up where I am currently ;) )

So with the last drop of money we have, we forfeit our pride and go the one of the big consulting companies like BCG. Here some Senior Partner cheer us up with a little smirk on his face saying, that we are not the first one who though it will be a piece of cake.

Nonetheless with the help of professionals our project is shaped into realistically looking roadmap: chemistry, form factor, A, B, C samples, equipment selection, layout development, utilities calculation, raw materials sourcing, scrap disposal methods, headcount calculation and many many more workstreams are defined.

This is now looking professional, but it also shows that the cost will be 5x what we expected at the beginning. So its time to meet investors again.

Round 2 - new start for some, end of the journey for the others

If we are again giving fiery speeches and try to spread enthusiasm, then it means we have learned nothing and we are done - no idiot will get big money twice.

So this time our pitch is lower, we don't gesture so vigorously, and many topics are explained by our BCG friend. We talk about the priceless experience we have got, about lessons learned, and that this time our pricing and estimations are accurate and realistic. We are humble and full of respect for the towering difficulties of this task. And the timeline changes from 1 to 3 years.

OK, so we get the money and hope that this time our consulting friends will lead us from the beginning to the SOP.

But very quickly we discover that consultants can provide networking, looking for qualified people, prepare preliminary estimations and forecasts and fill some very specific holes through their network of experts.

But they are not there to build our plant. After all this is our business - not theirs.

Alright, maybe this is not what we expected, but now at least we have defined workstreams, so start to build a team: raw material procurement, construction, process, design...

And the real, dull work begins.

All those workstreams are connected to each other - without design, there is no process, without process there is no equipment, without equipment there is no layout and utilities requirements, and without that we don't even know where to put this plant, due to the environmental constrains like access to water.

So all the teams starting to work, the machine slowly moves somewhere, but we are not entirely sure whether this movement is forward.

This stage is very frustrating, because things are getting delayed, but we don't even know why exactly - civil engineers are constantly requesting utilities, process constantly wants design, and equipment manufacturers are still negotiating terms and conditions of FAT and SAT prior of signing PO.

All this activity is visualized with hundreds of excel files, formulas, calculations - all of which is unreadable for investors.

Something is happening, but what exactly?

And here again we see a big challenge, as many activities which we think we knew, like sourcing or construction, reveal their battery-related characteristics.

Sourcing of raw materials and components

Battery manufacturing combines the most painful challenges from many industries - it is the same difficult to get PLC as in all other high volume robots-saturated manufacturing, but at the same time there is that raw materials headache known only in the chemical industry.

Basically many raw materials providers will not talk with us unless we can prove that we will be able to procure bulk amounts in the future. It is like discussing with investors for the third time: we must convince them that we are reliable partner - and its not easy to do this without running factory and previous experience. So even if we succeed, it still consumes time, and our design team along with external know-how provider need few hundreds kilos of the target material in order to start their trials.

We are in the dead lock: without raw materials we won't have cell design, without that there is no process, no equipment, no layout, utilities... Progress is stalling, managers are getting stressed, equipment suppliers grow inpatient and start reminding us about their lead time.

At the same time we still need to get design of the additional components of the cell like cans for cylindrical or prismatic, tabs, pouches, tapes etc. Those are small things but all of them needs to be covered.

Same story is with electrolyte and the collector foils - and the copper foil is not mechanically rolled kitchen foil - it is a special material designated for batteries, manufactured by the electrodeposition method. If we buy wrong one, then grease residing on it after mechanical process will lead to active material adhesion defect.

Cell design - scalability challenge

Another challenge is the cell design itself. Even if we have chosen a specialized company to design cells for us, they operate on the laboratory, or semi-pilot scale, with a lot of manual operations (controlling web tension during winding using operator's hand is completely normal, but it doesn't give much accuracy).

The exception is UKBIC, but again - this is a government funded entity and choosing semi-pilot lines like this one limits form factors you can select.

So even though progress is being made to design the cell, this doesn't necessary provide information required for manufacturing at scale. DFM - design for manufacturing will be impacted, which will later act detrimental to the OPEX of the manufacturing process.

Another challenge is the attitude of the design team. Typical R&D usually focus the most on electrodes themselves and electrolyte. They are looking for energy and power density, coulombic efficiency, DCIR and cyclability. Pouch, tabs, laser welding methods etc. are usually not the center of interest. Many projects were delayed particularly for this reason - minor components were not covered by anyone as they were located in the competence gap - in between responsibilities of the design, process and procurement departments.

I could write much more on this topic, but this article is getting long enough, so lets move on to another challenges...

Layout, scrap and safety

There is no factory without... factory, so its time to start looking for land. For that the first thing is to know footprint of the plant and all the auxiliary areas like, carparks, social spaces, utilities etc.

