A polymer membrane developed by researchers at KAIST and the Georgia Institute of Technology separates crude oil into light and heavy fractions at room temperature, without the boiling step that has defined oil refining for more than a century. The work, a peer-reviewed study published online on June 24, 2026, could chip away at crude oil distillation’s roughly 1,100 terawatt-hours of annual energy demand.
The team, led by KAIST Professor Dong-Yeun Koh with Professor Ryan Lively’s group at Georgia Tech, used a bare polyacrylonitrile (PAN) membrane with no engineered selective layer. As crude oil flows through, heavy hydrocarbons deposit on the inner pore walls and narrow them to channels smaller than 2 nanometres, a self-built sieve that lets naphtha, gasoline, and kerosene pass while holding heavier molecules back.
How a Membrane Built Its Own Sieve
Conventional membrane separation relies on a thin, specially designed selective layer coated onto a porous support. That coating is what does the molecular sorting, and it is also what makes industrial-scale fabrication expensive and defect-prone. The KAIST team removed the coating entirely.
The bare PAN support, a chemically stable polymer already mass-produced as a filtration substrate, has surface pores roughly 15 to 18 nanometres wide. Crude oil molecules sit at about 2 nanometres, far smaller than the openings, so on first contact nothing separates. Then the heavy fractions, including asphaltenes and long hydrocarbon chains, adhere to the inner pore walls. The deposits narrow the channels, and the narrowing stabilises at openings smaller than 2 nanometres that admit only lighter molecules.
This study reveals a new scientific principle in which a membrane interacts with a complex mixture and spontaneously forms its own separation channels.
Koh is a professor of chemical and biomolecular engineering at KAIST and a corresponding author on the Nature paper. He said the team had validated the technology using real crude oil supplied by HD Hyundai Oilbank. Within about five to ten minutes of operation, the permeate shifts from dark crude to a clear amber filtrate, visible evidence that the feedstock has carved its own sieve.
Twenty-Three Times the Throughput, Twenty-Eight Days Stable
The PAN membrane delivered crude oil permeation rates approximately 23 times higher than those of previously reported state-of-the-art crude oil membranes, per the institutional announcement describing the breakthrough. It held stable separation performance for 28 consecutive days of continuous operation. Georgia Tech’s release put the stability figure at four weeks, the same window. Earlier coated-membrane prototypes typically last about two weeks before fouling-related degradation forces replacement.
- ~23x higher permeation than previously reported crude oil membranes
- 28 days of stable continuous operation
- $5/m² approximate cost of the bare PAN support layer
- 1,100 TWh/year global energy used by crude oil distillation
Four weeks of continuous operation is a different conversation with a refinery engineer than a short-burst laboratory demonstration. The throughput gap of 23 times compounds with that stability window: a module that processes more crude per square metre per day, and stays online longer, pushes down the capital cost per barrel. The PAN polymer itself is a commodity chemical, mass-produced for water-treatment and textile applications, so the supply chain to scale up is already in place.
Savings from a Membrane Pretreatment
The membrane does not stand alone. It works as a pretreatment, sending the lighter fractions into conventional distillation at a lower throughput while the heavier residue still goes through the legacy column. The team’s process simulations, using real crude oil supplied by HD Hyundai Oilbank, mapped the savings.
| Metric | Reduction with membrane pretreatment |
|---|---|
| Energy consumption | 31.6% |
| Carbon dioxide emissions | 37.6% |
| Cooling water usage | 20.7% |
| Operating cost | 36% |
Distillation currently heats crude oil above 350 °C. Pre-filtering out the light fractions means less material has to reach that temperature, and less cooling water has to be circulated afterwards. The numbers above come from process simulations of a Korean refinery, with the membrane deployed as a front-end filter and the residue still routed through a conventional distillation column. The legacy column still handles the heavier residue.
Applied across Korea’s refining and petrochemical sector, the technology could cut greenhouse-gas emissions by approximately 10 million tonnes annually, equal to the output of roughly four million internal combustion vehicles. The membrane is a filter placed before distillation, with the heavier residue still routed through the legacy column. Refiners get a smaller, lower-temperature column load for the same end products.
Beyond Crude Oil and Refining
The same self-forming channel principle is not limited to crude oil. The researchers point to four adjacent separation tasks that share the fouling-as-feature chemistry.
- Purification of pyrolysis oil from waste plastics
- Recovery of solvents used in battery manufacturing
- Pharmaceutical purification
- Biofuel production
Each of these processes currently relies on energy-intensive distillation or extraction steps. The PAN membrane’s room-temperature operation could offer the same kind of pre-fractionation shortcut the team demonstrated for crude oil. The team observed similar fouling-driven channel formation in a second membrane material tested alongside PAN, a sign that the mechanism may extend to other porous polymers.
