Ambient, two-step CO₂-to-CaCO₃ conversion: alkali-activated H₂O₂-promoted carbonate CO₂ capture and direct calcite mineralization

Abstract

We demonstrate an ambient-temperature two-step CO₂-to-CaCO₃ conversion in which alkali-activated H₂O₂ accelerates carbonate CO₂ capture, followed by calcite mineralization upon CaCl₂ addition, without thermal stripping of the capture liquor. Alkali activation generates transient reactive oxygen species (ROS) that promote rapid CO₂ absorption at 30 °C into aqueous Na₂CO₃ and K₂CO₃ without persistent promoters, yielding bicarbonate-rich liquors that are directly mineralized to phase-pure calcite and release a CO₂ stream free of O₂. Single-reactor screening identified concentration windows that maximize capacity while maintaining homogeneity, with Na₂CO₃ optimal near 7–10 wt% and K₂CO₃ near 20–25 wt%. Multi-reactor staging prolonged cycle duration and moderated the trade-off between outlet CO₂-removal specification and first-stage carbonate-to-bicarbonate conversion, with K₂CO₃ formulations showing the smallest penalty at stricter cutoffs. The CaCO₃ products were identified as calcite by XRD and ATR-FTIR, with SEM/EDS results consistent with rhombohedral CaCO₃ and no detectable secondary elements. A consumables-only techno-economic analysis using four geography-specific reagent price decks shows that LCOC is reagent-dominated, with carbonate utilization emerging as the principal cost lever through its effect on stoichiometric make-up demand. Overall, the results establish an ambient-temperature capture-to-mineralization route in which alkali-activated H₂O₂-derived ROS accelerate carbonate capture and direct mineralization avoids thermal regeneration.

---

Sustainability spotlight

Decarbonizing point-source CO₂ capture demands low-energy routes that avoid amine regeneration and degradation. We demonstrate an ambient-temperature capture-to-conversion scheme in which alkali-activated, H₂O₂-derived ROS transiently promote carbonate CO₂ absorption at 30 °C and then self-dissipate, yielding bicarbonate liquors free from persistent promoter residues that precipitate phase-pure calcite with CaCl₂; multi-reactor staging maintains high removal while preserving carbonate utilization. The chemistry uses commodity inputs (Na₂CO₃/K₂CO₃, H₂O₂, CaCl₂), eliminates thermal stripping, and yields a utilizable CaCO₃ product and an O₂-free CO₂ stream, highlighting a simple path to lower thermal energy demand and clearer pathways for future cost reduction through improved utilization, regeneration, and waste-derived reagents. This advance supports SDG 13 (Climate Action), SDG 9 (Industry, Innovation and Infrastructure), and SDG 12 (Responsible Consumption and Production).

1 Introduction

The increasing atmospheric concentration of CO₂ is a major driver of global climate change, requiring action to reduce emissions and achieve net-zero targets by mid-century.^1–3 Carbon capture, utilization, and storage (CCUS) technologies are widely regarded as a central pillar of decarbonization, enabling substantial reductions in emissions from point sources and, potentially, from the atmosphere itself.^4,5 Within the portfolio of existing CO₂ capture methods, amine-based absorption, particularly with monoethanolamine (MEA), is the most mature and widely applied industrial method.^6–8 However, amine-based absorption systems suffer from well-documented disadvantages, including high energy requirements for thermal regeneration (typically 3.6–4.0 GJ t⁻¹ CO₂), as well as thermal and oxidative degradation, corrosivity, volatility, and aerosol emissions.^4,9 These challenges impose both economic and environmental costs, limiting large-scale deployment.^4,10 Consequently, there is considerable interest in developing CO₂ capture strategies that reduce energy penalties, minimize environmental impacts, and enable direct integration with carbon utilization pathways to produce valuable products.^11,12

Alkaline carbonate solutions, such as potassium carbonate (K₂CO₃) and sodium carbonate (Na₂CO₃), are promising alternatives to amines due to their low cost, low volatility, negligible toxicity, and high chemical stability.^13–15 K₂CO₃-based processes, such as the Benfield process (also known as the hot potassium carbonate process), have already been implemented for CO₂ and H₂S removal in more than 600 plants, demonstrating their industrial maturity.^16,17 Furthermore, carbonate-based systems have lower regeneration energy requirements than amines and avoid the degradation products associated with the latter.^14,18 Nevertheless, their main limitation is the relatively slow CO₂ absorption kinetics, particularly under ambient conditions.^14,19 Industrial processes such as the hot potassium carbonate process address this limitation by operating at elevated absorption temperatures (∼100–120 °C) and employing rate promoters, such as piperazine and amino acids, to accelerate the bicarbonate formation step.^20–22 These promoters are commonly classified into three groups: organic, inorganic, and enzymatic.^14,23 While effective for CCS, the persistence of organic promoters (such as amines or amino acids) leaves chemical residues in the resulting bicarbonate liquor. Integrating these promoted solutions directly into CCU processes can therefore complicate downstream conversion or necessitate additional separation steps to recover the capture agents.^24–26

Recently, we presented a novel kinetic promotion method for K₂CO₃ solutions that utilizes hydrogen peroxide (H₂O₂)-derived reactive oxygen species (ROS) as promoters for CO₂ absorption by K₂CO₃.²⁶ Mechanistically, we treat ROS as transient, in situ promoters that accelerate the otherwise slow bicarbonate-forming chemistry in carbonate capture under near-ambient conditions, while decomposing to benign products and therefore not persisting into downstream mineralization. This approach achieved CO₂ uptake rates at room temperature comparable to those of piperazine-promoted systems, while leaving a promoter-free bicarbonate-rich solution after ROS decomposition. In strongly alkaline media, where H₂O₂ is partially deprotonated as HO₂⁻ (pK_a ≈ 11.6 at 25–30 °C), small additions of H₂O₂ generate transient ROS under ambient conditions, including superoxide (O₂˙⁻), hydroxyl radicals (˙OH), and singlet oxygen (¹O₂).^27–30

H₂O₂ + OH⁻ ⇌ HO₂⁻ + H₂O (1)

In related systems, oxygen-centered radical species have been reported to react rapidly with CO₂ to give transient oxygenated intermediates; such chemistry offers a plausible kinetic basis for enhanced bicarbonate formation under strongly basic conditions.

