Critical factors to inhibit water‐splitting side reaction in carbon‐based electrode materials for zinc metal anodes

Abstract

Zinc metal anodes (ZMA) have high theoretical capacities (820 mAh g⁻¹ and 5855 mAh cm⁻³) and redox potential (−0.76 V vs. standard hydrogen electrode), similar to the electrochemical voltage window of the hydrogen evolution reaction (HER) in a mild acidic electrolyte system, facilitating aqueous zinc batteries competitive in next‐generation energy storage devices. However, the HER and byproduct formation effectuated by water‐splitting deteriorate the electrochemical performance of ZMA, limiting their application. In this study, a key factor in promoting the HER in carbon‐based electrode materials (CEMs), which can provide a larger active surface area and guide uniform zinc metal deposition, was investigated using a series of three‐dimensional structured templating carbon electrodes (3D‐TCEs) with different local graphitic orderings, pore structures, and surface properties. The ultramicropores of CEMs are the determining critical factors in initiating HER and clogging active surfaces by Zn(OH)₂ byproduct formation, through a systematic comparative study based on the 3D‐TCE series samples. When the 3D‐TCEs had a proper graphitic structure with few ultramicropores, they showed highly stable cycling performances over 2000 cycles with average Coulombic efficiencies of ≥99%. These results suggest that a well‐designed CEM can lead to high‐performance ZMA in aqueous zinc batteries.

INTRODUCTION

Aqueous zinc batteries based on metallic zinc anode are gaining the spotlight as next-generation power sources for application in compact mobile electronics and large-scale energy storage systems owing to extensive zinc resources and high theoretical capacity (820 mAh g⁻¹ and 5855 mAh cm⁻³) of zinc metal anode (ZMA).^1–3 In particular, the volumetric capacity of ZMA is much higher than those of sodium (~1129 mAh cm⁻³) and potassium (~586 mAh cm⁻³), respectively, which are worth noticing.^4–8 Considering the additional advantages of aqueous electrolytes, such as environmental compatibility, noninflammability, and fast ion transfer rate (~1 S cm⁻¹), ZMA-based rechargeable aqueous batteries (Z-RABs) are intriguing as power sources for e-mobilities, which recently progressed in electric vehicle markets.^9–11 However, the aqueous system has a limitation originating from the narrow operating voltage window of the water solution.^1,12,13 The voltage range of hydrogen evolution reaction (HER) between 0 and −0.829 V versus standard hydrogen electrode (SHE) is always higher than the redox potential of Zn²⁺/Zn in the range of −0.76 to −1.29 V versus SHE from acidic to alkaline conditions, indicating that ZMA is thermodynamically unstable in aqueous media.¹⁴ However, owing to the significantly high overpotential of the HER, ZMA with lower kinetic polarizations can be feasible in the water solution.^15–17 Further, this distinctive feature requires additional research for the efficient utilization of the water-splitting overpotential of ZMA.

The redox mechanism of ZMA has been studied in various aqueous electrolyte systems under alkaline and/or acidic conditions, wherein the alkaline electrolyte initiates zinc metal dissolution and nonionic conducting byproducts such as Zn(OH)₂.^18–21 The surface passivation layers hinder zinc ion diffusion at the interface between the electrode and the electrolyte, leading to diffusion-controlled metal reduction reactions which facilitates dendrite growth.^22,23 In contrast, a mild acidic electrolyte can suppress the formation of the Zn(OH)₂ passivation layers, demonstrating better electrochemical performances.^24–27 Accordingly, several strategies were applied in the subacidic electrolyte system—(a) coating a protective layer, (b) modifying surface properties, (c) introducing zincophilic sites, (d) controlling a crystallographic face, and (e) increasing the effective surface areas to mitigate water-splitting and/or guide homogeneous zinc metal deposition.^28–35 Previously reported studies suggest that high-performance Z-RABs can be realized with well-designed ZMAs.^4,36,37 However, issues such as side reactions and resultant electrode contaminations persisted.^38,39 To fabricate next-generation Z-RABs, the electrochemical decomposition of water molecules should be suppressed by investigating the key factor to promote the HER in repetitive Zn metal storage cycles.⁴⁰ In particular, a high-performance Z-RAB can be realized by reducing the excess zinc metal loading density.⁴¹ The anode-minimized ZMA system requires a multifunctional templating electrode material that has zincophilicity, high electrical conductivities, high mechanical and chemical stabilities, and three-dimensional (3D) structures with high active surface areas to reversibly accommodate zinc metals with high Coulombic efficiency (CE) during repetitive cycles.^28,42

