Direct air capture

Direct air capture (DAC) is a class of carbon dioxide removal (CDR) technology that uses chemical sorbents or solvents to extract CO₂ directly from ambient air, after which the captured CO₂ is either stored geologically (DACCS, direct air capture with carbon storage) or used in products such as synthetic fuels, beverages, or building materials.

DAC differs from point-source carbon capture, which pulls CO₂ from concentrated flue gases at power plants or industrial facilities. Because atmospheric CO₂ is dilute (around 420 parts per million), DAC is energy-intensive and currently expensive, with cost estimates ranging widely depending on the design, energy source, and scale.

Two main approaches dominate commercial development:

• Liquid solvent systems (e.g., Carbon Engineering) use hydroxide solutions and high-temperature regeneration.
• Solid sorbent systems (e.g., Climeworks, Global Thermostat) use amine-functionalized materials regenerated at lower temperatures.

DAC features prominently in modeled pathways to limit warming to 1.5°C. The IPCC's Sixth Assessment Report (AR6, 2021–2022) treats CDR, including DAC, as essential for offsetting residual emissions from hard-to-abate sectors and for achieving net-negative emissions later in the century.

Policy support has accelerated deployment. The U.S. Inflation Reduction Act of 2022 expanded the Section 45Q tax credit to up to $180 per tonne of CO₂ captured via DAC and permanently stored. The U.S. Department of Energy in 2023 announced funding for Regional DAC Hubs, including projects in Texas and Louisiana. The European Union's Net-Zero Industry Act and Industrial Carbon Management Strategy (2024) also reference DAC as a strategic technology.

Critics argue DAC may delay emissions reductions by offering a perceived technological escape (a "moral hazard" critique), consumes significant electricity and water, and remains unproven at gigatonne scale. Proponents respond that residual emissions from aviation, agriculture, and heavy industry make some form of large-scale CDR unavoidable, and that early deployment is needed to drive learning curves.