The trade-off in question dates back to the late twentieth century, when researchers plotting the performance of glassy polymers noticed a stubborn pattern: materials that let CO2 diffuse quickly tended to be poor at rejecting other gases, and highly selective materials tended to be slow. When Lloyd Robeson consolidated these observations into his famous upper-bound plots in 1991 and revisited them in 2008, the resulting curves became the de facto scoreboard for membrane science. Breaking the upper bound became a headline achievement, and entire families of polymers of intrinsic microporosity, thermally rearranged polymers, and mixed-matrix composites have been celebrated for doing exactly that. The new Perspective does not dispute the physics behind these curves, but it insists that the curves answer only one narrow question about one narrow class of materials.

The core of the authors’ argument is that membrane performance is constrained by several distinct physical limits, and the permeability–selectivity trade-off captures only some of them. First, there are inherent upper limits on selectivity itself: no material, however cleverly designed, can distinguish CO2 from other molecules with infinite precision, because sorption and diffusion selectivities are bounded by the thermodynamics and kinetics of the gas–membrane system. Second, permeance, the practical flux per unit area and pressure, does not scale indefinitely with thinning. As membranes approach molecular thicknesses, interfacial resistances, defects, and support-layer limitations begin to dominate, meaning that an ultrathin film cannot simply deliver the permeance predicted by dividing bulk permeability by thickness. Third, operating conditions, most notably CO2 partial pressure, fundamentally alter what a membrane can achieve, sometimes reversing the apparent advantages of one material class over another.

That last point may be the most consequential for real-world deployment. The authors show that materials with high CO2 sorption or reactive affinity, such as facilitated transport membranes built on amines, amino acid salts, ionic liquids, or even enzymes like carbonic anhydrase, shine precisely where the feed gas is dilute and high selectivity is non-negotiable. In direct air capture, where CO2 makes up roughly 420 parts per million of the atmosphere, the driving force for separation is vanishingly small, and only materials that chemically or strongly physically grab CO2 can achieve meaningful capture fractions. But the same high-affinity chemistry becomes a liability at elevated CO2 partial pressures. Carrier saturation, reaction stoichiometry, and plasticization mean that facilitated transport membranes can actually lose permeability and selectivity as the CO2 concentration in the feed rises, a counterintuitive result that simple upper-bound plots cannot predict.

Conversely, the analysis finds that conventional polymers and non-reactive liquids are the materials of choice when CO2 partial pressures are high and very high permeance is required. In natural gas sweetening or pre-combustion capture, where CO2 concentrations can reach tens of percent, the physical solubility and diffusivity of glassy and rubbery polymers provide robust, high-throughput separation without the saturation bottlenecks of reactive carriers. This reframing turns what looks like a materials competition into a matching problem: the best membrane for a given job depends on where that job sits on the partial-pressure axis. A single universal champion material, the authors suggest, is a category error, and research programs that optimize exclusively against the Robeson upper bound may be steering effort toward applications where their materials will never be economically relevant.

The thickness problem deserves particular attention because it sits at the heart of industrial feasibility. Permeance, not permeability, determines the membrane area, and therefore the capital cost, of a full-scale separation plant. The obvious route to high permeance is to make the selective layer thinner, and laboratory reports of films tens of nanometers thick are now common. Yet the Perspective highlights that permeance gains from thinning eventually stall. Surface barriers at the membrane–gas interface, which have been measured even in crystalline zeolites such as silicalite-1, impose a floor on resistance that no amount of thinning can remove. Physical aging in ultrathin glassy films densifies the free volume over time, eroding permeability after fabrication. Pinholes and defects become statistically unavoidable as area scales up, and the porous supports that mechanically hold ultrathin layers add their own resistance. The practical consequence is that permeance targets for post-combustion capture, which earlier techno-economic work by Merkel and colleagues placed in the range of thousands of gas permeation units, remain genuinely difficult even for materials with spectacular intrinsic permeability.

