Industries / Energy, water and waste / Carbon capture
Uses for captured carbon
Selling or using captured carbon dioxide: to push more oil out of old fields, to make fertilizer and drinks, and to make fuels, chemicals and building materials.
5
Milestones achieved, from 1772 to 2025.
2
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
1772
Water made to sparkle
Joseph Priestley, an English minister and chemist, collected the gas given off by fermenting beer, then called "fixed air" and now carbon dioxide, and worked out how to dissolve it in water. In 1772 he published a pamphlet in London, dedicated to the head of the Admiralty, on "impregnating water with fixed air" to give it the sparkle of the mineral water of Pyrmont. The Science History Institute calls the result the first carbonated water. Priestley hoped it would cure scurvy at sea, which it did not, and the Royal Society gave him its highest honour in 1773 for the work.
LessonA gas that would otherwise be thrown away found its first buyer in drinks, which still use it today.
Credit
Joseph Priestley
2012
Methanol from geothermal CO2
Carbon Recycling International (CRI) built its George Olah plant at Svartsengi, Iceland, in 2012. It takes CO2 that would otherwise be vented from a geothermal power plant next door, combines it with hydrogen made by splitting water with electricity from the grid, and makes methanol, a liquid fuel and chemical feedstock. The International Energy Agency (IEA) described it in 2019 as the largest CO2-based fuel plant in operation, converting about 5,600 tonnes of CO2 a year, after CRI expanded its output in 2015 from 1,000 tonnes to more than 4,000 tonnes of methanol a year. CRI reports that an audit by SGS Germany, under a protocol set by the certifier ISCC, found 80–90% lower CO2 emissions than a comparable amount of fossil fuel.
Credit
Carbon Recycling International
2019
Most use is old use
The IEA estimated in September 2019 that about 230 million tonnes of CO2 are used each year. About 130 million tonnes go into urea fertilizer, mostly CO2 made and used on site at ammonia plants, and 70–80 million tonnes into enhanced oil recovery, mainly in North America; food and drink, metal working and other uses take the rest. CO2 made into new fuels, chemicals and building materials was "negligible" by comparison. The IEA notes that carbon in fuels is released within a year and in most chemicals within ten years, while building materials hold it permanently.
LessonUsing CO2 is not the same as storing it, because in fuels and chemicals the carbon returns to the air.
Credit
IEA, using IHS Markit's estimate of 2015 demand
2023
Most US CO2 is natural
Under EPA's reporting rules for CO2 suppliers, US supply in 2023 was 42.9 million tonnes: 16.1 million captured at industrial plants and 26.8 million drawn from natural underground deposits. Of the total, 33.76 million tonnes went to enhanced oil and gas recovery and 4.59 million to food and drink.
LessonOil fields still get most of their CO2 from natural deposits drilled for it, so CO2 sold for oil recovery is not always captured CO2.
Credit
US EPA Greenhouse Gas Reporting Program
2025
One credit for every use
The US 45Q credit, enacted in 2008, at first paid only for CO2 captured at plants and stored or used for oil recovery; a 2018 law added direct air capture and use in products. The One Big Beautiful Bill Act, signed in July 2025, set the same amount for every end use: generally $85 a tonne for CO2 captured at a plant and $180 for direct air capture, when wage and apprenticeship rules are met, for equipment placed in service after 4 July 2025. A product maker must first have a lifecycle analysis (LCA), a count of emissions from raw material to disposal, approved by the IRS after review by the Department of Energy; by January 2026 the IRS had approved 26 of 54 requests, rejected 22 and was still reviewing 6. GAO noted that oil recovery and products earn income of their own, which raises the question of paying for work that would have happened anyway.
Credit
US Congress; account by GAO, August 2026
Watch for
Jet fuel from CO2
Twelve opened AirPlant One at Moses Lake, Washington, on 10 June 2026, with Alaska Airlines and Microsoft. It says the plant turns captured CO2, water and renewable electricity into jet fuel that meets the ASTM D7566 Annex A1 specification, that the fuel is "being delivered and sold for commercial aviation use", and that its lifecycle emissions are up to 90% lower than ordinary jet fuel. It gives no production figure.
