By Sandeep Unnikrishnan | LionVolt CTO
Key takeaways
- Energy density is one property among six, and they are always traded off against each other. Push one and you pay somewhere else.
- Cells are tuned per application. A grid-storage cell and a drone cell are engineered for nearly opposite things.
- A cell can look excellent on paper and not that great in a real-world application, because published numbers are typically measured under conditions your application will hardly see.
- The number worth optimising is not the peak figure. It is the total usable energy under certain conditions and across the lifetime of the cell.
- Chemistry decides which curve you sit on. Electrode architecture decides the shape of the curve.
——–
Every few weeks another battery announcement arrives, and it arrives with one number attached to it. 400 Wh/kg, 500 Wh/kg, and sometime a first laboratory result turned into a headline.
I understand why. One number is easy to compare and easy to remember. But if you are the engineer who has to put that cell into a drone, or a robot, or an aircraft, that figure on its own tells you very little about whether the battery is any good for the thing you are building.
So when somebody shows me a single number, my first question is not how big it is. Rather: what did you give up to get there?
I want to spend this article explaining why that is the better question. Not to argue that energy density does not matter, because it matters enormously. It is one leg of the animal, and the animal has about six.
The ceiling is real, but it comes in shades of grey
Let me start with the ceiling, because that is usually where the conversation opens, and the honest answer is that there are many shades of grey in there.
Look at automotive. There has been a strong move towards LFP, lithium iron phosphate, which is a cathode material paired against graphite on the anode side, sometimes with a bit of silicon mixed into the graphite to lift its capacity. It is less expensive, and more than half the industry has moved that way for affordable cars. At the other end you still have NMC chemistries in the more expensive segment, where the car needs performance and energy density.
Now, why do I call out these two chemistries? Because they cover the spectrum. LFP cells start below 200 Wh/kg at cell level, maybe 180, let’s say 200. NMC takes you up to something like 260 or 280 at the top of the range. To date, if you wanted to go beyond that with conventional materials, more and more silicon was added to the anode.
And why silicon? Because on the anode side, capacity is the limiting factor for the cell, so improving the anode is what increases the energy density. Graphite stores lithium by slotting ions between its layers, roughly one lithium atom for every six carbon atoms, which gives it a theoretical capacity of about 372 mAh/g. You cannot argue with that. It is a property of the material. Silicon holds far more lithium per gram, so silicon is what the industry reached for.
It worked, up to a point. But that is also where the bill arrives, at least with conventional silicon. A solid silicon particle swells substantially when it takes up lithium, on the order of 300 percent at full lithiation, and that expansion and contraction, cycle after cycle, damages the electrode. So, you buy energy density and pay in cycle life. Now, the industry is not sitting still here. Universities and several companies are engineering around the swelling, with porous particles and structured anode designs that give the silicon room to expand, and those designs can tame a good part of the problem. The interesting thing is the kind of fix that works: a structural one. Structure, though carries its own price, in cost and in the volume the engineered pores occupy. The trade gets softer. It does not go away.
So yes, there is a real ceiling. The materials have limits and we are close to them. Silicon is how the industry has pushed against that ceiling so far, and it bought real energy density at a price in cycle life that structure can soften but not remove. If you want to go meaningfully beyond it, you need another solution. I will come to that.
The ceiling is only half the story, and it is the half that gets told.
Energy density is one leg of a six-legged sheep
Here is what a single number hides. A cell is not optimised for one property. It is tuned across several, and it’s a compromise. And the tuning is application specific.
If I name the top KPIs today, energy density is one of them: how much energy you get per kilogram or per liter, which in a car means more kilometres on one charge, and in a drone means more flight time, more range, less mass and less space for the same job. Charging speed is another, and it is underrated. While a battery is being charged it is not available for use, so charging speed quality changes everything around the use case of the cell. Discharge power is the third, and in our markets that means takeoff and landing, or a robot lifting something heavy. Then cycle life, which depends entirely on the market. Manufacturability is another one not to forget, because a cell that works once in a laboratory is not a product. And cost, which sits on top of all of it.
I sometimes call this the five-legged sheep, and then I have to add a sixth leg for manufacturability halfway through the sentence, which tells you something about the problem already. You choose which leg the animal puts its weight on. The others carry less.
Now, why am I painting the whole spectrum here? Because these KPIs are always traded off. It is very hard to get a cell which does everything for a certain application: high energy, high power, high charging speed, long cycle life, manufacturable and cost-effective. People have been chasing that Holy Grail for as long as I have been in this industry, and every time, the tuning you do in the cathode, the anode, the electrolyte or the cell design costs you on one of the other legs.
Look at cycle life across applications and you can watch the trade being made in public.
| Application | Typical cycle life | What the cell is tuned for |
| Grid storage | 6,000 to 10,000 | Cycle life above everything, weight is less important. |
| Automotive | 2,000 to 3,000, minimum around 1,500 | A balance, tilted towards longevity |
| Aviation | 1,000 to 1,500 | Energy density, with life still mattering |
| Consumer electronics | 600 to 700 | Energy and volume, in a device replaced in three years |
| Light mobility, robotics and UAV | Hundreds of cycles. | A balance of energy density and power, almost above all else. |
Industry-wide illustrations of how cells get tuned.
