Introduction

Proton transfer in an aqueous environment is of fundamental importance for many physical, biological, chemical, and energy-related processes1,2. This process has been intensively investigated in acids by physical chemists to explain the anomalously high mobility of hydrated excess protons3,4,5. As suggested by von Grotthuss6,7,8, a “structural diffusion” mechanism requires a proton hopping through water that functions as a bridge, leading to a molecular-level description of hydrated proton intermediates, for instance, the Eigen cation H9O4+ and the Zundel cation H5O2+ complexes9,10,11. In contrast to the acidic aqueous solution with abundant solvated proton complexes, the neutral aqueous solution is a proton-deficient condition. Whether or not this neutral aqueous solution involves massive proton transfer has not received sufficient attention. This may be due to the assumption considered intuitively unlikely.

It should be noted that the life activities of plants and animals are conducted under near-neutral conditions and mostly buffered with naturally available bicarbonate (HCO3–) system12,13,14. For example, natural photosynthesis involving oxygen evolution reaction provides protons that transfer via proton channels for the facile synthesis of carbohydrates in the presence of CO2/HCO3– 15. Protons as messengers of intercellular communication in the nervous system can be regulated by the HCO3–/H2CO316. In addition, biomineralization occurs by mediating pH for the formation of calcified structures17,18. In the last decades, converting greenhouse CO2 to chemical feedstock through various catalytic processes, for instance, electrocatalysis has been implemented in CO2/HCO3–-buffered electrolytes19,20,21,22.

When approaching the analysis of the CO2/HCO3– system, physical chemists, physiologists, and electrochemists tend to focus on different aspects of their interests, but the basic principle lies in Bjerrum plot23,24,25,26, where the equilibrium between CO2, HCO3– and CO32–, as well as the rates of their interconversion, is governed by the solution pH (Supplementary Fig. 1). Based on the dissociation constant (pKa), the extra formation of either CO2 or CO32– depends on the variation in pH (Fig. 1a).

$${{{{\rm{HCO}}}}}_{3}^{-}+{{{{\rm{H}}}}}^{+}\leftrightharpoons {{{{\rm{H}}}}}_{2}{{{\rm{O}}}}+{{{{\rm{CO}}}}}_{2}$$
(1)
$${{{{\rm{HCO}}}}}_{3}^{-}+{{{{\rm{OH}}}}}^{-}\leftrightharpoons {{{{\rm{H}}}}}_{2}{{{\rm{O}}}}+{{{{\rm{CO}}}}}_{3}^{2-}$$
(2)
Fig. 1: Dissociation of HCO3–.
figure 1

The schematic of conversion of HCO3– via a acid–base route and b HCO3– dissociation. c Spontaneous pH increases in 3.0 M KHCO3 solution at 25 °C. The error bars indicate the standard deviation of three independent experiments. d CO2 release at 25 °C was measured by online differential mass spectroscopy (m/z = 44 for CO2) in 3.0 M KHCO3 solution.

Full size image

Otherwise, Eqs. (1) and (2) remain dynamic equilibrium for the reason that only the auto-ionization or hydrolysis of H2O provides rare H+ and OH– under neutral conditions. This well-established view reasons that HCO3– solution could resist change in pH, yet how proton transfer could happen should be beyond this scope.

Another fact is that both CO32– and HCO3– always coexist with cations in aqueous solutions for electroneutrality, for instance, alkali metal cations, such as Na+ or K+. For example, these cations in the electrolyte, as shown in combined ab initio/continuum simulation studies27, significantly affect the catalytic conversion rate in critical electrochemical processes. Relatively little is known about the effects of metal cations associated with the acid–base process for HCO3– anions. As of now, the crucial experimental evidence for the functions of metal cations in the seemingly simple yet mysterious HCO3– system is missing and needs to be addressed urgently.