Carbon-based materials are the initial preference for the 3D-templating electrode material owing to their versatile physicochemical material properties and in addition to their lightweight and mass-producible advantages.^43–46 The application of carbon-based electrode materials (CEMs) has recently been demonstrated in alkali metal batteries under organic electrolyte systems, where homogeneous and stable alkali metal deposition/dissolution cycles were guided.^47,48 In an aqueous medium for Zn metal storage, similar outcomes, such as reducing local effective current densities and suppressing volume expansion, are anticipated.^17,49–51 In addition, the minimal lattice mismatch of polyhexagonal carbon structures with zinc metal can guide uniform metal deposition.^2,34 Particularly in acidic conditions, CEMs with high chemical stability are inevitable because of the insufficient chemical stability of heavy metal-based electrode materials. However, CEMs can also promote electrochemical water-splitting, which is crucial for the application of CEMs for ZMA.⁵² Therefore, the critical factor to accelerate the water-splitting side reaction on CEMs should be discerned to develop a high-performance ZMA.⁵³

Herein, 3D-structured templating carbon electrodes (3D-TCEs) comprising entangled nanocarbon networks with different local graphitic microstructures, pore structures, and surface properties were prepared as a CEM platform to investigate their effects on the water-splitting side reaction during zinc metal deposition/dissolution processes. This systematic study provides information on the key factors for the water-splitting side reaction in CEMs for ZMA and the design of CEMs to achieve ZMA with high electrochemical performance.

EXPERIMENTAL SECTION Preparation of 3D-TCE-800, 3D-TCE-1600, and 3D-TCE-2400

Microbial-derived cellulose hydrogel precursors were prepared using a previously reported method,⁵⁴ where the nanostructured crystalline cellulose hydrogels fabricated from an aerobic microbe were incubated for 2 weeks at 30°C, and then thoroughly washed with an alkaline solution and distilled water until the hydrogels were neutralized. The hydrogels were immersed in tert-butanol for 12 h at 30°C to exchange the solvent and were lyophilized for 3 days at −50°C under <5 Pa. The nanostructured cellulose cryogels were pyrolyzed in a tubular furnace at a heating rate of 5°C min⁻¹ and a target temperature of 800°C for 2 h. A continuous nitrogen flow of 150 mL min⁻¹ was applied during the pyrolysis process. The resulting 3D-TCE-800 was thermally treated at 1600°C and 2400°C under an Ar atmosphere to prepare 3D-TCE-1600 and 3D-TCE-2400, respectively. The final products were stored in a vacuum oven at 30°C.

Characterization

The bare and ex situ morphologies of the 3D-TCE samples were analyzed by field emission scanning electron microscopy (FE-SEM) (S-4700; Hitachi). X-ray photoelectron spectroscopy (XPS) analysis was performed using Al Kα radiation (Nexsa; Thermo Fisher Scientific) to analyze the surface properties of the 3D-TCE samples. Raman spectroscopy was performed using a laser with a wavelength of 514 nm (LabRam ARAMIS IR2; Horiba). The laser was focused using a 100× optical lens. Additionally, the microstructure of the 3D-TCE samples was investigated by X-ray diffraction (XRD) analysis (SmartLab; Rigaku) with Cu-Kα radiation (λ = 0.154 nm) at 40 kV and 100 mA and through a high-resolution transmission electron microscopy (HR-TEM) (Tecnai 20; FEI). The pore structure of the samples was characterized by nitrogen adsorption and desorption isotherms at 77K (ASAP2020; Micromeritics). Before the measurement, the samples were degassed below 1.33 × 10³ Pa for 1 h and heated (5 K min⁻¹) to 393 K overnight. Ex situ FE-SEM and XPS measurements were conducted by extracting the 3D-TCE samples after the 10th and 100th deposition and dissolution cycles at an areal current rate of 0.5 mA cm⁻². The surface properties of 3D-TCE-800 and 3D-TCE-1600-O were characterized by Fourier transform infrared spectroscopy (FTIR; VERTEX 80 v; Bruker Optics).

