numbers, not adjectives — D. J. C. MacKay

11  Direct Air Capture

The idea in Direct Air Capture (DAC) is to run filtering or chemical processing plants which use technology to extract CO2 from the atmosphere. That would be great if it was scalable. But there are monumental obstacles to scaling DAC.

What really worries me is that many people, including scientist and policy makers, talk about DAC as though it is something which could feasibly be implemented at a scale which would make a significant impact on future climate.

The purpose of this page is to present facts and simple calculations, to show that even a tiny effect from DAC would require unimaginably extreme infrastructure investments. And thereby hopefully discourage casual discussion of DAC as a feasible tool.

Let’s set the bar really low, at the removal of 1% of atmospheric CO2 cumulatively over 25 years, ie by 2050. That is an annual removal of 0.04% of current CO2 levels. If DAC technology can’t feasibly reach this level, then its use is really of very marginal importance, and we would do well to focus on more promising strategies.

The problem that we are going to encounter is that our atmosphere is HUGE, it weighs 5.5×1018 kg. That’s close to a million tonnes per person. Removing some fraction of the CO2 is going to require that we build a processing plant (or multiple smaller plants) which is able to process at least that fraction of the atmosphere. And that, in a nutshell, is infeasible.

In addition to the extremely large quantity which needs to be treated, the process itself is also inherently difficult. There are two unavoidable physical difficulties: first we need to isolate the CO2, and then we need to compress it from its very dilute form, to something more practical for disposal. Both of these processes require energy. There are fundamental physical limits to the required energy, which no amount of research and development will be able to overcome.

In order to give some intuition for the scale of the chemical plants necessary, I am going to work out how big a chemical plant would have to be built by each person alive. By this, I’m not suggesting that each person necessarily builds a plant by themselves. I’m merely using it as a tool to figure out the scale of the endeavour. What would it mean for you?

11.1 A “realistic” Direct Air Capture plan

Caveat: I’ve quoted the word “realistic”, since the realism refers to the physical and engineering aspects, not the feasibility of implementation in the real world.

Goal: to remove 1% of atmospheric CO2 cumulatively over 25 years. Let’s assume we start building DAC factories now, and continuously build out capacity for the next 25 years. We also assume that we can build these factories instantaneously, and once they’re built they run continuously. Finally we assume that we can build and run these factories without emitting CO2. These are big assumptions. How big a DAC factory would we have to build per human per year to achieve the goal?

The atmosphere (in 2025) contains 410 tonnes of CO2 per human. So, you might think we need to remove 4.1 tonnes, but that’s not quite right. When we release CO2 into the atmosphere, only 45% of it stays in the air (the so-called airborne fraction), the rest dissolves in surface waters (where it causes acidification). But the atmosphere and the surface waters are in dynamic equilibrium, which means that if we remove CO2 from the atmosphere, then CO2 will be released from the surface water back into the atmosphere. So to lower the atmospheric content of CO2 by 1%, we need to remove 4.1/0.45 = 9.1 tonnes of CO2 per person.

The density of CO2 is 1.87 kg/m3, so 9.1 tonnes is 9100 kg / 1.87 kg/m3 = 4870 m3. The atmospheric concentration of CO2 is 425 ppm, so the required amount of CO2 is contained in 11.5 million m3 of atmosphere. Let’s say the chemical or filtering process is 50% efficient, then we need to process 22.9 million m3 per person.

So, how fast do we need to build? Over 25 years of continuous build out, we need to process 22.9 million m3, so the additional capacity needed per year is 2×22.9×106/25^2 = 73400 m3 of atmosphere per year. Since the density of air is 1.293 kg/m3, this corresponds to ~100 tonnes.

Conclusion: if we want to remove 1% of atmospheric CO2 by DAC by 2050, then every human being needs to build a new factory every single year for the next 25 years, with a capacity to process 100 tonnes of atmosphere annually, and run all of them continuously after they’ve been built.

A chemical plant capable of processing 100 tonnes of material annually is very substantial. Every year, every single human alive must finance, and build an additional plant. And over 25 years, these plants need to be maintained and run, and supplied with the required energy, all without producing any additional CO2, requiring substantial additional renewable energy supply.

I think most will agree, that the required measures are extreme, bordering on the absurd. Nevertheless, these are the incontrovertible facts for scaling DAC. And don’t forget, that we set the bar really low at 1% cumulative removal over 25 years.

You may concede that DAC technology is not yet mature enough to be practical. Note however, that in my reasoning I have said nothing whatsoever about the technology. My sole argument has been one of scale. No amount of research and development work is going to change the scale.

Releasing CO2 into the atmosphere, and subsequently extracting it again is a terrible idea. There may be some very limited circumstances, such as fossil fuel powered aircraft, where it might be useful, but the practical and economic obstacles remain monumental. Proponents of DAC should make clear, that they are talking about a hugely impractical niche technology with very limited impact.

As a general technology against climate change with a practical and significant impact at scale, DAC is completely infeasible.