Achieved whenan outside party, such as a fuel certifier or a state clean fuel program, publishes the volume delivered from the plant over a year.
Claimed by
Twelve, June 2026
A larger e-methanol plant
CRI says it delivered and installed a methanol reactor for Tianying's e-methanol project at Liaoyuan, Jilin Province, China, whose first phase is to make about 170,000 tonnes of methanol a year from hydrogen made with renewable power and CO2 captured from burning biomass, with start-up later in 2026. CRI says it will be the largest e-methanol plant in operation.
Achieved whenthe plant reports a full year of output and an outside party confirms the volume and the sources of its hydrogen and CO2.
Claimed by
Carbon Recycling International, July 2026
Open
01
Oil per stored tonne
For every oil field that reports CO2 storage under subpart RR, a yearly table of CO2 stored next to oil produced, giving barrels of oil per tonne stored and the net CO2 once the oil is burned, with every figure linked to its source.
The briefCredit · unclaimed02
Same product, different answers
A public set of worked lifecycle analyses for three CO2-based products (methanol, synthetic jet fuel and carbonated building aggregate), each run several times with different allowed choices of comparison product and data source, to show how much the 45Q "displacement factor" that sets the credit changes with each choice.
The briefCredit · unclaimed
Open 01 · the brief
Oil per stored tonne
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
For every oil field that reports CO2 storage under subpart RR, a yearly table of CO2 stored next to oil produced, giving barrels of oil per tonne stored and the net CO2 once the oil is burned, with every figure linked to its source.
Why it is still open
Methods exist. A 2019 report for the US National Energy Technology Laboratory by Vanessa Núñez-López and colleagues at the University of Texas set out a repeatable way to judge whether oil from a CO2 flood is "net carbon negative", and Azzolina and co-authors analysed historical floods statistically in 2015. EPA's table gives each field's stored CO2 by year to 2023, and state records give its oil output, but searches of the web, Crossref and GitHub found no public field-by-field table that puts the two together. Since 2025 the US pays the same credit per tonne for oil recovery as for dedicated storage, and GAO has asked whether that pays for work that would happen anyway.
Built with what exists
EPA's subpart RR table and annual monitoring reports for oil fields such as Wasson San Andres and Hobbs; public oil production records from the Railroad Commission of Texas and New Mexico's Oil Conservation Division; Python.
First step
Take the Wasson San Andres Field reports for 2016–2023 and match them with the field's yearly oil production from the Railroad Commission of Texas.
Done looks like
Every oil field reporting under subpart RR, for every reported year, with oil produced, CO2 stored, barrels per tonne and net CO2 under a stated emission factor for burned oil, and an outside reviewer reproducing two fields from the links.
Open 02 · the brief
Same product, different answers
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A public set of worked lifecycle analyses for three CO2-based products (methanol, synthetic jet fuel and carbonated building aggregate), each run several times with different allowed choices of comparison product and data source, to show how much the 45Q "displacement factor" that sets the credit changes with each choice.
Why it is still open
The IRS will not pay the credit for products until it approves an LCA, and by January 2026 it had rejected 22 of 54 requests. GAO recommended that the IRS and the Department of Energy clarify which datasets are acceptable for calculating carbon displaced; the agencies disagreed with five of its six recommendations. NETL's free toolkit ships with a single example system, and searches found no public comparison across data choices.
Built with what exists
NETL's CO2U LCA Guidance Toolkit, version 2.1, with its 45Q addendum and openLCA database; openLCA, free LCA software; published plant figures such as those the IEA gives for the George Olah plant as inputs.
First step
Install openLCA, load NETL's CO2U database version 2.1, and rerun its example system until the result matches NETL's.
Done looks like
Three products, each with at least three data choices, every run's displacement factor and its spread published with the model files, and an outside LCA practitioner reproducing one result.