Hydroxide-based scrubbers are among the most promising for direct air capture (DAC) of carbon dioxide4,5. Industrially mature approaches use aqueous KOH solutions6 or solid calcium hydroxide7 as the sorbent, but high-energy regeneration steps at 900 °C and the use of natural gas are often required6. These high regeneration temperatures arise from the significant lattice energy of the formed carbonate materials and contribute greatly to the costs of running a DAC process. An alternative approach that can significantly reduce regeneration temperatures is to disperse hydroxides in a porous material or polymer matrix8,9. For example, hydroxide-functionalized metal-organic frameworks have achieved promising DAC performance with much lower regeneration temperatures (roughly 100 °C), but these materials suffer from limited stabilities and high sorbent costs10,11,12. Motivated by these challenges, we sought a low-cost and robust hydroxide material that could combine DAC with low-temperature regeneration. We proposed that (1) new DAC sorbents could be synthesized by electrochemically inserting reactive hydroxide ions into a porous carbon electrode13,14,15,16, and that (2) the resulting electrically conductive sorbents could be heated using rapid Joule heating (also known as resistive heating) without a secondary conductive support2,3.

Preparation and characterization of charged-sorbents

The preparation of charged-sorbents is based on the charging of an electrochemical energy storage device (Fig. 1). During charging, electrolyte ions accumulate in the pores of a conductive porous carbon electrode; for example, anions accumulate in the electrode when charging positively (Fig. 1, step 1). After completing the charging process, this electrode is removed from the cell and is washed and dried to remove residual electrolyte and solvent to yield a charged-sorbent material. Our hypothesis is that the accumulated ions in the porous electrode can then serve as active sites for an adsorption process such as CO2 capture. Different electrode materials, electrolyte ions and solvents can be selected, allowing for the preparation of tailored sorbents for different applications.

Fig. 1: The preparation of charged-sorbents.
figure 1

A porous carbon electrode is charged in an electrochemical cell (step 1). The electrodes are removed from the cell, washed with deionized water and evacuated to remove solvent molecules (shown as green circles) to yield charged-sorbents (step 2). The activated carbon schematic is adapted from ref. 33, Springer Nature America.

Using this approach, we targeted a hydroxide-functionalized porous carbon as a new sorbent for DAC. This was achieved by using inexpensive activated carbon cloth as the electrode (Extended Data Fig. 1) and a 6 M KOH(aq.) solution (Fig. 2a). The activated carbon cloth was positively charged by applying a potential of +0.565 V versus standard hydrogen electrode (SHE) for 4 h, thereby accumulating reactive hydroxide ions within the carbon pores. Both capacitive and faradaic currents were observed during charging, suggesting that hydroxides are accumulated through both electric double-layer formation and oxidation of surface functional groups (Extended Data Fig. 2a,b and Extended Data Table 1). Following charging, the electrode was removed from the cell, and was washed to remove excess KOH and minimize salt formation on the surface of the carbon cloth (Extended Data Fig. 1). Finally, the material was dried to yield a positively charged-sorbent (PCS) bearing hydroxide ions, referred to as PCS-OH (Fig. 2a). Powder X-ray diffraction analysis of PCS-OH showed no reflections from crystalline KOH or related products, suggesting that the hydroxide ions are incorporated primarily within the nanometre-sized pores of PCS-OH (Extended Data Fig. 2c). Finally, the counterpart negatively charged-sorbent was prepared as a control sample by applying a potential of −0.235 V versus SHE for 4 h, to yield a negatively charged-sorbent replete with potassium ions (NCS-K).