Electrochemical characterization

Electrochemical tests were conducted using CR2032-type coin cells and Wonatech automatic battery cyclers in an incubator at 30°C. The working electrode was prepared by puncturing the 3D-TCE samples into a half-inch-sized piece. A metallic zinc foil (Alfa Aesar) was used as both the reference and counter electrodes, and a glass microfiber filter (GF/F; Whatman) was used as the separator. A mild acidic electrolyte was prepared by dissolving 1 M ZnSO₄ in distilled water. Galvanostatic discharge/charge measurements were conducted at different areal current densities with an areal capacity of 0.5 mAh cm⁻². Cyclic voltammetry (CV) and linear sweep voltage (LSV) tests were performed at a scan rate of 10 mV s⁻¹. To assemble symmetric cells, the corresponding 3D-TCE samples were assembled as half cells to deposit zinc metals. Zn metal is pre-deposited on the 3D-TCEs with the capacity of 5 mA h cm^‒2. And then, the same two Zn metal-containing 3D-TCEs extracted from the half-cells were re-assembled as a symmetric full cell.

Computational details

First-principles density functional theory (DFT) calculations were performed to obtain optimized atomic structures and energies using the Vienna Ab-initio Simulation Package software (VASP).⁵⁵ The projector augmented wave (PAW) method was used to treat electron-ion interaction with plane-wave energy up to 400 eV and the exchange-correlation function of Perdew–Burke–Ernzerhof.^56,57 The carbon network models were constructed based on a 5 × 5 supercell of hexagonal graphene and (6, 6), (5, 5), and (4, 4) armchair carbon nanotube structures combined with a 15 Å vacuum region along the surface normal direction. Gamma-centered k-point meshes of 3 × 3 × 1 and dipole corrections were employed. A constant value of +0.24 eV was added to H-binding energy derived from the electronic energy to take into account the zero-point energy and entropy changes.⁵⁸

RESULTS AND DISCUSSION

Microbial-derived natural polymer precursors were pyrolyzed at 800°C and thermally annealed at 1600°C and 2400°C, and the resulting CEMs were denoted as 3D-TCE-800, 3D-TCE-1600, and 3D-TCE-2400, respectively, corresponding to the thermal treatment temperatures. The specific fabrication process and optical images of their monolithic shapes are shown in Figure S1. The morphologies of the samples were characterized by FE-SEM, where 3D macroporous network structures comprising entangled nanofibers were observed (Figure 1A–C). The porous structures are induced from the original morphologies of the microbial-derived cellulose precursors (Figure S2). The size of macropores ranges from several hundreds of nanometers to a few micrometers; therefore, they can accommodate the deposited zinc metals with no volume expansion. HR-TEM images revealed that the nanofiber building blocks are of dissimilar carbon microstructures with diameters of approximately 20–30 nm (Figure 1D–F). The carbon nanofiber of 3D-TCE-800 has an amorphous structure with no distinct graphitic lattices (Figure 1D). The carbon domains slightly developed into graphitic structures upon thermal annealing at 1600°C, where a few layers of graphitic fringes were observed (Figure 1E). Further high-temperature annealing at 2400°C induced a profound transition for the carbon building blocks into significantly developed graphitic structures while maintaining the initial fibrous shape, where well-stacked and enlarged multiple graphitic layers were observed (Figure 1F). More magnified HR-TEM images of 3D-TCE-1600 and 3D-TCE-2400 elucidate their differences in graphitic microstructures (Figure S3).

Figure 1. Material properties of 3D-TCE samples. FE-SEM images of (A) 3D-TCE-800, (B) 3D-TCE-1600, and (C) 3D-TCE-2400; HR-TEM images of (D) 3D-TCE-800, (E) 3D-TCE-1600, and (F) 3D-TCE-2400. (G) XRD patterns and (H) Raman spectra of 3D-TCE-800, 3D-TCE-1600, and 3D-TCE-2400. (I) Nitrogen adsorption and desorption isotherms and (J) micropore size distribution curves of 3D-TCE-800, 3D-TCE-1600, and 3D-TCE-2400. XPS (K) C 1s and (L) O 1s spectra of 3D-TCE-800, 3D-TCE-1600, and 3D-TCE-2400. 3D-TCE, three-dimensional templating carbon electrode; FE-SEM, field-emission scanning electron microscopy; HR-TEM, high-resolution transmission electron microscopy; XPS, X-ray photoelectron spectroscopy; XRD, X-ray diffraction.