A grid-storage cell and a drone cell are not the same product with different labels. The grid operator has a backyard and does not care about size, so they will happily take a larger, less energy-dense chemistry/design that survives several thousand cycles. The drone engineer needs every gram in the air and will accept a cell that lives a fraction as long.
Which brings me to the question I keep coming back to. Why should one cell fit all applications?
For a long time the industry answered that with standardisation. Build one dominant format, make it in enormous volume, drive the cost down. It has its advantages, and it is a large part of why batteries got cheap. But the applications have diverged too far for it now. A cell that suits a car, a drone and a warehouse robot equally well is a cell that is mediocre in all three. Customise the cell to the application and you give the engineer room to be creative. A balance is often required to benefit from economies of scale.
Chemistry A or chemistry B: why you cannot have both
When I explain this without the jargon, it goes like this.
Assume there are two chemistries. Chemistry A gives me energy density. Chemistry B gives me cycle life. Take the drone case. The engineer wants energy density first, and will trade some cycle life for it depending on what the drone has to do, so I choose chemistry A. Go to grid storage, where they focus more on cycle life and at a system level and less about energy density, I choose chemistry B.
And then somebody always asks the obvious question: why can’t I have both?
That is where material science comes in, and people have been striving towards it for years. One answer was to stop taking the extremes. Do not go all the way to advanced chemistries for energy, do not go all the way to the stable chemistry for cycle life, combine them and sit in the middle. Some cycle life, some energy density.
But you can feel the trade underneath it. If I want the best cycle life I know of, I pair the stable LFP cathode with graphite, and I accept a charging rate of maybe 1 for that graphite, not more (with specialized designs in optimal scenario’s this could go up to 4C). I am tuning for one KPI. I lose heavily on energy density, because neither of those materials is the best choice for high energy at cell level.
Shift the cathode up and swap the graphite for lithium metal anode, and voila you get energy density.
What do you lose? The cycle life and charging speed crash down. In a conventional cell design, with the lithium metal sitting as a flat layer, the cycle life and charging speed are compromised.
Why you cannot improve everything at once
The reason is physical, and you can see it if you look inside. So bear with me, I am going to zoom into a cell.
If you cut open a battery cell, whether it is the flat pouch in your phone or the cylinder in a power tool, what you find inside is a stack of layers, often stacked as a sandwich or wound up like a roll of paper, which is why engineers call it a jelly roll. Unroll it and you see the stack. Unwind it and you see a stack of layers. Two metal foils, aluminium and copper, called current collectors. Coated onto them are the cathode and the anode. In the middle sits a separator, a porous polymer that behaves like a sponge filled with liquid electrolyte. That liquid soaks the cathode and the anode as well.
That construction is what makes a battery work. When you draw power from a battery, electrons leave the anode, flow out through the copper current collector and
Now zoom into the electrode coatings on the current collectors. Each one is made by mixing particles into a solvent, casting that onto the foil and drying it, which leaves a fluffy layer held together with binders and additives. To compact it, the layer goes through calendaring, which presses it under load. What comes out is typically a porosity of around 25 to 35 percent, depending on the manufacturer.
Look at that porosity for a moment. Too high a porosity is a waste of space. And you have to fill the voids with liquid, so you are adding the weight of the electrolyte on top of it. The empty space costs you twice.
So if you want more energy density, an obvious move is to compress the layers further. Mix smaller particles with bigger ones, bring the porosity down towards 25 percent. Energy density improves tremendously, because the layers are more compact, more of them fit into the same cell, and you need less electrolyte to fill what is left.
But doing that, what do you trade off? That is where the importance comes in. The ions now have to travel a far more tortuous path. Everything is squeezed together, so they do not have the freedom anymore. They have to find a way through. You have gained energy density and lost power and charging speed, in the same operation.
Push the other way and it mirrors exactly. Add more conductive material inside the cathode, open the pores up, make it easy for ions to move quickly, and you get your fast charging. You have just handed the energy density back.
This is not a failure of engineering effort. It is the shape of the problem.
Brilliant on paper, less so in an application
Let me give you the cleanest example I know, because it shows how far a paper number can drift from reality.
Do not change the chemistry at all. Keep every component of the cell the same. What is left to improve? You can change the geometry. We typically coat electrodes at around 100 micrometres, so why not half a millimetre? Why not a millimetre?
What you gain is the active material ratio. The current collectors provide a pathway for electron transport, but store nothing. Together with the separator and the cell packaging, they are so called parasitic losses. They add necessary inactive mass and volume and are reducing the cell-level energy density. So, you do not want them there but you cannot build a cell without them. Make the active layers thicker and that parasitic fraction shrinks, so on paper you see an enormous proportion of active material and very low parasitic losses. Best optimisation you have ever done.