In this study, we first present that the HCO3– spontaneously converts into CO2 and CO32– anions under ambient conditions tracked by using online differential mass spectroscopy and Raman spectroscopy. Then we show that ~100% of HCO3– and 81% of CO32– have an overall exchange of oxygen isotopes with H2O within 120 min at ambient conditions, although the optimal probability of overall exchange of oxygen isotopes is only 12.5%. In contrast to the acid or base-induced reactions in Eqs. (1) and (2), we find that the electrostatic interactions between hydrated cations and HCO3– facilitate a pH-independent dissociation of HCO3–, which can be seen as a reverse reaction of the acid–base neutralization (Fig. 1b).

$$\{{{{{\rm{HCO}}}}}_{3}^{-}\cdots {{{\rm{Cation}}}}\}\leftrightharpoons \{{{{{\rm{OH}}}}}^{-}\cdots {{{\rm{Cation}}}}\}+{{{{\rm{CO}}}}}_{2}$$
(3)
$${{{{\rm{HCO}}}}}_{3}^{-}+\{{{{{\rm{OH}}}}}^{-}\cdots {{{\rm{Cation}}}}\}\leftrightharpoons {{{{\rm{H}}}}}_{2}{{{\rm{O}}}}+\{{{{{\rm{CO}}}}}_{3}^{2-}\cdots {{{\rm{Cation}}}}\}$$
(4)

Unlike Eq. (1), where the conversion of HCO3– to CO2 depends on the neighboring H+ or H2O-ionized H+, here the formation of CO2 in Eq. (3) relies on the kinetics of HCO3– dissociation and the concentration of HCO3– anions. While the generation of CO32– in Eq. (4) is a resulting favorable molecular-level acid–base process. If the starting concentration of HCO3– is high, the dominant HCO3– dissociation results in a continuous generation of both CO2 and CO32–. We thus unveil a dramatic interconversion starting from HCO3– dissociation, followed by massive proton transfer via an H2O + CO2 ↔ HCO3– ↔ CO32– loop, reflecting a dramatic isotopic oxygen exchange. To understand the trend of alkali cations, we have screened K+, Rb+, and Cs+ cations, and found that hydrated K+ performs the best. In addition, isolating HCO3– from K+ dramatically impedes the HCO3– dissociation and oxygen exchange. This study shows evidence that cation enables HCO3– to produce acid–base encounter pairs acting as a driving force for proton transfer.

Results and discussion

Kinetics for HCO3 – dissociation

To investigate the kinetics for HCO3– dissociation, temperature-dependent Raman spectroscopy was implemented from 5 to 80 °C within 60 min in 3.0 M KHCO3 aqueous solution (Supplementary Figs. 2, 3). The specific vibrations for CO32– and HCO3– species can be detected, while the peak-area ratio between CO32– and HCO3– indicates the solution pH value28. The prominent peak at 1016 cm−1 was ascribed to the νC–OH vibration of HCO3–, while the peak at 1064 cm−1 was assigned to the νC–O vibration of CO32– (Supplementary Fig. 4). At 5 °C, the ratio of CO32–/HCO3– keeps identical thus a stable system. Interestingly, upon increasing the temperature to 25 °C, the ratio of CO32–/HCO3– increased over time, from an initial value of 0.06–0.13 within 60 min (Supplementary Fig. 5). This observation indicates a spontaneous conversion of HCO3– to CO32– at ambient conditions. This enhanced CO32–/HCO3– ratio correlates with an increased bulk pH from 8.5 to 9.0 (0.5 units) (Fig. 1c and Supplementary Fig. 4). Moreover, higher solution temperature leads to a higher dissociation rate, thus higher solution pH (Supplementary Fig. 6).

The increase in solution pH is most probably due to the dissociation-induced release of CO2, as described by Eq. (3). To confirm this, online differential mass spectroscopy was employed to monitor the CO2 gas. A decent signal at m/z = 44 became salient, with an increase in temperature from 5 to 25 °C, while decreasing along with cooling down the KHCO3 solution (Fig. 1d). As a result, the formation of CO32– while evolving CO2 gas can be observed at room temperature. If the freshly generated CO32– cannot be consumed by hydrated CO2 or H2CO3 as shown in Eq. (5), the increase in solution pH is unavoidable regardless of whether the HCO3– solution is in an open or closed system (Supplementary Fig. 7). Thus, bubbling CO2 through the HCO3– solution can decrease CO32– content while bubbling Ar can accelerate the alkalization (Supplementary Fig. 8).

$$\left\{{{{{\rm{CO}}}}}_{3}^{2-}\cdots {{{\rm{Cation}}}}\right\}+{{{{\rm{H}}}}}_{2}{{{{\rm{CO}}}}}_{3}\leftrightharpoons 2\{{{{{\rm{HCO}}}}}_{3}^{-}\cdots {{{\rm{Cation}}}}\}$$
(5)
$$\left\{{{{{\rm{OH}}}}}^{-}\cdots {{{\rm{Cation}}}}\right\}+{{{{\rm{H}}}}}_{2}{{{{\rm{CO}}}}}_{3}\leftrightharpoons \left\{{{{{\rm{HCO}}}}}_{3}^{-}\cdots {{{\rm{Cation}}}}\right\}+{{{{\rm{H}}}}}_{2}{{{\rm{O}}}}$$
(6)

In addition, downsizing the water size facilitates the release of CO2, thereby lowering the reverse reaction between hydrated CO2 and CO32–, resulting in higher CO32– content in Eqs. (4)–(6). This suggests that the smaller the size of a water droplet, the faster its alkalization rate (Supplementary Fig. 9).