Fig. 2: Preparation of hydroxide charged-sorbents and CO2 sorption.
figure 2

a, Scheme of charging activated carbon fabric ACC-5092-10 cloth in 6 M KOH through a three-electrode configuration. Scale bar, 0.5 cm. b, 1H solid-state NMR (9.4 T) spectra of PCS-OH and control samples, acquired at a magic angle spinning (MAS) rate of 12.5 kHz. c, CO2 adsorption (filled data points) and desorption (hollow data points) isotherms of PCS-OH and control samples at 25 °C. d, Low-pressure region of the CO2 adsorption isotherms from c. e, CO2 uptake of PCS-OH and control samples at 0.4 mbar and 25 °C. Standard deviations are calculated from ten independent samples for PCS-OH, and three independent samples otherwise. f, Adsorption microcalorimetry measurements of the differential molar adsorption heats curves related to the adsorption of CO2 at 30 °C on PCS-OH (grey) and blank cloth (green). The dotted horizontal line represents the standard molar enthalpy of liquefaction of CO2 at 30 °C of −17 kJ mol−1. The inset shows volumetric isotherms obtained by performing CO2 adsorption at 30 °C on PCS-OH (grey) and blank cloth (green) with the volumetric line coupled to the microcalorimeter. g, Dry, pure CO2 uptake curves at 40 °C and 1 bar CO2 for PCS-OH after activation under flowing dry N2 at 130 °C for 1 h (grey curve), after exposure to flowing dry air (roughly 21% O2 in N2) at 100 °C for 12 h (red curve), and after exposure to flowing dry air (roughly 21% O2 in N2) at 150 °C for 12 h (blue curve). h, Cycling capacities for 150 adsorption–desorption cycles for the PCS-OH in a simulated temperature–pressure swing adsorption process (Extended Data Fig. 6b). Adsorption 30 °C, 20 min; Desorption 100 °C, 20 min; dry 30% CO2 in N2 was used for both the adsorption and desorption steps.

Evidence for the incorporation of hydroxide ions in the pores of PCS-OH was obtained from 1H solid-state nuclear magnetic resonance (NMR) measurements (Fig. 2b). For PCS-OH, a strong resonance was observed at roughly 1.5 ppm, corresponding to OH species present in the pores. This assignment was supported by control measurements on the as-purchased activated carbon cloth (‘blank cloth’), a carbon cloth that was soaked in the electrolyte but not charged (‘soaked cloth’) and the negatively charged carbon cloth sample NCS-K, which all showed much weaker 1H resonances assigned to small amounts of residual H2O or OH species. The more positive 1H chemical shift in PCS-OH probably arises from differences in the ring-current effects from the charged aromatic carbon surfaces17,18. Further support for the accumulation of OH species in PCS-OH was provided by titration experiments, in which 1.2 mmol g−1 of HCl was required to neutralize PCS-OH, compared to 0.2 mmol g−1 for NCS-K (Extended Data Fig. 2d). Combustion analysis also supports an increased amount of hydrogen in PCS-OH compared to the controls (Extended Data Fig. 2e), consistent with the charge-driven accumulation of OH species in the electrode. Finally, the Brunauer–Emmett–Teller (BET) surface area of PCS-OH was 920 m g−1, 20% lower than the blank cloth (1,175 m2 g−1) (Extended Data Fig. 3a,b). We propose that the 20% reduction in surface area for PCS-OH compared to the blank cloth arises from two factors: (1) the accumulation of hydroxide ions in the pores reducing the accessible pore volume, and (2) the electrochemical oxidation of functional groups in the carbon (Extended Data Fig. 2a,b, Extended Data Table 1 and the section on ‘Preparation of charged-sorbents’) disrupting the carbon’s overall porosity. Despite the small decrease in the material surface area, PCS-OH remains highly porous (Extended Data Fig. 3c), suggesting that the accumulated hydroxide ions should be accessible as reactive sites for CO2 adsorption.

