Posted on 30 December 2025 by Ken Rice

This is a re-post from And Then There's Physics

I thought I’d written about this before, but can’t seem to find a post. Either, my searching ability is poor, or my memory is poor. I mostly wanted to highlight an interesting YouTube video by David Kipping that illustrates why Direct Air Capture (DAC) is thermodynamically challenging. I encourage you to watch the video (which I’ve put at the end of this post) but his basic conclusion is that thermodynamic constraints mean that implementing DAC at the necessary scale would require a significant fraction of all global electricity consumption.

I wanted, however, to work through some of the numbers myself and to do the calculation of how much DAC we would need to use in a slightly different way.

A key point is that given an atmospheric concentration of 400 ppm and a temperature of 300K, it takes a minimum of 19505 J to remove 1 mole of CO2. 1 mole of CO2 is 44g, so 1 tonne of CO2 has 22727 moles. Therefore, removing 1 tonne of CO2 requires a minimum of 4.43 x 108 J.

Typically, however, we emit so much that we tend to think in terms of gigatonnes of CO2 (GtCO2). Removing 1 GtCO2 would require a minimum of 4.43 x 1017 J.

Recent (optimistic?) projections now suggest that we’re currently heading for warming of about 2.5oC. Given a Transient Climate Response to Cumulative Emissions (TCRE) of 0.45oC per 1000 GtCO2, this suggests total emissions of about 5500 GtCO2 (we’re currently at around 2500 GtCO2). If we want to limit overall warming to 1.5oC, then we’d need net emissions to be about 3300 GtCO2. If we’re going to rely on DAC, this means removing 2200 GtCO2, which would require a minimum energy of 1021J.

To put this into contect, current (2024) global electricity consumption is about 31000 TWh, which is 1.1 x 1020 J. If we assume 3% growth in electricity consumption per year, then by the end of this century, we will have consumed a total of 3.5 x 1022 J of electrical energy between now and 2100. The minimum energy requirements for DAC would be about 3% of this total.

However, this assumes 100% efficiency, which seems unrealistic. More realistic estimates suggest something like 10% efficiency. Hence, if we want to use DAC to limit overall warming to 1.5oC, while emitting enough to reach 2.5oC, DAC would need to use something like 30% of all electricity consumed over the next 80 years.

I should stress that although I have checked some of these numbers, and I think my basic calculations are correct, I am happy to be corrected if anyone sees an error. Also, my estimates are approximate and I’m ignoring uncertainties, which could make things sightly better, or considerably worse.

I do think that the basic conclusion that DAC at scale would consume a significant fraction of future electricity is broadly correct. This probably suggests that it would be better to use this electrical energy to avoid these emissions, rather than using DAC to later remove what has already been emitted.