2.1. Point of zero charge determination
In Fig. 1, the PZC values obtained by the CV, SPECS and SPEIS techniques in the EQCM cell are compared for 0.1 mol L− 1 LiNO3 (Fig. 1a), Li2SO4 (Fig. 1b) and KI (Fig. 1c) with the YP-50F electrode material on the specific capacitance vs. potential plot (for detailed CV and SPECS methodology, see Supplementary information).
The experimental conditions for CV (for 5 and 50 mV s− 1 see Supplementary Information - Fig. S5) and SPECS were convergent, and the currents recorded from these techniques were converted to specific capacitance. For CV, 1 mV s− 1 was applied, and for SPECS, E = 10 mV with t = 10 s, providing an average scan rate of 1 mV s− 1. The PZC is determined as the lowest capacitance (current) value [35] in the intermediate potential range (directly correlated with the capacitance (CCV) calculated from CV technique based on Eq. S2). However, this method is not accurate since the voltammetric response results from the full current response, which is related to diffusion-limited processes and EDL formation. The PZC corresponds to an electrostatic interaction between the electrode and the electrolyte (EDL formation only). Diffusion-limited processes include those related to the oxidation‒reduction of compounds and the intercalation/insertion and residual processes resulting from side reactions (such as electrolyte decomposition). [36] The Faradaic current could cover the EDL response, as shown in the CV profiles (dashed black lines in Figs. 1a-c) near 0.15 V vs. SCE for LiNO3, 0.15 and 0.2 V vs. SCE for Li2SO4 and − 0.3 V vs. SCE for KI. This redox peak could be attributed to a quinone/hydroquinone redox pair grafted onto carbon. [37] Furthermore, as extreme potentials (Emax and Emin) were approached, the box-shaped CV profile was deformed by the increase in current related to the decomposition of the electrolyte.
C SPEIS(1mHz) curves for the LiNO3-based system were not symmetric for both polarization directions (see Supplementary Information – Fig. S6). The same situation was observed for systems with Li2SO4 (Fig. 1b) and KI (Fig. 1c). Furthermore, for the KI redox system, the measurement toward Emin was impossible to carry out correctly due to the high redox activity in the positive potential values (> 0 V vs. SCE).
Contrary to CV (Fig. S5) and SPEIS (Fig. S6), which are burdened with experimental limitations, SPECS is determined to be the most precise technique for PZC identification (Fig. 1; blue squares). The advantage of this technique includes a short implementation step time (in this case, 1 mV s− 1). Moreover, the potential shift is quite gentle and enables detailed data to be recorded in the entire potential range, increasing the recorded data resolution and accuracy. These potential steps lead to a smooth behavior of probable redox reactions and balanced ion redistribution in the pores. The superiority of SPECS over CV resides the possibility of quantifying the capacitive contribution of different charge-storing mechanisms, including that associated with EDL formation (see Supplementary Information). [36, 38]
The PZC in Fig. 1 is not a specific potential value for the ACs, with a clear inflection point on the capacitance curve. Here, we can distinguish a wide potential range (blue region) from − 0.15 to 0.09 V vs. SCE, with a comparable capacitance (82 F g− 1) for the LiNO3-based system (with ± 1% variation of the minimal specific capacitance value), providing a PZC range of 240 mV. For Li2SO4, this region is from − 0.04 to 0.05 V (PZC range of 90 mV) and for KI, this region is from − 0.05 to 0.06 V (PZC range of 110 mV). We propose to determine exact PZC value (if needed) in the middle of this low capacitance region, in these examples at 0 V vs. SCE. Interestingly, an increase in CSPECS(EDL) and CT can also be observed in the potential region where redox occurs (for all tested electrolytes). Since redox induces charge transfer in the specific adsorbed ions, it affects the formation of the EDL itself. Moreover, this response cannot be differentiated from CEDL, and an ambiguous increase is observed and is most likely related to quinone/hydroquinone redox activity.
The abovementioned SPECS technique provides insight into the charging mechanism of electrochemical capacitors by differentiating the capacitance contribution (see Supplementary Information), as shown in Fig. 2.
