Can a Joule-heated CO2 capture filter cut ventilation energy use?

A filter that captures CO2 and heats itself

The paper's proposal is a modular filter that removes indoor CO2 instead of relying only on fresh-air ventilation to dilute it. The filter is built by spray-coating an amine-functionalized solid sorbent onto a carbon fiber substrate. The carbon fiber serves two roles: it holds the sorbent, and it conducts electricity so the sorbent can be heated from within.

That internal heating is the key design choice. The sorbent is regenerated by Joule heating, meaning current passed through the conductive carbon fiber raises its temperature directly. The paper reports that 95% desorption was reached within 15 minutes at 100 °C. The authors contrast this with convective heating by externally supplied hot air, which they say can cause substantial thermal losses; the claim that Joule heating minimizes those losses is the authors' interpretation of the design, not a measured comparison in the evidence provided.

The filter is described as cell-based, with the number of cells and fiber packing density adjustable to match ventilation demand. In operation, the paper describes two modes: an internal circulation mode in which the filter removes indoor CO2, and a ventilation mode in which the filter is regenerated while desorbed CO2 is discharged outdoors and fresh air is brought in.

Why 100 °C beats cooler regeneration

A counterintuitive result anchors the material section: heating to a higher temperature used less total energy. The authors measured desorption at 60 °C, 80 °C, and 100 °C and used the time to reach 95% desorption as the comparison metric. At 100 °C that took 15 minutes; at 80 °C it took about 35 minutes; at 60 °C desorption was still incomplete after 120 minutes.

Combining the fitted desorption kinetics with the measured temperature–power density relationship, the authors estimated the energy density needed for 95% desorption as 39.6, 60.7, and 566.8 Wh m⁻² at 100 °C, 80 °C, and 60 °C respectively. On that basis, operating at 60 °C would require about 14.3 times more energy than at 100 °C. The higher instantaneous power at 100 °C is outweighed by the much shorter operating time.

The temperature choice also has a durability boundary. Thermal analysis showed no significant weight loss below about 150 °C, which the authors read as negligible amine degradation in that range. In a 200-cycle test at 100 °C regeneration, mass fell 2.36% overall while CO2 capture capacity per unit mass rose slightly, from 1.074 to 1.098 mmol g⁻¹. Separately, carbon fiber held stable thermal profiles across 100 Joule-heating cycles at 4–5 V with under 3% variation, but at 6–7 V, where peak temperatures exceeded 160 °C, peak temperature dropped roughly 5–15% over 100 cycles.

The coated filter captures less than the powder

The sorbent itself, a 50 wt% amine-impregnated silica nanopowder, reached a maximum CO2 adsorption capacity of 1.84 mmol g⁻¹ over 200 minutes at 1000 ppm CO2. The corresponding sorbent-coated carbon fiber filter reached 1.26 mmol g⁻¹, a 31.5% decrease. The authors attribute most of this gap to competitive adsorption between CO2 and water on amine sites under humid conditions, noting that capacity at 0% relative humidity was comparable to that at 40% relative humidity, which they take to mean the coating process itself causes negligible loss.

Humidity is therefore a real performance variable, not a footnote. In co-adsorption measurements, CO2 uptake rose modestly at 20% relative humidity relative to dry conditions, then declined at 40% and 60%. The authors explain the initial rise as a shift toward bicarbonate and carbonate species that improves amine utilization, and the later decline as water blocking pores and competing for sites. They conclude that above roughly 40% relative humidity, water competition may substantially reduce capture performance.

The lab-scale chamber that produced these numbers was a closed recirculating system, not an open ventilation system. The authors state explicitly that absolute capture quantities and the measured temperature and humidity variations should be read as relative performance indicators rather than direct predictions of in-situ behavior, because in a real system desorbed CO2 and moisture leave the building instead of accumulating in the test volume.

Optimization and a neural-network surrogate

The system-level work is simulation-based. The authors used Differential Evolution, an optimization algorithm, to find control parameters that minimize ventilation load while keeping indoor CO2 within limits. Because running that optimization for every indoor condition would be slow, they trained a deep neural network as a surrogate model to approximate the optimized control parameters from occupancy feedback.

