Cobalt Incorporated Graphitic Carbon Nitride as a Bifunctional Catalyst for Electrochemical Water-Splitting Reactions in Acidic Media

Non-noble metal-based bifunctional electrocatalysts may be a promising new resource for electrocatalytic water-splitting devices. In this work, transition metal (cobalt)-incorporated graphitic carbon nitride was synthesized and fabricated in electrodes for use as bifunctional catalysts. The optimum catalytic activity of this bifunctional material for the hydrogen evolution reaction (HER), which benefitted at a cobalt content of 10.6 wt%, was promoted by the highest surface area and conductivity. The activity achieved a minimum overpotential of ~85 mV at 10 mA/cm2 and a Tafel slope of 44.2 mV/dec in an acidic electrolyte. These values of the HER were close to those of a benchmark catalyst (platinum on carbon paper electrode). Moreover, the kinetics evaluation at the optimum catalyst ensured the catalyst flows (Volmer–Heyrovsky mechanism), indicating that the adsorption step is rate-determining for the HER. The activity for the oxygen evolution reaction (OER) indicated an overpotential of ~530 mV at 10 mAcm−2 and a Tafel slope of 193.3 mV/dec, which were slightly less or nearly the same as those of the benchmark catalyst. Stability tests using long-term potential cycles confirmed the high durability of the catalyst for both HER and OER. Moreover, the optimal bifunctional catalyst achieved a current density of 10 mAcm−2 at a cell voltage of 1.84 V, which was slightly less than that of the benchmark catalyst (1.98 V). Thus, this research reveals that the present bifunctional, non-noble metallic electrocatalyst is adequate for use as a water-splitting technology in acidic media.


Introduction
Energy is remarkably vital to the development of the global economy. However, the overuse of fossil fuel energy is objectionable for the environment. Thus, the development of sustainable, green, and environmentally friendly alternative energy sources is essential, and humans are constantly seeking such sources [1]. One of these alternative green energy sources is the hydrogen energy [2][3][4]. A water-splitting generator received great attention for producing molecular hydrogen because hydrogen reveals high energy density (143 KJ/g), plentiful availability, and zero carbon emission [5,6]. Although hydrogen can be produced from hydrogen-containing compounds (biomass, water, natural gas, etc.), hydrogen from water electrolysis is the cleanest and most sustainable method [7,8]. In such cases, hydrogen evolution reaction (HER) and oxygen evolution reaction (OER) occur at the cathode and anode half-reactions, respectively, and the method opens new strategies for obtaining electric energy from the most abundant resource (water) [9]. However, both HER and OER must overcome a large energy barrier (overpotential). Thus, the scalable industrial application is still under experiment, and until now, only a small number (about 4%) of
The g-CN powder is the same sample that was synthesized and characterized in a previous work [21]. It was synthesized by pyrolysis of melamine and exfoliation by strong sonication. A composite of cobalt on g-CN (Co@g-CN) was synthesized by the reduction of the cobalt precursor in a solution with sodium borohydride (NaBH 4 ): g-CN (50 mg) was dispersed in water (1.6 g L −1 , 30 mL) by sonicating for 1 h, and an aqueous CoCl 2 solution (0.01 M, 2.5, 5, 10, or 15 mL) was added to the g-CN dispersion and stirred overnight. Subsequently, an aqueous NaBH 4 solution (0.5 M, 2 mL) was added and further stirred for 1 h. The solid product was collected by filtering with a filter paper (2 µm pore). The calculated contents of cobalt in Co@g-CN were 2.9, 5.6, 10.6, and 15.0 wt%, respectively, depending on the added volume of the CoCl 2 solution. The products were then termed as Co(x)@g-CN (x = 3, 6, 11, 15).
The linear scan voltammetry (LSV) measurements were performed using a computercontrolled potentiostat on a Zahner electrochemical workstation (Zennium, 40442, Kronach, Germany) with a three-electrode system: a cell consists of an aqueous 0.5 M H 2 SO 4 solution (25 mL) as an electrolyte, a platinum wire as a counter electrode, an Ag/AgCl in 3 M KCl as a reference electrode, and a working electrode, which was prepared on a carbon paper by loading a catalyst ink (slurry, 2.0 ± 0.2 mg·cm −2 , 50 µL), that is, an ultrasonically dispersed and stirred mixture of Co(x)@g-CN (10 mg), isopropanol (200 µL), and Nafion binder (0.5 vol%, 100 µL). A working electrode of a benchmark catalyst, called Pt/Cp, was prepared by coating Pt of 1 nm thickness by 30 min vapor deposition on a surface of carbon paper with a 1 cm 2 active area. Electrochemical impedance spectroscope (EIS) measurement was performed using the same electrode system and electrolyte, at a potential of −0.25 V, a frequency range of 10 kHz-1 MHz, and an amplitude of 10 mV on the same instrument.