The calculated total capacitance (CT) of the system is identical to the CCV (Fig. 2) when identical measurement conditions are maintained (in this case, the average scanning rate of 1 mV s− 1). As expected for porous carbon material, the main contribution to the CT is made by porous capacitance (CP) due to the strongly developed specific surface area of the AC used (Fig. S1). The CP curves show a butterfly shape that is usually visible for organic systems, [26, 35] and fest repolarization processes are shown in the box-like shape. The geometric capacitance (CG) share is significantly lower for capacitive systems, i.e., LiNO3 (Fig. 2a) and Li2SO4 (Fig. 2b), resulting from the poorly developed electrode surface in direct contact with the electrolyte. Interestingly, for the redox-based system (Fig. 2c), CP and CG are almost equal in the potential range lower than PZC, in the capacitive region. Unfortunately, the computational method does not allow one to make such predictions (CP and CG) for redox activity region; thus, it was not applied. In the studied systems, the smallest capacitive contribution comes from ion diffusion (CD), showing that the main charge storage mechanism is based on EDL formation. Moreover, residual capacitance (CR) is responsible for any discrepancies from the ideal box-like shape characteristic for a capacitive response; the value of which increases in those potential regions where, for example, electrolyte decomposition, specific ion adsorption or redox processes occur. CR is responsible for the overall CT cyclic voltammogram shape deviating from the ideal CP capacitive curve.
Our study also included the influence of cell construction on the PZC values (see Supplementary Information), as various approaches are used in the literature. Unfortunately, a potential shift for electrochemical behavior was observed (Fig. S7) and resulted from different uncompensated resistance factors (Fig. S8). In addition, we placed the reference electrode in a spotlight (Fig. S9). In contrast to overall cell construction (volume of electrolyte, working electrode loading, distance between electrodes, etc.), it had a negligible influence; thus, for short-term experiments, no Cl− migration from the SCE electrode was tracked. The electrochemical tests of neutral-pH aqueous ECs were not susceptible to the REF type. However, cell construction had a large impact on individual electrode stability.
Notably, all tested aqueous solutions are characterized by a wide PZC potential range (Fig. 3) when determined using the SPECS method (min. capacitance value ± 1%). The LiNO3-based system displays mostly capacitive charge storage behavior [33, 41] and also possess the widest PZC potential range (~ 240 mV). Thus, the region of ion reorganization/mixing is not limited by the potential value itself, and both the Li+ cations and the NO3− anions have the same affinity toward the electrode. In contrast, KI is a pure redox system [42–45] characterized by a narrow PZC potential range (110 mV). Beyond this narrow potential range, other mechanisms occur, such as specific ion adsorption on the electrode surface. [8] Considering these facts and the ambiguous Li2SO4 charge storage mechanism, [34] it can be stated that it is a pseudocapacitive electrolyte. Its PZC range is rather narrow (90 mV), which can indicate SO42− specific adsorption on the electrode surface. Additionally, the SO42− ion flux during positive electrode charging has been excluded thus far. [34, 46]
Studies on the PZC determination methods are rather limited in the literature; however, authors usually relate it to one specific measure [25–27]: minimum conductance value or minimum capacitance value. These values are obtained by different techniques, varying in the experimental setups and experimental electrochemical conditions. For the ACs studied, YP-50F and DLC30, the wide PZC region is thought to result from both electrolyte/electrode matching and active solvent molecules during EDL formation. [47] AC has a wide variety of surface functional groups, high tortuosity, 3-dimensional porosity, and specific structure. Planar metallic electrode (without defects and with low specific surface area) correlates to a wide PZC region (Fig. S10), like wide potential window for ionic liquid on glassy carbon electrodes. For the two ACs studied, the PZC region is much narrower than that for the planar resonator (Fig. S11). However, for those materials studied with D2O, no PZC difference is observed, confirming a particular electrolyte ion/electrode interaction when inorganic salt is present.
2.2. EQCM data evaluation
The previous section addressed the proper and accurate PZC determination. However, in addition to the PZC determination for 0.1 mol L− 1 LiNO3, Li2SO4, KI and 0.1 mol L− 1 LiNO3 in D2O (Fig. 3), the significance of these data has not yet been quantified. Therefore, basic electrochemical tests (cyclic voltammetry at 5 mV s− 1) were performed to study ion flux onto the carbon electrode during the charging process (validation of experimental set-up is presented in Fig. S12). To thoroughly interpret the influence of the PZC value on the evaluation of the data, a Li2SO4-based system was selected; this system: i) is broadly described in the literature, ii) has a complex charging mechanism and iii) has promising electrochemical performance (very long cyclability), as presented in Fig. 4.
