Low-carbon tech needs much fewer materials than it used to; this matters for resource extraction in the future
Improvements in material efficiency + recycling = super-circularity.
The concept of a “circular economy” has always sounded nice and aspirational: running an economy on refurbished and recycled materials so that the amount of new materials you need to extract is close to zero.
I have to be honest and say that I was always a bit skeptical of this vision: wonderful in theory, but just not how the world will ever work.
Of course, a fossil-fueled economy is as far from circular as you can get. Fossil fuels are dug up, burned once, and the process repeats. It requires continuous extraction, with no hopes of recycling.
Low-carbon technology gives us some opportunity to get closer to a circular model. Many materials can be recycled, even if they’re not being recycled right now (see my previous post on battery recycling stats).
Yes, we will need a large ramp-up period of mineral extraction as the world shifts to renewables, batteries, and electric cars, but the hope is that we then reach a better equilibrium where materials for new solar panels and turbines are coming from old ones that have reached the end of their life. Total circularity seems unlikely, but maybe we could get close.
But I think this circularity discussion actually underestimates how much of a difference low-carbon technology could make in the model of the resource economy.
That’s because we underestimate, forget, or are not aware of the massive improvements in material efficiency of these technologies. A solar panel built today uses far fewer materials than one a decade ago, and the same goes for batteries.
My vision of circularity has always been a 1-to-1 conversion or loop: the best we could achieve would be turning one old solar panel into one new solar panel, or using my old phone battery as the battery in my new phone.
Sure, the recovery rate of materials might be 80% or 90% rather than 100%, so one old panel provides a little less than a full new one. But the overall aim is to get as close to “one in, one out” as possible.
But, if we’re able to achieve very high recovery rates for valuable materials, then a solar panel built 20 years ago can now supply the materials for far more than one today. It’s not just “circular” — it’s what we might call “super-circular” or, to use environmental language, “regenerative.”
That’s because it now takes far less silicon, lithium, silver, cobalt, glass, or other materials to produce a solar panel, turbine, or battery than it did in the past. Material intensity has improved massively (and this has been one reason why prices have plummeted).
It’s surprisingly hard to find material intensity data for these technologies. If anyone has access to good public data on this, please let me know—I’d love to build a dataset that people can use and explore.
But I did find data on polysilicon and silver for solar panels, which I’ll use as an example.
In 2004, one watt of solar PV needed around 16 grams of poly-silicon. By 2023, this had dropped to 2 grams. One-eighth of the amount.
A solar panel installed in 2004 will be reaching the end of its life sometime this decade. Now, if we could recover most of that silicon (which isn’t common today, but scientists are making progress on methods to recycle it back into silicon suitable for new panels), then theoretically it could be enough to make eight new panels.1 Realistically, recovery rates wouldn’t reach 100%, so let’s assume it’s only 80%—that would still be enough for six new panels.
With the 16 grams of polysilicon, you could have made one watt of solar power in 2004, 2 to 3 watts in 2014, and 6 to 8 watts today.
⚠️ Note that for simplicity — and to make the key message more soluble — I’m going to assume the wattage of a solar panel hasn’t changed over time. In other words, the change in silicon needed per “panel” is the same as the change per watt. This is so I can help people visualise the transformation.
Or take the example of silver. The amount of silver needed per watt of solar fell by 20% for every doubling of global cumulative capacity.2 That’s very similar to the “learning curve” of solar costs: every doubling in cumulative capacity led to a 20% drop in costs.
In 2010, solar used around 55 milligrams (mg) per watt. By 2020, this was around 20mg. And recent figures from Jenny Chase (legendary solar analyst at BNEF) have it at around 10 to 13 mg in 2023.3
So, the silver used in one solar panel built in 2010 would be enough for around five panels today (or four if recovery rates are just 80%).
By the time a 2010 panel reaches the end of its life — in 2035 or 2040 — silver consumption might be as low as 5mg. In this case, it’d be enough for 10 panels (or “only” 8 if the recovery rate is 80%).
High recycling and recovery rates will be key
This is very different from the standard vision of “circularity,” where, at best, one product at the end of its life is reincarnated into one new product.
Instead, it could look something like the model below—at least for some minerals.4 For other materials, efficiency improvements might be much lower, so the amount of steel used decades ago would still only be enough for one panel today.
We might be able to achieve some degree of circularity even while demand for products such as solar panels and batteries continues to grow.
Even if this super-circular model is too optimistic, improvements in material efficiency could, at the very least, offset the amount of material that’s lost in recycling processes that are below 100%. If recycling rates recover just 80% of the material, as long as the material efficiency has improved by 20%, there will be enough material in one panel, turbine, or battery to make another one. No extra minerals needed.
I’ve heard Michael Liebreich make this point several times before on his Cleaning Up podcast.
Of course, this model depends on achieving high recycling rates, especially recycling materials to a “good enough” grade that they can be reused in low-carbon technologies.
I admit, this is a significant contingency. But if we can overcome it, there’s enormous potential to reduce pressure on new mineral extraction for decades to come.
Material consumption data is hard to come by, and academics are constantly lagging behind
You’ll notice that I focused on polysilicon and silver here. That’s because they were the two minerals that I could find public reports and discussions on.
There is a dire lack of publicly available data on the material intensity of different technologies. How much cobalt and lithium do lithium-ion batteries use compared to a decade ago? I couldn’t find any good numbers. If you know of datasets that document the changes in material use of low-carbon technologies over time, let me know! I’d love to build a small dataset that others could use.
This lack of transparency is probably one of the reasons why so few people are aware of how dramatic the improvements in material efficiency have been.
Even academics and analysts writing key papers on the state of the energy transition and material requirements can’t keep up. I was only aware of the amazing silver story because of this exchange between Seaver Wang and Jenny Chase.
The fact that researchers can’t keep up with developments in low-carbon energy is, in many ways, a good thing. It means things are moving quickly. But it also means that a lot of the literature is too pessimistic, using outdated assumptions on costs, and the amount of materials we’ll need in the future.
Hoseinpur, A., Tang, K., Ulyashin, A., Palitzsch, W., & Safarian, J. (2023). Toward the recovery of solar silicon from end-of-life PVs by vacuum refining. Solar Energy Materials and Solar Cells, 251, 112181.
Preet, S., & Smith, S. T. (2024). A comprehensive review on the recycling technology of silicon based photovoltaic solar panels: Challenges and future outlook. Journal of Cleaner Production.
I'm being a bit conservative here because recovery rates closer to 90% are more likely":
Li, L., Zhang, X., Li, M., Chen, R., Wu, F., Amine, K., & Lu, J. (2018). The recycling of spent lithium-ion batteries: a review of current processes and technologies. Electrochemical Energy Reviews, 1, 461-482.
This is from her fantastic (and fun) book, Solar Power Finance Without The Jargon (Second Edition).
This is a great concept Hannah, thank you for sharing! There will most likely be an asymptote of diminishing returns at some point, but the concept is robust. We just need adequate recycling companies and the benefit of recycling must be significant compared to landfilling. That seems to be one of the major challenges. As there are less raw materials in these renewable energy products, the economic value of them is lower, which reduces incentive to recycle.
Maybe we could get to a nice situation where my 2011 5KW solar array is worth more as scrap material than the cost of a new 5KW array.
Unlikely.
But the most important resource - especially for roof top - is land. Which is why when the FIT ends in 2031, I will probably replace the 5KW array with a Peroskovite-Silicon multi junction array generating 10KW.