The clean energy transition won’t run out of minerals.

Introduction

What do climate deniers, Guardian journalists, nuclear fanboys and resource startup CEOs have in common?

They like to exaggerate the impact of material use in cleantech.

The most common themes centre on the criticality of particular materials to a technology, and state that those resources cannot be sustainably extracted to meet the needs of a modern global economy.

Both assumptions are false as we will explore below. I want to write this because all too often the false narrative convinces otherwise rational people that solar panels are toxic or that we haven’t enough lithium/copper/nickel for EVs.

In short, the mineral use of cleantech is not a reason to keep burning fossils, return to a pre-modern life, fantasise about a nuclear renaissance, or buy shares in a mining startup.

We need to increase momentum on the transition to clean electricity and transport rather than slow down. Yes, cleantech is made from matter, but as we will see it is less material intensive than the status quo.

Myth busting

For climate deniers and nuclear frauds the truth comes second to a good meme. They manage to spit out enough falsehoods that some of them stick and are bandied about as fact. This filters into even progressive media trying to stay sober and even-handed on the situation without doing the basic research.

Let’s bust some common myths before getting to the real work.

Solar - takes 1-2 years to pay back energy inputs. Outside experimental uses, solar does not contain lead, mercury, arsenic, platinum, palladium, selenium, indium, chromium or rare earths. You could eat silicon solar panels and the copper cable would be the most toxic part. One manufacturer uses cadmium telluride cells (both elements are toxic before processing) but accounts for only a few percent of the market, does not sell to households, and recycles panels at the end of life.

Solar modules are in order of mass: glass, aluminium, copper, glue, silicon, silver, phosphorus, boron. The last two are measured in parts per billion within the silicon.

Wind - takes 6-12 months to pay back energy inputs. Does not contain meaningful amounts of cobalt, silver or any rare minerals. About 20% of turbines use "rare earth" generators, but a) optional, and b) not scarce.

Wind turbines are in order of mass: steel, polymer, glass, aluminium, copper.

Waste is miniscule, even with poor recycling. The amount of waste produced in the supply of a household's yearly electricity consumption by wind & solar is comparable to a plastic fork.

Per MWh the mined minerals needed for coal power are around 1000 times more than for wind power. The ash landfilled from coal is several hundred times more massive than wind inputs (not even accounting for recycling of metals). The nitrous oxide pollution from coal power (restricted to a very small concentration by law) weighs more than the mineral inputs to wind power per MWh.

The Stone Age didn’t end for a lack of stone

We’ve never run out of a mineral resource before.

Beyond substituting a scarce resource for a common one, we actually progress toward superior materials. Copper was replaced by bronze even though it was more difficult to make, bronze was then replaced by iron, then steel, aluminium, titanium, carbon fibre…

Each of these superior materials requires more challenging production and often more difficult manufacturing techniques which are the main impediments to their widespread use. Over time we get better methods and superior materials become commonplace.

I read some CPI stats in a German newspaper years ago about the real cost of consumer goods over a few decades. Almost every consumer good decreased in real price over time - clothing, food, communications, travel. Technological advancements drove those benefits through greater efficiency and labour productivity. The exceptional items whose price rose in real terms were north sea cod (over-exploited) and silver (hoarding for speculative reasons).

Peak Oil

We’ve seen these emotional responses over scarcity before, and they proved to be wrong. Simplistically, when one projects an increase in demand against known reserves it will always show that we are about to run out. The problem with this kind of forecast is that new production or exploration techniques are unknown, consistently biasing potential production below what is possible. I’ll explore this more in the Supply section below.

New oil production techniques embarrassed Peak Oil advocates in the past, yet we are approaching peak oil anyway due to superior substitution. In the next few years electric vehicles are expected to have a lower cost of ownership than petrol cars. This puts half of oil demand at threat from a product that is easier to use, more reliable, cleaner, quieter, and soon - cheaper.

Peak Extraction

The hand wringing over potential scarcity uses emotional references without a wider perspective. For example “copper demand is set to DOUBLE: CRITICAL to clean technologies”, or “each ton of silicon [or steel] uses 2 TONS OF COAL to be smelted”.

We aren’t seeing the full picture here. When minerals are in low demand it’s rather easy for demand to change by multiples, but they are still minor players in overall mining activity.