CO2 adsorption isotherm measurements (Fig. 2c,d,e) showed enhanced CO2 capture at low pressures for PCS-OH relative to the control samples. This enhanced low-pressure uptake is typical of hydroxide-functionalized sorbents10,11, supporting CO2 chemisorption by the accumulated hydroxide ions. PCS-OH showed increased CO2 uptake at pressures relevant to DAC. At 0.4 mbar and 25 °C, the CO2 capacity for PCS-OH is 0.26 ± 0.06 mmol g−1 (ten independent samples, Extended Data Fig. 4), which is significantly larger than the capacities of the control samples under these conditions (Fig. 2d,e). Experiments on samples synthesized using less-positive charging potentials and shorter charging times gave smaller CO2 uptake at low pressures due to the incorporation of fewer hydroxide ions (Extended Data Figs. 2d and 5a,b). PCS-OH also showed enhanced low-pressure CO2 uptake compared to a ‘dripped cloth’ sample, which was preparing by adapting an approach from the literature for impregnating activated carbon with metal hydroxide solutions (Fig. 2e, Extended Data Fig. 5c,d and Methods)13. Finally, our charged-sorbents are highly tuneable materials because the electrode (and electrolyte) can readily be varied in the synthesis. As a second example, powdered YP80F activated carbon was first prepared into a free-standing electrode film before synthesizing a charged-sorbent referred to as PCS-OH (YP80F) (Extended Data Fig. 5e and Methods). Enhanced low-pressure CO2 uptake was again observed, demonstrating the generality of the charged-sorbent approach (Extended Data Fig. 5f,g).

To investigate the nature of CO2 sorption in PCS-OH we performed microcalorimetry tests that allowed the direct quantification of the heat released during CO2 uptake measurements (Fig. 2f). For the blank carbon cloth control, the measured CO2 adsorption heat is between −28 and −20 kJ mol1, consistent with CO2 physisorption19. A large increase in the adsorption heat is observed for PCS-OH relative to the blank carbon, consistent with CO2 chemisorption. The measured adsorption heats gradually decrease from a value of −137 kJ mol1 at zero coverage (extrapolated value) to −33 kJ mol1 at a coverage of 0.8 mmol g−1 indicative of bicarbonate formation at a distribution of hydroxide sites. The adsorption heats compare well with previous reports for bicarbonate formation in porous materials and metal oxides10,11,20, lending support that the electrochemically inserted hydroxides in PCS-OH serve as CO2 chemisorption sites to enhance low-pressure uptake.

Further to the promising gas sorption and calorimetry results, thermogravimetric analysis (TGA) measurements (using a heated furnace) indicate that PCS-OH has good thermal and oxidative stability. The sorbent was heated to 150 °C under flowing dry air at atmospheric pressure for 12 h. Dry, pure CO2 adsorption isobars before and after air exposure showed very similar capacities and uptake kinetics, confirming the oxidative stability of the material under these conditions (Fig. 2g). This promising oxidative stability is in contrast to the behaviour of amine-based sorbents, which typically show poor oxidative stability, a recurring challenge for their application for DAC21,22. We do, however, note that 10% of its capacity was lost after sample storage for 14 months (Extended Data Fig. 6a). As a further stability assay before DAC tests, PCS-OH was subjected to 150 adsorption and desorption cycles using TGA under concentrated CO2 conditions (30% CO2 in N2) (Fig. 2h and Extended Data Fig. 6b). Consistent with the thermal stability test, PCS-OH showed a stable cycling capacity (assuming negligible N2 uptake), with minimal performance loss after 150 cycles. Low regeneration temperatures of roughly 100 °C can be used (Extended Data Fig. 6c), supporting the idea that dispersal of the reactive hydroxides in the carbon pore network prevents the formation of stable bulk carbonates. Overall, these data support that PCS-OH shows remarkable stability and promising affinity towards CO2 at pressures relevant to DAC.