X₂CO₃ + CO₂ + H₂O ⇌ 2XHCO₃ (X = Na or K; promoted by transient ROS) (2)

For clarity, eqn (1) and (2) summarize the chemistry at the process level: alkaline peroxide activation generates transient promoter species, and the overall promoted capture step converts carbonate to bicarbonate; however, the present study does not seek to distinguish among the possible underlying elementary ROS-mediated pathways.

Our previous work demonstrated that ROS, including superoxide, can facilitate the carbonate-to-bicarbonate conversion.²⁶ Earlier work by our group demonstrated in situ superoxide generation upon introducing H₂O₂ to NaOH and KOH solutions.^31,32 In this study, our approach omits the addition of excess caustic, relying instead on the hydroxide inherently present at the alkaline pH of K₂CO₃ (or Na₂CO₃) solutions to enable peroxide activation. The self-dissipating nature of ROS offers a unique advantage: the resulting bicarbonate-rich liquor can be directly repurposed for downstream utilization without additional purification. Economically, commodity H₂O₂ can be competitive with organic promoters and avoids amine-derived degradation products. To circumvent the substantial energy penalty associated with conventional CCUS technologies, carbonate mineralization has been explored as a low-energy chemical conversion route.^33,34 Within the literature, carbonate mineralization approaches are broadly divided into direct gas–solid mineralization at elevated temperatures and pressures, and indirect aqueous mineralization at ambient conditions.^13,34 For positioning the present work within the broader aqueous alkaline CO₂-capture literature, two process archetypes are especially relevant: (i) chemical regeneration (looping) approaches, in which the alkaline capture capacity is restored while CaCO₃ precipitates (e.g., causticization/looping schemes);^35,36 and (ii) single-pass capture-and-conversion routes. In the latter, the bicarbonate liquor is consumed stoichiometrically to form CaCO₃ (and a corresponding brine), a step that can be coupled with the release of a concentrated CO₂ stream without thermal stripping.¹³

Barzagli et al. reported a similar two-step process in which CO₂ is first absorbed into an aqueous Na₂CO₃ solution, and the resulting bicarbonate-rich liquor is then reacted with CaCl₂ at room temperature to yield CaCO₃ and high-purity CO₂.¹³ This method eliminates the need for thermal stripping by chemically transforming NaHCO₃ into CaCO₃. However, its reliance on dilute Na₂CO₃ solutions, limited CO₂ removal thresholds (∼80%), and the intrinsically slow kinetics of unpromoted carbonate absorption present challenges for scale-up. Improving the kinetics and enabling the use of more concentrated solutions could enhance the practicality of such ambient-temperature capture-conversion schemes.

In this work, we present an improved two-step CO₂ capture and utilization process that combines ROS-promoted carbonate absorption with subsequent CaCO₃ precipitation. In the first step, CO₂ is absorbed into aqueous alkali carbonate solution containing a small concentration of H₂O₂ (typically one-tenth of the carbonate concentration), enabling rapid CO₂ uptake at 30 °C and high bicarbonate conversion while preventing precipitation. A transient O₂ signal occurs only in promoted runs during CO₂ absorption, typically peaking early but not confined to the onset, and is used solely as a diagnostic of promoter activity.

The promotion is self-terminating as CO₂ loading reduces alkalinity toward the HO₂⁻/H₂O₂ boundary, yielding a bicarbonate-rich liquor without persistent promoter residues. Both K₂CO₃ and Na₂CO₃ were investigated to assess how carbonate solubility and cation identity influence ROS-promoted uptake and downstream precipitation, and to quantify the associated cost-performance trade-offs. In the second step, the bicarbonate-rich solution, free from persistent promoters, is treated with CaCl₂ to precipitate pure calcite CaCO₃ and release CO₂, as summarized in eqn (3).

2XHCO₃ + CaCl₂ → CaCO₃↓ + 2XCl + CO₂ + H₂O (X = Na or K) (3)

This single-pass design simplifies operation, avoids thermal regeneration, and yields phase-pure calcite and a CO₂ stream with O₂ below the detection limit. In our bench configuration this gas is N₂-balanced due to the inert sweep; the headspace composition thus reflects the sweep rather than the chemistry. To sustain CO₂-removal efficiency setpoints of 80%, 90%, and 100%, cycles were terminated at the corresponding CO₂ outlet threshold. Through single- and multi-reactor configurations, we evaluate the kinetics, bicarbonate conversion, and product quality, demonstrating the potential of this approach as a sustainable, low-energy CCU pathway.

2 Experimental

2.1 CO₂ absorption and conversion setup

Single-reactor CO₂ absorption experiments were conducted using the system described in our previous work.²⁷ A stirred reactor containing an aqueous carbonate/H₂O₂ solution was supplied with a CO₂/N₂ gas mixture (15 vol% CO₂ in N₂) at a constant flow rate of 0.2 L min⁻¹, controlled by a mass flow controller (Aalborg DPC 17). The gas was introduced through a sintered glass diffuser to ensure efficient dispersion. The reactor was immersed in a thermostated bath at 30 ± 1 °C.

Outlet CO₂ and O₂ concentrations were logged every 20 s using an NDIR CO₂ sensor (Senseair K33 BLG) and a fluorescence-based O₂ sensor (LuminOx UV Flux). The first 2 min of recorded data correspond to an effective deadtime arising from gas transport lag in the sampling line. Baseline tests at 15 vol% CO₂/N₂ gave standard deviations of ±0.047–0.158 vol% after deadtime, well below the manufacturer accuracy (±0.5 vol% + 3% of reading). Instrument specifications are therefore taken as error bounds. The CO₂ removal efficiency (E) was calculated according to eqn (4):

where C_i and C_o denote the inlet and outlet CO₂ concentrations, respectively.

To provide non-ambient controls, we performed a single-reactor temperature series (25, 30, 50, and 70 °C; ±1 °C) for 30 wt% K₂CO₃ + 3 wt% H₂O₂ (15 g) under the same feed conditions as above (15 vol% CO₂/N₂, 0.2 L min⁻¹). CO₂ capacity (mmol CO₂ per g solution) was obtained by integrating the inlet-outlet CO₂ difference (C_i − C_o), over time.

For multi-reactor tests, two or three identical reactors were connected in series. The feed gas flow rate was adjusted to 0.35 L min⁻¹ (two reactors) or 0.5 L min⁻¹ (three reactors). The lead reactor was maintained at 30 ± 1 °C, while downstream reactors were operated at room temperature. Runs were continued until the train-averaged CO₂ removal efficiency fell below predefined thresholds (80%, 90%, or 100%). Each condition was tested over three consecutive cycles. Between cycles, the lead reactor was removed for CaCl₂ treatment, the train was advanced by one position, and the tail stage was replenished with fresh absorbent.