Elevating the concentration of HCO3– anions should have faster initial hydrolysis and then advance the conversion of HCO3– to CO32– and CO2 if the view of HCO3– dissociation dominates. As shown in Supplementary Fig. 10, higher HCO3– concentration gets a higher conversion rate. These kinetic factors can prove that the HCO3– is not stable at ambient conditions and tends to spontaneously convert to CO32– and CO2.

Dramatic oxygen exchange reflects proton transfer

If we assume that this is just a simple HCO3– dissociation, leading to CO2 release and an increase in CO32– content, we may miss out on interesting findings. To provide molecular-level insight into the dissociation process, time-dependent Raman spectra were recorded in 3.0 M KHCO3-containing H2O at 25 °C. The vibrations at 633, 672, and 1016 cm−1 were respectively ascribed to γCO-H, δCO2, and νC–OH for HCO3– (Supplementary Fig. 11a), while the vibration at 1064 cm−1 was assigned to νC–O for CO32– anions29,30. Adding 3.0 M KHCO3 in H218O caused a red shift of the isotopic oxygen groups and isotopic splitting of νC–O for CO32– (Supplementary Figs. 11, 12 and Supplementary Note 1). For instance, relative to νC–OH at 1016 cm−1, the νC−18OH for HC18O3– was fast produced and stabilized at 970 cm−1. Meanwhile, isotopic splitting of νC-O was observed with C18O32– at 1005, C18O2O2– at 1025, and C18OO22– at 1045 cm−1 for CO32- 31,32, while the νC–16O peak disappeared at 1064 cm−1 for CO32– after 120 min of solution ageing (Fig. 2a). These results show that ~100% of HCO3– has an overall exchange of oxygen isotopes with H2O while ~81% of CO32– performs the same (Fig. 2b). Further, both crystallizing KHC18O3 from KHC16O3 in H218O and re-dissolving the re-crystallized KHC18O3 in H216O forming KHC16O3 evidenced the oxygen exchange behavior (Supplementary Figs. 13, 14). Notably, the oxygen exchange was not observed in the 3.0 M K2CO3 solution owing to little HCO3– as a critical medium (Supplementary Fig. 15).

Fig. 2: Oxygen exchange in HCO3– aqueous solution.
figure 2

a Time-dependent Raman spectra of 3.0 M KHCO3 in H218O at 25 °C. b The fractions of different carbon species in 3.0 M KHCO3, 3.0 M RbHCO3, and 3.0 M CsHCO3 within 2 h of solution ageing.

Full size image

The oxygen exchange has to be accompanied by proton transfer (Supplementary Fig. 16). Since the oxygen exchange only takes place at the CO2 hydration step, the optimal probability of complete oxygen exchange calculated is only 12.5%, assuming that the 18O abundance of H218O is 100% (Supplementary Figs. 17, 18). Even the solubility of CO2 in an aqueous solution is only about 30 mM, two orders of magnitude less relative to the applied concentration of HCO3–, ~100% of HCO3– and ~81% of CO32– have completed oxygen exchange suggesting the multiple proton transfer processes via H2O + CO2 ↔ HCO3– ↔ CO32– loop (Supplementary Fig. 19). This is unexpected under such a proton-deficient neutral condition.

HCO3 – dissociation in the hydrated K+ shell

To confirm whether the counter-cation is innocent, 3.0 M MHCO3 (M = K+, Rb+, and Cs+) was added into H218O for time-dependent Raman spectra. Cation-dependent HCO3– dissociation and oxygen exchange were observed (Supplementary Fig. 20). The ratios of CO32–/HCO3– in the presence of K+, Rb+, and Cs+ are 0.57, 0.48 and 0.42, respectively. In addition, the highest concentration of 18O labeled HC18O3– and C18O32– species was observed in KHCO3 solution, following the order: K+ > Rb+ > Cs+ (Fig. 2b and Supplementary Fig. 21). Considering the limited solubility of LiHCO3 and NaHCO3, the oxygen exchange of HCO3- in 1.0 M LiHCO3, NaHCO3, KHCO3, RbHCO3, and CsHCO3 in H218O was also examined. As shown in Supplementary Figs. 22, 23, the impact of K+ on the dissociation of HCO3– was particularly pronounced. Unlike the acid–base route33, the cation identity plays a key role in the conversion between CO2, HCO3– and CO32–.

