Characterization of self-doped BM-TNA and BP-TNA electrodesFigure 1(a) and (b) show the horizontal and cross-sectional FE-SEM images of the growth shape of the self-doped BM-TNA after annealing on a Ti mesh at 600 °C. The structure of the open TiO2 nanotube array exhibited approximate outer and inner diameters of 138.0 and 75.3 nm, respectively, and a length of 9.7 μm. Similarly, Fig. 1(c) and (d) show the FE-SEM images of the self-doped BP-TNA annealed at 600 °C on a Ti plate. The results suggested that the anodization potential and annealing temperature were the key parameters in determining the crystallographic structure of the TiO2 surface during the synthesis of BM-TNA and BP-TNA. In addition, the XPS results shown in Fig. 1(e) indicate Ti2p and O1s peaks with high binding energy. According to the visualization, the peaks for Ti2p and O1s were equally identified at 459.5 eV and 530.75 eV, respectively, for both the BP-TNA and BM-TNA electrodes. Figure 1(f) shows that the anatase and rutile peaks as the crystal structure were annealed in the order of XRD. This structural aspect was described in a previous study40, and anatase peaks formed at 300 and 450 °C. In addition, mixed peaks (i.e., anatase and rutile) were observed at 600 °C. The behaviors of the self-doped BM-TNA, BP-TNA were confirmed through the applied cathodic reduction followed by the anatase peak at 450 °C, and the anatase and rutile peaks at 600 °C for the TiO2 surface structure according to the change in annealing of the catalyst. These self-doped materials had excellent photochemical efficiency at peaks consisting of anatase and rutile peaks, which hindered reconversion.Fig. 1: Characterization of the BM- and BP-TNA electrodes via SEM, XPS, and XRD analyses.SEM images of a, b BM-TNA and c, d BP-TNA; e XPS signals and f XRD intensities. The length of the scale bars are as follows: a, c 500 µm; b, d 5 µm.Full size imageEffect of the novel catalyst and photoelectrochemical activity via UV-lights sourceTo explore the potential of the catalyst, we evaluated the oxygen evolution reaction (OER) properties by comparing the activities of the BM-TNA and BP-TNA catalysts, as shown in the electrochemical impedance spectroscopy (EIS) results (Fig. 2(a)). EIS analysis confirmed the occurrence of charge transfer, which was evaluated under an open circuit potential with an amplitude of 10.0 mV over a scan frequency of 100.0–10.0 kHz in brackish water (i.e., 3000 mg L−1). The BM-TNA exhibited a much higher transfer resistance than the BP-TNA catalysts, indicating the rapidly induced electron transfer efficiency of the BM-TNA.Fig. 2: Nyquist plot, and degradation of benzoic acid using BM- and BP-TNA electrodes under UV.a Nyquist plot (BM-TNA/BP-TNA) and b, c benzoic acid degradation by BM-TNA and BP-TNA photoelectrochemical processes under different forms of UV light ([NaCl]0 = 3000 ppm; scan rate = 50 mV s−1 for LSV/CV; scan frequency = 100 kHz–10 mHz with 10 mV rms sinusoidal modulation versus open circuit potential for EIS; [Benzoic acid]0 = 0.1 mM; constant voltage = 1 A; pHi = 7.0).Full size imageTo further evaluate the capacity of the two catalysts, degradation performance of BA as a model substrate within the PEC system under UV-A, B, and C lamps with and without the BM-TNA and BP-TNA catalysts was investigated as shown in Fig. 2(b) and (c). When the organic substance was irradiated with only the UV lamps, BA was hardly decomposed under UVA/B, while degradation slightly increased under the UVC light (Fig. 2(a)). However, in Fig. 2(b) and (c), it was confirmed that the morphology of the anodic catalyst (i.e., BM-TNA, BP-TNA) accelerated the decay of BA, and efficiency was much more significant for BM-TNA with its large specific surface area.In the combined operation with the BM-TNA electrode annealed at 600 °C, degradation rate was significant with the reaction rate of the PEC system as follows: (k (UVC-PEC) 0.0444 ± 0.0017 min−1) > (k (UVB-PEC) 0.0269 ± 0.001 min−1) > (k (UVA-PEC) 0.0108 ± 0.0004 min−1). In contrast, degradation using the BP-TNA electrode equally annealed at 600 °C showed much lower results with the following reaction rates: (k (UVC-PEC) 0.0145 ± 0.0004 min−1) > (k (UVB-PEC) 0.0119 ± 0.0004 min−1) > (k (UVA-PEC) 0.0053 ± 0.0001 min−1). Significance of the novel BM-TNA catalyst was further confirmed through comparison with a similar BM-TNA electrode prepared at 450 °C (Supplementary Fig. 2). The novel electrode annealed at 600 °C exhibited greater mineralization efficiency, and the enhancement was distinguishable under all UV-A, B, and C lights. The measured energy consumption under UV-A, B, and C with a reaction time of 120 min was 0.035, 0.056, and 0.074 kWh, respectively.The quenching effect of the BM-TNA catalyst in the presence of saltwater was also confirmed, as shown in Supplementary Fig. 3(a). When excess methanol and tert-butanol were applied, a dynamic delay in BPA degradation by the BM-TNA was observed. The