Here’s a scenario to compare the extractive intensity of the clean alternative to the status quo. All passenger cars are electric, all buildings heated with gas have their systems replaced with heat pumps, and thermal coal is replaced with wind & solar as well as the new electricity demand from electrification.

This isn’t an aggressive decarbonisation scenario; all the hard to abate sectors are left alone. Gas power still generates in the peaks. I did not study feedback loops e.g. coal and oil make up one third of maritime freight, which would further reduce oil consumption. Population and consumption standards are static. Cleantech resource consumption is annualised over operational life.

The chart is ordered as per the legend so bauxite is the last properly visible mineral. The order is fossils first to make the drop easier to see, then other minerals ordered by declining production mass.

Total mineral extraction has gone down 55% in the Limited Transition Scenario. If you squint you can make out the relatively tiny increases in iron, bauxite, coking coal, and other minerals.

This makes me think that terrestrial mineral extraction will most likely peak in the next few decades.

Nevertheless production of several minerals needs to increase significantly over the next 30 years.

Does it seem like a lot? Unachievable? Is scarcity imminent?

This kind of growth in production is normal. Solar grade silicon production increased 3000% in the '00s. Copper production doubles every 25 years like clockwork. Lithium production tripled in the last decade. This means lithium and copper are already on trend, and silicon will have to moderate its growth in production.

However this doesn’t mean we won’t have price shocks. Steep rates of growth in supply and demand are hard to match and it is likely there will be mismatches that result in both high and low prices. Silicon has had this feature over the past 20 years with repeated price shocks every 5 years or so.

The Decarboniferous

I’m a big fan of Hans Rosling (who isn’t?) - his empirical perspectives and his enthusiasm are infectious.If you haven’t seen his talk on the washing machine I recommend you spend 10 minutes watching it. In it he admonishes his students for suggesting the global poor should not be able to increase their resource consumption. His example is on the washing machine which has done so much for human development - how could you deny someone access to such a life changing appliance? In his conclusion he suggests rich nations focus on being more efficient and getting green energy first, before lecturing anyone else. But Rosling frames this transition as a challenge rather than an opportunity, in his conclusion the worlds’ poor are still stuck using fossil fuels while the noble westerners strive to become more efficient and greener. The energy transition began with an ecological imperative and we have framed cleantech as an inferior substitute that we have to bear with. However, the far greater resource efficiency of wind, solar, EVs, and heat pumps will make them a more affordable, superior substitute.

In Afghanistan poppy growers are increasing yields with solar powered bore pumps - it’s simply cheaper and more reliable than a diesel pump. While this is a provocative example, I’m sure many more mundane crops are being irrigated with this modern productivity multiplier - it just doesn’t make for a clickable headline.

Sub-saharan Africa already has more smartphone users than people with access to centralised electricity. Consider: if these folks aren’t on a gas pipeline what kind of stove will they cook with in the future? Will their first vehicle be an electric motorbike or a petrol one if they have a solar array at home but no reliable petrol supply?

I think historians in 100 years will look back at this period of time not only as one of decarbonisation, but a proliferation of access to basic energy services with corresponding increases in human development and productivity. It will seem obvious in hindsight for someone in 2122 “What were those neanderthals thinking?! Extracting billions of tonnes of fossil fuels simply wasn’t scalable for a modern society.”.

At this point in time, assuming we stabilise the global population at 11 billion, of which almost everyone is middle class, the only need for mining will be to replace lost materials and to produce fertiliser.

Demand side dynamics

Demand for minerals is somewhat flexible, and has interactions with other industries that use the same minerals. If supply is constrained how could we best distribute resources? In some circumstances a price shock will impact other industries more and they will reduce demand, but cleantech can also modify demand.

In all cases cleantech is not the primary consumer of particular mineral resources.

Scrimping

Scrimping (or thrifting) is both an ongoing process and a response to price shocks. It is the practice of making things thinner at the expense of strength, conductivity, or reliability. Scrimping’s advantage is that it can be done incrementally without changing technology or methods significantly, but the disadvantage is that you can’t scrimp to nothing. It might also make recycling more challenging or uneconomic.

Solar managed to reduce silicon consumption per watt by 80% over the last 20 years.

Substitution

Let’s cover off material substitution from a material perspective and clean technology perspective.

Nickel

Around two thirds of nickel consumption is for stainless steel consumer goods,and the majority of that is austenitic (300 series) steel.