To elucidate the role of K+, we employed 18-crown-6 (18C6) as a chelating agent to isolate the influence of K+ 34,35. To preclude the Raman signal interference from 18C6 (Supplementary Fig. 24), the vibrations of γCO-H and δCO2 were used to investigate the isotopic oxygen exchange. In the absence of 18C6, the γCO-H and δCO2 of HCO3– exhibited a red shift ~27 cm−1 due to the isotope effect (Fig. 3a). However, with 18C6, the γCO-H and δCO2 vibrations remain almost unchanged (Fig. 3b). Interestingly, unlike that bubbling Ar can accelerate the solution alkalization owing to CO2 release in the absence of 18C6 (Supplementary Fig. 8), with 18C6 in HCO3– solution, bubbling Ar failed to increase the CO32–/HCO3– ratio (Supplementary Fig. 25), implying that no significant dissociation of HCO3– into CO2 that could escape from the solution. We thus hypothesize that the electrostatic interactions between hydrated cations and HCO3– lead to the dissociation of HCO3– into CO2 and OH–.

Fig. 3: The role of K+ in HCO3– dissociation.
figure 3

a Time-dependent Raman spectra of 3.0 M KHCO3 without 18C6 and b with 18C6. c RDF and integrated RDF of C–O in KHCO3 (denoted as with K+) and HCO3 (denoted as without K+) at 25 °C. d Simplified sketch showing the interplay between HCO3− and K+ in the hydrated K+ shell.

Full size image

Ab initio molecular dynamic simulations were conducted to support our hypothesis (Fig. 3c). Indeed, the C–OH bond of HCO3– has been stretched in the presence of K+, where the HCO3– dissociation happens (Supplementary Movie 1 and Supplementary Figs. 25–27). In contrast, without K+, HCO3– remains relatively stable, although the hydrolysis and ionization of HCO3– may somehow take place (Supplementary Movie 2 and Supplementary Figs. 28–30). The radial distribution function (RDF) peak for C–O of HCO3– at 1.28 Å splits into two peaks at 1.26 and 1.37 Å (Fig. 3c). Notably, the integrated coordination number for C–O at 1.37 Å is 1, suggesting that a third of C–O has been elongated, consequently resulting in the dissociation of the C–OH bond within the first solvation shell of K+ at 25 °C (Fig. 3d). Additionally, the coordination situation of the water molecules and HCO3– in the first solvation shell of K+ was summarized in Supplementary Fig. 31a. The probability of five water molecules and one HCO3- anion occupying the first solvation shell was pronounced in KHCO3 electrolyte at 25 °C. Consequently, we could conclude that the K+ was hydrated by simultaneously surrounding H2O and HCO3–, and the K+ induced the dissociation of HCO3– (Supplementary Fig. 31b).

HCO3 –-mediated proton transfer requires cations

To prove the interaction between hydrated K+ and HCO3– that contributes to proton transfer, we designed a local change of pH at the electrode/solution interface through proton-producing oxygen evolution reaction on a rotating ring-disk electrode (RRDE) (Fig. 4a). To ensure an identical change in local pH for comparison, we applied the same current density on the model Co3O4 disk electrode (Supplementary Figs. 32–35). The local pH at the disk electrode interface can be in situ measured by the pH-sensing IrOx ring electrode and quantified based on the reaction–convection–diffusion theory36,37 (Supplementary Figs. 36, 37).

Fig. 4: HCO3–-mediated proton transfer with K+ cations.
figure 4

a Schematic of RRDE measurements in 3.0 M KHCO3 with/without 18C6. b Ring open circuit potential and corresponding pH response in 3.0 M KHCO3 in the presence/absence of 18C6 at 25 °C when different current density was applied on the disk electrode. The resistance was 6 ± 0.5 Ω in 3.0 M KHCO3 and 50 ± 1.0 Ω in 3.0 M KHCO3 containing 3.0 M 18C6.