After determining PZC using the techniques mentioned (average PZC = 0 V), mass change calculations were performed to establish PZC and two other artificially imposed values (PZC1 and PZC2). This results in a recalculation where the PZC is the following: (1) -300 mV from PZC (PZC1) and resembles the PZC determined in the Swagelok cell (Fig. S7), (2) PZC = 0 V vs. SCE and is considered as the proper value determined for the studied system (Fig. 1) and (3) + 300 mV from PZC (PZC2). In total, this provides a wide range of potential ΔE = 600 mV in which PZC is considered, imposed, and discussed.
In Fig. 4a, the hydrogen evolution potential (HER) is shown; the HER of this electrolyte equals − 0.681 V vs. SCE, and oxygen evolution potential (OER) of this electrolyte equals + 0.549 V vs. SCE. In accordance with Faraday’s law (Eq. 1), the mass change (related to dissolution or deposition) is linearly correlated with the amount of charge passing through the electrochemical setup. From the slopes of the mass:charge curves (Fig. 4b-c), both adsorption and desorption data show multistep processes ongoing at the electrode/electrolyte interface (determination of ionic species is done based on Tab. S3). According to the literature, one can assume the adsorption of Li+ as well as its solvated forms, i.e., Li+ ∙ 2H2O (with Δm = 42.97 u) and OH− (with Δm = 17.01 u), during negative and positive polarizations, respectively (Tab. S3). [48] Fig. 4b shows that, the adsorption process is followed by the desorption process. In Fig. 4c two-step process is observed only for PZC1, with an initial desorption and then adsorption of mass difference (15 u). For negative polarization, the ionic species with a molecular weight between Li+ and hydrated Li+ ∙ 3H2O are responsible for the mass change. However, for positive polarization, where SO42− or OH− are anticipated, neither was found. Therefore, the processes that occur at the positive electrode/electrolyte interface are not related to single ion adsorption. This is potentially related to the carbon oxidation process that can continuously occur, such as specific SO42− ion adsorption on the YP-50F electrode, SO42− redox reactions [57] or ion reorganization. Thus, the matching of ion flux for sulfate-based electrolytes is not straightforward.
The electrolytic solution of 0.1 mol L− 1 LiNO3 was studied using EQCM for the first time (Fig. 5). This electrolyte has been described as a capacitive electrolyte (even at low concentrations); [33] therefore, ion fluxes based on Li+ and NO3− were anticipated.
Since the negative polarization range was strongly extended from the PZC in all measurements, LiNO3-based systems display adsorption followed by desorption while approaching increasingly lower potential values. When the 0.1 mol L− 1 solution was prepared in DI water, no nitrate anion flux was detected (Fig. 5a), thus leading to the conclusion that the NO3− anion, like SO42−, was more prone to redox processes than EDL formation. However, the same solution was prepared using D2O (Fig. 5b) – characterized by a higher molecular weight (see Supplementary Information). In the second case, NO3− adsorption and desorption were recorded, accompanied by an opposite sorption process of solvated cation (Li+ ∙ 2D2O). PZC region in this case shifts towards more positive potential values, slightly narrowing its potential range (Fig. 3). Solvation shell of lithium cation seems to be reasonable when combined with an acid radical at this concentration. Moreover, a good correlation between the PZC values determined via SPECS and SPEIS and the mass change curve bending can be observed from the mass change curves recorded by EQCM. This result further confirmed the necessity of proper PZC determination and its insight into the charging mechanism; dividing the mass change curve properly for different regimes where either cation or anion adsorption is the leading process in EDL formation. These experiments confirmed that the nitrate-based electrolyte is capacitive and that NO3− ion flux can be detected. When H2O was used as a solvent, the OH− anion was more likely to form the EDL than NO3− was. These results also showed that pH, conductivity, ionic species (type and their activity), and solvent molecules influenced the charging mechanism in aqueous-based EC, emphasizing the importance of treating each system individually.