CO2 capture mechanism

To further investigate the mechanistic pathway responsible for strong CO2 binding in PCS-OH, we collected 13C solid-state NMR spectra for PCS-OH and the control samples after dosing with 13CO2 gas at 0.9 bar (Fig. 3a). All samples showed strong resonances at 119 ppm and weaker resonances at 125 ppm, which are assigned to physisorbed CO2 in the carbon nanopores and free 13CO2 gas, respectively23. The chemical shift difference between these species is consistent with a ring-current shielding for the in-pore physisorbed species, with the shift difference of −6 ppm similar to a recent NMR study on other carbon cloths24,25. In contrast to the controls, the spectrum of PCS-OH showed an extra resonance at 156.0 ppm, assigned to chemisorbed (bi)carbonate species (Fig. 3b)26,27,28,29. Note that carbonate and bicarbonate species are often in fast exchange on the NMR timescale, so we cannot readily discriminate between these species30. Regardless, observation of this chemisorption resonance provides strong evidence that the hydroxide sites incorporated through our electrochemical synthesis chemically react with CO2, and corroborate the findings from microcalorimetry (Fig. 2f). The chemical shift of 156.0 ppm is similar to bulk potassium bicarbonate (161.4 ppm, Extended Data Fig. 7b), with the smaller value in PCS-OH partly arising from ring-current effects and supporting that the (bi)carbonate species reside in the carbon nanopores (Fig. 3b). Quantitative measurements showed that the amount of physisorbed CO2 in the four sorbents was similar (around 1 mmol g1), whereas only PCS-OH chemisorbed CO2 (0.95 mmol g1 at 0.94 bar) (Fig. 3c). The observation of 0.95 mmol g1 of chemisorption aligns with the microcalorimetry measurements, which show a decrease in adsorption heat below −30 kJ mol1 once this CO2 loading is reached.

Fig. 3: CO2 binding mechanism from solid-state NMR experiments.
figure 3

a, Quantitative 13C solid-state NMR spectra of sorbents with 13CO2 gas dosing at the pressure of 0.9 bar, acquired at a MAS rate of 12.5 kHz (see Extended Data Fig. 7a for a control experiment with no CO2 dosing). b, Proposed mechanism for CO2 capture by positively (hydroxide) charged-sorbent. Schematic adapted from ref. 33, Springer Nature America. c, CO2 uptake of sorbents by means of chemical and physical adsorption calculated from resonances at 156 and 119 ppm of a, respectively.

The NMR results further allow us to estimate a lower limit for the hydroxide content in PCS-OH of 0.95 mmol g1 (assuming a 1:1 reactivity of CO2 and hydroxide), which is comparable to the value of 1.2 mmol g1 from titration (Extended Data Fig. 2d). This further leads to an estimated molecular formula for PCS-OH of (OH)C86. Similar NMR spectroscopy experiments on PCS-OH prepared at a lower charging potential (0.365 V versus SHE, instead of our optimized value of 0.565 V versus SHE) provided a much lower limiting hydroxide content of 0.38 mmol g1 and an estimated molecular formula of (OH)C218 (Extended Data Fig. 7c). The lower hydroxide content of samples prepared in this way led to lower CO2 uptakes at low pressures (Extended Data Fig. 5a), supporting a positive correlation between hydroxide content and low-pressure CO2 uptake. Overall, the NMR spectroscopy experiments strongly support that reactive hydroxide ions can be installed in porous carbon through our electrochemical synthesis to achieve greatly enhanced reversible CO2 sorption.

Demonstration of DAC and regeneration by Joule heating

The promising low-pressure CO2 uptake of PCS-OH through chemisorption motivated DAC tests. DAC performance was initially evaluated under simulated dry air with 400 ppm CO2 at 30 °C, with desorption conducted under 100% N2 at 130 °C. These measurements showed a CO2 capacity of roughly 0.2 mmol g1, similar to the isotherm measurements, which was stable over repeated adsorption and desorption cycles (Fig. 4a). PCS-OH can still do DAC (Extended Data Fig. 8a) at comparable capacity after 14 months, despite the small capacity lost seen under pure CO2 conditions. The DAC kinetics for PCS-OH are also promising, with a comparable CO2 capture rate to several benchmark sorbents, albeit with a lower saturation capacity (Extended Data Fig. 8b,c).

