How Carbon Capture Works In Practice

Ordinary air is only about 0.04 percent carbon dioxide by volume, a concentration so dilute that pulling meaningful amounts of it out is closer to filtering a single flavor of dye out of an enormous swimming pool than scooping something visible off the surface. That dilution problem is exactly why carbon capture, whether from a power plant's exhaust stream or directly from open air, depends on chemistry engineered to selectively grab CO2 molecules and let everything else pass through. The technology exists and works at the chemistry-lab scale; the harder question, the one this article spends most of its time on, is whether it can work economically and reliably at the scale the climate problem actually demands.

Two very different starting concentrations, two very different challenges

Carbon capture generally falls into two broad families that face very different engineering problems, and understanding the split explains almost everything else about why some projects succeed and others struggle. Point-source capture pulls CO2 from a concentrated exhaust stream, like a power plant or cement factory smokestack, where CO2 concentrations can run from 4 percent up to 15 percent or more, dramatically higher than open air. Direct air capture (DAC) by contrast pulls CO2 straight from the atmosphere at that roughly 0.04 percent concentration, a far more dilute and energy-intensive target.

Think of it like the difference between recovering gold from a vein of ore versus panning it out of an entire riverbed. Both are technically the same extraction problem, but one starts with a resource already concentrated by geology, while the other requires processing an enormous volume of low-yield material to get the same amount out. This is why point-source capture, used at facilities like Boundary Dam in Saskatchewan, Canada and various industrial plants, has generally required less energy per ton of CO2 captured than direct air capture.

DAC plants like Climeworks' Orca facility in Iceland, which began operating in 2021 with a design capacity of roughly 4,000 tons of CO2 per year, need more energy per ton precisely because the target gas is so much more diluted in ordinary air. Climeworks scaled that concept up considerably with its Mammoth facility, which started operations near Reykjavik on May 8, 2024, and carries a nameplate capture capacity of up to 36,000 tons of CO2 annually, roughly ten times the size of Orca (Climeworks, 2024). Both Icelandic plants pair their DAC units with a mineralization partner called Carbfix, which injects the captured CO2 underground where it reacts with basalt rock and turns into solid carbonate minerals within a few years, a storage method distinct from the injection-into-porous-rock approach used at most industrial CCS sites.

As of the first quarter of 2025, the International Energy Agency's CCUS project database tracked just over 50 million tonnes of carbon capture and storage capacity in operation worldwide, only slightly higher than a year earlier, against roughly 37 billion tons of CO2 humanity emits annually ([IEA, 2025](https://www.iea.org/data-and-statistics/data-product/ccus-projects-database)). Direct air capture makes up a sliver of even that small total.

How the actual separation chemistry works

Most carbon capture systems rely on one of two chemical approaches: liquid solvents or solid sorbents, and both hinge on the same basic trick of chemistry, a substance that bonds weakly enough with CO2 to be reversed with heat, but strongly enough to actually grab it in the first place. Liquid solvent systems, most commonly amine-based chemicals, work by flowing exhaust gas up through a tower where it contacts a liquid solvent, typically a water solution containing 20 to 30 weight percent of an amine compound such as monoethanolamine (MEA). The amine chemically bonds with CO2 molecules while letting nitrogen, oxygen, and other gases pass through largely unreacted, a technology platform that has been used industrially since the 1930s and can achieve captured-CO2 purity above 99 percent (ACS Omega review, 2022).

That CO2-rich, or "rich," amine solution is then pumped to a separate regenerator vessel and heated to roughly 100 to 120 degrees Celsius (212 to 248 degrees Fahrenheit). That heat breaks the chemical bond between the amine and the CO2, releasing concentrated CO2 gas for compression and storage while regenerating "lean" amine that gets pumped back to the absorber for another cycle. It is essentially a two-step handshake and release: grab in one vessel, let go in another, using temperature as the switch.