The degree of graphitic ordering of the samples was characterized by XRD and Raman spectroscopic analyses. The XRD patterns of 3D-TCE-800 and 3D-TCE-1600 show similar broad graphite (002) peaks centered at 24.0°, which corresponds to a d-spacing of ~3.70 Å (Figure 1G). A sharp graphite (002) peak was observed at 26.0° (d-spacing: ~3.42 Å) in the XRD pattern of 3D-TCE-2400. According to the Scherrer equation, the crystalline domain sizes (L_c) of 3D-TCE-800, 3D-TCE-1600, and 3D-TCE-2400 samples were calculated to be 0.85, 0.86, and 10.92 nm, respectively, which correlates with the HR-TEM result, showing the crystallographic transition at annealing temperatures above 1600°C. The Raman spectra of 3D-TCE-800 and 3D-TCE-1600 also suggest similar D and G bands at ~1355 and ~1601 cm⁻¹, respectively, which originate from the disordered A_1g breathing and E_2g vibration modes of the polyhexagonal carbon structures (Figure 1H). The broad D and G bands are overlapped, and their lateral graphenic domain size (L_a) was calculated from the deconvoluted D to G intensity ratio (I_D/I_G). The I_D/I_G ratio of 3D-TCE-800 and 3D-TCE-1600 were 1.49 and 2.09, respectively (Figure S4).⁵⁹ The small L_a and L_c values in nanometers indicate that the samples have typical amorphous carbon structures with short-range graphitic order. In contrast, the Raman spectrum of 3D-TCE-2400 shows a very sharp and highly intense G band, which is several times higher than its corresponding D band (Figure 1H). From the deconvoluted I_D/I_G value, the L_a of 3D-TCE-2400 was 24.14 nm, which is ≥10 times larger than those of 3D-TCE-800 and 3D-TCE-1600. In addition, the presence of a highly intense 2D band at ~2700 cm⁻¹ substantiates the well-ordered graphitic structure of 3D-TCE-2400.

The specific open surface area and pore structure of the samples are characterized by nitrogen adsorption and desorption isotherm tests in the relative pressure region in the range of 0–1.0 (Figure 1I). The nitrogen adsorption curves of all the samples show a large amount of adsorption in the lower relative pressure region of <0.5, indicating that they have high specific open surface areas for monolayer adsorption.⁶⁰ The initial curve shapes resemble the IUPAC type-I microporous structure corresponding to pores of <2 nm. The micropores could originate from the chinks and/or junctions between the disordered crystallographic carbon domains. Therefore, the higher adsorption quantity at the initial adsorption region indicates the presence of additional micropore volume. With increasing relative pressure, the adsorption gradually increases in a linear shape in the relative pressure region of 0.05–0.8. A gentle slope indicates that the samples have a broad range of mesopores. At a higher relative pressure region of >0.8, the slopes increased steeply. This corroborates the 3D macroporous structures observed in the FE-SEM images. Therefore, the samples exhibited multimodal porous structures. The pore size distribution data obtained by the Barrett–Joyner–Halenda method supports a broad range of pore distributions (Figure S5). In addition, the micropore distribution plots obtained by the Horvath-Kawazoe method revealed that the micropores of all the samples were predominantly distributed in pore sizes of <1.0 nm (Figure 1J). Further, 3D-TCE-800 has a much higher micropore volume, particularly in the ultramicropore section at approximately 5 Å. With high-temperature annealing, the ultramicropore is slightly larger (6–7 Å), and the pore volume is significantly reduced. Specific Brunauer–Emmett–Teller surface areas of 3D-TCE-800, 3D-TCE-1600, and 3D-TCE-2400 are ~421.8, ~236.4, and ~172.3 m² g⁻¹, respectively, where the micropore surface area of 3D-TCE-800 is ~221.2 m² g⁻¹, which is ~7.0 and ~13.5 times higher than those of 3D-TCE-1600 and 3D-TCE-2400, respectively. In contrast, the specific mesopore surface areas of 3D-TCE-800 and 3D-TCE-1600 are similar—200.6 and 204.2 m² g⁻¹, respectively, and that of 3D-TCE-2400 (155.9 m² g⁻¹) is slightly lower. The textural properties of the 3D-TCE series samples are summarized in Table 1.