Fig. 4: DAC tests and Joule heating regeneration.
figure 4

a, Cycling capacities of DAC for ten adsorption–desorption cycles for PCS-OH from TGA measurements. Adsorption 30 °C, 60 min, 400 ppm CO2 in dry air; Desorption 130 °C, 60 min, 100% N2. The cycled capacity (difference) is shown. b, DAC measurements in a sealed box filled with air at 37% RH. The mass of PCS-OH is 120 mg. c, Schematic of Joule heating for CO2 release (scale bar, 0.5 cm, sample dimensions 2 × 1 cm). d, 13C solid-state NMR spectra of sorbents with 13CO2 gas dosing at 0.9 bar: after exposure to the air for 20 min without Joule heating (control sample, red curve) and after 20 min of Joule heating around 90 °C in air (blue curve). MAS rate of 12.5 kHz. e, Proposed mechanism for CO2 release from positively (hydroxide) charged-sorbent by means of Joule heating. Schematic adapted from ref. 33, Springer Nature America. f, DAC experiments for PCS-OH, with 20 min of Joule heating regeneration under nitrogen between cycles. PCS-OH was added into the box at a time of 25 min (dashed line) for each cycle. The mass of PCS-OH in f is 33 mg.

As a more realistic DAC test, freshly activated PCS-OH and three control samples were subjected to ambient air (37% relative humidity, RH) in a sealed container equipped with a CO2 sensor. A large decrease in CO2 concentration was observed for PCS-OH, with a decrease from 500 ppm to around 25 ppm (a 475 ppm decrease) in 25 min. By contrast, the CO2 concentration decreased much less for the control samples (Fig. 4b). These measurements strongly support that our electrochemical synthesis enhances carbon capture at low partial pressures and enables DAC.

The electrically conductive nature of our PCS-OH sorbent opens the door to regeneration by direct Joule heating26, a strategy that leads to rapid regeneration times compared to traditional heating methods2. In contrast to previous approaches that required the use of electrically conductive supports2, here we directly attach electrodes to our conductive sorbent for Joule heating (Fig. 4c). A d.c. voltage of 7–8 V was applied across a piece of PCS-OH (2 × 1 cm), resulting in the rapid heating of the material to roughly 90 °C within roughly 1 min. Solid-state NMR experiments on samples predosed with 13CO2 gas showed that Joule heating led to the complete release of adsorbed CO2 species in this time (Fig. 4d,e).

We then carried out proof-of-concept DAC cycles with ambient air with varying RH (Fig. 4f) and air with 11% RH (Extended Data Fig. 8d), with regeneration by Joule heating in nitrogen. The regenerated PCS-OH was then reused in the next DAC cycle, with a total of four cycles completed. In these experiments, a smaller piece of PCS-OH was used compared to that in Fig. 4b, so that CO2 uptake was limited by the capacity of the sorbent, rather than the quantity of CO2 in the sealed container. The experiments show reversible CO2 capture both in ambient conditions, as well as conditions with controlled RH at 11% (Fig. 3f and Extended Data Fig. 8d). Our results show that our charged-sorbents can be regenerated by direct Joule heating without the need for a secondary conductive support material.

We finally quantified the impact of RH on the CO2 capture performance of PCS-OH, as water deteriorates the CO2 capacity of many OH-based materials10,27. Whereas PCS-OH could capture CO2 in humid air (Fig. 4b,f), we found that water co-adsorption is detrimental to the material’s CO2 uptake. Under DAC conditions, the CO2 capacity decreased from roughly 0.14 to 0.08 mmol g−1 as the RH increased from 11 to 38% (Extended Data Fig. 9a–f). Consistent with this, solid-state 13C solid-state NMR experiments at 0.9 bar CO2 and RHs of 53 and 85% showed decreased CO2 chemisorption (0.19 and 0.21 mmol g−1, respectively, compared to 0.95 mmol g−1 at 0% RH, Extended Data Fig. 9a–f). Joule heating regenerates the CO2 capacity of the sorbent under humid conditions (Fig. 4e,f). Therefore, the reduction in CO2 capacity is probably due to the filling of the hydrophilic pores with H2O (H2O uptake isotherm in Extended Data Fig. 9g), blocking access of CO2 to reactive OH sites10, rather than due to sorbent degradation.

