Battery-Based Carbon Capture Halves Energy Use in Delaware Study

A battery that captures carbon instead of storing charge

Most direct air capture (DAC) systems work like a sponge that has to be cooked. A filter material soaks up carbon dioxide, and then heat, often steam, drives the gas back out so the material can be reused. A team at the University of Delaware is trying a different approach. Its process runs on electricity alone, at room temperature, inside a device that looks and works much like a rechargeable battery. The researchers report that it uses about half the energy of conventional systems.12

The work appeared in the peer-reviewed journal Nature Energy. It was carried out with RepAir Carbon, a startup that licenses the technology and has already built a pilot-scale version.12 Ars Technica first covered the study in late September, and the university announced it in late October. Both reports describe the same basic claim: removing CO₂ from air takes less energy, and could therefore cost less, than with the reversible-filter designs that lead the field today.1112

The reports agree on the science. Where they differ is in how they treat the cost projections, and that difference matters for how the result should be read.

How the cell works

Each side of the cell has a nickel hydroxide electrode, a chemistry close to that of nickel-metal hydride batteries.13 At the cathode, the cell produces hydroxide, which reacts with incoming CO₂ to form carbonate or bicarbonate. Those ions move through a separator membrane to the anode. The pH is lower there, so the reaction runs backward and releases CO₂ as a concentrated gas.1314 When one electrode is spent, the voltage is reversed and the electrodes swap jobs, so capture can continue while the first electrode recharges.1213

The key difference is that the voltage, not heat, releases the captured carbon. Thermal regeneration is a large part of why DAC costs so much. Electrochemical designs avoid it.1314 The university says the process needs less than one megawatt-hour per ton of CO₂ removed.12 Industry analyses put total DAC energy demand at 1.5 to 2.5 MWh per tonne. One of those analyses estimates 1.5 to 2.0 MWh per tonne for existing electro-swing approaches.3 If the Delaware figure holds up outside the lab, it would be a real gain over earlier electrochemical designs as well as over thermal ones.

According to the university, RepAir ran its pilot device in Israel for more than 5,000 hours, over six months, which points to reasonable durability.12 That matters because sorbent degradation is a known operating cost for current systems. One analysis estimates that amine panel degradation alone accounts for 8 to 12 percent of operating costs in solid-sorbent plants.9

The $92 question

The economic case comes from the paper itself. Several authors work with RepAir, and part of the study sets out what the technology could cost at scale.11 The team puts the cost of an initial small pilot plant at about $566 per ton of CO₂. It then applies the cost-learning rate seen in the lithium-ion battery industry, plus standard scaling assumptions, over several generations of plants. On that basis, it projects that a plant 1,000 times the pilot's capacity could reach about $92 per ton.11

That figure falls just under $100 per ton, the long-standing industry target at which carbon capture could see much wider use.1112 The study's lead author, James Buchen, said the technology puts those numbers "within reach."12

This is where the coverage splits. Ars Technica presented the $92 figure as a projection and placed it next to Climeworks, which is aiming for $250 to $350 per ton by 2030 and has shown how hard it is to meet optimistic scaling forecasts.11 The university release takes a more confident tone.12 Wider industry analysis is more doubtful still. One recent cost study calls $100 per tonne highly aspirational. It argues the figure is very unlikely before 2040 without a basic physics breakthrough, and treats $200 to $250 per tonne as a credible early-2030s target.3 Notably, that analysis names electro-swing designs with near-zero heat demand as the kind of advance that could change the picture.3 The Delaware cell belongs in that category.

The reasonable reading is that this is a meaningful energy result paired with an optimistic cost model. The energy numbers rest on measured data from a pilot that ran for months.12 The $92 figure assumes this chemistry will get cheaper at the same rate lithium-ion cells did. That rate came from huge consumer demand for phones, laptops and electric vehicles, and carbon removal has no comparable market.

