Direct Air Capture: How It Works, What It Costs, and Where It Fits

The atmosphere is roughly 0.04% carbon dioxide by volume — about 420 parts per million. That small fraction, accumulated from two centuries of fossil fuel combustion and land use change, is the target of a new class of industrial machines that attempt to reverse the process: pull CO₂ directly out of ambient air, concentrate it, and either bury it underground or use it as a raw material. The technology is called direct air capture, and it works. The question that determines its relevance to climate strategy is not whether it works but whether its costs can fall fast enough and far enough to matter at the scale of the problem.

What Direct Air Capture Is — and Is Not

Direct air capture (DAC) — a technology that uses chemical sorbents or solvents to bind CO₂ molecules from outdoor air, then releases and concentrates the CO₂ under heat or pressure to produce a pure stream that can be permanently stored or used. The term distinguishes it from point-source carbon capture, which extracts CO₂ from concentrated exhaust streams at power plants or industrial facilities, and from carbon capture and storage (CCS), which is usually applied to point-source systems.

DAC is a carbon removal technology, not an emissions reduction technology. It does not reduce how much CO₂ enters the atmosphere from burning fossil fuels. It removes CO₂ that is already in the atmosphere. This distinction matters for how it fits into climate strategy: emissions reduction prevents concentration from rising; carbon removal reduces the concentration after the fact.

The Intergovernmental Panel on Climate Change (IPCC) Sixth Assessment Report (2021) modeled 230 scenarios consistent with limiting warming to 1.5°C above pre-industrial levels. In the majority of those scenarios — specifically those without immediate, dramatic reductions in all emissions sectors simultaneously — carbon dioxide removal (CDR) from some combination of natural and engineered sources plays a role in reaching net zero. DAC is one of several CDR options, alongside afforestation, soil carbon sequestration, bioenergy with CCS, and ocean-based approaches.

Source: IPCC Sixth Assessment Report, Working Group III (2022), Chapter 12 — "Cross-sectoral perspectives," documenting the role of carbon dioxide removal in 1.5°C mitigation pathways. Available at ipcc.ch.

Source: National Academies of Sciences, Engineering, and Medicine (2019), "Negative Emissions Technologies and Reliable Sequestration: A Research Agenda" — comprehensive review of DAC and other CDR technologies, their costs, scalability, and uncertainties.

The Chemistry: Solid Sorbent Systems

Climeworks, the Swiss company operating the Orca and Mammoth plants in Iceland, uses a solid sorbent approach. The chemistry relies on amine-functionalized materials — organic molecules containing nitrogen groups (amines, −NH₂ or −NH) that react selectively with CO₂ under specific temperature and humidity conditions.

Amine sorbent — a solid material whose surface is coated with amine functional groups. When ambient air flows over the sorbent at low temperatures (near 20–40°C), CO₂ molecules bind to the amine groups through chemisorption. When the sorbent is heated to 80–100°C in a low-pressure environment, the CO₂ releases, producing a concentrated CO₂ stream. The sorbent is then cooled and the cycle repeats.

The process runs in structured cycles: adsorption (CO₂ binds from airflow, typically 30–60 minutes), then desorption (heat releases CO₂, typically 10–30 minutes), then back to adsorption. The entire process is analogous to a sponge that absorbs water when cold and dry, then releases it when heated — except the "water" here is CO₂ and the "sponge" is a specially engineered solid material.

Climeworks' Mammoth facility, which came online in Iceland in May 2024, uses a modular design of collector units stacked in rows. Each collector module passes air through filter cartridges containing the solid sorbent. The facility has a nameplate capacity of 36,000 tonnes of CO₂ per year — approximately the annual emissions of 7,800 average US passenger cars. Captured CO₂ is injected underground into basaltic rock, where it mineralizes to calcium carbonate through a reaction with the rock's minerals within 1–2 years, providing near-permanent storage.

Source: Climeworks (2024), "Mammoth Plant Fact Sheet" — official technical specifications for the Mammoth DAC facility in Hellisheiði, Iceland, including capacity, sorbent type, and storage partner CarbFix. Available at climeworks.com.

Source: Sanz-Pérez, E.S. et al. (2016), "Direct Capture of CO₂ from Ambient Air," Chemical Reviews, Vol. 116 — comprehensive review of solid sorbent and liquid solvent DAC chemistry, thermodynamics, and material performance.

