A small lab-scale cell with all components placed in it. The inlet and outlet ports carry the flow of gas that interacts with the cell.
Credit: Buchen et al/Nature Energy
A New Direct-Air-Capture Design Could Cut the Energy Cost of Removing CO2
A lab-scale direct-air-capture device removed carbon dioxide from ambient air while using approximately 0.8 megawatt-hours of electricity per tonne of captured CO2. That is below the roughly 1.5 to 3 megawatt-hours per tonne used by current facilities that capture carbon directly from the air.
The researchers tested a device made by stacking nine cells, each with an area about half the size of a letter-sized sheet of paper. A blower forced air through winding channels in plates surrounding each cell. The design uses components borrowed from fuel-cell technology.
From a Lab Device to a Pilot Plant
Several researchers involved in the study are also part of RepAir Carbon, a startup based on the technology. The paper examines whether the process could become economically viable at larger scales.
The researchers estimate that a first small-scale pilot plant could cost approximately $566 per tonne of captured CO2. They then apply a learning rate drawn from the lithium-ion battery industry, along with the cost reductions typically associated with scaling up large plants.
Under those projections, a plant with 1,000 times the pilot’s capacity could capture carbon dioxide for approximately $92 per tonne. That would be substantially lower than the $250 to $350 per tonne Climeworks has said it expects to achieve by 2030.
Why a Lower Cost Matters for Direct Air Capture
Reaching $100 per tonne has long been a goal in the carbon-capture industry. A cost that low could encourage broader adoption of direct air capture, although the researchers’ projections still depend on successfully scaling the technology through several generations of pilot plants.
As more direct-air-capture technologies are pursued, the researchers argue that the chances of at least one reaching its cost target may increase.
Nature Energy, 2026. DOI: 10.1038/s41560-026-02129-z.
Source: arstechnica.com