The CO₂ absorption-conversion configuration is illustrated in Fig. 1, which schematically shows the staged multi-reactor absorption setup (Step 1) and the subsequent CaCO₃ precipitation stage (Step 2). In Step 2, the bicarbonate-rich solution was reacted with a stoichiometric CaCl₂ solution at ambient temperature, while N₂ was supplied at 0.2 L min⁻¹. This process yielded solid CaCO₃, an O₂-free CO₂ stream carried by the N₂ sweep, and a brine (NaCl or KCl, depending on the starting carbonate). The CaCO₃ precipitate was collected by filtration, washed with deionized water, and dried at 60 °C prior to characterization.

Fig. 1 Schematic representation of the experimental two-step CO₂ capture-to-CaCO₃ process. In the absorption step, a 15 vol% CO₂/N₂ stream is delivered via a mass flow controller and solenoid valve to stirred reactors connected in series, generating a bicarbonate-rich solution. R₁ denotes the first reactor in the series, whereas R_n denotes the final reactor, with n = 2 or 3 depending on the number of reactors employed. In the CaCO₃ production step, the bicarbonate-rich solution is reacted with CaCl₂ under N₂ flow to yield CaCO₃ and the corresponding salt brine.

For representative multi-reactor runs, pH was measured to track alkalinity changes during CO₂ loading and subsequent mineralization. Measurements were performed for 7 wt% Na₂CO₃ + 0.7 wt% H₂O₂ and 25 wt% K₂CO₃ + 2.5 wt% H₂O₂ at the 90% train-averaged CO₂-removal threshold. pH in reactor 1 was monitored at four stages: in the fresh starting solution before cycle 1, in the CO₂-loaded liquor at the end of cycle 1 immediately before CaCl₂ addition, in the corresponding liquor at the end of cycle 2, and in the recovered brine after CaCO₃ precipitation.

2.2 ¹³C NMR analysis of carbonate/bicarbonate speciation

Post-absorption solution samples were analyzed by ¹³C NMR spectroscopy using a Bruker Avance NEO 500 MHz spectrometer. Bicarbonate conversion was quantified from a linear regression of carbonate/bicarbonate chemical shifts (δ, ppm) against their molar fraction (χ), following an established method (R² = 0.9996, RMSE = 0.008).^26,37 The bicarbonate conversion was quantified with reference to the carbonate/bicarbonate equilibrium corresponding to eqn (2).

Analytical uncertainty was taken as the calibration residual (±0.8% conversion), since the propagated peak-fit error (<0.1%) was negligible. Error bars in Fig. 4 and 6 reflect this uncertainty.

2.3 Characterization of CaCO₃ precipitates

The solid CaCO₃ obtained after absorption-conversion cycles was characterized by X-ray diffraction (XRD, Bruker D8 Advance), scanning electron microscopy with energy-dispersive X-ray spectroscopy (SEM/EDS, Thermo Fisher Scientific Apreo 2), and attenuated total reflectance Fourier-transform infrared spectroscopy (ATR-FTIR, Shimadzu IRAffinity-1S).

3 Results and discussion

3.1 Single-reactor screening of carbonate/H₂O₂ systems

Screening experiments were first conducted in a single stirred reactor to obtain CO₂ uptake curves and capacity, track O₂ evolution as a promoter diagnostic, and establish precipitation limits at 30 °C. Fig. 2 depicts the CO₂ absorption performance of two representative solutions from this single-reactor screening: 7 wt% Na₂CO₃ + 0.7 wt% H₂O₂ and 20 wt% K₂CO₃ + 2 wt% H₂O₂, operated at a flow rate of 0.2 L min⁻¹ of a 15 vol% CO₂/N₂ mixture, contacting 15 g of solution. Panels (a andc) show the outlet CO₂ concentration and the corresponding removal efficiency, and panels (b andd) show the simultaneously measured O₂. Dashed curves in panels (a–d) represent the corresponding unpromoted (neat) Na₂CO₃ and K₂CO₃ solutions of equivalent carbonate concentration, included for comparison. The promoted systems show distinctly higher CO₂ removal efficiency and measurable O₂ evolution, in clear contrast to the unpromoted baselines, which exhibit minimal O₂ and a rapid breakthrough. Both systems begin at ∼100% removal (outlet CO₂ ≈ 0%). The Na-based liquor shows an early CO₂ breakthrough (≈3 min) and a modest, broad O₂ peak (≈2.5 vol%) that decays to a zero baseline within the first half of the run. The K-based liquor maintains quantitative removal for ∼9 min, exhibits a taller and narrower O₂ burst (peak ≈ 6.5 vol%) coincident with that period, followed by a slowly decaying non-zero O₂ baseline while CO₂ removal efficiency declines gradually over the remainder of the 45 min run.

Fig. 2 Single-reactor CO₂ absorption behavior and temperature control experiments. Panels (a and b): 7 wt% Na₂CO₃ + 0.7 wt% H₂O₂; panels (c and d): 20 wt% K₂CO₃ + 2 wt% H₂O₂, each at 30 °C using 15 vol% CO₂/N₂ (0.2 L min⁻¹) and 15 g solution. (a and c) CO₂ outlet concentration (red) and removal efficiency (black); (b and d) O₂ outlet signal. Dashed curves represent the corresponding unpromoted solutions. (e) CO₂ absorption capacity versus time for 30 wt% K₂CO₃ + 3 wt% H₂O₂ at 25, 30, 50, and 70 °C (same gas feed and solution mass).

For Na₂CO₃, the profiles are readily divided into three regions: (i) an initial interval of quantitative capture with a rising O₂ signal, indicating abundant HO₂⁻/ROS at very high alkalinity, (ii) a transition where both the O₂ level and the removal efficiency decline as CO₂ loading lowers alkalinity, and (iii) a bicarbonate-rich interval in which O₂ is at baseline, below the analyzer's detection limit, and the CO₂ removal efficiency declines to low levels. This behavior is consistent with the promoter being effectively “off” as the solution nears the HO₂⁻/H₂O₂ acid–base boundary (pK_a ≈ 11.6 at 25–30 °C), where the HO₂⁻ fraction and the ROS-driven O₂ release become negligible.