Replacing austenitic with ferritic stainless steels is cheaper and they use very little or no nickel. Duplex stainless steels have 50-90% less nickel content, instead using higher chromium and molybdenum fractions. Both of these options are fine for your fancy looking fridge.

For specialty nickel alloys substitution with titanium or tungsten alloys may be viable if high nickel prices are expected, but they can most likely afford the nickel super alloys.

Copper

Copper is mostly used as wire and tube, especially in buildings.

Copper tube for water pipes can be replaced with cross-linked polyethylene tube at lower cost and often lower labour. Copper tube in heat exchangers can be replaced with aluminium tube making them lighter and cheaper, but they need coating to resist corrosion.

Copper wire can be replaced by aluminium, and this is already widespread for electricity distribution. Aluminium conductors are cheaper and half as heavy as copper, but have a 50% larger volume which makes them unappealing in volume sensitive consumer products.

Manual manufacturing techniques such as soldering and winding are easier with copper than with aluminium, but aluminium is easier to cast, extrude and weld. Especially for small uses, the higher difficulty of terminating aluminium wire makes copper more convenient.

Rare Earth Magnets

Magnets are optional for electric car motors or wind turbine generators. The higher performance is offset with higher cost.

Substitution could be for cheaper, weaker magnets such as alnico but this would be offset by the larger volume of material required and low power density.

The simplest substitution is copper coil, which is the dominant design for stationary motors and generators already.

The largest use of rare earth magnets is hard drives, especially for internet servers. This is largely unsubstitutable without increasing cost and reducing energy efficiency. This is the strategic interest that China has in controlling rare earth supply. The previous rare earth price shock was ameliorated once US and Japanese hard drive manufacturers committed to opening factories in China.

From a cleantech perspective, rare earths are not a critical input and there is no strategic interest in rare earth supply when recycling and substitution are available. They are just critical for watching cat videos online.

Now rather than a material perspective, let's look by technology.

Wind turbines

Wind turbine generators have three existing material configurations (rotor-stator).

• Permanent Magnet-Copper
• Copper-Copper
• Copper-Aluminium

Copper-copper is the most common, but permanent magnets are popular in offshore wind where higher reliability can save awkward maintenance work.

Enercon has turbines with aluminium stators and some induction motors are the same.

In a standard copper-copper configuration, the cost of copper represents 0.6% of total wind turbine capex suggesting that substitution is not a priority.

A potential superior substitution for copper is superconductors, but the technology is currently only used in military and medical applications. The power density and efficiency would be excellent, but superconductors need cryogenic cooling and are quite expensive.

Underground cables and overhead lines are already aluminium so no substitution is required.

Solar Modules

Silver

Used for front contacts, silver is worth the extra cost because a thinner conductor shades the cell less. Solar demand for silver is around one tenth of annual production.

Silver makes up a couple of percent of module cost, but very little in terms of system cost. Scrimping has been successful over time, and combined with increasing efficiency, silver use per watt is falling.

One technology bucking the trend is bifacial panels which use silver contacts on both sides in order to increase productivity.

Potential substitutes are copper or transparent oxide conductors for the cell face, and copper or aluminium rear contacts. Contacts can tunnel through from the rear to reduce shading and allow cheaper materials. Side-on cells have also been tried which have no front contacts, but it wasn’t successful.

A notable substitutor is Sundrive, an Australian startup which has demonstrated a commercial format silicon cell with copper contacts and high efficiency (>25%), suggesting they have a clever geometry which minimises the shading issue.

One confounding factor for silver use is that the price is highly sensitive to speculation. Appreciation of silver prices may stimulate hoarding, making silver more volatile. Solar recycling could be a counteracting force against this volatility. Currently panels have the aluminium and copper removed and the glass sandwich is stockpiled until volumes are worth processing. In future, silver price spikes are likely to trigger solar panel recyclers to process their stocks.

Copper

Copper cabling is used to connect solar modules and to run back to the inverter. While copper cables are convenient for household installations, they are overkill for a utility scale project. Often the cables are longer than necessary and they can limit array design due to losses.

According to an old estimate copper makes up 4-5% of the capex of a solar farm, suggesting substitution is potentially viable. Subject to the labour cost of termination, aluminium cables could reduce cost, reduce losses, and allow for larger arrays.

EVs

Cathodes

Lithium ion batteries generally use one or more “ferrous” elements: nickel, cobalt, iron, and manganese which are next to each other on the periodic table.