Measured against a struggling industry

The gap between DAC forecasts and actual results is wide, and that is why the energy result matters. Climeworks has raised more than $1 billion.1 Its flagship Mammoth plant removed only a small fraction of its rated capacity in its first year.29 The company has since reported progress. Net removal in the first half of 2026 was more than five times the level of a year earlier, and operating costs per tonne fell by more than half. It did not publish a cost per tonne.6 Occidental's Stratos project saw its capital costs rise to about $1.2 billion, putting its theoretical removal cost near $600 per ton.9

Contract prices tell a similar story. Implied prices in recent Frontier offtake deals range from about $440 per tonne for CarbonCapture to roughly $989 for Heirloom. BloombergNEF has estimated current industry capture costs at around $900 per tonne.6 The International Energy Agency puts first-of-a-kind DAC costs at $500 to $1,900 per tonne. It expects that better materials and scale could bring them to about $300 by mid-century.15 Against those figures, even a two-generation path to the low hundreds of dollars would be notable.

RepAir is a small company in this field. It has a disclosed $15 million Series A, far less than the larger players.6 Total funding for carbon dioxide removal startups has passed $3.6 billion since 2021. Most deals remain at seed or Series A stage, and the sector still depends heavily on the US 45Q tax credit of $180 per ton for DAC.1 Any company claiming to approach $100 per ton would need far less of that subsidy, which is commercially valuable on its own.

A crowded field of low-energy ideas

Delaware is not the only group working on this problem. Researchers at Empa in Switzerland are working on ultra-thin amine coatings that capture less CO₂ per cycle but cycle in under a second. Because the coatings have little mass to heat, they need less energy per molecule captured.16 Empa says it does not yet have the equipment needed to measure these sub-second processes properly.16 A study in Science reported a membrane-separated electrode design that aims to increase throughput while cutting the electricity needed for point-source electrochemical capture. That work has no named scale-up partner and is still at the research stage.20 Startup Parallel Carbon pairs DAC with water electrolysis and says it could capture CO₂ for under $100 per metric ton by 2030.18 In Australia, CSIRO is building a center designed to move DAC materials from the lab to pilot scale faster.15

For heavy industry, the near-term picture looks different. Capture can be cheap from concentrated emission streams such as ethanol production, at as little as $15 to $35 per ton.7 The capture step accounts for nearly three-quarters of total carbon capture, use and storage costs, so shared pipeline clusters such as Norway's Northern Lights cannot fix the economics alone.17 Buchen himself said DAC will not solve emissions from large sources like furnaces and power plants. He described it as better suited to scattered sources such as agriculture and construction.12

The battery and grid connection

The battery comparison goes beyond the name. The device uses nickel hydroxide electrodes from established battery chemistry.13 Its cost model borrows the lithium-ion learning curve.11 It runs entirely on electricity, which fits a grid adding more renewable power. That is a different design choice from plants that need a steady supply of high-temperature heat. Other electric-only startups are already using this. Phlair, for example, is designing its system to follow intermittent solar output.6 One account of the Delaware work describes the combination of electrochemical circuitry and gas separation as a possible dual-purpose framework. It adds that membrane durability and gas collection rates still need work at scale.14

This may be the most useful point for climate-tech investors. Over the past decade, battery manufacturing has built up supply chains, membrane know-how and cell-stacking methods. If carbon capture becomes mainly an electrochemistry problem, it can draw on all of that. Whether that produces the steep cost declines seen in batteries is the open question.

The verdict

The Delaware–RepAir result is one of the stronger DAC claims this year. The energy reduction is backed by a pilot that ran for months, published in a top journal, by a team that has already licensed the technology.12 The $92-per-ton figure should be treated as a modeled target, not a demonstrated cost. It depends on cost-learning assumptions borrowed from an industry with very different demand.11 The next milestone is a bigger plant with a published cost per tonne. Climeworks' experience shows how often estimates in this sector change once plants are built and running.69