For K₂CO₃, the mechanism is essentially the same, albeit persisting longer due to higher solubility and higher carbonate/H₂O₂ loadings, which sustain promoter activity in the homogeneous regime. The O₂ peak falls within the 100%-removal window and is followed by a slowly decaying O₂ baseline starting at ∼2 vol%, consistent with the slower decline in CO₂ removal efficiency.

To benchmark the 30 °C operating condition against lower and elevated temperatures, an additional single-reactor temperature series (25, 30, 50, and 70 °C) was performed for 30 wt% K₂CO₃ + 3 wt% H₂O₂, and the resulting CO₂ absorption capacity profiles (mmol CO₂ per g solution) versus time are shown in Fig. 2e. This formulation was selected as a high-loading, high-signal case to isolate temperature effects under otherwise identical conditions and is used here as a control comparison rather than as a proposed optimum composition. The 25 and 30 °C traces are similar, with 25 °C showing only a marginally higher capacity, whereas capacity decreases at 50 °C and declines further at 70 °C. This monotonic reduction with temperature is consistent with faster H₂O₂ decomposition and increased self-quenching of ROS at elevated temperature, which shortens the effective promotion window for carbonate-to-bicarbonate conversion. Temperatures below 25 °C were not pursued further because they increased the tendency toward premature precipitation and would require active cooling, introducing an additional energy penalty.

Table 1 summarizes the single-reactor screening, listing the concentration-time windows tested and the onset of precipitation during operation at 30 °C and upon post-run cooling to room temperature. The performance outcomes across the homogeneous window, reported in Table 1, informed the selection of multi-reactor conditions (Section 3.2).

Table 1 Screening space and phase behavior at 30 °C

System	Conc. (wt%)	Homogeneous during run at 30 °C (max time observed)	Precipitation during operation (first observed)	Cooling-induced precipitation after run
a Outside Fig. 3 but observed in screening; included here for completeness.
Na2CO3	5	60 min	None	None
Na2CO3	7	60 min	None	None
Na2CO3	10	55 min	None during ≤55 min	Appears after ≥35 min runs
Na2CO3	15	≤30 min	By ∼30 min	Appears after ≥20 min runs
Na2CO3	20^a	n/a	By ∼20 min	n/a
Na2CO3	30^a	n/a	<10 min	n/a
K2CO3	10	60 min	None	None
K2CO3	20	60 min	None	None
K2CO3	25	50 min	By 60 min	Appears after ≥50 min runs
K2CO3	30	40 min	By ∼45 min	Appears after ≥40 min runs

---

Fig. 3 maps the specific CO₂ capacity (mmol CO₂ per g solution) versus run time (0–60 min) and concentration for (a) Na₂CO₃ (5–15 wt%) and (b) K₂CO₃ (10–30 wt%) at 30 °C, across the concentration-time windows summarized in Table 1. Na₂CO₃ loadings above 15 wt% (20–30 wt%) were excluded, as these solutions precipitated prematurely in operation or on cooling, which placed them outside the homogeneous operating window. Surfaces smoothed from discrete runs indicate apparent capacities of ∼0.4–2.1 mmol g⁻¹ for Na and ∼0.5–3.8 mmol g⁻¹ for K (from the color scales).

Fig. 3 Specific CO₂ capacity (mmol CO₂ per g solution) as a function of run time and carbonate concentration at 30 °C for (a) Na₂CO₃ (5–15 wt%) and (b) K₂CO₃ (10–30 wt%).

For Na₂CO₃, capacity rises steeply during the first 15–20 min across all concentrations, reaching ∼1.0 mmol g⁻¹. At lower loadings (5–7 wt%), the increase slows thereafter and plateaus near ∼1.4 mmol g⁻¹ after ∼30 min, reflecting the intrinsic limit of these dilute solutions. At higher loadings, capacity continues to grow with time, reaching a maximum near 10 wt% where it peaks at ∼2.1 mmol g⁻¹ at 55–60 min. Beyond this mid-range, more concentrated solutions level off earlier due to precipitation, reinforcing 9–11 wt% as the region of highest attainable capacity. By contrast, K₂CO₃ increases more monotonically with concentration and time; at 30 wt% the run is truncated by in-run precipitation beginning at ∼45 min, at which point the capacity is ∼3.1–3.3 mmol g⁻¹. Thus, the observed ceiling at 30 wt% is set by time-to-solids, not by an intrinsic capacity limit. Within the homogeneous window, higher K₂CO₃ concentrations and longer times consistently deliver greater capacities.

Several compositions remain homogeneous at 30 °C but precipitate on cooling to room temperature (Na₂CO₃: 10–15 wt%; K₂CO₃: 25–30 wt%), distinguishing the in-run operating window from the post-run window and informing the process sequence. To avoid unintended solids, the absorber liquors should be kept at 30 °C through CaCl₂ addition, especially near the upper concentration/time edges, so that the only intentional precipitation is CaCO₃ during the second step of the 2-step process.

Fig. 4 links capacity to base utilization by reporting the extent of carbonate-to-bicarbonate conversion at the end of each single-reactor run at 30 °C, determined from ¹³C NMR speciation of endpoint aliquots. Run times are shown above the bars. Panel (a) covers Na₂CO₃ (5–20 wt%) and panel (b) K₂CO₃ (10–30 wt%). For dilute-to-intermediate compositions, these runs fall within the homogeneous operating window, whereas the highest-loadings illustrate endpoint behavior near or beyond the precipitation boundary identified in Table 1.

Fig. 4 Extent of carbonate-to-bicarbonate conversion (%) at 30 °C, determined by ¹³C NMR of endpoint aliquots for single-reactor runs of (a) Na₂CO₃ solutions and (b) K₂CO₃ solutions. Run times (shown above bars) correspond to the endpoint aliquots used for NMR analysis. At higher carbonate loadings, these runs approach or exceed the precipitation limits summarized in Table 1.

For Na₂CO₃, the most dilute liquors (5 and 7 wt%) reach ∼90% HCO₃⁻ after 60 min, indicating that nearly all of the initially charged carbonate is utilized. At 10 wt% the conversion is ∼80% after a 50 min run, and it declines at higher loadings: 15 wt% levels off near 45% by 30 min, whereas 20 wt% reaches only ∼25% over the same period. These results, together with the capacity maximum near 9–11 wt% in Fig. 3 and the earlier onset of in-run precipitation at higher Na₂CO₃ loadings, indicate that approximately 10 wt% Na₂CO₃ offers the best balance between attainable capacity and carbonate utilization.