Operating-condition effects extend beyond partial pressure alone. Humidity, for instance, cuts both ways: water can enhance CO2 transport in amine-containing facilitated transport membranes by enabling bicarbonate formation, yet it can swell and degrade ionic-liquid membranes or suppress transport in polybenzimidazoles. Temperature shifts the balance between sorption and diffusion, generally favoring diffusion selectivity at lower temperatures and reactive kinetics at higher ones, and thermal stability limits vary enormously between polymeric, ionic-liquid, and inorganic systems. Pressure ratio, the ratio of feed to permeate CO2 partial pressures, caps the achievable stage cut and purity regardless of how selective the membrane is, a constraint well known to process engineers but often absent from materials-level benchmarking. The authors argue that any credible evaluation framework must fold these variables in from the start, rather than treating them as afterthoughts in a techno-economic analysis performed after the material is already fixed.

Applying this expanded framework yields a set of practical operating windows that map material classes onto industrially relevant separations. For flue-gas carbon capture, with CO2 partial pressures around 0.1 to 0.15 bar, facilitated transport membranes and high-sorption materials hold a genuine edge, provided their stability against humid, oxygen-containing streams can be solved. For biogas upgrading, where CO2 levels near 40 percent meet methane at elevated total pressure, plasticization-resistant polymers and mixed-matrix membranes containing metal–organic framework fillers are better matched to the duty. For direct air capture, the thermodynamic analysis of dilute-gas separation suggests that only membranes combining very high selectivity with high CO2 affinity, potentially including supported ionic liquids and biocatalytic systems, can approach competitive energy footprints, and even then the required membrane areas are daunting. The Perspective also flags potentially disruptive directions: water itself as a separation medium, rubbery organic frameworks aiming at ultrapermeability, and enzymatic liquid membranes that exploit nature’s fastest CO2 conversion chemistry.

What makes the analysis timely is the sheer expansion of the membrane application space. A decade ago, CO2 membranes were largely discussed in the context of natural gas processing and, aspirationally, post-combustion capture. Today, proposals span biogas purification to biomethane, hydrogen purification with CO2 co-capture, and gigatonne-scale direct air capture, each with distinct partial pressures, impurity profiles, and cost sensitivities. Funding agencies and startups alike are pouring resources into novel materials, and the risk identified by Saffer-Meng and colleagues is that of systematic misallocation: brilliant materials optimized for the wrong regime, benchmarked against a curve that cannot see the difference between the regimes. By articulating inherent selectivity ceilings, thickness-dependent permeance constraints, and partial-pressure effects as first-class evaluation criteria, the Perspective offers the field a more honest compass.

The message to membrane scientists is ultimately an enabling one rather than a discouraging one. The upper bound remains a useful map of the diffusion–solubility landscape for glassy polymers, and surpassing it still matters for the applications where those polymers fit. But the frontier of carbon capture now lies in regimes the upper bound never charted: dilute feeds demanding extraordinary selectivity, concentrated feeds demanding plasticization resistance, and ultrathin architectures demanding defect-free scaling. Recognizing the true fundamental limits of each regime, the authors conclude, is the fastest way to direct ingenuity where it can actually move the needle on the economics of separating carbon dioxide from the air, the smokestack, and the digester.

Subject of Research: Fundamental limits of carbon dioxide separation membranes

Article Title: Evaluating the fundamental limits of carbon dioxide separation membranes

Article References: Saffer-Meng, M., Lopez, K. P., & Straub, A. P. (2026). Evaluating the fundamental limits of carbon dioxide separation membranes. Nature Chemical Engineering. https://doi.org/10.1038/s44286-026-00441-9

Image Credits: AI Generated

DOI: 10.1038/s44286-026-00441-9

Keywords: carbon capture, CO2 separation, membranes, Robeson upper bound, facilitated transport, direct air capture, flue gas, biogas upgrading, ionic liquids, permeance, selectivity, chemical engineering