The Chemistry: Liquid Solvent Systems

1PointFive, a subsidiary of Occidental Petroleum, operates the Stratos DAC plant in West Texas, which came online in 2024. It uses a liquid solvent approach originally developed by Carbon Engineering (a company Occidental acquired in 2023). The chemistry is based on potassium hydroxide (KOH), a strongly alkaline solution that reacts with CO₂ from air to form potassium carbonate (K₂CO₃).

Potassium hydroxide (KOH) — a strongly alkaline compound that reacts with CO₂ in the reaction: 2 KOH + CO₂ → K₂CO₃ + H₂O. The potassium carbonate is then reacted with calcium hydroxide (Ca(OH)₂) in a caustic recovery process that regenerates KOH and produces calcium carbonate (CaCO₃) pellets. Those pellets are heated in a calciner to above 900°C, releasing pure CO₂ and regenerating calcium oxide (CaO), which is slaked back to Ca(OH)₂ to continue the cycle.

The liquid solvent approach uses an air contactor — a large structure resembling a cooling tower — where air flows over the KOH solution. The contactor can process large volumes of air continuously because the liquid sorbent does not require cycling off-line for regeneration. However, the calcination step (heating CaCO₃ to 900°C) requires substantial high-temperature heat, which is energy-intensive and difficult to supply from renewable electricity alone.

The Stratos plant in Texas has a nameplate capacity of approximately 500,000 tonnes of CO₂ per year at full buildout, though early operations were closer to 30,000 tonnes as the facility ramped up.

Feature	Solid Sorbent (Climeworks Mammoth)	Liquid Solvent (1PointFive Stratos)
Sorbent material	Amine-functionalized solid	KOH aqueous solution
Regeneration temperature	80–100°C (low-grade heat)	900°C (high-grade heat, calcination)
Air contactor	Modular filter cartridges	Large cooling-tower-style contactors
Current capacity	36,000 tCO₂/year	~30,000 tCO₂/year (ramp-up phase)
Target capacity	Scaling via modular replication	500,000 tCO₂/year at full buildout
CO₂ purity	>99%	>99%
Energy source advantage	Compatible with low-temp renewables	Large-scale industrial process; requires high-temp heat

Source: Keith, D.W. et al. (2018), "A Process for Capturing CO₂ from the Atmosphere," Joule, Vol. 2 — the technical paper describing the liquid solvent DAC process developed by Carbon Engineering, including energy requirements and cost analysis.

Source: 1PointFive / Occidental Petroleum (2024), "Stratos Plant Technical Overview" — capacity and technology specifications for the Stratos DAC facility in Ector County, Texas.

The Energy Problem

The central challenge of DAC is thermodynamic. CO₂ is already at its lowest chemical energy state in air — it is the waste product of combustion. Reversing this requires energy. The theoretical minimum energy to separate CO₂ from air at 420 ppm is approximately 8.4 kJ per mole of CO₂, or about 191 kJ per kilogram. In practice, real systems require far more than the thermodynamic minimum.

Climeworks' solid sorbent process uses approximately 1,500–2,000 kWh of heat and 500–600 kWh of electricity per tonne of CO₂ captured — totaling roughly 2,000–2,600 kWh per tonne. The Mammoth plant uses geothermal energy from Iceland's Hellisheiði power plant, which provides both electricity and heat at low carbon intensity. This matters: if DAC machines are powered by fossil-fuel electricity, the CO₂ emitted to generate that electricity can approach or exceed the CO₂ captured — producing near-zero or negative net removal.

A 2019 analysis by Fasihi et al. in Joule estimated the energy requirements for mature solid sorbent DAC at approximately 1,460–1,750 kWh per tonne at 2050 scale with improved materials and process integration. For reference, global electricity generation in 2024 was approximately 29,000 TWh. Removing 1 billion tonnes of CO₂ per year via DAC at current energy intensity would require approximately 2,000–2,600 TWh — roughly 7–9% of current global electricity production, entirely from low-carbon sources.

This is not a fictional or marginal energy requirement. It is a fundamental constraint that places a ceiling on how much CO₂ DAC can remove without displacing other electricity uses.

Source: Fasihi, M., Efimova, O., Breyer, C. (2019), "Techno-economic assessment of CO₂ direct air capture plants," Journal of Cleaner Production, Vol. 224 — comprehensive cost and energy analysis for DAC at scale, covering both solid sorbent and liquid solvent approaches through 2050.