The highest energy density comes from using cobalt/nickel mixtures so these are preferred for consumer electronics and electric vehicles alike. The issue is that cobalt is well known as a conflict mineral. The Democratic Republic of Congo produces about two thirds of global cobalt, of which around half is produced in larger regulated mines. The remainder comes from unregulated small scale miners whose activities have adverse effects on the environment and human health. Even worse, the small-scale miners are exploited by militia groups seeking to control lucrative tantalum, tin, cobalt and diamond mines. As such many multinational corporations have been pressured into reducing cobalt consumption or finding alternative supplies. Without proper regulation I’m worried this won’t solve DRC’s issues, but it at least reduces the geopolitical importance of their minerals. The DRC government has moved to centralise sales from small miners but this mainly is to avoid cash going to militants. Ideally more widespread use of regulated, mechanised mining would do away with small-scale mining, but unfortunately this may have to wait until easily accessible surface deposits are depleted.

Cobalt substitutions include “high nickel” cathodes that have mostly nickel and some manganese and aluminium. Nickel exploration and production is being increased by major miners in response.

The other substitute we should talk about is lithium iron phosphate (LFP). LFP cells are the lowest energy density, but still appropriate for mid range cars and of course stationary applications. They are also cheaper, more stable, and not reliant on nickel or cobalt.

Tesla expects to use high nickel chemistries for performance cars and trucks where energy density is important. They already use LFP in their mid-range Chinese models and stationary applications.

Copper

Copper is used in the motor (similar to wind generators), cabling, and to collect current in the battery cells.

It could be worthwhile to substitute aluminium cabling to save weight, but it probably doesn’t account for much weight or cost.

Substitution of copper in the motor is unlikely, except to use permanent magnets for manufacturers who don’t already. Reliable manufacturing and operation is more important than material cost.

Substitution is also less likely in the cells as something like aluminium foil would reduce volumetric energy density. This disadvantage may not be important in cars with smaller batteries.

Lithium

Although lithium is abundant, people are still interested in substitution. Sodium in particular has been suggested as a drop in alkali metal replacement. I’m skeptical because lithium is a superior ion, but CATL is reporting a commercial sodium ion product and they are a credible GWh scale battery producer. Their sodium ion cell is also notable for an iron based cathode.

Graphite

Graphite is quite common, but perhaps limited by processing and quality. It can be synthesised which results in a high purity product at a higher cost. Synthesis of other carbon materials with potential in battery anodes such as nanotubes and graphene is done at a non-commercial scale.

Silicon is also being explored as an anode material as it can hold more lithium ions per mol. It is, however, not dimensionally stable.

Lastly, lithium metal is being explored as a direct anode with new electrolytes under research.

As before, substitution is progressing based on superior material properties rather than material scarcity.

Scavenging

A special kind of substitution.

When a material becomes especially valuable it is not only substituted in new goods, but encourages the recycling of old stocks. Copper is a familiar case; old electrical systems replaced copper with aluminium to make money, and copper tubes and wire are often stolen from disused homes or those under construction.

Stocks is the important word here - cleantech material "consumption" is not permanent. The materials can be recycled at end of life and are merely held as material stocks in the meantime.

Supply side dynamics

I’m going to explore copper on the supply side because it’s a hot commodity, part of the exuberance of “critical mineral scarcity”, remains in a similar use today as many decades ago, and there is good data on it.

Iron, aluminium and silicon are too abundant to be much fun, and the other minerals have pretty limited data. Silver is too volatile and speculative to get anything meaningful out of the data even though it’s there.

Ore grade

First let’s explore the concentration of copper within ore. This has been falling over time which is a source of concern for some, and a source of speculation for others.

I got this chart from the source below, who in 2017 said that copper mine production was set to peak in 2019. They have a chart that looks exactly like peak oil - demand keeps growing but supply tapers off.

https://moiglobal.com/copper-mining-articulating-a-contrarian-thesis/

The impact of a falling ore grade is that more ore must be extracted and processed for each ton of metal, and this increases the cost assuming all else remains the same.

On the other hand, where a lower grade becomes economical because of higher prices, then the potential resources expand, i.e. there is much more ore with 1% grade than with 2% grade. Merely forecasting supply based on current grade requirements and current reserves with that grade severely underestimates potential resources at lower grades and higher volumes.