The HER and OER activities were recorded by linear sweep voltammetry (LSV) at a scan rate of 10 mV·s −1 , and the electrode durability was tested by a cyclic voltammogram (CV) response. Potentials were converted to the reversible hydrogen electrode (RHE) scale as denoted by E RHE . Then CVs were conducted at a scan rate of 10 mV/s between 0.10 to −0.11 V vs. E RHE , where a quasi-rectangular shaped curve was obtained. The current density was calculated by J = I A , where I is a current and A is an active area of an electrode. The overpotential (η) was calculated according to the following formula: η [V] = 0.000 − E RHE for HER and η [V] = E RHE − 1.230 for OER. Then, the slope b of the Tafel plot was calculated from a Tafel equation (Equation (1)).
where a is an overpotential at current density (J = 1). The hydrogen and oxygen gases generated at respective electrodes were quantified, using a laboratory-made instrument, by the downward displacement of water in which electrodes prepared as the cathode (HER), anode (OER), and Pt as a reference electrode were used at an applied potential of 1.8 V.

Structural Characteristics of Co@g-CN
On a 2D g-CN sheet synthesized by a pyrolysis condensation reaction of melamine [21], Co was deposited via the reduction reaction of Co precursor (See Figure 1A). FTIR spectra of Co@g-CN with different Co contents were measured and compared with g-CN ( Figure 1B). The characteristic IR bands of g-CN are a stretching N-H vibration mode of the terminal amine at 3100-3300 cm −1 , C=N, C-N, N-C 3 , and NH-C 2 stretching vibration modes of the heptazine unit at 1640, 1410, 1322, and 1250 cm −1 , respectively, and an out-of-plane bending vibration mode of the heptazine unit at 810 cm −1 [21]. These bands were observed even after loading the Co, and the relative band intensities were also nearly unchanged. These results indicate that the loading of Co does not disturb the vibrational modes of g-CN with a heptazine unit ring consisting of different carbon-nitrogen bonds. ( Figure 1B). The characteristic IR bands of g-CN are a stretching N-H vibration mode of the terminal amine at 3100-3300 cm −1 , C=N, C-N, N-C3, and NH-C2 stretching vibration modes of the heptazine unit at 1640, 1410, 1322, and 1250 cm −1 , respectively, and an outof-plane bending vibration mode of the heptazine unit at 810 cm −1 [21]. These bands were observed even after loading the Co, and the relative band intensities were also nearly unchanged. These results indicate that the loading of Co does not disturb the vibrational modes of g-CN with a heptazine unit ring consisting of different carbon-nitrogen bonds. A crystal structure of Co@g-CN was analyzed using an XRD pattern. As shown in Figure 1C, Co(11)@g-CN exhibited two distinctive diffraction peaks at 2θ = 27.99° and 13.36°, in good agreement with those of g-CN. A weak peak indexed as (100) equivalent to d spacing of 0.662 nm can be ascribed to the in-plane distance of the g-CN sheet, and a strong peak (002) with d spacing of 0.318 nm is attributed to the interlayer distance between the g-CN sheets [21]. However, it must be noted that no peaks of cobalt species in Co(11)@g-CN were observed, although they were observed at 2θ = 44, 51, and 75° [22]. The reason should be the low content of Co [23,24]. However, the decrease in intensity and the slight broadening of (002) plane indicate the coordination of Co to g-CN by the host-guest framework forming Co-N bonds [10]. From these results, even though the FTIR A crystal structure of Co@g-CN was analyzed using an XRD pattern. As shown in Figure 1C, Co(11)@g-CN exhibited two distinctive diffraction peaks at 2θ = 27.99 • and 13.36 • , in good agreement with those of g-CN. A weak peak indexed as (100) equivalent to d spacing of 0.662 nm can be ascribed to the in-plane distance of the g-CN sheet, and a strong peak (002) with d spacing of 0.318 nm is attributed to the interlayer distance between the g-CN sheets [21]. However, it must be noted that no peaks of cobalt species in Co(11)@g-CN were observed, although they were observed at 2θ = 44, 51, and 75 • [22]. The reason should be the low content of Co [23,24]. However, the decrease in intensity and the slight broadening of (002) plane indicate the coordination of Co to g-CN by the host-guest framework forming Co-N bonds [10]. From these results, even though the FTIR spectra did not confirm the existence of Co, XRD spectra indicated the coordination of Co in the framework of g-CN.
