Is it true that battery energy density improves 5-8% per year?
I've heard representative of Tesla Motors make this claim. Is it true? Does this represent an average or is it a consistent trend each year? Do these improvements increase the cost of the battery? What has been the trend, if any, regarding energy to weight ratio?
7 Answers
TL;DR: Yes, but the trend isn't going to continue into the distant future.
There are great answers here, particularly Leigh Christie's. I'd like to point out that this trend is most likely not going to continue into the distant future. It is not feasible to get much more energy into a material than 1 eV per atom. Most solids have atomic weights of 30 GeV/c^2 [*] which yields E/m = 3x10^-11 c^2. When you convert this into the more human Watt-hour/kg, you get 850 W-hr/kg. We're already at 200 W-hr/kg, which means that we can gain a factor of 4 before we hit this rough limit. With a growth rate of 5%, we hit this limit in 25 years and at 8% this is hit in 17 years. This is similar to the end of Moore's law where the atomic size is limits the gain in speed that arose from miniaturizing integrated circuits.
[*] Li-Ion batteries are composed of Li Co O2 with mass with a mass of 80 GeV/c^2.
There are great answers here, particularly Leigh Christie's. I'd like to point out that this trend is most likely not going to continue into the distant future. It is not feasible to get much more energy into a material than 1 eV per atom. Most solids have atomic weights of 30 GeV/c^2 [*] which yields E/m = 3x10^-11 c^2. When you convert this into the more human Watt-hour/kg, you get 850 W-hr/kg. We're already at 200 W-hr/kg, which means that we can gain a factor of 4 before we hit this rough limit. With a growth rate of 5%, we hit this limit in 25 years and at 8% this is hit in 17 years. This is similar to the end of Moore's law where the atomic size is limits the gain in speed that arose from miniaturizing integrated circuits.
[*] Li-Ion batteries are composed of Li Co O2 with mass with a mass of 80 GeV/c^2.
As we all know, 5-8% growth is exponential growth. While this is much "slower" than Moores law, it's still the kind of growth that predicts fantastic things for our future given enough time. For example, if we wait 48 years, 8% compounded annually would give a battery that has a gravimetric energy density that exceeds that of gasoline!
But when we actually look at the curves for battery chemistries they are almost always linear. So why, then, do they say that "battery energy densities improve at 8% per year"? The answer lies in this graph:
![]()
The basic idea is that we continually discover new chemistries that change the slope of the curve. And all of these chemistries seem to allow for linear-like growth for that particular chemistry. It's also pretty safe to say that there have been several long stretches of time since the early 1900s where it would appear that there was little or no improvement at all since the "jumps" that come along with the discovery of new chemistries happens rather suddenly (at least in contrast to the slower linear growth we see for each chemistry). That said, I would imagine legendary researchers such as Dr. John B. Goodenough would yell at me for saying it's sudden!
If we zoom out on the graph and show figures all the way back to the 1900s. Like this graph (source: Thermodynamic analysis on energy densities of batteries):
![]()
The graph that Tesla's J.B. Straubel showed a while back exhibits this behaviour perfectly. While the improvements are fairly erratic from year to year, from this graph we can deduce that the improvement ranges, on average, from 5 to 8% per year. Again, however, we see this long linear stretch which leads us, in the short term, to think that "batteries are not getting that much better"... only to be surprised the next time a "leap improvement" occurs when a new chemistry or energy storage paradigm comes along.
![]()
Note: I have not actually done a curve fit, so I can not comment on the exact percentage. But given that it's doubling roughly every 9-14 years, I'd say 5-8% sounds about right!
But when we actually look at the curves for battery chemistries they are almost always linear. So why, then, do they say that "battery energy densities improve at 8% per year"? The answer lies in this graph:
The basic idea is that we continually discover new chemistries that change the slope of the curve. And all of these chemistries seem to allow for linear-like growth for that particular chemistry. It's also pretty safe to say that there have been several long stretches of time since the early 1900s where it would appear that there was little or no improvement at all since the "jumps" that come along with the discovery of new chemistries happens rather suddenly (at least in contrast to the slower linear growth we see for each chemistry). That said, I would imagine legendary researchers such as Dr. John B. Goodenough would yell at me for saying it's sudden!
If we zoom out on the graph and show figures all the way back to the 1900s. Like this graph (source: Thermodynamic analysis on energy densities of batteries):
The graph that Tesla's J.B. Straubel showed a while back exhibits this behaviour perfectly. While the improvements are fairly erratic from year to year, from this graph we can deduce that the improvement ranges, on average, from 5 to 8% per year. Again, however, we see this long linear stretch which leads us, in the short term, to think that "batteries are not getting that much better"... only to be surprised the next time a "leap improvement" occurs when a new chemistry or energy storage paradigm comes along.
These answers are all excellent. I would like to add that there are certain thresholds in energy density, and that we should inspect what you mean by battery.
Obviously most of us think chemical batteries when we use the phrase battery, but I like to define a battery as any portable means of energy storage. Fossil fuel is a battery. Uranium is a battery.
Now, there are two factors that matter. First, there is the energy density of the 'battery' (the material we are using to store energy). Second, there is the ability to take energy out of the battery and do useful work. That is, there is a set cost of retrieving energy from the battery. For a fossil fuel powered car, for instance, you need pistons and a heavy metal system to get energy out. Since this is a mobile application, this extra weight matters. There's also the efficiency of retrieving the energy. A car, for instance, has a peak efficiency of just 30% when running on conventional fuel.