The simulation compared three strategies: conventional demand-controlled ventilation (DCV), a CO2-filter-integrated version using average occupancy (CDCV), and a deep-learning-assisted version (DL-CDCV). The paper reports that the deep-learning-assisted system could achieve up to 28.3% energy savings relative to the conventional system. Weather inputs were based on typical July–August and December–January data for South Korea, and energy calculations for the filter system assumed a 200 g sorbent-coated filter with regeneration energy estimated from desorption duration and the measured temperature–power density relationship.

These are model results, not building measurements. The authors state that the system-scale deep learning model was trained on simulated data without validation against real occupancy measurements, and that the system-level analysis used simplified boundary conditions. Occupancy scenarios were limited to 1–4 persons based on a typical residential setting.

Winter savings, summer penalty at loose targets

The most decision-relevant result is a split by season and CO2 target. In winter, at a 600 ppm target, the CDCV system achieved approximately 25.0% net energy savings, which the authors attribute to the large enthalpy difference between indoor and outdoor air. As the CO2 setpoint rises, the relative benefit shrinks, because ventilation demand falls while regeneration energy stays roughly constant.

In summer at 1000 ppm, the CDCV system consumed about 30.6% more total energy than the DCV system. The authors explain the asymmetry through the smaller summer indoor–outdoor enthalpy difference: there is less ventilation energy to save, while the regeneration load remains. They state that this trade-off held consistently across all CO2 setpoints and both seasons examined, and that the qualitative outcome is governed by the balance between ventilation and regeneration loads rather than by any single assumed condition.

The abstract's 14.9–25% range and the 600 ppm condition therefore describe the favorable regime, not a universal result. The same system that saves energy under a strict CO2 target and a large enthalpy difference can be a net energy cost under a loose target in summer.

What energy parity at 1000 ppm would require

The authors ran a feasibility analysis to identify what would have to change for the filter system to break even at high CO2 setpoints. They report two pathways to energy parity: increasing sorbent adsorption capacity by 1.38 times, or reducing regeneration energy by 43.5%. They frame this as requiring simultaneous advances in sorbent material design and electric-to-thermal conversion efficiency.

A techno-economic assessment compared total daily cost against the conventional DCV system across summer and winter and three filter replacement intervals of 7, 15, and 30 days. The CDCV cost was decomposed into capital expenditure for installation, ventilation energy cost, Joule-heating regeneration electricity cost, and periodic consumable replacement cost, while the DCV system incurred only ventilation-related energy costs. The paper reports break-even points of 18, 17, and 17 days in summer and 56, 46, and 43 days in winter across those replacement intervals.

The pressure-drop question was handled by simulation rather than measurement. Because direct differential pressure measurement across the filter was not feasible in the lab apparatus, the authors used CFD and obtained simulated differential pressures from 0.393 Pa (3 cells, 1 m s⁻¹) to 14.144 Pa (9 cells, 3 m s⁻¹), which they say represents less than 3.4% variation in fan power demand and was therefore excluded from energy calculations.

Where the energy answer holds and stops

The paper's own framing is that it assesses building-scale feasibility through simplified simulations grounded in lab-scale experiments, rather than reproducing actual indoor environments. That places the 14.9–25% savings and the 30.6% summer penalty in the category of model-based projections under stated weather, occupancy, and filter assumptions, not measured building performance.

The practical reading is a conditional one. The filter-integrated system is advantageous when ventilation loads are large, meaning low CO2 setpoints and large indoor–outdoor enthalpy differences, notably winter. It incurs a penalty in the opposite regime, where regeneration load dominates. The authors identify the priority for closing that gap as raising sorbent capacity by 1.38 times or cutting regeneration energy by 43.5%, and they note that future building-scale experiments under real occupancy, plus safety validation of regeneration in occupied environments, would be needed before real-world adoption.

For a reader deciding whether this approach is worth pursuing, the paper's most useful contribution is the threshold structure rather than a single savings number: the benefit depends on the CO2 target and the season, and the material-level lever that would extend the favorable regime is quantified.

Sources used in this answer