On an XPS spectrum of Co(11)@g-CN, as seen in a survey spectrum of Figure 2A, additional peaks of the Co element were observed, as well as C, N, and O elements, indicating the loading of Co on a structure of g-CN. The fine spectra and their deconvoluted peaks of each element are shown in Figure 2B. Co 2p exhibited two intense binding peaks at 783.1 and 799.8 eV, which are attributed to 2p 3/2 and 2p 1/2 of cobalt, respectively. These binding energies are higher than those of free metallic cobalt and cobalt oxide [23,24], but they can be attributed only to metallic cobalt coordinated to the center of the nitrogen atoms in the heptazine ring (Co-N x ) [10], as illustrated in Figure 1A. Meanwhile, the peaks at 787.7 and 805.9 eV are satellite peaks of 2p 3/2 and 2p 1/2 peaks, respectively. The result shows that Co is successfully incorporated in the structure of the graphitic framework. In comparison with XPS peaks of g-CN [21], the same number of deconvoluted peaks of C 1s and N 1s were observed, but both the C 1s and N 1s binding energies of the C=N-C bond and an N 1s binding energy of C-NH x bond were remarkably shifted from the corresponding binding energies of g-CN [21] because of the influence of the Co-N x coordination on the binding energies of NC conjugation. Additionally, O 1s was deconvoluted in three peaks in comparation to one peak of g-CN at 287.2 eV [21]. The new peaks are attributed to CO 2 and H 2 O being simply adsorbed to Co@g-CN. CO 2 is easily adsorbed on the amine group and forms the NHCOO − group based on an equation of CO 2 + 2 R-NH 2 R-NHCOO − + R-NH 3 + [25]. Then, peaks of the NHCOO − group were found in C 1s and O 1s, but C 1s of CO 2 was assumed to overlap on the C1s of any group. Thus, the XPS spectra confirmed that Co was incorporated in the graphitic framework through Co-N x bonding [24].
Molecules 2021, 26, x FOR PEER REVIEW 5 of 13 spectra did not confirm the existence of Co, XRD spectra indicated the coordination of Co in the framework of g-CN.
On an XPS spectrum of Co(11)@g-CN, as seen in a survey spectrum of Figure 2A, additional peaks of the Co element were observed, as well as C, N, and O elements, indicating the loading of Co on a structure of g-CN. The fine spectra and their deconvoluted peaks of each element are shown in Figure 2B. Co 2p exhibited two intense binding peaks at 783.1 and 799.8 eV, which are attributed to 2p3/2 and 2p1/2 of cobalt, respectively. These binding energies are higher than those of free metallic cobalt and cobalt oxide [23,24], but they can be attributed only to metallic cobalt coordinated to the center of the nitrogen atoms in the heptazine ring (Co-Nx) [10], as illustrated in Figure 1A. Meanwhile, the peaks at 787.7 and 805.9 eV are satellite peaks of 2p3/2 and 2p1/2 peaks, respectively. The result shows that Co is successfully incorporated in the structure of the graphitic framework. In comparison with XPS peaks of g-CN [21], the same number of deconvoluted peaks of C 1s and N 1s were observed, but both the C 1s and N 1s binding energies of the C=N-C bond and an N 1s binding energy of C-NHx bond were remarkably shifted from the corresponding binding energies of g-CN [21] because of the influence of the Co-Nx coordination on the binding energies of NC conjugation. Additionally, O 1s was deconvoluted in three peaks in comparation to one peak of g-CN at 287.2 eV [21]. The new peaks are attributed to CO2 and H2O being simply adsorbed to Co@g-CN. CO2 is easily adsorbed on the amine group and forms the NHCOO − group based on an equation of CO2 + 2 R-NH2 ⇌ R-NHCOO − + R-NH3 + [25]. Then, peaks of the NHCOO − group were found in C 1s and O 1s, but C 1s of CO2 was assumed to overlap on the C1s of any group. Thus, the XPS spectra confirmed that Co was incorporated in the graphitic framework through Co-Nx bonding [24]. A TEM image of Co(11)@g-CN in Figure 3a showed the sheet-like structure of g-CN [21] and the grain texture with an average size of 11.72 ± 2.56 nm on the sheet. An EDS on the FE-SEM image of Co(11)@g-CN (Figure 3b,c) confirmed the existence of the Co atom A TEM image of Co(11)@g-CN in Figure 3a showed the sheet-like