Current Costs and the Learning Curve

The current cost of CO₂ removal via DAC is approximately $400–$1,000 per tonne of CO₂. Climeworks' Mammoth plant has a stated capture cost of approximately $1,000 per tonne of CO₂ at its current scale. This makes DAC roughly 10–100 times more expensive than the social cost of carbon used in most policy analyses ($50–$200 per tonne) and 20–200 times more expensive than nature-based carbon removal approaches (afforestation, soil carbon: $5–$50 per tonne).

Whether DAC costs will fall substantially depends on how strongly it follows a learning curve — the empirical pattern observed in many manufactured energy technologies (solar panels, lithium-ion batteries, wind turbines) where costs fall as a power-law function of cumulative deployed capacity.

Learning rate — the percentage by which unit costs fall for every doubling of cumulative production capacity. Solar photovoltaic modules have achieved a learning rate of approximately 20–24% over five decades. Lithium-ion batteries have shown similar rates. DAC's learning rate is uncertain because the technology is so early-stage that there are too few doublings to estimate reliably.

An analysis published in Energy and Environmental Science in 2023 by Hanna et al. estimated that if DAC follows a 15% learning rate (conservative) and cumulative capacity doubles from 2024's ~100,000 tonnes to 1 million tonnes by 2030, costs could fall to $300–$500 per tonne. Reaching $100–$150 per tonne — a level sometimes cited as cost-competitive for certain applications — would require cumulative capacity on the order of 100–200 million tonnes per year, which under current growth rates is decades away.

The U.S. Department of Energy launched a "Carbon Negative Shot" initiative in 2021 with a target of $100 per tonne of CO₂ for DAC by 2035. As of 2025, progress toward that target has been modest; the DOE's own assessments acknowledge that achieving $100 per tonne within a decade requires breakthrough improvements in sorbent materials and regeneration processes, not just deployment scaling.

Source: McQueen, N. et al. (2021), "A review of direct air capture (DAC): scaling up commercial technologies and exploring new lowcost approaches," Progress in Energy, Vol. 3 — systematic review of DAC technologies, cost trajectories, and the conditions required for cost reduction to $100/tonne.

Source: U.S. Department of Energy (2021), "Carbon Negative Shot" — official DOE initiative documentation for the $100/tonne DAC cost target. Available at energy.gov.

The Scale Challenge: Current Removal vs. Annual Emissions

Global CO₂ emissions from fossil fuels and industry in 2024 were approximately 37.4 billion tonnes of CO₂, according to the Global Carbon Project. All operational DAC plants combined remove approximately 0.01 million tonnes per year — about 0.000027% of annual emissions.

Put differently: if every DAC plant currently in operation ran continuously for a year, the total removal would be offset by global emissions in less than 9 minutes.

This comparison is not an argument against DAC. It is a calibration. The climate problem requires removing or preventing dozens of billions of tonnes per year. DAC's current contribution is in the tens of thousands of tonnes. The technology needs to scale by roughly five to six orders of magnitude (a factor of 100,000 to 1,000,000) to be a significant contributor to net zero.

For comparison, the wind power industry grew from negligible capacity in 1990 to approximately 2,100 GW of global installed capacity by 2024 — roughly a 4,000-fold increase. Wind energy now generates about 7% of global electricity. If DAC follows a comparable 35-year growth trajectory from 2024, it could plausibly remove 0.5–2 billion tonnes per year by 2060 — meaningful but still below the scale of nature-based sinks or the reductions needed from phasing out fossil fuels.

Source: Global Carbon Project (2024), "Global Carbon Budget 2024," Earth System Science Data — annual measurement of global CO₂ emissions and the carbon cycle. Available at globalcarbonproject.org.

Where DAC Fits in Climate Strategy

The climate strategy framework distinguishes between mitigation (reducing emissions) and carbon dioxide removal (extracting CO₂ already emitted). DAC belongs entirely in the second category. This framing matters for understanding its role.

Mitigation should come first. The emissions pathway over the next 20 years is the dominant determinant of peak warming. Carbon removal cannot compensate for continued high emissions — the rate of atmospheric accumulation from ongoing emissions is far faster than any plausible rate of removal.

DAC is relevant for specific, well-defined purposes:

Residual emissions offsetting: Some emissions are technically and economically very difficult to eliminate — aviation fuel combustion, certain chemical manufacturing processes, ruminant livestock methane. DAC can provide genuine carbon removal offsets for these residual emissions, unlike most voluntary carbon market credits that have faced credibility challenges.