Table 1 Textural properties of the 3D-TCE series samples

Samples	La (nm)	Lc (nm)	SBET (m2 g−1)	Smic (m2 g−1)	Smeso (m2 g−1)	O/C ratio	O (%)	H (%)
3D-TCE-800	1.4880	0.8493	421.8	221.2	200.6	0.068	4.89	1.01
3D-TCE-1600	2.0875	0.8614	236.4	32.2	204.2	0.030	0.36	0.16
3D-TCE-2400	24.1374	10.9205	172.3	16.4	155.9	0.002	0.08	0
3D-TCE-800-M	1.4942	0.8502	112.8	7.0	105.8	0.017	1.65	1.28
3D-TCE-1600-O	2.0820	0.8612	305.0	63.8	241.2	0.048	4.61	0.26

Abbreviations: 3D-TCE, three-dimensional templating carbon electrode; BET, Brunauer–Emmett–Teller.

The surface properties of the samples were analyzed by XPS (Figure 1K–L). In the deconvoluted XPS C 1s spectra, several carbon bonds, such as C═C, C‒C, C‒O, and O‒C═O, were confirmed in all the samples, where the sp²-hybridized C═C bonding is present in their main structure.⁴⁷ In addition, the deconvoluted XPS O 1s spectra confirm that C‒O and C═O bonds are the major configurations of the oxygen functional groups. The chemical structures of the samples were similar; however, they showed a considerable difference in their oxygen content. The O/C ratios of 3D-TCE-800, 3D-TCE-1600, and 3D-TCE-2400 were calculated as 0.068, 0.030, and 0.002, respectively, from the XPS results. Furthermore, elemental analysis (EA) was also performed to calculate the oxygen contents, and the oxygen weight ratios of 3D-TCE-800, 3D-TCE-1600, and 3D-TCE-2400 were 4.89, 0.36, and 0.08 wt%, respectively. These results indicate that the oxygen functional groups were removed by thermal annealing at 1600°C, and most of the remaining oxygen groups were further removed at a higher annealing temperature of 2400°C. The hydrogen content was also reduced with annealing temperature and the hydrogen weight ratios of 3D-TCE-800, 3D-TCE-1600, and 3D-TCE-2400 were 1.01, 0.16, and 0 wt%, respectively (Table 1).

Electrochemical analysis of the carbon-based samples was performed in a coin-type half-cell configuration based on a two-electrode system with a zinc foil as both the reference and counter electrodes in an electrolyte solution of 1 M ZnSO₄ dissolved in distilled water. In the first galvanostatic Zn metal deposition process, with a cut-off capacity of 0.5 mAh cm⁻² at an areal current density of 2 mA cm⁻², the voltages of the 3D-TCE-1600 and 3D-TCE-2400 abruptly decreased in the negative voltage range, and Zn metal nucleation overpotential was observed in the initial areal capacity region, indicating the presence of few side reactions (Figure 2A and  Figure S6). In contrast, the voltage profile of the 3D-TCE-800 linearly decreased, where an areal capacity of ~0.19 mAh cm⁻² was observed before Zn metal deposition. Accordingly, a poor CE value of 43.65% was exhibited for 3D-TCE-800, which is much lower than those of 3D-TCE-1600 (~91.50%) and 3D-TCE-2400 (~94.10%). The linear sweep voltage (LSV) profile progressed from 1.1 to −0.2 V versus Zn²⁺/Zn, which confirms the presence of a large amount of side reactions at approximately 0.4 V for the 3D-TCE-800, which contrasts the volage profiles of 3D-TCE-1600 and 3D-TCE-2400 (Figure 2B). The CE values of all the samples gradually increased with cycling; however, the average CE value (~89.19%) of 3D-TCE-800 calculated using the values in 2nd–100th cycles showed a large gap compared with those of 3D-TCE-1600 (98.66%) and 3D-TCE-2400 (98.72%) (Figure 2C). Further, a large CE gap was maintained even when different areal current densities were applied for the measurements. To discern the origin of the poor CE values in the 3D-TCE-800, electrochemically active surface areas (ECSAs) of the samples were characterized in the cathodic voltage range of 0.9–1.5 V versus Zn²⁺/Zn (Figure 2D). The cyclic voltammetry (CV) curve of 3D-TCE-800 shows a significantly higher ECSA than those of 3D-TCE-1600 and 3D-TCE-2400, which can be a critical factor in revealing a large number of side reactions in 3D-TCE-800. The electrolyte wettability tests for 3D-TCE-800 and 3D-TCE-1600 reveal that they have good wettability with the aqueous electrolyte (Figure S7). The electrolyte droplets were fully absorbed in 3D-TCE-800 and 3D-TCE-1600 after 4 and 10 s, respectively. Therefore, the large difference in the ECSA could originate from their own material properties rather than electrolyte affinity. In the material properties shown in Figure 1, the critical differences between 3D-TCE-800 and other samples were observed to be ultramicropore volume and oxygen content, which could be directly associated with the large ECSA. In general, the sp²-hybridized polyhexagonal carbon structure has hydrophobicity, wherein the ultramicropores of the carbon materials are inactive for an aqueous electrolyte system. However, oxygen functional groups can endue the carbon structure with polar properties, extending the ECSAs by activating the ultramicropores. Therefore, 3D-TCE-800 has multitudinous ultramicropores and sufficient polarity, which leads to a distinctive water-splitting side reaction. Further, specific experimental evidence for the effects of ultramicropores on 3D-TCE-800 will be discussed in the following section.