Solid sorbent systems, increasingly favored for direct air capture because they handle dilute gas streams more efficiently than liquid amines do, work on the same absorb-then-release principle but use a solid material instead of a liquid one. A porous solid, sometimes with amine groups chemically anchored to its surface, works something like a specialized sponge: air passes over or through it, CO2 selectively sticks to active sites on the material's surface, and once the sorbent is saturated it gets heated or exposed to a vacuum to release the captured CO2 as a concentrated stream, similar in concept to how a household dehumidifier's silica-gel desiccant can be dried out in an oven and reused instead of thrown away.

Why does capturing the CO2 require so much energy just to release it again?

Breaking the chemical bond between the solvent or sorbent and the captured CO2 requires significant heat, because the same chemical affinity that makes a material good at grabbing CO2 also makes it energy-intensive to let go of it again. That regeneration step, the heating stage, is the single biggest energy cost in the whole process. Industry analyses estimate that heating the solvent for regeneration accounts for roughly 70 to 80 percent of a conventional amine-scrubbing system's total operating cost, which is the core economic obstacle the entire carbon capture industry keeps running into (ScienceDirect thermal degradation review, 2013).

Comparing the four main carbon capture approaches

Not every capture project uses the amine-scrubbing approach described above. Four broad approaches dominate real-world deployment today, and they differ significantly in where they capture CO2, how mature the technology is, and how energy-intensive they are per ton captured.

Approach	Where CO2 is captured	Core mechanism	Technology maturity
Post-combustion amine scrubbing	After fuel is burned, from flue gas	Liquid amine solvent bonds with CO2, then heat releases it in a regenerator	Commercially deployed since the 1930s for gas processing; used at Boundary Dam since 2014
Pre-combustion capture	Before fuel is fully burned, from syngas	Fuel is converted to hydrogen and CO2 in a gasifier, then CO2 is separated from hydrogen before combustion	Fewer commercial power projects; more common in industrial hydrogen and ammonia production
Oxy-fuel combustion	During combustion, using pure oxygen instead of air	Burning fuel in oxygen instead of air produces a flue gas that is mostly CO2 and water vapor, simplifying separation	Demonstrated at pilot and small commercial scale; not yet widely deployed at large power plants
Direct air capture (DAC)	Directly from ambient atmospheric air, at any location	Solid sorbents or liquid solvents pull CO2 from open air at roughly 0.04 percent concentration	Newest and least mature at scale; largest single plant (Climeworks Mammoth) captures up to 36,000 tons per year

Post-combustion capture is the most commercially proven approach for existing power plants, while direct air capture remains the least mature but is the only method that can remove CO2 already in the atmosphere anywhere on Earth.

What happens to the CO2 after it's captured

Captured CO2 does not simply vanish once separated from the exhaust or air stream; it has to be compressed, transported, usually by pipeline, and either stored permanently underground or put to industrial use. Permanent storage typically means injecting compressed CO2 deep underground into geological formations, like depleted oil and gas reservoirs or saline aquifers (porous rock layers saturated with salty water rather than freshwater or oil), at depths and pressures where the CO2 behaves as a dense, supercritical fluid, a phase that is neither fully liquid nor fully gas, that is unlikely to migrate back up to the surface.

The clearest real-world proof that this kind of underground storage can work for decades comes from Norway's Sleipner field in the North Sea, where the operator, then Statoil and now Equinor, has been separating CO2 from natural gas extracted at the platform and injecting it into the Utsira Sand saline aquifer, more than 800 meters (2,600 feet) below the seabed, continuously since 1996. That project began injecting roughly one million tons of CO2 per year specifically because Norway introduced a carbon tax that made capturing and storing the CO2 cheaper than paying to vent it, and by recent counts more than 23 million tonnes of CO2 have been injected and permanently stored there through a single extended-reach well (Sleipner project documentation via MIT, ongoing). Nearly three decades of continuous seismic monitoring at Sleipner has become one of the field's most-cited pieces of evidence that engineered underground CO2 storage can hold for long timescales without leaking back out.

Not all captured CO2 goes toward permanent storage, though. Some is instead used industrially, for example in enhanced oil recovery, where operators inject CO2 into aging oil wells to help push out additional oil that conventional pumping can no longer reach. This use case is more controversial from a climate standpoint, since the CO2 injected can help extract and burn more oil, meaning the net climate benefit depends heavily on how much of the injected CO2 stays underground permanently versus how much additional fossil fuel combustion the technique enables downstream.