Over comparable run durations, K₂CO₃ shows a markedly smaller fraction of residual carbonate. At 10 wt% the conversion is ≥95% at 60 min; at 20 wt% it is ∼90% at 60 min. Even at 25 wt% conversion is ∼80% by 50 min, whereas at 30 wt% it remains ∼60–70% by 35–40 min. This pattern echoes the monotonic capacity rise and delayed precipitation mapped in Fig. 3 and Table 1.

Overall, the single-reactor screening results (Table 1, Fig. 3 and 4) define the operating windows used in the multi-reactor experiments discussed next. Na₂CO₃ should be maintained near 7–10 wt% to ensure both high base utilization and a homogeneous liquor at 30 °C, whereas K₂CO₃ can be increased to 20–25 wt% without significantly reducing conversion and carbonate residuals or triggering premature solids.

3.2 Multi-reactor operation of carbonate/H₂O₂ systems

We implemented cyclic absorption in two- and three-reactor trains to improve CO₂-removal efficiency while keeping the first reactor under the conditions established in the single-reactor screening. The aim of staging is not to increase total throughput, but to maintain high carbonate-to-bicarbonate conversion in the lead reactor while holding a tight outlet CO₂ specification throughout the run. Fig. 5 presents three successive cycles at 30 °C for four carbonate/H₂O₂ formulations: 7 wt% Na₂CO₃ + 0.7 wt% H₂O₂, 10 wt% Na₂CO₃ + 1 wt% H₂O₂, 20 wt% K₂CO₃ + 2 wt% H₂O₂, and 25 wt% K₂CO₃ + 2.5 wt% H₂O₂. Panels (a, c, e and g) show the train-averaged CO₂ removal efficiency for the three cycles, and panels (b, d, f and h) show the corresponding O₂ outlet traces for cycle 2, time-aligned with the CO₂ data. The 7 wt% Na and 20 wt% K cases ran as three-reactor trains at 0.5 L min; the 10 wt% Na and 25 wt% K cases ran as two-reactor trains at 0.35 L min⁻¹ to avoid early precipitation observed when adding a third stage at those concentrations. Each cycle was terminated when the train-averaged CO₂ removal efficiency reached a set threshold of 80%; the panels display these cut-off runs. After each cycle, the lead reactor was removed for CaCl₂ treatment, the train advanced one position, and a fresh solution was loaded at the tail.

Fig. 5 Cyclic CO₂ absorption in two- and three-reactor trains at 30 °C for four carbonate/H₂O₂ formulations: (a and b) 7 wt% Na₂CO₃ + 0.7 wt% H₂O₂, (c and d) 10 wt% Na₂CO₃ + 1 wt% H₂O₂, (e and f) 20 wt% K₂CO₃ + 2 wt% H₂O₂, and (g and h) 25 wt% K₂CO₃ + 2.5 wt% H₂O₂. Panels (a, c, e and g) show train-averaged CO₂ removal efficiency over three cycles; panels (b, d, f and h) show corresponding O₂ outlet traces for cycle 2. Each cycle was terminated at 80% CO₂ removal. Three-reactor trains (a, b, e and f) operated at 0.5 L min⁻¹; two-reactor trains (c, d, g and h) at 0.35 L min⁻¹.

In panels (a, c, e and g), each cycle starts with a plateau at complete CO₂ removal, followed by a decline until the 80% threshold is reached. Cycle 1 is longest, whereas cycles 2 and 3 are shorter and nearly equal in duration, as they start with two partially spent stages and only one fresh stage, demonstrating replicability.

The O₂ outlet profiles (panels b, d, f and h) indicate the sequential operation of stages. Each O₂ peak marks the point at which a downstream stage first receives appreciable CO₂ slip while still at high pH (HO₂⁻-rich). As absorption drives the local pH toward the HO₂⁻/H₂O₂ boundary, promoter activity in that stage subsides and its O₂ signal declines until the next stage enters operation. Once the final stage crosses this boundary, residual capacity collapses and the train-averaged removal falls, resulting in the end of the cycle. In (panels b and f), three well-resolved peaks are observed, consistent with three-reactor trains operating at high pH. For two-reactor trains, the 25 wt% K₂CO₃ + 2.5 wt% H₂O₂ case (panel h) similarly exhibits two distinct peaks. By contrast, in the 10 wt% Na₂CO₃ + 1 wt% H₂O₂ case (panel d), the lower sweep flow (0.35 L min⁻¹), together with the greater O₂ peak intensity from Na₂CO₃-treated H₂O₂ (relative to K₂CO₃ at equal mass), causes the two stage contributions to merge into a broad, slightly delayed hump rather than two distinct peaks.

Fig. 6 shows how first-stage conversion in multistage systems changes as a function of efficiency threshold. The bars report the carbonate-to-bicarbonate conversion from ¹³C NMR measured at the end of cycle 2 for 7 and 10 wt% Na₂CO₃ and 20 and 25 wt% K₂CO₃, each run at three thresholds: 80% (cyan), 90% (purple), and 100% (grey). Tightening the outlet specification shortens the cycle and therefore reduces conversion, but the magnitude of that penalty depends strongly on formulation. 20 wt% K₂CO₃ remains highly converted across thresholds (about 95% at 80%, ∼90% at 90%, ∼85% at 100%). 25 wt% K₂CO₃ shows a moderate decline (∼80%, ∼70%, ∼60–65%). 7 wt% Na₂CO₃ stays high at 80 and 90% (about 85–90%) but drops at 100% (∼70%). 10 wt% Na₂CO₃ is the most sensitive, falling from about 75–80% at 80% to ∼50–55% at 90% and ∼35–40% at 100%. These results show that K-based liquors tolerate stricter outlet specifications because they remain homogeneous and promoter-active at higher concentration, whereas Na-based liquors suffer a larger conversion penalty when the threshold is tightened.

Fig. 6 First-stage carbonate-to-bicarbonate conversion at the end of cycle 2 for 7 and 10 wt% Na₂CO₃ solutions and 20 and 25 wt% K₂CO₃ solutions, each promoted with H₂O₂, as a function of efficiency cutoff (80%, 90%, and 100%).