Figure 2. Electrochemical performances of 3D-TCE samples. (A) Galvanostatic zinc metal deposition profiles with a cut-off capacity of 0.5 mAh cm−2 at an areal current density of 2 mA cm−2, (B) LSV curves at a scan rate of 10 mV s−1, (C) average CE bar graphs calculated from the CE values between 2nd and 100th cycles, (D) CV curves characterized in the cathodic voltage range of 0.9–1.5 V versus Zn2+/Zn at a scan rate of 10 mV s−1, (E) cycling performances in repetitive zinc metal deposition/dissolution cycles at an areal current density of 2 mA cm−2, and (F) symmetric cell test profiles comprising two same Zn/3D-TCEs of 3D-TCE-800, 3D-TCE-1600, and 3D-TCE-2400. Ex situ FE-SEM images of 3D-TCE-800 observed after (G) 10th and (H) 100th zinc metal deposition/dissolution cycles; and ex situ FE-SEM images of 3D-TCE-1600 observed after (K) 10th and (L) 100th zinc metal deposition/dissolution cycles. Ex situ XPS survey spectra of (I) 3D-TCE-800 and (M) 3D-TCE-1600 characterized after the 100th cycle, and ex situ XPS Zn 2p spectra of (J) 3D-TCE-800 and (N) 3D-TCE-1600 characterized after the 100th cycle. 3D-TCE, three-dimensional templating carbon electrode; CE, Coulombic efficiency; CV, cyclic voltammetry; FE-SEM, field-emission scanning electron microscopy; LSV, linear sweep voltage; XPS, X-ray photoelectron spectroscopy.

In consecutive Zn metal deposition/dissolution cycles with a cut-off capacity of 0.5 mAh cm⁻² at an areal current density of 2 mA cm⁻², 3D-TCE-1600 and 3D-TCE-2400 revealed highly stable cycling behaviors over 2000 cycles with high average CE values of ~99.28% and 99.36%, respectively (Figure 2E). The stable cycling performances were also confirmed in the higher areal current density of 5 mA cm⁻² and higher cut-off capacity of 1 mAh cm⁻² (Figure S8). In contrast, 3D-TCE-800 showed poor cycling performance with an average CE value of ~97.27%, which unexpectedly terminated after 200 cycles. A significant difference in the cycling performances of the samples was also observed in the symmetric cell tests; the symmetric cells comprised two identical 3D-TCEs, including electrochemically deposited Zn metals (Figure 2F). The 3D-TCE-1600 or 3D-TCE-2400-based symmetric cells showed stable charge‒discharge profiles with no voltage changes over 2000 cycles; however, the 3D-TCE-800-based symmetric cell exhibited a continuous increase in the overpotential from the early cycling stage, which destroyed after ~760 cycles. Similar results were also obtained when a higher areal current density of 5 mA cm⁻² and cut-off capacity of 1 mAh cm⁻² were applied for the symmetric cell tests (Figure S9).