The backbone of the entire plant is the process and the process is dependent on cell design. But lets try to stop complain and start doing something productive. So we arrange space for mixing, electrode, assembly and formation areas. Some of those spaces will require dryroom conditions, some of them will be cleanrooms. Cleanrooms will require overpressure to prevent contamination flowing in. At this point we must know very battery specific things like: avoiding temperature gradients with respects to height in the cell formation, special anti fire measures, limiting numbers of airshowers so that environment conditions are easier to met, special floor for heavy AMRs.

As every process is handled by different equipment supplier, its tough to get uniform information from them on how to arrange our layout. Another thing is to make sure that we didn't forget about inter-process links, so we won't find out that our processes are not compatible logistic-wise (for example assembly conveyor using different tray than formation one).

At this stage we need to think about how to deliver components into production line and how to remove scrap. If our dry scrap transportation route leads through cleanroom area then obviously we can expect troubles. But the problem is that at this stage we don't even know what type of scrap we can expect. The finished cell is obvious, but there is more than 50 different type of scrap in our production line and without prior experience we won't be able to define it.

For scrap we also need to designate the storage area and incorporate solutions in case we are handling anything toxic, flammable or explosive (and we are).

Lets also not forget that our equipment must be somehow inserted inside the building, so if we won't plan this ahead we will have to disassemble sections of walls.

Safety is a big concern and the biggest impact on it comes directly from the factory's layout. Will we separate formation from assembly completely, or we put them in the single building? Will we store wet and dry scrap together? Will we use centralized electrolyte supply system? How tight we are going to pack aging racks on formation and how high we are going to stack them? All those questions have great impact on safety later on, and unfortunately almost all of them must be addressed on the early stage of the project, where our knowledge is very limited.

Utilities

This topic is closely related with layout as the most utilities we are going to use are there to sustain desired environmental conditions (RH, temp.) and they are influenced by the volume of the production area. However we are also use a large amount of utilities for processes like oven drying of the web, vacuum drying of the electrodes, high temperature aging, and of course charging of the cells.

A common trap here is not taking into consideration the frequency of air-shower or cargo gate usage, as with every opening of the door we are exchanging outside air with the dry one - but how on Earth can we predict that on that early stage of designing?! I think you are already start to see some pattern here ;)

Another thing is to predict the number of personnel kept inside the plant, and especially dryroom - if for example we take optimistic scenario that the quality of the cells will be ensured only by 100% automated inline inspections, then later in case of need for human cosmetic inspection we may exceed the number of people inside the dryroom, and the people generate a lot of moisture in the air - I also don't have to remind you that number of employees will also affect air shower usage frequency.

And as layout impacts utilities, the utilities impact layout as well - we need to place DHU, chillers, compressors etc. Everything is interconnected here.

This means that if we have hired wrong people, who are afraid to take risk on their shoulders, and always awaiting for full scope information from others before making any decision, then we have a bunch of people who are passive, bringing no solutions and playing blaming game for lack of progress - very toxic situation, which wears down the most active members of the team.

Documentation and process preparation

Everything is moving forward, and so the process. We obviously haven't finished design, we don't have layout, equipment is not finalized but especially if we are going to supply our cells for automotive OEMs, we need to start preparing all the required documents: DFMEA, PFMEA, CP, SOP, list of CTQ and CTP parameters... All of this will be required from us and without that we won't pass any audit.

Obviously at this stage we don't know how our machines will look like, as the PO has not been submitted yet, we don't know any parameters, let alone CTP or CTQ.

Nonetheless there will be never a right time to start preparing documentation - so its better to start right away, even if it is more thought exercise than real documentation creation.

I think I owe you some explanation here, as you can ask: what is the point of doing this, if the quality of documentation will be low. That is true, but we need to prepare our engineers for future cooperation with equipment suppliers. If we will not introduce regime of parameter based discussion, tabular form information exchange with: SV, RV, analogue, digital, online, offline categorization of parameters, discipline of revision control (change management), then the discussion between our and equipment supplier's engineers will be chaotic chit-chat bringing more and more confusion.

You don't believe me? And what if I tell you that cell manufacturing line has more than 30 000 online parameters?

At the same time we must think about preparation of the process. Design team is preparing cell specification, but parallelly we need to prepare manufacturing specification, which is used for cell stack-up analysis. Every manufacturing operation has its margin of error which corresponds into tolerances. However if this is not controlled, then quickly we end up with ridiculous variation of the cells' parameters. This is controlled by defining every sub process tolerances - again, without knowledge of the machines we won't know what tolerances can be realistically achieved.

Technical cleanliness

Confused already? Wake up as we haven't finished yet!