The answer that MES offers is deceptively simple in concept. A typical system consists of two chambers separated by an ion-exchange membrane. In the anodic compartment, water or organic substrates are oxidized to release electrons. In the cathodic compartment, electroactive microorganisms intercept those electrons either directly from the cathode surface, where they form biofilms, or indirectly via hydrogen gas and redox mediators generated at the electrode. Once inside the cell, the electrons enter microbial metabolism and serve as reducing power for fixing carbon dioxide. Because the microbial catalysts are alive, they replicate themselves, operate under mild near-ambient conditions, and tolerate feedstock variability in a way that expensive metal catalysts cannot. And because microbial metabolism is enormously diverse, MES can in principle reach C3 and longer-chain molecules that remain stubbornly out of reach for conventional electrochemistry.

The contrast with existing CO2 conversion technologies is stark. Thermocatalytic routes such as the Sabatier reaction and reverse water-gas shift chemistry require high temperatures, high-purity hydrogen as a reductant, and durable metal catalysts that suffer from carbon deposition and poisoning. Electrochemical CO2 reduction on copper and other catalysts can make carbon monoxide, formate, and C2 products under ambient conditions, but selectivity for C3 and more complex molecules remains poor, catalysts deactivate, and long-term stability is inadequate. Meanwhile, green hydrogen produced by water electrolysis currently costs between 4.5 and 6.0 US dollars per kilogram, meaning that simply reducing CO2 with hydrogen often yields chemicals worth less than the hydrogen consumed. Microbial catalysts sidestep many of these constraints, using self-assembled enzymatic pathways such as the Wood-Ljungdahl route to weave carbon dioxide into multi-carbon products with remarkable specificity.

The performance numbers reported in recent studies are striking. Acetate, the workhorse product of MES, is routinely produced with coulombic efficiencies exceeding 90 percent, meaning that more than nine out of every ten electrons supplied by the circuit end up stored in the target molecule. Most impressively, a continuous thermophilic hydrogen-mediated system using the acetogenic bacterium Thermoanaerobacter kivui has achieved acetate concentrations of up to 29.4 grams per liter, roughly 490 millimolar, from carbon dioxide. That is an order of magnitude beyond typical laboratory titers and begins to approach concentrations relevant to industrial separation. For context, commercial acetic acid is produced today by methanol carbonylation in plants rated at 200,000 to 650,000 tonnes per year, so MES still has far to travel in scale, current density, and process intensification, but the trajectory of improvement is unmistakable.

Beyond acetate, the product spectrum widens considerably. When acetate and ethanol accumulate in the reactor, chain-elongating microbes take over, running reverse beta-oxidation pathways that stitch short-chain intermediates into C4 through C8 medium-chain fatty acids such as butyrate and caproate, which command far higher market prices as feed and chemical precursors. Solventogenic metabolism can be triggered by tuning reactor operating conditions, reducing accumulated organic acids to ethanol, butanol, and 2,3-butanediol. Reductive branches of the tricarboxylic acid cycle yield lactate and succinate. Photo-bioelectrochemical systems using the purple bacterium Rhodobacter sphaeroides go further still, simultaneously converting CO2 into biomass and hydrogen gas, while MES-integrated setups direct CO2-derived carbon into intracellular polyhydroxybutyrate, a biodegradable plastic. In each case, the electron source is the electrode rather than sugar, decoupling production from agricultural feedstocks.

One of the most commercially mature applications is biogas upgrading. Anaerobic digestion plants in Germany, Denmark, and the Netherlands already supply 10 to 20 percent of renewable power in parts of the European Union, but raw biogas contains only 50 to 70 percent methane, with the remainder mostly CO2 plus troublesome impurities such as siloxanes and sulfur compounds. Conventional pressure swing adsorption can polish biogas to roughly 97 percent methane for pipeline injection, but contaminants foul the adsorbents and raise costs. MES offers an elegant pre-treatment: raw biogas is sparged into the cathode chamber, where methanogenic archaea electrochemically reduce the CO2 fraction to additional methane. Recent work with biogas from an operating anaerobic digestion plant achieved 95 percent methane in the upgraded gas at a methane production rate of 8.8 liters of CH4 per square meter of catalyst per day. Because gaseous impurities dissolve into the liquid phase during this step, the downstream PSA unit faces a lighter, cleaner load, and the captured CO2 is not merely discarded but converted into fuel.