Using a desorption temperature of 90 °C and the low measured heat capacity of PCS-OH of 0.62 J g−1 K−1 (Extended Data Fig. 9h), we estimate a minimum electrical energy consumption for sorbent heating of 6.5 and 11.4 GJ ton−1 CO2 captured for RH values of 11 and 38%, respectively (equivalent to 1,800 and 3,200 kWh ton−1 CO2). These values are comparable to those reported for a range of DAC processes31. For example, when using aqueous KOH solutions, the capture of 1 ton of CO2 required either 8.8 GJ natural gas or 5.3 GJ natural gas and 77 kWh electricity in limiting cases6. A key advantage of charged-sorbents is that full electrification of the DAC process is possible, thereby avoiding issues with natural gas use and leakage, which can offset a significant fraction of the captured CO2 in traditional DAC processes due to the high global warming potential of methane32. Although this may justify the higher operating energy costs for charged-sorbents, future work should be carried out to improve their energy efficiencies. The most straightforward way to achieve this is to increase the sorbent CO2 capacities, something that we are working towards in our laboratories. With further optimization and scaling, we anticipate that Joule heating regeneration will lead to a rapid temperature–pressure swing DAC process driven by renewable electricity.


Charged-sorbents are a new low-cost materials class with highly tuneable chemical structures. These materials are synthesized through a battery-like charging process that accumulates ions in the sorbent pores, with these species then serving as reactive sites in an adsorption process. Targeting DAC of carbon dioxide as a representative separation, we achieved the electrochemical insertion of hydroxide ions into activated carbon electrodes, with the resulting charged-sorbent material showing enhanced CO2 uptake at low pressures due to chemisorption to form (bi)carbonates. Our materials capture CO2 directly from ambient air and can be regenerated at low temperatures of 90–100 °C. The electrical conductivity of charged-sorbents enables material regeneration by direct Joule heating, with no need for separate heating equipment. The promising oxidative stabilities combined with rapid Joule heating regeneration should enable a DAC process that requires renewable electricity as the only input. Moreover, the low cost of activated carbons and the electrolytes used here is promising from an applications standpoint. Beyond DAC, we anticipate that the readily tuneable nature of our charged-sorbent materials, in which the electrolyte and electrode can easily be varied, will lead to a family of new materials with a wide range of applications.



Activated carbon fabric ACC-5092-10 cloth was purchased from Kynol. The cloth was activated for 1 h at 100 °C in a vacuum oven before use. Potassium hydroxide (99%), potassium bicarbonate (99%) and sodium hydroxide (99%) were purchased from Sigma-Aldrich. All chemicals were of analytical grade and directly used as received without further purification.


All electrochemical measurements were performed at room temperature using a BioLogic SP-150 potentiostat and a Biologic BCS-800 Series. A coiled platinum wire (BASi, catalogue no. MW-1033) was used as a counter electrode. In three-electrode measurements, the reference electrode was Hg/HgO (ALS, catalogue no. RE-61AP) with 0.1 M KOH filling solution; the filling solution was exchanged routinely to keep the potential constant. Potentials were converted to the SHE using the correction ESHE = EHg/HgO + 0.165 V.

Elemental analysis

C, H and N wt% were determined by means of CHN combustion analysis using an Exeter Analytical CE-440, with combustion at 975 °C.

Volumetric gas sorption measurements

N2 isotherms were collected using an Autosorb iQ gas adsorption analyser at 77 K. The BET surface area was determined by the BET equation and Rouquerol’s consistency criteria implemented in AsiQwin. All pore size distribution fittings were conducted in AsiQwin using N2 at 77 K on carbon (slit-shaped pores) quenched solid density functional theory model. CO2 sorption isotherms were also collected on an Autosorb iQ gas adsorption analyser. Isotherms conducted at 25, 35 and 45 °C were measured using a circulating water bath. Samples were activated at 100 °C in vacuum for 15 h before gas sorption measurements.