The Iceland approach: mineralizing CO2 into rock instead of just storing it

Iceland's basalt-rich geology gives the Climeworks projects there, Orca and the larger Mammoth facility, a storage method that differs meaningfully from most other capture sites. Rather than simply injecting supercritical CO2 into a porous reservoir and relying on caprock to keep it from migrating, the Carbfix process mixes captured CO2 with water and injects the mixture into basalt formations, where it reacts chemically with the rock's calcium, magnesium, and iron content to form solid carbonate minerals within a few years, a much faster mineralization timescale than the geological storage used at sites like Sleipner. Basalt is common in Iceland's volcanic terrain but is far less common under most industrial regions worldwide, which is part of why this particular storage pathway has not scaled beyond a handful of favorable sites.

The tradeoff worth being honest about is that mineral storage in basalt currently costs more per ton than injecting CO2 into a conventional saline aquifer, and Iceland's abundant, comparatively cheap geothermal energy is part of what makes running an energy-hungry DAC plant there economically feasible in the first place. A DAC plant built in a region without cheap low-carbon electricity would face a much steeper energy bill for the same capture volume, since running fans to move enormous volumes of air and heating sorbent for regeneration both require continuous power input.

Why carbon capture remains contested as a climate strategy

Carbon capture is not a silver bullet, and climate scientists broadly treat it as a complement to, not a substitute for, reducing emissions at the source. Even well-run capture facilities typically capture somewhere between 60 and 95 percent of a given exhaust stream's CO2, not all of it, and the energy required to run capture equipment can itself generate additional emissions if that energy comes from fossil sources rather than renewables or nuclear power.

The gap between announced and actual performance is not hypothetical, and Boundary Dam is the case study most often cited. SaskPower's Boundary Dam 3 unit in Saskatchewan began capturing CO2 from a coal-fired power plant in 2014 with a target capture rate of 90 percent of the plant's emissions. According to SaskPower's own long-term data, the facility's actual capture rate through the end of 2023 has run closer to 57 percent, well below the original target, in large part because the capture equipment has only been available to operate during roughly 80 percent of the hours the power plant itself has run, meaning flue gas was vented uncaptured during the remaining downtime (IEEFA analysis of SaskPower data, 2024). SaskPower and the International CCS Knowledge Centre first publicly acknowledged in a 2022 paper that BD3 had not sustained its planned 90 percent capture rate since 2015.

That pattern is why credible discussions of carbon capture tend to treat announced capacity and capture-rate figures cautiously until they are verified by independent, ongoing monitoring rather than by a facility's original design specifications. It also underscores a broader industry reality: the IEA's CCUS project database shows that while eight new capture and storage projects began operating globally in 2024, most were small, some as low as 5,000 tonnes of CO2 per year in capacity, meaning total global operational capacity grew only marginally that year even as more offtake agreements for future carbon removal were being signed (IEA CCUS commentary, 2025).

What people get wrong about scaling carbon capture up

A common misconception is that carbon capture is primarily a chemistry problem still waiting to be solved. It is not. The core chemistry, amine solvents grabbing CO2 and releasing it under heat, solid sorbents doing the same thing with less water handling, has been understood and industrially deployed since the 1930s for uses like natural gas purification, long before climate applications existed. What has not been solved is the economics and reliability of running that chemistry continuously, at gigaton scale, without diverting so much energy into the capture process itself that the net climate benefit shrinks.

Another common assumption is that once a facility is built, it simply runs at its advertised capture rate indefinitely. Boundary Dam's multi-year gap between its 90 percent design target and its 57 percent realized rate shows why that assumption is risky to build climate policy around. Equipment downtime, maintenance needs, and the simple fact that capture equipment and the power plant it's bolted onto do not always run on the same schedule, all eat into real-world performance in ways a design brochure does not capture.