main oxidant tended to degrade under the influence of hydroxyl radicals during electrolysis, and tert-butanol is effective for eliminating •OH k (•OH + tert-BuoH = 6.0 × 108 M−1S−1). However, in the PEC system without a scavenger, the decomposition of BPA was monitored, indicating that reaction with UV-A light induced the formation of hydroxyl radicals as a powerful oxidizing agent due to the nature of the OER catalyst. Concurrently, the decomposition efficiency was monitored with BA as the target compound to identify the strong ROS generated in the PEC system (Supplementary Fig. 3(b)). When BA was decomposed, the formation of by-product as 4-HBA was confirmed, and an excellent reduction in TOC was observed.Removal of organic compounds and catalytic stability of the PEC systemThe removal efficiencies of various organic pollutants in the PEC system were represented by pseudo-first-order rate constants, as shown in Fig. 3(a). Representative organic contaminants in brackish water were selected for investigation (i.e., BPA, 4CP, CMT, BA, PH, NIB, AMP, and SMX). The reaction rates of the organic pollutants in the system were as follows: (k (4CP) = 0.0447 ± 0.0019 min−1) > (k (PH) = 0.0381 ± 0.0028 min−1) > (k (CMT) = 0.0361 ± 0.0012 min−1) > (k (BPA) = 0.0291 ± 0.001 min−1) > (k (AMP) = 0.0270 ± 0.003 min−1) > (k (BA) = 0.0181 ± 0.0012 min−1) > (k (SMX) = 0.0173 ± 0.0002 min−1) > (k (NIB) = 0.0127 ± 0.0032 min−1). The PEC oxidation efficiencies, including aromatic compounds (such as electron-donating group (EDG) and electron-withdrawing group (EWG)) for BM-TNA catalysts exhibit different substrate specificities41,42. For instance, the phenolic compounds of EDG more easily released protons into the solution under •OH- induced oxidation, and were more susceptible to PEC anodization43,44. Specifically, the positive redox potential of 4CP exhibited faster degradation than PH (+0.86 VNHE for PH versus +0.8 VNHE for 4CP), which may contribute to the significant resistance against the oxidation45,46. In contrast, the EWG (i.e., BA and NIB) dynamically hindered the degradation of BA and NIB via benzene ring substitution24.Fig. 3: Organic degradation via BM-TNA electrode, and evaluation of electrode stability.a Organic compound removal efficiency of BM-TNA under the photoelectrochemical system ([bisphenol-A]0, [4-chlorophenol]0, [sulfamethoxazole]0, [cimetidine]0, [benzoic acid]0, [acetaminophen]0, [nitrobenzene]0, [phenol]0 = 0.1 mM, [NaCl]0 = 3000 ppm; pHi = 7.0), and b repetition test of bisphenol-A decay ([bisphenol-A]0 = 10 µM, [NaCl]0 = 3000 ppm; pHi = 7.0).Full size imageThe stability and durability of the BM-TNA catalyst system were evaluated via 20 repeated BPA degradation cycles in the PEC cell (Fig. 3(b)). The results indicate the long-term stability of the catalyst through the similar decomposition efficiency to the initial value of BPA when operating continuously at an initial pH of 7 for 20 h. In addition, the catalysts annealed at 600 °C maintained the initial color (i.e., blue) of the photocatalyst activity electrode. This result is linked to Supplementary Fig. 4, in which the BPA removal efficiency of the PEC system was evaluated as a function of pH (pH 3–9). The efficacy of BPA at pH 3, 5, 7, and 9 were 99.3%, 94.4%, 93.9%, and 87.3%, respectively. This result suggests that the kinetic retardation and inhibition of radical formation of the PEC cell under alkaline conditions (above pH 9) increased the OH- content of the aquatic system.Evaluation of the FCDI process operation via deionization and SECThe performance of the FCDI operation was evaluated under three carbon mass loadings (5, 10, and 15 wt%) and three applied voltages (0.5, 0.8, and 1.1 V), and was optimized based on the SEC. In FCDI, activated carbon plays the role of ion adsorption owing to its exceptional surface area, electrical conductance, and ideal adsorption isotherm47. Accordingly, a higher carbon mass loading in the flow-electrode solution increases the surface area available for ion adsorption. In addition, activated carbon particles act as a bridge for charge transportation. In an ideal homogeneous solution, in which the distribution of the activated carbon particles is even throughout the flow-electrode solution, charge transportation is also uniform with no fluctuation in the charge efficiency. However, flow-electrode solutions are heterogeneous and dynamic in nature, thus raising the issue of charge percolation. The deformation in the electrode material distribution during flow causes a decrease in the particle and charge connectivity48. Similarly, increasing the carbon mass loading allows the formation of strongly defined charge-percolation pathways, thereby improving the solution conductivity and accelerating charge transportation49,50. This relationship was confirmed to be consistent with previous studies that reported a positive correlation between the mass loading and conductivity of the flow-electrode