Cells manufacturing is unique for one more reason: metallic contamination is one amongst the most hazardous things which can happen to the cells. Its severity lies in the fact that it is difficult to detect. There are methods of monitoring voltage drop (k-value) after formation, but in many cases they will pass unnoticed. This can lead to slow process of dendrite formation and at some point can result with field issue.

For that reason Brass, Zinc, SS and many other alloys should be maximally reduced from the process. Even SS bearings are often replaced with the ceramic one. In some companies they even change Zink coated air ducts into polymer one!

Anode and cathode areas must be separated to avoid any cross contamination.

The air movement inside machines should not be a turbulent one, and should be directed from top to bottom (no horizontal flow).

Special care should be given to the CDA installation, as any mishandling during pipeline installation will result with air contamination which will require months of flushing the circuit with necessity of clever cutting out sections of the pipelines and trying to flush contaminants out.

There are many those little things like premature HEPA filters installation, or lack of internal air purification circuit (in that case OPEX is skyrocketing).

Yes, cleanliness is one amongst the most challenging topics in the li-ion cells factory - let alone that I have listed almost 200 contaminants/scrap types on the production line. Every place of occurrence must be thoroughly analyzed and prevention/detection means need to be implemented.

It would be much easier if there would be a good standard to follow, but VDA19 is not fully compatible with cells. For example particles extraction methods are not suitable for wet cells (rinsing? Air blowing?).

Equipment suppliers - design freeze, FAT and SAT

Finally the cherry on top: discussion with equipment suppliers. Those guys will be your best friends for the time being. You know almost nothing - they know a lot. You are green - they have long experience with biggest players from this market.

So if you start this dialogue from the position of requests, you will regret that very quickly. The truth is that if you don't have large experience in the cells manufacturing, and they have delivered Gigawatts of equipment production capacity. The only time period where you negotiate is during price bargaining, later it is more following than setting directions.

So in our scenario we try to discuss how the cells are made on their machines, and we are getting hints how to adjust our design to meet their machines expectation. Hey! Shouldn't be the other way around? We wanted to be so innovative! Well in that case they will say: we cannot guarantee quality of the final product, and this is not the thing we can live with at this moment. In this scenario we don't have much corporate leverage to negotiate from the position of strength thus design freeze is rather formality where the equipment is shown and you can do some really minor changes than anything significant.

Submitting PO is also the moment when we need to set up FAT and SAT criteria, but if we haven't even defined process and quality parameters and their tolerances, then how on Earth we can define Cm and Cmk indicators? Of course we know that for CTP and CTQ Cmk/Cpk will be 1.33/1.67 as this is pretty standard knowledge, but to calculate them we need LSL and USL, so defining Cmk/Cpk without tolerances is more theater than a real engineering activity, its like saying we will fit between A and B without defining what is A and B.

Creative chaos

So as we can see all the workstreams are going their own way, with many interdependencies, where very often A is a function of B, B is a function of C and C is a function of A - the chicken and egg problem. This is very frustrating and requires flawless teamwork, as if any member of the project stops contributing, the entire thing changes into marble machine, which burns money at a staggering rate.

At the same time it is not easy to evaluate the progress using standard project assessment methods, for that clear milestones with relations between them would be necessary, so that if the milestone A is not reached we couldn't require that the work dependent on it would start. But this would work only if the timeline were extended to 5-6 years, and nobody will give financing for pilot line which will be set up within 5 years.

So all the activities must go in parallel, managers must withstand enormous pressure and all the engineers must maintain their work discipline.

Its a hardship, hell actually...

But if we are able to survive the first months of this utter chaos, then some sort of work routine kicks in. Every day we push our work a little more forward, everyone has forfeit his pride long ago, so cooperation is less ego-based, as we start to realize what monumental the task is, and that no-one can complete it by his own.

Everyone just quietly work towards the ramp-up phase.

Job well done - happy ending?

But those who think that setting up of the cell plant is the most difficult task are actually wrong. Batteries kill many startups at the pre-production stage - that is true, but the post SOP period is the real trial by combat.

At the initial stage yield is low, and the raw material cost a fortune. Every single day of high scrap generation is a significant loss, which tempts to widen the quality specs - to give us at least little bit more breathing space.

But the process stage is a completely different story and will not be described in this article - Its already too long ;)

At the end of this article I would like to thank whole OLA Electric Team, for enduring with me throughout this most labor-intensive, chaotic period.

We are lucky anyway, as our enterprise has Battery Innovation Center, where R&D has started to design cells long time ago, we have Indian Government support in a form of PLI scheme and we have this strong negotiating position thanks to the reputation of OLA and willingness of any partner to get into colossal Indian market.

We can reach far, because we stand on the shoulders of a Giant.