A third misconception worth flagging is treating direct air capture and point-source capture as interchangeable solutions to the same problem. They solve different problems. Point-source capture reduces emissions from a specific facility that is still operating and still burning fuel. Direct air capture, by contrast, can in principle remove CO2 that was emitted decades ago, anywhere in the world, since atmospheric CO2 mixes globally, which is why DAC is often discussed as a tool for actual carbon removal rather than merely emissions reduction. But that flexibility comes at the cost of dramatically higher energy intensity per ton captured, since DAC has to process roughly 350 times more dilute a gas stream than a coal plant's flue gas.

The takeaway

Carbon capture works by using chemistry, liquid amine solvents or solid sorbents, that selectively bonds with CO2 molecules while letting other gases pass through, then applying heat or pressure changes to release that CO2 again for compression, transport, and either underground storage or industrial use. The technology's biggest practical challenge is not the underlying chemistry, which has been well understood for nearly a century, but the combination of CO2's dilution in open air and the steep energy cost of the regeneration step that releases captured CO2 from its solvent or sorbent. Real-world projects like Sleipner in Norway show that underground storage itself can hold reliably for decades, while Boundary Dam in Canada shows that hitting a facility's advertised capture rate consistently, year after year, is a separate and much harder engineering problem than the underlying chemistry suggests. Anyone evaluating a carbon capture claim should ask not just "does the chemistry work" but "has this specific project sustained its promised capture rate over multiple years of independently verified operation."

---

Read next:

• How biochar stores carbon in soil
• Why methane matters for climate change
• Why grid storage is hard

FAQ

Q: What's the difference between direct air capture and point-source carbon capture?

Point-source capture pulls CO2 from concentrated exhaust streams like power plants, where concentrations can reach 4 to 15 percent or more, while direct air capture pulls CO2 from ordinary atmospheric air at only about 0.04 percent concentration, making it significantly more energy-intensive per ton captured. Point-source capture reduces a specific facility's ongoing emissions, while direct air capture can in principle remove CO2 already in the atmosphere regardless of where or when it was emitted.

Q: How does the captured CO2 actually get released from the capturing material?

Heat, typically 100 to 120 degrees Celsius (212 to 248 degrees Fahrenheit) for amine solvents, or pressure and temperature changes for solid sorbents, breaks the chemical bond between the CO2 and the material that captured it. This releases concentrated CO2 gas for compression while the solvent or sorbent is regenerated and reused for another capture cycle. This regeneration step accounts for roughly 70 to 80 percent of a conventional amine system's operating cost.

Q: Where does captured CO2 actually go once it's separated?

Most permanently stored CO2 is compressed and injected deep underground into geological formations like saline aquifers or depleted oil and gas reservoirs, a practice with a multi-decade track record at sites like Norway's Sleipner field, which has stored more than 23 million tonnes since 1996. In Iceland, Climeworks' projects instead mix captured CO2 with water and inject it into basalt rock, where it mineralizes into solid carbonate within a few years.

Q: Does carbon capture technology capture 100 percent of emissions from a facility?

No. Even well-run facilities are designed to capture somewhere between 60 and 95 percent of a given exhaust stream's CO2, and real-world performance can fall well short of design targets. SaskPower's Boundary Dam 3 in Canada, for example, was designed for a 90 percent capture rate but has averaged closer to 57 percent through 2023, largely because the capture equipment is not always available while the power plant is running.

Q: Is carbon capture considered a replacement for cutting emissions at the source?

No. Climate scientists and international assessments generally treat carbon capture as a complement to emissions reduction, not a substitute, particularly since capture equipment itself requires energy that can generate additional emissions if it is not powered by low-carbon sources. Global operational CCS capacity, just over 50 million tonnes as of early 2025 according to the IEA, remains a tiny fraction of the roughly 37 billion tons of CO2 emitted globally each year.

Q: How big is the world's largest direct air capture plant right now?

Climeworks' Mammoth facility near Reykjavik, Iceland, started operations on May 8, 2024, and has a nameplate capture capacity of up to 36,000 tons of CO2 per year, making it roughly ten times larger than the company's earlier Orca plant, which began operating in 2021 with a capacity of about 4,000 tons per year.