solution48.Correspondingly, the ion removal efficiency of the system was enhanced as the mass loading and applied voltage increased, as shown in Fig. 4(a). Under an applied voltage of 0.5 V, the efficiency was, on average, seven-fold and eleven-fold higher when the carbon mass loading increased to 10 and 15 wt%, respectively. For 0.8 and 1.1 V, the efficiencies increased by approximately eightfold and thirteen-fold, and twofold and threefold, under 10 and 15 wt%, respectively. The obtained results were consistent with those of previous studies. One previous study reported a sharp increase in the desalting performance when the carbon mass loading was increased from 0 to 10 wt%51, while another reported a similar trend in process performance as the mass loading increased from 20 to 25 wt%52. Ion removal was quantified in detail as shown in Supplementary Figs. 5–7, with (a) and (b) for the three figures representing the sodium and chloride ion concentrations in the permeate solution, respectively. A comparable reduction can be observed, with the slope of the concentration gradient becoming sharper as mass loading increased. (c) and (d) represent the ionic concentrations within the flow-electrode (slurry) solution, which were confirmed to increase as the ions within the feed stream adsorbed to the flow-electrode solution. The influences of the mass loading and applied voltage were further assessed via solution conductivity, SAC, and SAR, as shown in Supplementary Figs. 8–10. Regarding conductivity, a mass loading of 5 wt% resulted in nominal removal under all applied voltage conditions, with the maximum deionization reaching only ~20%. However, higher mass loadings of 10 and 15 wt% resulted in considerable enhancements. Quantitatively, the deionization performance under a mass loading of 10 wt% increased by 1.3-, 1.6-, and 1.8-fold under applied voltages of 0.5-, 0.8-, and 1.1 V, respectively, and by 1.4-, 1.8-, and 2.0-fold under a mass loading of 15 wt%. Similarly, increasing the two operational parameters enhanced the SAC, and the linear trend in SAR demonstrated the consistent performance of the FCDI system.Fig. 4: Deionization performance and relative energy consumption of the FCDI system.a Ion removal efficiency, and b specific energy consumption of the flow-electrode capacitive deionization system (activated carbon mass loading = 5, 10, 15 wt%, applied voltage = 0.5, 0.8, 1.1 V, flow-electrode electrolyte = 1 M).Full size imageHowever, the parallel increase in viscosity and the decrease in solution fluidity are common major obstacles associated with increasing mass loading. Owing to the trade-off relationship, the loading in FCDI systems is commonly restricted to within 20 wt%53,54,55 when no additional surface modification methods are followed for electrode material synthesis. Consistently, the mass loading condition in this study was limited to 15 wt%, and solution viscosity and clogging in the flow channel were not observed. As shown in Fig. 4(b), the validity of the mass loading and applied voltage was further assessed via SEC and quantified based on the total amount of ions removed during operation. The SEC followed a similar trend under all conditions, with the value increasing under higher loading rates. The change in the applied voltage followed comparable trends under mass loadings of 10 and 15 wt%. Despite the improvement in the deionization performance as the mass loading increased, as shown in Fig. 4(a), the reason for the higher SEC was the concurrent enhancement in the charge percolation of the system52. The strengthened electrical conductance and synchronous reduction in cell resistance led to the enhancement of current efficiency56,57, resulting in a higher projected SEC. Specifically, the SEC values under a mass loading of 10 wt% were highly similar to those under 5 wt%. The changes in the values were below twofold for all applied voltage conditions, with the maximum being 1.6-fold. However, increasing the mass loading to 15 wt% increased the SEC values by 1.7, 2.4, and 2.5-fold under applied voltages of 0.5, 0.8, and 1.1 V, respectively. As the initial objective of the study was to obtain permeate within the suggested TDS range of potable water set by the World Health Organization or domestic water58, the sequential two-adsorption operation was deemed sufficient to satisfy the standards based on the ion removal efficiency, as shown in Fig. 4(a). Therefore, a carbon mass loading and applied voltage of 10 wt% and 0.8 V were determined to be the optimal operational parameters, respectively. Comparably, a higher applied voltage (1.1 V) resulted in nominal changes in ion removal, with a much larger SEC. Similarly, a carbon mass loading of 15 wt% resulted in a higher deionization efficiency, which was also within a nominal range of ~1.2-fold with a much higher SEC.Evaluation on the performance of PEC-FCDI dual processThe feasibility of the PEC-FCDI dual system was