structure of g-CN [21] and the grain texture with an average size of 11.72 ± 2.56 nm on the sheet. An EDS on the FE-SEM image of Co(11)@g-CN (Figure 3b,c) confirmed the existence of the Co atom in Co(11)@-gCN. Moreover, an inset table in EDS indicates the existence of 11.6 wt% Co. This value is comparable to the calculated value (10.6 wt%) in Co(11)@-gCN. Moreover, the atom ratio of C:N in g-CN, which can be evaluated from Figure 1A, is 18.0:25.5. Since the atom ratio of C:N: Co calculated from the atom % in Figure 3c (inset) was 18.0:26.9:0.77; Co(11)@g-CN achieved 77% Co loading of the estimated complex structure. in Co(11)@-gCN. Moreover, an inset table in EDS indicates the existence of 11.6 wt% Co. This value is comparable to the calculated value (10.6 wt%) in Co(11)@-gCN. Moreover, the atom ratio of C:N in g-CN, which can be evaluated from Figure 1A, is 18.0:25.5. Since the atom ratio of C:N: Co calculated from the atom % in Figure 3c (inset) was 18.0:26.9:0.77; Co(11)@g-CN achieved 77% Co loading of the estimated complex structure. As seen in Figure 4a, nitrogen adsorption-desorption curves of Co(x)@g-CN (x = 6, 11, and 15) at different cobalt contents belonged to the type II isotherm, showing hysteresis at high relative pressure due to the porous structure of the materials. The specific surface area, pore volume, and pore diameter were analyzed from the BET isotherms, and the Barret, Joyner, and Halenda (BJH) plots are shown in Figure 4b. As seen in Table 1, Co(11)@g-CN exhibited the largest specific surface area and pore volume and the smallest pore diameter compared to Co(x)@g-CN with other Co contents. These results show the preferable variation of the surface properties of g-CN with the addition of cobalt at the optimum content of 10.6 wt%.   As seen in Figure 4a, nitrogen adsorption-desorption curves of Co(x)@g-CN (x = 6, 11, and 15) at different cobalt contents belonged to the type II isotherm, showing hysteresis at high relative pressure due to the porous structure of the materials. The specific surface area, pore volume, and pore diameter were analyzed from the BET isotherms, and the Barret, Joyner, and Halenda (BJH) plots are shown in Figure 4b. As seen in Table 1, Co(11)@g-CN exhibited the largest specific surface area and pore volume and the smallest pore diameter compared to Co(x)@g-CN with other Co contents. These results show the preferable variation of the surface properties of g-CN with the addition of cobalt at the optimum content of 10.6 wt%. in Co(11)@-gCN. Moreover, an inset table in EDS indicates the existence of 11.6 wt% Co. This value is comparable to the calculated value (10.6 wt%) in Co(11)@-gCN. Moreover, the atom ratio of C:N in g-CN, which can be evaluated from Figure 1A, is 18.0:25.5. Since the atom ratio of C:N: Co calculated from the atom % in Figure 3c (inset) was 18.0:26.9:0.77; Co(11)@g-CN achieved 77% Co loading of the estimated complex structure. As seen in Figure 4a, nitrogen adsorption-desorption curves of Co(x)@g-CN (x = 6, 11, and 15) at different cobalt contents belonged to the type II isotherm, showing hysteresis at high relative pressure due to the porous structure of the materials. The specific surface area, pore volume, and pore diameter were analyzed from the BET isotherms, and the Barret, Joyner, and Halenda (BJH) plots are shown in Figure 4b. As seen in Table 1, Co(11)@g-CN exhibited the largest specific surface area and pore volume and the smallest pore diameter compared to Co(x)@g-CN with other Co contents. These results show the preferable variation of the surface properties of g-CN with the addition of cobalt at the optimum content of 10.6 wt%.    