Overshoot correction: Most 1.5°C scenarios involve a temporary "overshoot" — CO₂ concentrations rise above the target level before being drawn back down by large-scale carbon removal after mid-century. DAC, combined with bioenergy with CCS and enhanced weathering, is one of the tools modeled for this post-peak removal phase.

CO₂ as industrial feedstock: Captured CO₂ can be used to make synthetic fuels (combining CO₂ with hydrogen from electrolysis to make methanol or synthetic jet fuel) or building materials. These utilization pathways are not carbon removal — the CO₂ eventually re-enters the atmosphere — but they can substitute for fossil-sourced CO₂ in industrial chemistry.

Source: Fuss, S. et al. (2018), "Negative emissions — Part 2: Costs, potentials and side effects," Environmental Research Letters, Vol. 13 — systematic analysis of CDR technologies including DAC, comparing costs, potentials, and risks across approaches.

Source: Smith, P. et al. (2016), "Biophysical and economic limits to negative CO₂ emissions," Nature Climate Change, Vol. 6 — analysis of theoretical and practical limits to carbon removal approaches, including land and energy constraints.

Frequently Asked Questions

Q: How is DAC different from carbon capture at power plants?
A: Point-source carbon capture (at a power plant or industrial facility) extracts CO₂ from a concentrated exhaust stream — typically 4–15% CO₂ — and prevents it from reaching the atmosphere. This is far less energy-intensive than DAC because the CO₂ is already concentrated. DAC captures CO₂ from ambient air at 0.042% concentration, which requires more energy to achieve the same mass of CO₂ extracted. DAC is also unique in being able to draw down existing atmospheric CO₂, not just prevent future emissions.

Q: Is Iceland the only viable location for DAC?
A: Iceland is advantageous because of cheap geothermal energy and the basalt geology suitable for CO₂ mineralization storage. But DAC can operate anywhere with access to low-carbon energy and CO₂ storage. Texas (Stratos plant) demonstrates that DAC is possible in diverse geologies. Desert regions with high renewable energy potential are theoretically attractive due to low humidity (which improves sorbent performance in some designs) and abundant solar power for electricity.

Q: What happens to the CO₂ after it is captured?
A: Two main pathways. Geological storage: the CO₂ is compressed and injected deep underground into geological formations — saline aquifers or depleted oil fields — where it mineralizes over decades to centuries. This is considered near-permanent removal. Utilization: the CO₂ is sold for industrial use — carbonating beverages, growing plants in greenhouses, producing synthetic fuels, or making concrete. Utilization is not permanent removal because the CO₂ re-enters the atmosphere when the product is used or disposed of.

Q: Could DAC ever be cost-competitive without subsidies?
A: At current costs ($400–$1,000 per tonne), DAC is not economically viable without policy support. The 45Q tax credit in the United States pays $180 per tonne for DAC-captured CO₂ geologically stored, which covers 15–45% of current costs. At a hypothetical $100–$150 per tonne cost with carbon prices of $50–$100 in major jurisdictions, DAC could become viable for specific applications — particularly for airlines and shipping companies seeking credible offsets for hard-to-abate emissions.

Q: How many DAC plants would need to exist to remove 1 billion tonnes of CO₂ per year?
A: At Mammoth's capacity (36,000 tonnes per year), removing 1 billion tonnes would require approximately 27,800 plants of that size. At the scale of the Stratos facility at full buildout (500,000 tonnes), it would require approximately 2,000 plants. Building 2,000 Stratos-scale facilities would require investment in the hundreds of billions of dollars and enormous quantities of steel, concrete, and energy infrastructure. This illustrates why scale-up timelines are measured in decades.

Q: Does DAC compete with other land uses?
A: One of DAC's advantages over nature-based carbon removal is its small land footprint. A DAC plant removing 1 million tonnes of CO₂ per year requires roughly 1–3 km² of land, depending on design. An equivalent amount of removal through afforestation (planting trees) might require 800–2,000 km² of productive land, depending on location and tree species. This makes DAC complementary to, not competitive with, reforestation — the two approaches make different demands on land and water resources.

---

Read next:

• Renewable Energy Explained: Solar and Wind
• How the Internet Actually Works
• How Large Language Models Work