Consider again energy density. A standard chemical battery doesn't have very high energy density, at least not relative to fossil fuel or uranium. However, like fossil fuel it is easy to use the energy in a chemical battery. When we consider both factors at the same time, it becomes apparent that fossil fuels are very useful for mobile applications, chemical batteries a bit less useful, and uranium not very useful.
Although uranium has a stupidly high energy density, it's not very easy to get work out of it, so we can't use it to power our cars (well, we can, actually, but that's another story).
Now, back to your question. This combination of factors (energy density and ease of using energy / efficiency) get better for chemical batteries every year. Fossil fuels engine, however, have already been optimized. Even a fantastically improved fossil fuel engine would only get you a marginally better engine, since gas engines are so good already. At some point, chemical batteries (or a future equivalent) will be more efficient than fossil fuels, either as a result of energy density, or a result of better usage efficiency. When this happens, there will be 0 reasons to use gas powered cars, and they will fall out of use entirely.
But wait, what about cost? Well, like any economy of scale, costs for chemical batteries will undoubtedly fall in time. A rosy future for chemical batteries. And sure, as demand falls for fossil fuels the price for them will drop, but at that point I doubt it will matter all that much. The infrastructure for chemical batteries will be in place, and those driving fossil fuel cars will be reviled to the point that social considerations will permanently drive home the tombstone for gas powered cars.
I definitely ended up on a tangent to your question, but I hope you found it interesting none the less.
Obviously most of us think chemical batteries when we use the phrase battery, but I like to define a battery as any portable means of energy storage. Fossil fuel is a battery. Uranium is a battery.
Now, there are two factors that matter. First, there is the energy density of the 'battery' (the material we are using to store energy). Second, there is the ability to take energy out of the battery and do useful work. That is, there is a set cost of retrieving energy from the battery. For a fossil fuel powered car, for instance, you need pistons and a heavy metal system to get energy out. Since this is a mobile application, this extra weight matters. There's also the efficiency of retrieving the energy. A car, for instance, has a peak efficiency of just 30% when running on conventional fuel.
Consider again energy density. A standard chemical battery doesn't have very high energy density, at least not relative to fossil fuel or uranium. However, like fossil fuel it is easy to use the energy in a chemical battery. When we consider both factors at the same time, it becomes apparent that fossil fuels are very useful for mobile applications, chemical batteries a bit less useful, and uranium not very useful.
Although uranium has a stupidly high energy density, it's not very easy to get work out of it, so we can't use it to power our cars (well, we can, actually, but that's another story).
Now, back to your question. This combination of factors (energy density and ease of using energy / efficiency) get better for chemical batteries every year. Fossil fuels engine, however, have already been optimized. Even a fantastically improved fossil fuel engine would only get you a marginally better engine, since gas engines are so good already. At some point, chemical batteries (or a future equivalent) will be more efficient than fossil fuels, either as a result of energy density, or a result of better usage efficiency. When this happens, there will be 0 reasons to use gas powered cars, and they will fall out of use entirely.
But wait, what about cost? Well, like any economy of scale, costs for chemical batteries will undoubtedly fall in time. A rosy future for chemical batteries. And sure, as demand falls for fossil fuels the price for them will drop, but at that point I doubt it will matter all that much. The infrastructure for chemical batteries will be in place, and those driving fossil fuel cars will be reviled to the point that social considerations will permanently drive home the tombstone for gas powered cars.
I definitely ended up on a tangent to your question, but I hope you found it interesting none the less.
Aswath gave a very nice answer. Unfortunately for us, there is no miniaturization problem equivalent to that in semiconductor/transistor technology that was the underlying assumption behind Moore's law. Instead, batteries (and electrochemical energy storage in general) is filled with tough materials and chemistry problems without easy answers nor often obvious directions. So far batteries have done a decent job keeping pace, but a lot of the extended battery life has been due to more efficient devices, not longer lasting batteries.
In '65 a guy who went by the name Moore, predicted that the density of transistors in ICs would double every two years. And to this date, it is valid, with unnoticeable("unnoticeabl
While I could not directly draw a parallel in the battery industry, I guess you could say that Moore's law is valid for most innovations in technology (which hasn't yet reached saturation), be it semiconductors or batteries, just that the frequency of doubling is different. What those reps said was true; I searched a bit, and internet says that battery energy density doubled over the last 10 years. And does so every ten years. And considering the rep's stats(averaged 5 and 8 ---> hence 6.5),
time taken for energy density to double=log(2)/log(1+0.065
So, I guess he was pretty accurate.
EDIT: Another reason could be that batteries (ignoring car batteries and the ilk in this case) are intimately linked with electronic devices. And as electronic industry marches towards increased density, thereby contributing a smaller footprint, the battery industry inevitably is forced to keep up with the exponential increase.
While I could not directly draw a parallel in the battery industry, I guess you could say that Moore's law is valid for most innovations in technology (which hasn't yet reached saturation), be it semiconductors or batteries, just that the frequency of doubling is different. What those reps said was true; I searched a bit, and internet says that battery energy density doubled over the last 10 years. And does so every ten years. And considering the rep's stats(averaged 5 and 8 ---> hence 6.5),
time taken for energy density to double=log(2)/log(1+0.065
So, I guess he was pretty accurate.
EDIT: Another reason could be that batteries (ignoring car batteries and the ilk in this case) are intimately linked with electronic devices. And as electronic industry marches towards increased density, thereby contributing a smaller footprint, the battery industry inevitably is forced to keep up with the exponential increase.