Full size image

To demonstrate the pH trend, a set of current densities at 1.0, 5.0, and 10.0 mA cm−2 have been applied, with and without 18C6 in 3.0 M KHCO3 solution (Fig. 4b). The pH changes on the ring electrode mirror the modulation of applied current density, and it was shown that the higher the current density, the greater the pH change. It is different that without 18C6 to isolate the influence of K+, very little pH change can be detected. While, in the presence of 18C6, the K+-isolated HCO3– accumulates protons at the electrode interface. For instance, at 10.0 mA cm−2, a difference in measured potential ∆E (6 versus 106 mV) on the ring electrode has been determined (Fig. 4b), which correlates to the difference in pH variations (0.1 units versus 1.8 units) owing to the effect of K+ cations (Supplementary Fig. 38). This is due to the formation of acid–base encounter pairs allowing proton transfer to be efficient, but the K+-isolated HCO3– alone is not the case so that cation-chelated HCO3– electrolyte exhibits substantial proton accumulation at the electrode interface (Fig. 4a).

The discovery of the role of cations in HCO3–-mediated proton transfer urges us to re-examine the underlying mechanism of a generally noted HCO3– in natural and artificial aqueous systems. Although accepting or donating protons is a well-known feature for HCO3– anions, we believe that the cation-induced HCO3– dissociation is the driving force for the proton transfer and the fast interconversion between CO2, HCO3–, and CO32–. This study highlights that the HCO3– anions can supply OH– species and CO2 assisted by the coordination of cations. Our findings have the potential to provide additional insights into the electrochemical CO2 reduction regarding interfacial CO2 feeding, interfacial pH and cation effect27,38,39. Beyond the well-known dissociation constant (pKa), metal cation should be considered as a universal descriptor for the HCO3–-involved processes in the fields of respiratory, mineralization, and catalysis.

Methods

Overview of experiments

We employed multiple analytical techniques to investigate the dissociation and oxygen exchange of bicarbonate (HCO₃−) in H218O or H₂¹⁶O aqueous solutions containing LiHCO₃, NaHCO₃, KHCO3, RbHCO3, or CsHCO3. Raman spectroscopy was used to study the dissociation and oxygen exchange of HCO3− in various bicarbonate solutions. Online differential mass spectrometry was applied to detect CO2 gas in the HCO3− aqueous solution. Rotating ring-disk electrode tests were conducted in 3.0 M KHCO₃ solutions, with and without 18C6, to reveal the role of K⁺ in HCO3− dissociation and HCO3−-mediated proton transfer. Ab initio molecular dynamics simulations were performed to further support the crucial role of cations, considering the systems with/without cations and with different types of cations.

Chemicals

High-purity NaHCO3 (99.99% trace metal basis), KHCO3 (≥99.95% trace metals basis), CsHCO3 (99.9%), K2CO3 (99.995% trace metals basis), Na2SO4 (≥99.0%), H2SO4 (99.999%), and 18-crown-6 (18C6) (≥99.0%) were purchased from Sigma-Aldrich. LiHCO3 and RbHCO3 were prepared from Li2CO3 (99.99% trace metal basis, Sigma-Aldrich) and Rb2CO3 (99.8% trace metals basis, Sigma-Aldrich), separately. H218O (18O abundance of 99%) was obtained from Sigma-Aldrich. CO3O4 (99.99%) was purchased from Aladdin. K3IrCl6 (99.99%) was procured from Alfa Aesar. All the solutions without isotopes were freshly prepared using the ultrapure deionized water (18.2 MΩ cm−1 at 25 °C, Milli‐Q).

Raman spectroscopy

Raman spectra were recorded with an XploRA PLUS confocal micro-Raman spectrometer (Horiba Jobin Yvon) equipped with a 532 nm Nd-YAG laser and a ×50 objective. The detailed instrumental parameters were as follows. The filter was 50%, the grating was 1200 (750 nm), the slit was 100 μm, the hole was 100 μm, the acquisition time was 30 s, the accumulation time was 2 times, and the range was 100–4000 cm−1. The laser and grating were calibrated by Si wafer before measurements. Each Raman spectrum was obtained by signal focusing mode (RTD time is 1.0 s). The Raman peak was fitted in the Gaussian–Lorentzian function.

Online differential mass spectrometry

Online differential mass spectrometry characterization was carried out using a Hiden HPR-40 quadrupole mass spectrometer (Hiden Analytical). The product information was acquired from KHCO3 solutions along with the temperature changes. The temperature of the KHCO3 solution was calibrated by a temperature gauge. During the process, ionization of the detected species was achieved using an electron energy level of 70 eV, with an emission current of 400 μA. The detected species were then analyzed using a secondary electron multiplier, with a detector voltage set at 1200 V.