Thermogravimetric gas sorption measurements

Thermogravimetric CO2 adsorption experiments were conducted with a flow rate of 60 ml min−1 using a TA Instruments TGA Q5000 equipped with a Blending Gas Delivery Module. Samples were activated under flowing N2 for 30 min at various temperatures before cooling to 30 °C and switching the gas stream to CO2 mixtures. Cycling experiments were carried out on a Mettler Toledo TGA/DSC 2 Stared system equipped with a Huber mini chiller. For tests with high-concentration CO2, the adsorption and desorption of CO2 were performed at 30 and 100 °C for 20 min under 30% CO2 and 70% N2 with a flow rate of 140 ml min−1, respectively. For DAC tests, adsorption was carried out at 30 °C for 60 min, with 400 ppm CO2 in dry air, and desorption was carried out at 130 °C for 60 min with 100% N2.

Adsorption microcalorimetry measurements

The simultaneous measurement of the heat of adsorption and the adsorbed amount of carbon dioxide was performed by means of a heat flow microcalorimeter (Calvet C80 by Setaram), connected to a high-vacuum (residual pressure less than 10−4 mbar) glass line equipped with a Varian Ceramicell 0–100 mbar gauge and a Leybold Ceramicell 0–1,000 mbar gauge. Before the measurement, both PCS-OH and blank carbon cloth (roughly 150 mg before activation) were activated for 24 h under high vacuum (residual pressure less than 10−3 mbar) at 100 °C (temperature ramp 3 °C min−1). The adsorption microcalorimetry measurements were performed at 30 °C by following a well-established step-by-step procedure described in detail elsewhere34. This procedure allowed, during the same experiment, for the determination of both integral heats evolved (−Qint) and adsorbed amounts (na) for small increments of the adsorptive pressure. The partial molar heats obtained for each small dose of gas admitted over the sample were computed by applying the following ratio: ΔQintna, kJ mol−1. The (differential) heats of adsorption are then reported as a function of CO2 adsorbed amount, to obtain the (differential) enthalpy changes associated with the proceeding adsorption process. The equilibration time in the microcalorimetric measurement was set to 24 h for small equilibrium pressures (less than 30 mbar), whereas it was reduced to 2 h for larger doses for PCS-OH. The equilibration time was reduced to 2 h (regardless of the equilibrium pressure) for the bare carbon cloth, as equilibration is expected to occur faster in absence of specific adsorption sites.

X-ray diffraction

Powder X-ray diffraction patterns were collected on a Malvern Panalytical Empyrean instrument equipped with an X’celerator Scientific detector using a non-monochromated Cu Kα source (λ = 1.5406 Å). The data were collected at room temperature over a 2θ range of 3–80°, with an effective step size of 0.017°.

Scanning electron microscopy (SEM)

Samples were mounted onto a stainless-steel SEM stub using adhesive carbon tape. SEM imaging was performed on a Tescan MIRA3 FEG-SEM. Analysis of SEM images was conducted using the FIJI ImageJ software.

NMR spectroscopy

Solid-state NMR experiments were performed with a Bruker Advance spectrometer operating at a magnetic field strength of 9.4 T, corresponding to a 1H Larmor frequency of 400.1 MHz. A Bruker 4 mm HX double resonance probe was used in all cases. 1H NMR spectra were referenced relative to neat adamantane (C10H16) at 1.9 ppm and 13C NMR spectra were referenced relative to neat adamantane (C10H16) at 38.5 ppm (left-hand resonance). All the NMR tests were conducted with a sample magic angle spinning rate of 12.5 kHz. A 90° pulse-acquire sequence was used in each experiment. For 13C NMR experiments, recycle delays were set to be more than five times the spin-lattice relaxation time for each sample to ensure that the experiments were quantitative.

Charged-sorbents with different water contents were prepared for the NMR characterization. The sorbents were kept in a closed container for 24 h under different RHs. Saturated Mg(NO3)2 solutions were used to maintain 53% RH at 25 °C, respectively28.

Titration measurements

First, 88 mg of sample was immersed in 2 ml of deionized water and sonicated for 20 min at 25 °C. The pH value was then recorded with a pH meter (Insmark IS128C, calibrated with buffer solutions before use) at 25 °C as the initial point. Second, 100 µl of HCl (0.1 M) was slowly added. The mixture was sonicated for 20 min at a constant 25 °C and the pH of the solution was recorded. The second step was repeated until the end of the titration. There was no weight loss due to evaporation during the titration.