The researchers describe the membrane platform as a versatile base for next-generation molecular separations across multiple industries. Korean media reported the bare PAN support runs around $5 per square metre. A plastic pyrolysis operation, for example, might favour a different polymer than crude oil refining. Koh said the team’s next step is to extend the platform to plastic recycling, biofuel purification, and other sustainable chemical processes that support carbon neutrality.
Membrane research has historically chased ever-thinner engineered coatings to do the molecular sorting. The KAIST experiments show the separating structure can emerge from the crude oil itself. The team frames the work as a move from designing a perfect sieve to letting the mixture build one.
Three to Five Years to Commercialisation
Professor Koh told Korean media that, under conservative assumptions, payback for the membrane system runs about three to five years. He added that the technology could be commercialised within three to five years if pilot-stage modules hold up across different crude oils and over long-term operation. Material risks are low because PAN is already commercialised. The KAIST team will next verify scalability, testing large-area membranes and modules for industrial stability across different crude oil types.
Professor Ryan Lively of Georgia Tech called the productivity jump significant enough that ‘industry should seriously consider adopting the technology.’ His group’s write-up of the collaboration’s findings describes the result as a step toward industrial adoption. Lively, the Thomas C. DeLoach Jr. Endowed Professor at Georgia Tech’s School of Chemical and Biomolecular Engineering, served as an advisor and corresponding author on the Nature paper.
The membrane ships as a modular filtration unit that bolts onto existing refinery pipelines, so adoption does not require replacement of the distillation column itself. Refiners can also use the membrane units to incrementally expand throughput beyond the limit set by legacy distillation hardware. Andrew Livingston, vice president of research and innovation at Queen Mary University of London, who was not involved in the study, called the work a terrific piece of research that rewards curiosity. His group added that the work tackles the first, hard-to-achieve separation of heavy hydrocarbons, while most work to date has focused on lighter oils. Refiners now have an alternative to capital-intensive distillation expansions.
Who Funded the Work
The research was supported by the Korean Ministry of Science and ICT through the National Research Foundation of Korea’s Basic Research Program for outstanding early-career researchers and its Engineering Research Center program. Professor Lively’s work at Georgia Tech was supported by the Thomas C. DeLoach Jr. Endowed Professorship. HD Hyundai Oilbank supplied crude oil for the experiments and characterised the feedstocks. Co-first authors Dr. Jihoon Choi and Dr. Hyeokjun Seo of KAIST led the experimental work alongside their advisors.
The study, “Crude Oil Fractionation by Means of Mesoporous Polyacrylonitrile Membranes,” appears in Nature with DOI 10.1038/s41586-026-10677-3. The author list spans KAIST’s Department of Chemical and Biomolecular Engineering, the Korea Research Institute of Chemical Technology, HD Hyundai Oilbank, and Georgia Tech’s School of Chemical and Biomolecular Engineering, with Koh, Lively, and Jae W. Lee of KAIST as co-corresponding authors.
Frequently Asked Questions
When will this crude oil membrane reach refineries?
Professor Koh estimated commercialisation within three to five years if pilot modules operate stably across different crude oils. The first installations would likely pair the membrane with existing distillation columns as a pretreatment step; full replacement of distillation is not on the table.
Does this membrane replace crude oil distillation?
Not yet. The membrane enriches the lighter fractions, including naphtha, gasoline, and kerosene, but the permeate still contains heavier components that must go through conventional distillation. The team’s simulations treat the membrane as a front-end filter, with energy savings of 31.6% compared to distilling whole crude.
What is polyacrylonitrile (PAN), and why does it work?
PAN is a chemically stable, inexpensive polymer already manufactured at scale as a support layer in industrial filters. The KAIST team found that when crude oil flows through bare PAN, heavy hydrocarbons deposit on the pore walls and narrow the openings to channels smaller than 2 nanometres. The fouling itself becomes the separating layer.
Who funded the research?
The Korean Ministry of Science and ICT funded the work through the National Research Foundation of Korea’s Basic Research Program for early-career researchers and the Engineering Research Center program. Georgia Tech’s contribution was supported by the Thomas C. DeLoach Jr. Endowed Professorship.
What other chemical separations could this approach tackle?
The researchers named purification of pyrolysis oil from waste plastics, recovery of solvents used in battery manufacturing, pharmaceutical purification, and biofuel production as adjacent applications where the same fouling-as-feature mechanism could apply.