Ex situ FE-SEM observation for the 3D-TCE series samples was conducted after the 10th Zn metal deposition process of 0.5 mAh cm⁻² (Figure S10). The discharge products of the series samples are 2D-like shapes, which are similar to each other. In contrast, a significant difference was observed after stripping the deposited Zn metals. Ex situ FE-SEM images of 3D-TCE-800 obtained after the 10th and 100th cycles in the half-cell configuration revealed that many byproducts were formed in the internal area of the macroporous structure (Figure 2G,H). Ex situ energy-dispersive X-ray spectroscopy (EDS) mapping data for the byproduct observed in the ex situ FE-SEM images revealed that it is mainly composed of oxygen and Zn (Figure S11). In addition, the ex situ XPS survey spectrum and Zn 2p spectrum indicated that the byproducts were mainly Zn(OH)₂ (Figure 2I,J). The reaction forming byproducts between the Zn cation and two hydroxide anions is Zn²⁺ + 2OH⁻ → Zn(OH)₂ (Figure 2I,J). Zn(OH)₂ forms when the HER occurs with water-splitting, which demonstrates that the large irreversible capacities of 3D-TCE-800 observed in Figure 2A,C,E originate from the HER and resultant Zn(OH)₂ formation. This interpretation correlates with the LSV curves shown in Figure 2B. In contrast, the ex situ FE-SEM images of the 3D-TCE-1600 obtained after the 10th and 100th cycles showed few byproducts (Figure 2K,L). The ex situ XPS survey spectrum shows less intense Zn 2p and O 1s peaks (Figure 2M), and the Zn 2p spectrum shows that the discharge product is the same as that of Zn(OH)₂ (Figure 2N). The ex situ characterization results revealed that HERs occurred in both 3D-TCE-800 and 3D-TCE-1600 during repetitive Zn metal deposition/dissolution cycles; however, the 3D-TCE-800 caused a higher activity for the HER, suffering from poor CEs and clogging with incessantly deposited Zn(OH)₂ byproducts.

To discern the effects of ultramicropore and oxygen functional groups on the HER, the structure-modified-3D-TCE-800 with low ultramicropore volume and low oxygen contents was prepared by thermally annealing 3D-TCE-800 under an Ar/H₂ mixture gas (0.96/0.04 v/v) flow at 800°C for 2 h. Further, additionally oxidized 3D-TCE-1600 was prepared by thermal treatment of 3D-TCE-1600 under air atmosphere at 450°C for 4 h. The structure-modified-3D-TCE-800 and oxidized-3D-TCE-1600 were referred to as 3D-TCE-800-M and 3D-TCE-1600-O, respectively. After the thermal annealing process, the 3D-TCE-800-M maintained its original macroporous structure; however, its ultramicropore volume and oxygen content were significantly reduced (Table 1 and Figure S12). The specific micropore surface area of 3D-TCE-800-M was reduced by ~7.0 m² g⁻¹, which is >30 times lower than that of 3D-TCE-800 (Table 1 and Figure S12B). In addition, the O/C ratio and oxygen content of 3D-TCE-800-M obtained from XPS and EA analyses were reduced by 0.017 and 1.65 wt%, respectively (Figure S12C,D). Meanwhile, 3D-TCE-1600-O retained its original morphology and micropore structure, whereas the oxygen content increased by 4.61 wt% after the oxidation process (Figure S13 and Table 1). The Fourier-transform infrared (FTIR) spectra of 3D-TCE and 3D-TCE-O further confirm that they have similar oxygen configurations mainly composed of C‒O and C═O bonds at 1012 and 1727 cm⁻¹, respectively (Figure S14).

In the electrochemical analysis results using the CV method, the ECSAs of 3D-TCE-800-M were significantly reduced compared with those of 3D-TCE-800 (Figure 3A). This is owing to the large reduction in the ultramicropore volume, which significantly affected the HER in the Zn metal deposition process (Figure 3B). In contrast, the CV and LSV profiles of 3D-TCE-1600-O are similar to those of 3D-TCE-1600, indicating that the oxygen functional groups are insignificant for the HER during the Zn metal deposition process (Figure 3C,D). In the cycling performance test, the ultramicropore-removed 3D-TCE-800-M showed an enhanced average CE value of ~98.9% and long-term cycling stability over 1000 cycles (Figure 3E). In contrast, the additionally oxidized 3D-TCE-1600-O showed stable cycling behavior with an average CE value of ~98.96% over 2000 cycles, which is similar to that of the original 3D-TCE-1600 (Figure 3F). This result suggests that (1) the water-splitting side reaction triggered by the HER is a major factor in determining the electrochemical performance of CEMs for ZMA and (2) the amorphous carbon structure and oxygen functional groups are insignificant in promoting the HER, while (3) ultramicropores can catalyze the HER.