“According to USGS data, since 1950 there has always been, on average, 40 years of copper reserves and over 200 years of resources left.”

https://copperalliance.org/sustainable-copper/about-copper/cu-demand-long-term-availability/

This chart shows we’ve had a pretty consistent amount of reserves for copper over quite a long time without any impact from ore grade. In 1950 ore grade was about twice as good as 2017.

Given that reserves have grown in line with production we can also see from this chart that copper production has doubled three times over the period.

For perspective; the amount of copper in the upper crust of Earth contains about 100,000 times as much copper as we currently account for in reserves. We could theoretically keep doubling copper consumption every 25 years for 400 years and still have 40 years of reserves and a stupendous stock of copper metal. Given that population and per capita consumption are expected to stabilise within the century this seems extravagant.

Inflation

We can see that copper price has bounced between $2 and $4 per pound in real terms over 80 years. This chart cuts off the fall in price to $2 in 2016 and the recent rebound to $4 in 2020. Try to avoid recency bias when reading the chart, just because the recent prices were high doesn't make the trend go up. In 1980 the past made it look like an upwards trend too.

If copper price is reasonably likely to fall to $2 again within 5 years I’d be pretty careful about believing the resource shill and buying shares in a company that produces copper at $3/lb.

Although it’s quite volatile we can see that consumers are not paying more for copper products over time despite the degradation in ore quality over the same period.

Inflation helps mitigate some of the increased work associated with lower ore quality. This could explain why substitution for copper as described above has not taken place yet.

Productivity

The last supply side factor is productivity. Over these long time periods the production processes incrementally improve and new processes become available or economical.

A notable process that is slowly gaining market share is electrowinning which can work with low grade ores or even old mine tailings. This process also does without coking coal as an input so can assist with reducing CO2 emissions in copper production.

There is an interesting feedback loop between cleantech and copper. Remote mines that previously would be diesel powered at great expense can be more cheaply powered with some fraction of renewable energy in the mix. This would allow existing mines to operate for longer, or could make previously uneconomic resources viable to extract.

The combination of productivity and general inflation are likely to keep prices reasonable despite the reduction in ore concentration over time.

Conclusion

The challenge we have is not the lack of mineral elements. Cleantech is primarily made from minerals far more abundant than 11 billion people can reasonably use.

Rather, we are rate limited. The exploration, extraction and processing needed for what were previously uninteresting minerals face uncertainty in how much demand will grow and when. As with any other commodity, supply is likely to lag demand, allowing prices to spike. And, from that spike we can also expect an oversupply at some later date as production capacity overshoots.

If price pressures are sustained it will also spur new developments in supply and demand. Valuable resources will be diverted to where they are most useful, and more effective production techniques will be given the chance to be tested.

If we expect price shocks what could the impact on cleantech be? Simply that the transition will be stepwise, not smooth. With tailwinds cleantech will have a growth spurt, then slow down for a few years as the supply of inputs catches up. Power plants and vehicles are durable assets so it is quite normal for plans to wait on the right opportunity.

In terms of the environment we are simply extracting far less stuff as we transition away from fossil fuels. Even if we don’t account for the air pollution, and even if we continue to mine in reckless and damaging ways this means less overall damage to the environment from mining.

In terms of conflict over resources, this too will be reduced. Single-use energy sources create tension as they must continuously be delivered and are not found ubiquitously. Nations won’t be able to hold their neighbours to ransom with a gas pipeline if everyone has high efficiency heat pumps. Delinquent states will have little power if oil is no longer the lifeblood of the global economy (and will also find themselves invaded less often).

The clean energy transition is a wonderful opportunity in resource productivity rather than a hindrance.

Appendix - Abundance

The most common 10 elements in the earth’s crust account for almost all the matter. All of the stuff in green is incredibly common.

The platinum group metals down the bottom are rare, a billion times less common than silicon.

Everything in the middle is more or less “uncommon”, you probably have most of them in your home. Rare-earth elements are in this group, the name "rare" comes from when they were not widely discovered.

From an energy perspective we can get away with just the super abundant elements: Aluminium conductors, sodium ion batteries, silicon semiconductors, iron structures.

Used sparingly, the fancier elements can work too, but bulk semiconductors made from indium selenide, or structures made from tungsten aren't really scalable.

Disclaimer: this is my own opinion and does not represent my employer in any way.