Rotating ring-disk electrode tests

The rotating ring-disk electrode (RRDE) tests were conducted on a Pine Research Instrumentation, which consisted of a glass carbon disk with a disk diameter of 5.0 mm and an Au ring with an inner diameter of 6.5 mm and an outer diameter of 7.5 mm. The disk electrode was loaded with a commercial Co3O4 catalyst. Typically, 10 mg of Co3O4 catalyst was dispersed in 1.0 mL of a mixed solvent consisting of 500 μL ethanol, 450 μL H2O, and 50 μL Nafion117 solution. After 60 min of ultrasonication, 10 μL of the electrocatalyst ink was cast onto the glassy carbon disk and dried in the air. The mass loading of the Co3O4 catalyst was 0.5 mg cm−2, based on a geometric electrode area of 0.196 cm². The IrOx ring-pH sensing electrode was prepared in a standard three-electrode system by a Biologic potentiostat (SP-300), according to the previous report40. Ir metal was first electrodeposited onto the Au ring electrode in a 50 mL neutral solution of 4.0 mM K3IrCl6 and 0.5 mM Na2SO4 at 60 °C. An Ir electrode was used as the counter electrode and Ag/AgCl as the reference electrode in a single-compartment cell. The RRDE was rotated at 2500 rpm while a constant potential of −0.3 V vs. Ag/AgCl was applied for 6 min, followed by cycling the potential 50 times between −0.75 and 0.2 V vs. Ag/AgCl at 100 mV s−1. Afterward, the Ir/Au catalyst film was taken out and cleaned with deionized water. The conversion of electrodeposited Ir metal into IrOx was performed in 50 mL of 0.5 M H2SO4 using cyclic voltammetry. The potential range was set between −0.66 and 0.73 V vs. Hg/HgSO4 reference electrode for 300 cycles at 0.2 V s−1, with an Ir electrode as the counter electrode. The RRDE was rotated at 2500 rpm during the process. Finally, the RRDE was removed from the solution, cleaned with deionized water, and placed in a 60 °C oven for 1 h.

For the RRDE tests in a single-compartment cell, the disk and ring were simultaneously controlled by a Biologic potentiostat (SP-300). The Co3O4 disk served as the working electrodes and the IrOx ring acted as a sensor, while a Pt electrode was used as the counter electrode and two Ag/AgCl electrodes as the references. The electrolyte consisted of 50 mL of 3.0 M KHCO3, with or without 3.0 M 18C6. Chronopotentiometry was performed on the disk electrode at current densities of 1.0, 5.0, and 10 mA cm−2, respectively, while the open circuit potential of the ring was recorded. For the calculation of pHRing, according to Supplementary Fig. 36, the ring pH was determined from the OCPring. All electrochemical measurements were carried out at 1600 rpm in an Ar (99.999%) purged system. Electrochemical impedance spectroscopy (EIS) was conducted in a standard three-electrode system, with a frequency scan range from 100 kHz to 100 MHz and a sinusoidal wave amplitude of 10 mV. The measured resistance was 6 ± 0.5 Ω in 3.0 M KHCO3 and 50 ± 1.0 Ω in 3.0 M KHCO3 containing 3.0 M 18C6.

Molecular dynamics (MD) simulations

This work applied the Vienna ab initio simulation package (VASP) code for all calculations. The electronic exchange-correlation energy was described through the Perdew–Burke–Ernzerhof functional within the generalized gradient approximation (GGA). The projector-augmented wave (PAW) method was used to describe the ionic cores. An energy cutoff of 450 eV was utilized for a plane wave basis set. K-points were set to 1 × 1× 1 for structure optimization and energy calculations. The calculations proceeded until the force and energy converged to <0.02 eV Å−1 and 10−5 eV, respectively.

For the KHCO3 systems, simulations were performed within an (11.87 × 11.87 × 11.87) ų box containing 50 water molecules and 2 KHCO3. Ab initio molecular dynamics (AIMD) simulations were conducted to assess the formation of CO2 and OH– in the canonical ensemble (NVT) with a Nose Hoover thermostat. The simulations were run at 298 K for 5 ps, using a time step of 1 fs. For the system without K+, simulations were conducted in the same box with 50 water molecules and 2 HCO3 to assess CO32– and H+ formation under the same conditions. During the AIMD simulations, the system remained electroneutral.