On a UV-visible absorption spectrum of Co(11)@g-CN, in comparison with that of g-CN (Figure 5a), while the main and shoulder bands of g-CN were seen at 331 nm and around 400 nm, respectively, an absorption band at a shorter wavelength (330 nm) of Co(11)@g-CN decreased its intensity. These absorption bands can be assigned to π-π* and n-π* electronic transitions of C=N-C in the heptazine ring and the free electron of 2p [26]. Thus, the variation in the intensity after the addition of Co may be attributed to the formation of the Co-N interaction, which can decrease the intensity of the π-π* band. The influence of Co loading on the optical property of g-CN was observed on the fluorescence spectra, where an emission band at 440 nm with a shoulder band around 460 nm was observed at a 275 nm excitation for both g-CN and Co@g-CN, as shown in Figure 5b. While supplementary excitation bands were detected at 330 and 360 nm, the relevant emission bands displayed quite weak intensities at 440 nm. It should be noted that the fluorescence intensities of g-CN at 275 nm (excitation) and 440 nm (emission) reduced after the incorporation of metallic Co. Since a 275 nm band can be assigned the π-π* electron transition, the fluorescence property is related to the π-π* transition, but not the n-π* transition. This decrease in fluorescence may participate in the energy transfer [27,28]. The transferred energy contributes to the charge separation, promoting the electron mobility within the structure of the graphitic carbon nitride. Incidentally, the band gap (2.60 eV) of g-CN minimized to 2.38 eV in Co(11)@g-CN, as evaluated from the UV-visible absorption spectra (Figure 5c), indicated that metal doping decreases the bandgap of Co(11)@g-CN. On a UV-visible absorption spectrum of Co(11)@g-CN, in comparison with that of g-CN (Figure 5a), while the main and shoulder bands of g-CN were seen at 331 nm and around 400 nm, respectively, an absorption band at a shorter wavelength (330 nm) of Co(11)@g-CN decreased its intensity. These absorption bands can be assigned to π-π* and n-π* electronic transitions of C=N-C in the heptazine ring and the free electron of 2p [26]. Thus, the variation in the intensity after the addition of Co may be attributed to the formation of the Co-N interaction, which can decrease the intensity of the π-π* band. The influence of Co loading on the optical property of g-CN was observed on the fluorescence spectra, where an emission band at 440 nm with a shoulder band around 460 nm was observed at a 275 nm excitation for both g-CN and Co@g-CN, as shown in Figure 5b. While supplementary excitation bands were detected at 330 and 360 nm, the relevant emission bands displayed quite weak intensities at 440 nm. It should be noted that the fluorescence intensities of g-CN at 275 nm (excitation) and 440 nm (emission) reduced after the incorporation of metallic Co. Since a 275 nm band can be assigned the π-π* electron transition, the fluorescence property is related to the π-π* transition, but not the n-π* transition. This decrease in fluorescence may participate in the energy transfer [27,28]. The transferred energy contributes to the charge separation, promoting the electron mobility within the structure of the graphitic carbon nitride. Incidentally, the band gap (2.60 eV) of g-CN minimized to 2.38 eV in Co(11)@g-CN, as evaluated from the UV-visible absorption spectra (Figure 5c), indicated that metal doping decreases the bandgap of Co(11)@g-CN.

Electrochemical Water-Splitting Activity of Co@g-CN Catalysts
It was revealed from the characterization of Co@g-CN that Co incorporated in g-CN varies surface properties of g-CN and enables the charge transfer from g-CN. These effects should influence the water adsorption and the increase in the conductivity, which can enhance the catalytic activity of the HER and OER. Thus, the electrochemical activities of catalysts (Co@g-CN) towards the HER and OER were examined using a three-electrode system in an acidic electrolyte medium (0.5 M H2SO4 (aq.). The HER activities of the prepared catalysts and a benchmark catalyst (Pt/Cp) are presented in Figure 6a. The overpotentials of catalysts towards HER activities at a standard current density (10 mA·cm −2 ) are summarized in Table 2. Although g-CN exhibited poor electrocatalytic performance for HER, the incorporation of the small amount of Co on the g-CN sheet decreased the overpotential. The Co(11)@g-CN required an overpotential of 85 mV vs. ERHE to achieve a current density of 10 mA·cm −2 , although the overpotential of Co(15)@g-CN was the same as that of a benchmark catalyst(65 mV).