The pH evolution of representative reactor-1 liquors at the 90% cutoff supports the proposed self-terminating promotion mechanism (Table 2). The fresh 7 wt% Na₂CO₃ + 0.7 wt% H₂O₂ and 25 wt% K₂CO₃ + 2.5 wt% H₂O₂ liquors start at pH 10.80 and 11.40, respectively, consistent with sufficiently alkaline conditions for peroxide activation. By the cycle-termination point, the pH decreases to 9.00 (cycle 1) and 9.20 (cycle 2) for the Na-based liquor and to 9.47 (cycle 1) and 9.28 (cycle 2) for the K-based liquor. These values are well below the HO₂⁻/H₂O₂ acid–base boundary discussed above, consistent with attenuation of the O₂ signal and loss of promoter activity as CO₂ loading proceeds. After CaCl₂ addition, the recovered brines show still lower pH values, 8.10 for the Na system and 6.25 for the K system, confirming that mineralization is carried out on CO₂-loaded liquors that are no longer strongly alkaline. The close agreement between the cycle-1 and cycle-2 cutoff pH values also supports the reproducibility of the stage-swap protocol at the 90% threshold.

Table 2 pH at key process stages for reactor 1 in representative multi-reactor runs operated at the 90% CO₂-removal threshold

Process stage	7 wt% Na2CO3 + 0.7 wt% H2O2	25 wt% K2CO3 + 2.5 wt% H2O2
Fresh starting solution before cycle 1	10.80	11.40
End of cycle 1, before CaCl2 addition	9.00	9.47
End of cycle 2, before CaCl2 addition	9.20	9.28
Post-mineralization brine	8.10	6.25

---

3.3 Product characterization: CaCO₃ phase and morphology

Calcium carbonate was precipitated by adding CaCl₂ to the bicarbonate-rich liquors at 30 °C; the temperature was maintained during dosing to prevent cooling-induced precipitation of the absorber prior to conversion.

To quantify the efficiency of the CaCl₂-induced precipitation, the isolated CaCO₃ yield was determined gravimetrically. After addition of stoichiometric CaCl₂ to the post-absorption liquor at 30 °C, the precipitated CaCO₃ was collected by filtration, washed with deionized water, and dried at 60 °C to constant mass. The isolated yield was calculated as the ratio of recovered dry CaCO₃ mass to the theoretical CaCO₃ mass corresponding to the CaCl₂ charge (1 mol Ca²⁺ → 1 mol CaCO₃). On this basis, the isolated CaCO₃ yields were 97.9% for the 7 wt% Na₂CO₃ + 0.7 wt% H₂O₂ case and 98.7% for the 25 wt% K₂CO₃ + 2.5 wt% H₂O₂ case.

Fig. 7 presents ATR-FTIR spectra and XRD patterns of the solids, and Fig. 8 presents SEM micrographs, for the two cases carried through the study: 7 wt% Na₂CO₃ + 0.7 wt% H₂O₂ and 25 wt% K₂CO₃ + 2.5 wt% H₂O₂.

Fig. 7 (a) ATR-FTIR spectra and (b) XRD patterns of CaCO₃ precipitates obtained from 7 wt% Na₂CO₃ + 0.7 wt% H₂O₂ and 25 wt% K₂CO₃ + 2.5 wt% H₂O₂ solutions at 30 °C.

Fig. 8 SEM micrographs of CaCO₃ precipitates from (a–c) 7 wt% Na₂CO₃ + 0.7 wt% H₂O₂ and (d–f) 25 wt% K₂CO₃ + 2.5 wt% H₂O₂ solutions at 30 °C.

The ATR-FTIR spectra in Fig. 7a are consistent with calcite within the plotted range. Both samples exhibit characteristic bands near 870–875 cm⁻¹ and 712–714 cm⁻¹, assigned to the out-of-plane and in-plane bending modes of carbonate, respectively. Bands characteristic of vaterite or aragonite are not evident within the plotted ranges. The XRD patterns are consistent with rhombohedral calcite, showing the expected diffraction peaks and no additional peaks attributable to other CaCO₃ polymorphs or crystalline impurities. Within the detection limits of FTIR and XRD, the products from the Na-derived and K-derived liquors are indistinguishable and correspond to phase-pure calcite.

Morphology is consistent with this assignment. The SEM micrographs in Fig. 8 show aggregates of well-faceted rhombohedral crystallites for both liquors. At lower magnification the solids appear as dense, grape-like agglomerates. Higher-magnification views reveal individual crystals bounded by flat faces and stepped edges typical of calcite. Needle-like forms associated with aragonite and spherical cauliflower morphologies associated with vaterite are not observed. Qualitatively, both products comprise micron-scale crystallites with moderate polydispersity; the K-derived material tends to show slightly sharper faceting, while the Na-derived sample appears somewhat more aggregated, yet the dominant habit in both cases remains rhombohedral calcite. Elemental compositions determined by EDS (Tables S1 and S2) further support this assignment, showing only Ca, O, and C, with no detectable secondary elements, consistent with CaCO₃. Additional SEM/EDS outputs supporting Fig. 8 and the calcite assignment are provided in the SI (Fig. S1–S6). Representative SEM micrographs for the Na- and K-derived products (Fig. S1 and S4) confirm the same rhombohedral calcite habit observed in Fig. 8. Corresponding EDS spectra (Fig. S2 and S5) show only Ca, C, and O signals, while elemental maps (Fig. S3 and S6) confirm spatially uniform Ca/O/C distributions.

Collectively, ATR-FTIR, XRD, and SEM show that step-2 precipitation at 30 °C yields phase-pure calcite, with no detectable metastable polymorphs, for both Na₂CO₃- and K₂CO₃-derived liquors. This consistency shows that differences in alkali, concentration, and reactor staging during the absorption stage do not lead to mixed-phase CaCO₃ during precipitation at 30 °C, simplifying product handling and underscoring the robustness of the capture-to-conversion sequence.

3.4 Techno-economic implications of the multireactor operation

3.4.1 Scope. Fig. 9 reports the consumables-only levelized cost of capture (LCOC; USD per tCO₂), excluding capital and energy contributions, for four geography-specific reagent price decks (US, EU, Northeast Asia, and Southeast Asia; all from 2025 Q3). Because the present scheme is a non-regenerative, capture-to-mineral route, this consumables-only LCOC should be interpreted as a conservative upper bound based on stoichiometric reagent purchase. We report this boundary-transparent metric to isolate how measured carbonate utilization and staging translate into reagent consumption, while allowing absolute LCOC to reflect regional reagent-price differences rather than site-specific CAPEX and energy assumptions or uncertain coproduct and policy values. This approach follows TEA best-practice emphasis on explicit system boundaries and intended use for early stage/screening comparisons.³⁸ Four representative liquors were considered, Na₂CO₃ (7 and 10 wt%) and K₂CO₃ (20 and 25 wt%), at train-averaged outlet specifications of 90% and 95% CO₂ removal. The 95% values were obtained by linear interpolation between the measured carbonate utilizations at 90% and 100% removal (see Section 3.2). Detailed geography-specific price inputs are provided in Table S3, utilization factors in Table S4, and the underlying calculation assumptions are summarized in the SI.

