Carbon Membrane as Thin as 2nm Sieves CO2 Past Benchmarks in Lab Tests

A sheet of carbon as thin as 2 nanometres can pull carbon dioxide out of mixed gases faster than existing membranes while holding back nitrogen and methane, a team led by Tianjin University and Switzerland’s EPFL reported in Nature Sustainability on 23 October. The membrane let CO2 through at up to 5,000 gas permeation units, a rate the authors say “surpasses existing benchmarks”.

Most industrial carbon capture today washes exhaust gas through chemical solvents, which then have to be heated to release the CO2. Membranes promise to do the same job by filtering, but they have faced what the paper calls “a longstanding trilemma”: a membrane has to be extremely thin to be fast, have precisely sized pores to be selective, and be strong enough not to tear. This study claims all three, in the lab.

A graphene base lets the carbon layer stay this thin

The membrane has two layers. The team spread a polymer over a sheet of graphene, a single layer of carbon atoms, using what it calls a “super-spreading effect” to form a smooth, even coat. Heating the polymer in a controlled way turned it into a carbon layer as thin as 2 nanometres, riddled with tiny, evenly sized pores that let CO2 pass while holding back larger molecules.

The result is stiff for its size, with a Young’s modulus of about 26.8 gigapascals. In sieving tests it let through more than 60 times as much CO2 as nitrogen, the main split in power plant exhaust, and more than 70 times as much CO2 as methane, the split that matters for cleaning natural gas. It also separated hydrogen from CO2, which is relevant to syngas.

The paper’s supplementary information adds detail. The membrane is two to three orders of magnitude thinner than earlier carbon sieve membranes, which the authors credit for a “record-breaking” CO2 flow. At 100% humidity, a tough condition because water can clog pores, it still passed CO2 at more than 2,400 units with a CO2-to-methane selectivity above 50. The team also made a centimetre-scale version, and stiffness measurements after 24 hours under gas pressure stayed within the original range.

A permeation unit is a standard measure of how much gas passes through a membrane for a given pressure. For comparison, the EPFL group led by Kumar Varoon Agrawal reported atom-thin porous graphene last year with a CO2-to-nitrogen selectivity of up to 21 and flow of up to 4,050 units, though in different tests. The Tianjin group, led by Zhongyi Jiang and Guangwei He, separately published a different membrane design in Nature Energy earlier this month.

An outside reviewer sees promise for climate technology

Nature Sustainability published a News & Views article by Alexander Knebel of Friedrich Schiller University Jena alongside the paper. He describes “a carbon membrane as thin as only two molecular layers that sieves and captures CO2 with very high flux and selectivity, holding promise for climate change mitigation technologies”.

The membrane breaks the usual trade between speed and selectivity

Membrane makers normally have to choose. Thinner films pass gas faster but tend to let the wrong molecules through or break; tighter pores improve selectivity but slow everything down. Getting both a flow of up to 5,000 units and selectivity above 60 from a film that also survives handling is why the authors describe it as overcoming the trade-off, and why the reviewer reached for the phrase “very high flux and selectivity”.

The humidity and temperature results matter almost as much as the headline. Flue gas is wet, and the membrane kept sieving at full humidity; the team also tested it from minus 40C to 150C. At minus 40C its CO2 flow fell by more than half while its selectivity rose, and at 150C its hydrogen flow roughly tripled. Those are the conditions an engineer would ask about first.

The gap is scale. A 2nm film on a centimetre-sized support is a long way from the large areas of membrane a power station would need, and a 24-hour stiffness check says little about years of service. The published summary gives no cost for making the graphene base. Until the group or a partner shows a larger module running for months, this remains a materials result.

That gap matters for the UK. A bid to put a 1.2GW data centre at Drax in North Yorkshire rested its “carbon negative” claim on carbon capture with chemical scrubbers, the kind of chemical absorption this membrane is pitched as an alternative to. It joins a run of lab-to-pilot clean-tech results this month, from flexible perovskite solar modules to enzyme-embedded plastics and a subsea geothermal test.

How we sourced this

The full paper is behind a paywall; this report draws on its abstract, its publicly available supplementary information and the News & Views standfirst. No press release for this paper was available from Tianjin University or EPFL. The article page lists Alexander Knebel among the paper’s peer reviewers.

The figure to watch is not another record flow rate but the first membrane module measured in square metres, running on real flue gas.

Sources

• Guo, S., Shen, Y. et al., “Efficient carbon dioxide separation with molecularly thin bilayer carbon membranes”, Nature Sustainability, 23 October 2026 (doi:10.1038/s41893-026-01973-x)
• Supplementary Information for the paper (PDF)
• Knebel, A., “A thin carbon membrane to capture carbon”, News & Views, Nature Sustainability, 23 October 2026
• Kocaman, C. et al., “Scalable room temperature incorporation of CO2-selective ångström-scale pores in graphene for carbon capture”, Nature Communications, 2025
• Tianjin University news, 8 October 2026, on the group’s separate Nature Energy membrane paper (in Chinese)

Related stories

• UK Rejects York and North Yorkshire’s AI Growth Zone Bid Built Around a 1.2GW Data Centre at Drax
• Flexible Perovskite Solar Modules Reach 20.2% at Commercial Size
• Enzyme-Embedded Plastics Broke Down in the Deep Sea, Study Says