The review also highlights an emerging application with obvious public appeal: power-to-protein. In these schemes, renewable electricity splits water to generate hydrogen, formate, or methanol, which feed bioreactors cultivating protein-rich microorganisms for food and feed. Single-cell protein is not new; companies such as Unibio and Calysta have commercialized fermentation-based production, but their processes traditionally rely on sugar substrates that compete with food supply. A techno-economic assessment of solar-driven microbial protein production found that photovoltaic-powered systems could achieve protein yields per unit of land up to an order of magnitude higher than conventional agriculture, and that estimate assumed conservative solar-to-electricity and power-to-chemical conversion efficiencies of 5 percent or less. Because electricity delivers reducing power to microbes far more efficiently than photosynthesis delivers it to crops, the land-use arithmetic of protein production could be transformed, freeing farmland while feeding a growing population.

None of this means MES is ready for prime time, and the authors are candid about the obstacles. Most laboratory studies still rely on small H-type reactors whose distant electrodes and ion-exchange membranes impose severe ohmic resistance and overpotentials, often pushing cell voltages above 3 volts and crushing energy efficiency. The oxygen evolution reaction at the anode is kinetically sluggish, particularly on carbon-based electrodes, and acts as a bottleneck for the entire system. Mixed microbial consortia, while robust, tend to foul membranes and default to acetate rather than more valuable products, whereas pure cultures of Shewanella, Sporomusa, Geobacter, or Rhodobacter offer precision but demand sterility. The field is responding with nanostructured and conductive-polymer-coated cathodes such as polyaniline-deposited graphite felt, synthetic-biology strains with engineered electron-transfer and metabolic pathways, artificial redox mediators like neutral red, and scalable reactor geometries including bubble columns and 3D-printed electrodes designed to improve hydrogen delivery.

The most pragmatic near-term strategy may be integration rather than replacement. Because MES alone is unlikely to reach commercial viability at current productivities, the authors describe two-stage processes in which MES first converts CO2 to acetate, which is then recovered and fed to a second, optimized bioprocess that converts it into long-chain alkyl esters or high-value isoprenoids. Such hybrid configurations sidestep the selectivity limits of mixed-culture MES while still exploiting its unique ability to fix carbon with electricity. Coupled with direct air capture and low-carbon power, an integrated MES platform could even operate as a carbon-negative factory, drawing down atmospheric CO2 while selling chemicals, fuel, and protein. The remaining challenges, from current density to product recovery costs, are formidable but increasingly quantified, and for the first time the road from laboratory biofilm to industrial biorefinery looks less like a leap of faith and more like an engineering schedule.

Subject of Research: Microbial electrosynthesis using renewable electricity to convert CO2 into value-added chemicals, biogas, and protein

Article Title: Renewable electricity–driven microbial electrosynthesis for high-value CO2 valorization: recent trends

Article References: Lee, C. H., Kim, M., Kong, D. S., Son, H., & Kim, J. R. (2026). Renewable electricity–driven microbial electrosynthesis for high-value CO2 valorization: recent trends. Advances in Industrial and Engineering Chemistry, 2(1), Article 4. https://doi.org/10.1007/s44405-026-00044-1

Image Credits: AI Generated

DOI: 10.1007/s44405-026-00044-1

Keywords: microbial electrosynthesis, CO2 valorization, renewable electricity, acetate production, biogas upgrading, single-cell protein, polyhydroxybutyrate, bioelectrochemical systems, Wood-Ljungdahl pathway, power-to-protein, carbon capture and utilization, electroactive bacteria