Fig. 9 Consumables-only (upper-bound) levelized cost of capture (LCOC, USD per tCO₂) for representative Na₂CO₃ and K₂CO₃ liquors at 90% and 95% CO₂ removal, evaluated using four geography-specific reagent price decks (US, EU, Northeast Asia, and Southeast Asia; 2025 Q3). Values are derived from measured carbonate utilization at 90% and 100% CO₂ removal and linear interpolation to 95% (Section 3.2); supporting price inputs, utilization factors, and calculation assumptions are provided in the SI (Tables S3 and S4).

3.4.2 Definition. The consumables LCOC is defined as:where u(spec) is the measured carbonate utilization at the selected outlet specification, and the reagent-price terms correspond to the selected geography-specific price deck. The CaCl₂ term is fixed at 2.522 t tCO₂⁻¹ (1 mol mol⁻¹ basis). The carbonate requirements are 2.408 t Na₂CO₃ or 3.140 t K₂CO₃ per tCO₂, each scaled by 1/u. The H₂O₂ term is modelled as 10 wt% of the carbonate dose (normalized to 100% H₂O₂), and scaling with 1/u. No coproduct credits are applied for the CO₂ product stream, the CaCO₃ product, or brine handling/avoidance; accordingly, the values in Fig. 9 represent a conservative reagent-purchase upper bound rather than a net commercial cost.

3.4.3 Results. Across all four geography-specific price decks, tightening the outlet specification from 90% to 95% increases the consumables-only LCOC for every formulation. This increase directly reflects the reduction in first-stage utilization measured in Section 3.2, whereas the absolute LCOC level varies by geography because the underlying reagent prices differ among the regional decks. Across all four decks, sodium formulations consistently show lower absolute costs owing to their lower stoichiometric mass and lower unit price, whereas potassium formulations are consistently more costly. Notwithstanding this higher absolute cost, the 20 wt% K₂CO₃ case shows a smaller relative penalty upon tightening the outlet specification from 90% to 95%.

Because both the carbonate and promoter terms scale as 1/u, maintaining high first-stage utilization through reactor staging and appropriate stage-swap timing is the primary lever for controlling consumables cost at tighter outlet specifications. Accordingly, despite variation in absolute LCOC across geography-specific reagent price decks, stoichiometric make-up demand emerges as the dominant cost lever within this consumables-only framework. Future TEAs should therefore prioritize strategies that reduce stoichiometric make-up demand, including (i) improving first-stage utilization at tight specifications, (ii) substituting purchased CaCl₂/alkali with low-cost waste brines or by-product alkalinity,³⁴ and (iii) evaluating reagent regeneration/looping^35,36 and explicit credit scenarios for CaCO₃ and brine management.³⁴

4 Conclusions

This work demonstrates an ambient-temperature aqueous CO₂ capture-to-conversion route in which ROS-promoted carbonate absorption is coupled to direct chemical precipitation of CaCO₃, without heat-driven regeneration. Transient ROS formed in situ from H₂O₂-activated alkali carbonates substantially enhance CO₂ uptake at 30 °C, enabling rapid conversion of carbonate to bicarbonate without the need for persistent promoters. The promoter activity diminishes as the alkalinity approaches the HO₂⁻/H₂O₂ equilibrium boundary, marking a self-terminating mechanism that leaves a clean bicarbonate solution suitable for downstream mineralization.

Single-reactor screening defined the concentration and time windows that maximize capacity while avoiding premature precipitation, identifying optimal conditions near 7–10 wt% Na₂CO₃ and 20–25 wt% K₂CO₃. Multi-reactor staging was shown to sustain high CO₂-removal efficiencies while preserving carbonate utilization, thereby reducing the trade-off between outlet specification and conversion. K₂CO₃ systems exhibited greater operational stability and higher conversion than their Na-based counterparts, consistent with their higher solubility and broader homogeneous operating window.

Reaction of the bicarbonate-rich liquors with CaCl₂ yielded phase-pure rhombohedral calcite, confirmed by XRD, ATR-FTIR, SEM, and EDS analyses, regardless of alkali type or reactor configuration. The process consistently produced dense, well-faceted crystallites with no detectable vaterite or aragonite, underscoring the robustness of the capture-to-conversion sequence.

A consumables-only techno-economic analysis across four geography-specific reagent price decks revealed that the levelized cost of capture (LCOC) is strongly governed by carbonate utilization, with higher first-stage conversion and optimized staging providing the most effective means to minimize cost penalties at tighter CO₂ specifications. The present economics are therefore reagent-dominated, with the main cost-reduction levers being improved carbonate utilization and access to low-cost calcium/alkalinity sources and/or reagent looping. These constitute clear priorities for future scale-up studies, alongside scenario-based inclusion of coproduct and policy crediting. Overall, this study establishes a simple, ambient-temperature pathway that integrates accelerated CO₂ absorption with mineralization, eliminates the need for thermal regeneration, and yields a pure solid carbonate product suitable for utilization or storage.

Conflicts of interest

There are no conflicts to declare.

Data availability

All data supporting the findings of this study are available within the article and its supplementary information (SI). Additional underlying datasets, including raw and processed CO₂/O₂ outlet time-series for single- and multi-reactor experiments, original ¹³C NMR, XRD and ATR-FTIR files, SEM micrographs and EDS outputs, and the consumables-only LCOC calculation workbook, are available from the corresponding author upon reasonable request. Supplementary information; supporting SEM/EDS characterization of the CaCO₃ products, including SEM micrographs, EDS spectra, elemental maps, and elemental analysis tables, as well as supporting information for the consumables-only LCOC analysis, including regional reagent price decks, utilization factors, and calculation assumptions. See DOI: https://doi.org/10.1039/d6su00210b.