investigated by treating brackish water to produce organic contaminant-free permeate that meets the freshwater TDS standards. To investigate the efficacy of the system more accurately, the brackish water was prepared as a complex solution including multiple monovalent and divalent salts to reflect the composition of real brackish water as detailed in Supplementary Table 1. As shown in Fig. 5, the organic mineralization of BPA was evaluated at the PEC stage by measuring the TOC level. During the 60 min operation, an extremely sharp decrease in the TOC to 20% of the initial concentration was observed in the first 20 min. Subsequently, a three-log removal of BPA was monitored, with the final, complete degradation of the target pollutant occurring within 40 min of operation. No further formation of oxidation by-products was detected during the PEC process.Fig. 5: Operation of the PEC-FCDI dual system.PEC-FCDI dual system (PEC: [bisphenol-A]0 = 0.1 mM, pHi = 7.0; operation time = 60 min) (FCDI: [NaCl]0 = 1947 ppm, [MgCl2]0 = 428 ppm, [Na2SO4]0 = 278 ppm, pHi = 7.0; activated carbon mass loading = 10 wt%, applied voltage = 0.8/−0.8 V, flow-electrode electrolyte = 1 M; first-adsorption time = 40 min, desorption time = 30 min, second-adsorption time = 20 min).Full size imageA subsequent FCDI stage was conducted for the purpose of deionization, and performance was evaluated based on the quantification of the removal of individual ions. During the first adsorption phase, a rapid decrease in the ion concentration generally occurred within 20 min. Specifically, the contents of sodium, magnesium, chloride and sulfate ions were reduced by 71, 72, 75.6 and 75.2%, respectively, indicating similar deionization efficiencies for both cations and anions. The adsorption rate eventually decelerated and reached a plateau as the pores of the activated carbon electrode became fully saturated with ions transported from the feed solution; thus, a desorption phase was required for the pores and charge balance of the system to recover to the initial state38. The desorption phase was conducted by reversing the applied voltage, and was operated for 30 min. The ions adsorbed in the micro-/ macropores of the electrode particles were discharged via electrostatic repulsion and transported across the ion exchange membranes to the feed stream. The ionic concentration of the flow-electrode slurry solution gradually decreased during the 30 min discharge operation, and all the ions adsorbed during the adsorption phase were discharged from the solution. After 30 min, the slurry solution was fully recovered for the operation of the second-adsorption phase. The final adsorption phase was conducted for 20 min with the permeate from the first adsorption phase as the feedwater solution. Under the low initial ion concentration, a similar rapid decline was observed, with almost two-log removal within minutes of operation, and complete deionization of the feedwater was ultimately achieved. Lastly, the energy consumption of the FCDI system was monitored, and the total combined SEC of the two-adsorption phases was 0.031 kWh g−1 ion removal, which was lower than that evaluated during the process optimization (Fig. 4), thereby further strengthening the feasibility of the technology for real water applications.The treatment of brackish water for drinking or domestic use has mostly been conducted using brackish water RO or NF processes. However, critical drawbacks due to the low removal of trace organic pollutants, membrane surface fouling by the attachment of organics, and high energy consumption due to the pressure-driven nature of the technology indicate the limitations of the processes. In contrast, oxidation via the PEC system can achieve complete mineralization of trace organic compounds, while the subsequent deionization sequence of FCDI can attain ion desalination within the brackish water source under low energy consumption.The novel self-doped BP-TNA and BM-TNA electrodes prepared by annealing at 600 °C revealed excellent charge transfer efficiencies, and effective degradation of eight model organic pollutants, with the BP-TNA showing superior performance. Organic mineralization was stably maintained at high levels throughout a wide pH range of 3–9, and a continuous experiment of seven repetitive cycles with uniform performance demonstrated the stability of the electrode. In addition, solution desalination was observed during the subsequent FCDI stage, and particularly, the increase in carbon mass loading improved charge-percolation pathways, which led to superior solution conductivity and charge transportation efficiency of the system.The use of the hybrid PEC-FCDI system is expected to be a superior alternative to conventional processes. Owing to its exceptional performance and low energy consumption levels, the system is promising for practical application in the field of brackish water treatment for both potable or domestic use. More