A Tafel plot obtained from LSV of HER can be employed for estimating the ratedetermining step [1,15]. Figure 6b displays Tafel slope plots of g-CN, Co@g-CN, and Pt/Cp electrocatalysts, and numerical values of the slope are listed in Table 2. The experimental

Electrochemical Water-Splitting Activity of Co@g-CN Catalysts
It was revealed from the characterization of Co@g-CN that Co incorporated in g-CN varies surface properties of g-CN and enables the charge transfer from g-CN. These effects should influence the water adsorption and the increase in the conductivity, which can enhance the catalytic activity of the HER and OER. Thus, the electrochemical activities of catalysts (Co@g-CN) towards the HER and OER were examined using a three-electrode system in an acidic electrolyte medium (0.5 M H 2 SO 4 (aq.). The HER activities of the prepared catalysts and a benchmark catalyst (Pt/Cp) are presented in Figure 6a. The overpotentials of catalysts towards HER activities at a standard current density (10 mA·cm −2 ) are summarized in Table 2. Although g-CN exhibited poor electrocatalytic performance for HER, the incorporation of the small amount of Co on the g-CN sheet decreased the overpotential. The Co(11)@g-CN required an overpotential of 85 mV vs. E RHE to achieve a current density of 10 mA·cm −2 , although the overpotential of Co(15)@g-CN was the same as that of a benchmark catalyst (65 mV).
A Tafel plot obtained from LSV of HER can be employed for estimating the ratedetermining step [1,15]. Figure 6b displays Tafel slope plots of g-CN, Co@g-CN, and Pt/Cp electrocatalysts, and numerical values of the slope are listed in Table 2. The experimental slope of Co(11)@g-CN was the smallest for Co(x)@g-CN and similar to that of a benchmark catalyst [29]. The theoretical Tafel slope values are~120,~40, and~30 mV/decade, corresponding to Volmer, Heyrovsky, and Tafel reactions, respectively [1]. Compared to the theoretical values, the experimental slope of Co(11)@g-CN, as well as the benchmark catalyst, follows the Volmer-Heyrovsky step in the HER. That is, in the first step, H + in the electrolyte and e − form an adsorbed H atom (H ads ) on an electrocatalytic active site [30], and in the second step, the H ads reacts with H + and e − to yield H 2 through an electrochemical desorption process [11].
H ads + H + + e − → H 2 (Heyrovsky) On the other hand, the OER electrocatalyst is an energy-uphill reaction involving four-electron transfer processes in an acidic environment, but the correlation among the kinetics equations is extremely complicated to solve [11]. Therefore, when Co(11)@g-CN and Pt/Cp electrocatalysts were applied for OER activity, overpotentials of 530 and 680 mV, respectively, were obtained at 10 mA cm −2 (see Figure 6c and Table 2). The consistency of the OER Tafel slopes between Co(11)@g-CN and Pt/Cp catalysts was also achieved, as seen in Figure 6d and Table 2. EIS measurements were performed to further characterize Co@g-CN according to conductivity. Nyquist plots obtained from EIS of Co@g-CN are shown in Figure 6e, and resistances were evaluated using a Randles circuit model, as listed in Table 2. Co(11)@g-CN showed the lowest charge transfer resistance (Rct) of 27.0 Ω, which was small compared to g-CN (6.06 kΩ), indicating that the incorporation of Co into g-CN enhanced the conductivity, although the variation of the internal resistance (Rs) was less, depending on the content of Co. However, the excess addition of Co inhibited the conductivity increase, showing a certain limit for the advancement of activity. The stability of an electrocatalyst is another important factor for its suitability for practical applications. Thus, the LSV polarization measurement of Co(11)@g-CN was performed after 2000 continuous cycles (scan rate: 10 mV·s −1 ) for both HER and OER. The observed catalytic decay compared to the initial activity was very small (see Figure 6f,g), indicating better stability, and therefore, the effective durability for repeatedly using the Co(11)@g-CN electrocatalyst in acidic media.