References

1. S. J. Davis, N. S. Lewis, M. Shaner, S. Aggarwal, D. Arent, I. L. Azevedo, S. M. Benson, T. Bradley, J. Brouwer, Y. M. Chiang, C. T. M. Clack, A. Cohen, S. Doig, J. Edmonds, P. Fennell, C. B. Field, B. Hannegan, B. M. Hodge, M. I. Hoffert, E. Ingersoll, P. Jaramillo, K. S. Lackner, K. J. Mach, M. Mastrandrea, J. Ogden, P. F. Peterson, D. L. Sanchez, D. Sperling, J. Stagner, J. E. Trancik, C. J. Yang and K. Caldeira, Science, 2018, 360, eaas9793.
2. M. R. Allen, D. J. Frame, C. Huntingford, C. D. Jones, J. A. Lowe, M. Meinshausen and N. Meinshausen, Nature, 2009, 458, 1163–1166.
3. S. A. Matlin, G. Mehta, S. E. Cornell, A. Krief and H. Hopf, RSC Sustain., 2023, 1, 1704–1721.
4. M. Bui, C. S. Adjiman, A. Bardow, E. J. Anthony, A. Boston, S. Brown, P. S. Fennell, S. Fuss, A. Galindo, L. A. Hackett, J. P. Hallett, H. J. Herzog, G. Jackson, J. Kemper, S. Krevor, G. C. Maitland, M. Matuszewski, I. S. Metcalfe, C. Petit, G. Puxty, J. Reimer, D. M. Reiner, E. S. Rubin, S. A. Scott, N. Shah, B. Smit, J. P. M. Trusler, P. Webley, J. Wilcox and N. Mac Dowell, Energy Environ. Sci., 2018, 11, 1062–1176.
5. E. S. Sanz-Pérez, C. R. Murdock, S. A. Didas and C. W. Jones, Chem. Rev., 2016, 116, 11840–11876.
6. D. Y. C. Leung, G. Caramanna and M. M. Maroto-Valer, Renew. Sustain. Energy Rev., 2014, 39, 426–443.
7. G. T. Rochelle, Science, 2009, 325, 1652–1654.
8. S. Y. W. Chai, L. H. Ngu and B. S. How, Greenhouse Gases: Sci. Technol., 2022, 12, 394–427.
9. N. Yi, M. Fang, W. Di, Z. Xia, T. Wang and Q. Wang, Environ. Sci. Technol., 2021, 55, 5152–5160.
10. K. Veltman, B. Singh and E. G. Hertwich, Environ. Sci. Technol., 2010, 44, 1496–1502.
11. H. Sun, S. Sun, T. Liu, J. Zeng, Y. Wang, Z. Yan and C. Wu, ACS Catal., 2024, 14, 15572–15589.
12. C. Hepburn, E. Adlen, J. Beddington, E. A. Carter, S. Fuss, N. Mac Dowell, J. C. Minx, P. Smith and C. K. Williams, Nature, 2019, 575, 87–97.
13. F. Barzagli, C. Giorgi, F. Mani and M. Peruzzini, J. CO2 Util., 2017, 22, 346–354.
14. G. Hu, N. J. Nicholas, K. H. Smith, K. A. Mumford, S. E. Kentish and G. W. Stevens, Int. J. Greenhouse Gas Control, 2016, 53, 28–40.
15. H. Knuutila, H. F. Svendsen and M. Anttila, Int. J. Greenhouse Gas Control, 2009, 3, 143–151.
16. T. N. G. Borhani, A. Azarpour, V. Akbari, S. R. Wan Alwi and Z. A. Manan, Int. J. Greenhouse Gas Control, 2015, 41, 142–162.
17. H. Thee, K. H. Smith, G. da Silva, S. E. Kentish and G. W. Stevens, Chem. Eng. J., 2012, 181–182, 694–701.
18. K. Yu, W. A. Mitch and N. Dai, Environ. Sci. Technol., 2017, 51, 11522–11536.
19. S. Shen, Y. Yang and S. Ren, Fluid Phase Equilib., 2014, 367, 38–44.
20. J. T. Cullinane and G. T. Rochelle, Ind. Eng. Chem. Res., 2006, 45, 2531–2545.
21. H. Thee, N. J. Nicholas, K. H. Smith, G. da Silva, S. E. Kentish and G. W. Stevens, Int. J. Greenhouse Gas Control, 2014, 20, 212–222.
22. K. Gustafsson, R. Sadegh-Vaziri, S. Grönkvist, F. Levihn and C. Sundberg, Int. J. Greenhouse Gas Control, 2021, 110, 103434.
23. R. Ramezani and R. Di Felice, Green Energy Environ., 2021, 6, 83–90.
24. P. Wang, A. Pei, Z. Chen, P. Sun, C. Hu, X. Wang, N. Zheng and G. Chen, Nat. Commun., 2025, 16, 1–10.
25. X. M. Hu, H. Q. Liang, A. Rosas-Hernández and K. Daasbjerg, Chem. Soc. Rev., 2025, 54, 1216–1250.
26. E. Shirman and Y. Sasson, Greenhouse Gases: Sci. Technol., 2024, 14, 1037–1048.
27. J. L. Roberts, M. M. Morrison and D. T. Sawyer, J. Am. Chem. Soc., 1978, 100, 329–330.
28. L. J. Csyányi, Z. M. Galbács and L. Horváth, Inorg. Chim. Acta, 1981, 55, 1–4.
29. D. E. Richardson, H. Yao, K. M. Frank and D. A. Bennett, J. Am. Chem. Soc., 2000, 122, 1729–1739.
30. J.-M. Lin and M. Liu, Spectrochim. Acta, Part A, 2009, 72, 126–132.
31. U. Stoin, A. I. Shames, I. Malka, I. Bar and Y. Sasson, ChemPhysChem, 2013, 14, 4158–4164.
32. U. Stoin, Z. Barnea and Y. Sasson, RSC Adv., 2014, 4, 36544–36552.
33. L. Ji, H. Yu, K. Li, B. Yu, M. Grigore, Q. Yang, X. Wang, Z. Chen, M. Zeng and S. Zhao, Appl. Energy, 2018, 225, 356–366.
34. A. Sanna, M. Uibu, G. Caramanna, R. Kuusik and M. M. Maroto-Valer, Chem. Soc. Rev., 2014, 43, 8049–8080.
35. Q. Shu, L. Legrand, P. Kuntke, M. Tedesco and H. V. M. Hamelers, Environ. Sci. Technol., 2020, 54, 8990–8998.
36. D. W. Keith, G. Holmes, D. St. Angelo and K. Heidel, Joule, 2018, 2, 1573–1594.
37. R. Bar and Y. Sasson, Anal. Chim. Acta, 1982, 142, 345–347.
38. A. W. Zimmermann, J. Wunderlich, L. Müller, G. A. Buchner, A. Marxen, S. Michailos and R. Schomäcker, Front. Energy Res., 2020, 8, 5.

---