Then you put that cell on test and start charging and discharging, and you find you can only use maybe 10 or 20 percent of the cathode and the anode. At any decent charging or discharging speeds your application actually needs, the ions can never go in that deep. You have built a cell that carries a great deal of material it will never utilise.
On paper it looks great. In the application, it is much less useful.
De-rating: the energy that shrinks when you use it
There is a related effect that engineers meet the first time they put a new cell under real load.
Some cell manufacturers claim energy-dense cells, and the moment you put current through them, the usable energy density drops. In a cell this is called unintentional power or capacity de-rating.
The published figure was measured under ideal specs: a stated charge rate, a stated discharge rate, a stated temperature. Put your real duty cycle through the same cell and the promise crashes. Legally, they are right. They stated the conditions. But when it comes to an application, it is not ideal.
Which is why the first question about a number is never how big it is. It is: measured how? Cell level or material level. At what capacity. At what charge and discharge rate. At what temperature. On the best cell they ever built, or repeatably across a production batch. Nominal energy, or the energy you can actually use.
And then there is the question, the one I think matters from a sustainability perspective, and hardly anyone asks it.
Total usable energy: the overlooked metric
Here is the number I would not ignore: total usable energy. The energy a cell delivers per cycle, multiplied by the cycles it survives. Both halves count, and the peak figure only tells you one of them.
Now, read that from the energy side, because this is the part that gets missed. Every step up in energy density is more energy per cycle. For the application like drones, that means longer flight or a longer journey between charges, fewer stops at the charger for the same amount of work, and less downtime while it sits plugged in.
For the math, it means the cell does not need the same cycle count to come out ahead over its lifetime. A cell that carries more per cycle can live a shorter life in cycles and still deliver more usable energy in total, with less of that life spent charging. The trade only stops paying when cycle life collapses altogether. A cell with a spectacular headline number that survives five cycles has delivered almost nothing by the time it is scrap, and since sustainability in batteries mostly comes down to how much work you get out of the materials you dug up, that is worth avoiding.
That is the framing I would encourage any engineer to adopt. Take the peak number and the cycle count as inputs, and judge the total usable energy over the lifetime of the cell. It sits closer to total cost of ownership than to a specification, and it is a much better predictor of whether you will still be happy in two years.
Charging speed belongs in the same conversation, because it changes the application built around the cell. A Tesla carries something like a 100 kWh pack so it can drive 500 or 600 kilometres in one go. Now, if I had a battery that charged three times faster, I would put in 30 to 35 kWh and charge it in five minutes instead. Why would I carry the weight of a 100 kWh pack, and pay for it, for a journey I make once a year?
Smaller packs, faster charging. Same technology, different tuning, and a completely different business model. If you had optimised for energy density alone, you would never have found it.
So what does better actually mean?
It means better at the job.
For a drone, better means energy density and discharge power, because flight time and payload are the key product demand currently. For a robot working shifts, it means energy and turnaround together, because a robot on charge is a robot not working. For a car, it depends entirely on whether you are building for the motorway or the city.
There is no better battery in the abstract. There is only a battery better matched to what the application has to do.
Chemistry picks your curve. Architecture changes its shape.
Here is the part that took me a while to appreciate, and it is why I work on what I work on.
Everything above describes trade-offs inside one way of building a cell. You compress the electrode or you do not. You go thicker or thinner. Within a single architecture you are sliding along a curve, and every gain has a matching loss waiting for you.
Chemistry changes which curve you are on. That is real progress, and it is why the industry chases new materials so hard. But it is still a curve, with the same shape of compromise.
The more interesting question is whether you can change the shape itself. If the fight between energy and power comes largely from how far ions have to travel through a dense, tortuous electrode, then restructuring that electrode in three dimensions, rather than pressing a flat layer harder, changes the terms of that particular trade. More surface in the same footprint. Shorter paths for the ions. Energy and power stop pulling against each other quite so directly.
That is the bet we made at LionVolt. Our 3D anode architecture will move the envelope rather than slide along it, and it is also what lets us work with lithium metal, the anode material with roughly ten times the theoretical capacity of graphite, 3862mAh/g in a liquid electrolyte, in cells we build today, and compatible with solid-state electrolytes as they become a commercial reality. We’ve got the technology and the roadmap and are ready to embrace the future.
I am not going to give you our headline number here, because that would undo the whole argument I just made. If you want to take one thing away from this, take the questions instead. When a supplier hands you a figure, ask what it cost them everywhere else. Ask under what conditions and which cycle number it was measured. Any supplier worth working with will have those answers ready, and that includes us.
Ask us the same questions. When we put a number in front of you, it will come with the conditions it was measured under , and I would rather be judged on those than on the headline.
So yes, energy density deserves its importance. It just has to be put in context.
In the next piece in this series I want to go properly inside the cell, and look at why the physical structure of an electrode decides so much of what the battery is allowed to do.