The amounts of hydrogen and oxygen arising over time were measured using a setup shown in Figure 7A. The volume of hydrogen gas increased with time, and 1.5 mL was collected at 7 h; similarly, the volume of oxygen gas also increased with time in an equimolar ratio, as theoretically expected ( Figure 7B The amounts of hydrogen and oxygen arising over time were measured using a setup shown in Figure 7A. The volume of hydrogen gas increased with time, and 1.5 mL was collected at 7 h; similarly, the volume of oxygen gas also increased with time in an equimolar ratio, as theoretically expected ( Figure 7B), indicating the high performance of the present catalyst (Co(11)@g-CN).
Thus, the catalytic activity of g-CN was enhanced in the presence of metallic cobalt. This activity demonstrates that the incorporation of metallic Co in g-CN influences the HER. Specifically, Co coordinates chemically with N atoms in the heptazine moieties and  Table 3. Despite the present electrode being slightly inferior to that of Pt/Cp, it was better for the HER activity and somewhat good for the OER activity, compared to some advanced catalysts [10,18,19,22]. Moreover, most reports [19,30,31] were performed under alkaline conditions, different from the present work, which was conducted using acidic conditions. The present condition is preferable for minimizing the cost of performance by using only a cell with an acidic medium instead of two cells with separate acidic and basic media.   Thus, the catalytic activity of g-CN was enhanced in the presence of metallic cobalt. This activity demonstrates that the incorporation of metallic Co in g-CN influences the HER. Specifically, Co coordinates chemically with N atoms in the heptazine moieties and generates Co-N active sites, which significantly increases HER and OER reactions. Notably, the HER activity of the Co@g-CN catalyst with adequate Co content was close to that of the Pt/Cp catalyst, effectively minimizing the overpotential. The catalytic performance of the present catalyst is also compared with previously reported works in Figure 7C and Table 3. Despite the present electrode being slightly inferior to that of Pt/Cp, it was better for the HER activity and somewhat good for the OER activity, compared to some advanced catalysts [10,18,19,22]. Moreover, most reports [19,30,31] were performed under alkaline conditions, different from the present work, which was conducted using acidic conditions. The present condition is preferable for minimizing the cost of performance by using only a cell with an acidic medium instead of two cells with separate acidic and basic media. To determine the overall water-splitting performance, the voltage differences (∆V) between the HER and OER were calculated by deducting HER overpotential from OER overpotential at the current density of 10 mAcm −2 , as indicated in Figure 7D. The Co(11)@g-CN catalyst afforded a current density of 10 mAcm −2 at a cell voltage of 1.84 V in the acidic electrolyte, which was slightly less than that of the benchmark catalyst (Pt/Cp, 1.98 V). Although the voltage differences of some reported works at the same current density were lower [10,16,18,32,33] or similar [19] (see Table 3), the present work has merit since the water splitting was performed only in an acidic medium and used only mono-metallic conductive metal.

Conclusions
In this work, the HER and OER activities of Co@g-CN electrocatalyst were reported and compared with pristine g-CN and a benchmark catalyst (Pt/Cp). Since the encapsulation of cobalt in g-CN effectively enhanced the kinetics of water adsorption and exposed favorably active sites that can result from the increase in the surface area and the electron transfer capacity of the g-CN catalyst, the activity increased compared to that caused by the pristine g-CN. In this regard, the relevant catalyst exhibited good catalytic activities for both the HER and OER in 0.5 M H 2 SO 4 (aq.). The optimum catalyst (Co@g-CN) displayed the better activity, which exceeded the g-CN electrode for the HER. For the OER, the same electrode achieved lower overpotential relative to the benchmark catalyst (Pt/Cp) at a current density of 10 mAcm −2 . Furthermore, the catalyst showed better overall water-splitting cell voltage of 1.84 V to drive a current density of 10 mAcm −2 as a bifunctional catalyst. Additionally, the recyclability of the catalyst was better, since the cobalt in g-CN increased the active site for water adsorption. Considering the electrocatalytic activities and simple synthesis approach of the catalyst, this work provides insights to design and develop an economical and valid bifunctional catalyst for overall water splitting in acidic media. Additionally, the Co@g-CN catalyst holds great potential in the fields of electrocatalyst and energy conversion.

Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Data Availability Statement: All data has been made available through the manuscript.