13CO2 dosing for solid-state NMR experiments

Freshly activated samples (75 °C, vacuum oven, 24 h) were packed into 4 mm NMR rotors in air and then evacuated for a minimum of 10 min in a home-built gas manifold29. 13C-enriched CO2 gas (Sigma-Aldrich, less than 3 atom% 18O, 99.0 atom% 13C) was then used to dose the samples with gas at room temperature until the gas pressure stabilized, before the rotors were sealed inside the gas manifold with a mechanical plunger.

DAC tests in sealed chambers

The DAC tests were carried out in a sealed box (volume roughly 600 ml) with a CO2 sensor (Aranet4) to record the concentration of CO2, temperature and RH at every 1 min interval. Before each cycle, the box was exposed to fresh air until the CO2 concentration, RH and temperature stabilized. The sorbent was then placed in the box, which was sealed during measurements.

Joule heating

After each DAC adsorption step, the sorbent was extracted from the box and connected with an external power source for Joule heating. A BioLogic SP-150 potentiostat was used to vary the electrical input. A constant voltage was applied and adjusted during the experiment to achieve a sample temperature in a range of 85–95 °C under an N2 atmosphere. The temperature was monitored using a thermocouple at a single contact point. After Joule heating regeneration, the electrode was reused for another DAC adsorption cycle.

Preparation of charged-sorbents

Three-electrode method

The charged-sorbents were prepared in a three-electrode configuration with a home-made cell, a Hg/HgO (in 0.1 M KOH) reference electrode and a platinum wire counter electrode. The activated carbon fabric ACC-5092-10 cloth (2 × 2 cm) was charged with a constant potential for 4 h (0.565 V versus SHE for the PCS-OH and −0.235 V versus SHE for NCS-K, respectively) in 40 ml of 6 M KOH(aq.) by means of chronoamperometry (Extended Data Fig. 2). After completing the charging process, the charged cloth was removed and held by plastic tweezers and rinsed with deionized water from a wash bottle for 5 min in total on both sides. Next, 500 ml deionized water in total was used to wash off the residual KOH solution. The rinsed cloth was then placed in the vacuum oven at 75 °C for 24 h to remove the remaining water.

For the soaked control sample, the activated carbon fabric ACC-5092-10 cloth (2 × 2 cm) was soaked in 40 ml of 6 M KOH for 4 h. After soaking, the same rinsing and drying processes used for the charged-sorbents were carried out. As a further control, a dripped cloth sample was prepared by adapting a literature protocol13. Here, 200 μl of 6 M KOH was dripped onto the surface of 340.0 mg ACC-10. Subsequently, the dripped samples were left in a Schlenk flask connected to a vacuum to let the samples dry for 72 h at room temperature.

Two-electrode Swagelok method

Free-standing carbon films were prepared by adapting the published method in ref. 35. In brief, YP80F activated carbon powder (95 wt%) (Kuraray Chemical) was mixed with polytetrafluoroethylene binder (5 wt%) (Sigma-Aldrich, 60 wt% dispersion in water) in ethanol. The resulting slurry was kneaded and rolled to give a carbon film of roughly 0.25 mm thickness, followed by removing residual solvent at 100 °C in vacuum for at least 24 h. Disc-shaped electrodes were then cut from the carbon films using a 0.25 inch hole punch. Symmetrical Swagelok electrochemical cells were then prepared in Swagelok PFA-820-6 fittings with stainless-steel current collectors, YP80F film electrodes (for both the positive and negative electrodes), 6 M KOH(aq.) electrolyte and a glass fibre separator (Whatman glass microfibre filter (GF/A)). Cells were charged at a constant cell voltage of 0.8 V for 4 h in two-electrode mode, and the positive electrode was then extracted, washed and dried, as above, to yield a charged-sorbent referred to as PCS-OH (YP80F). Three samples from three independent electrochemical cells were combined to provide sufficient material for gas sorption measurements.