Figure 3. Electrochemical performance of structure-modified 3D-TCE-800 (3D-TCE-800-M) and thermally oxidized 3D-TCE-1600 (3D-TCE-1600-O). CV curves characterized in the cathodic voltage range of 0.9–1.5 V versus Zn2+/Zn at a scan rate of 10 mV s−1 of (A) 3D-TCE-800 and 3D-TCE-300-M, and (C) 3D-TCE-1600 and 3D-TCE-1600-O. LSV curves of (B) 3D-TCE-800 and 3D-TCE-800-M, and (D) 3D-TCE-1600 and 3D-TCE-1600-O at a scan rate of 10mV s−1. Cycling performances of (E) 3D-TCE-800 and 3D-TCE-800-M, and (F) 3D-TCE-1600 and 3D-TCE-1600-O in repetitive zinc metal deposition/dissolution cycles at an areal current density of 2 mA cm−2. 3D-TCE, three-dimensional templating carbon electrode; CV, cyclic voltammetry; LSV, linear sweep voltage.

The relationship between ultramicropores and HER activity was investigated using DFT calculations. According to the experimental analysis, 3D-TCE-800 comprises amorphous carbon networks with various defect sites and local structural distortions. To mimic these structural characteristics, active site models were prepared based on perfect graphene and single-walled carbon nanotube (CNT) structures of three sizes ((6, 6), (5, 5), and (4, 4)) to which V₂-defects (585 and 555777) were applied (Figure 4A). The H-binding free energy ($$\unicode{x02206}{G}_{{\rm{H}}}$$) is a well-known theoretical descriptor for HER activity; specifically, a $$\unicode{x02206}{G}_{{\rm{H}}}$$ value close to zero (thermoneutral condition) indicates a high HER activity.^61,62 For each model, $$\unicode{x02206}{G}_{{\rm{H}}}$$ was calculated for the ideal graphitic carbon site and four different sites around the V₂ defects, as shown in Figure 4B and Table S1. Notably, $$\unicode{x02206}{G}_{{\rm{H}}}$$ was 1.65 eV for the perfect graphene model, and gradually increasing values of 0.92, 0.80, and 0.61 eV were obtained for the CNTs with the (6, 6), (5, 5), and (4, 4) size, respectively. This result indicates that a substantial structural distortion of the carbon network can be advantageous for HER; however, the ideal graphitic carbon sites have H-bindings that are too weak to yield sufficient HER activity.

Figure 4. First principles calculation data using the DFT method. (A) Atomic structures of graphene and carbon nanotubes with and without V2-defects, and potential H-binding sites for HER. (B) Distribution of the H-binding free energy values according to the type of carbon network. (C) Correlation between H-binding free energy and corresponding structural reorganization energy during the H-binding process. CNT, carbon nanotube; DFT, density functional theory; HER, hydrogen evolution reaction.

The overall enhancements were observed in the $$\unicode{x02206}{G}_{{\rm{H}}}$$ values of the defective carbon sites compared to those of the ideal graphitic sites. Furthermore, the same tendency was observed—the larger the structural distortion, the stronger the H-binding. The averaged $$\unicode{x02206}{G}_{{\rm{H}}}$$ values for graphene, (6, 6), (5, 5), and (4, 4) models were 0.86, 0.32, 0.22, and 0.02 eV, respectively. Notably, several defective sites exhibited $$\unicode{x02206}{G}_{{\rm{H}}}$$ values very close to zero, which indicates that sufficiently high HER activity can originate from carbon-based materials with structural defects and distortions, consistent with previous studies.^63–65 The tendency of enhanced H-binding at defective sites can be correlated with the structural reorganization energy of the carbon networks. During the H-binding process, the carbon sites undergo structural change from planar to tetrahedral geometry, and energy is required for structural reorganization of carbon network. The re-organization energy for each site was calculated by comparing the energies of the carbon structures before and after H-binding (Figure S15), and its correlation with $$\unicode{x02206}{G}_{{\rm{H}}}$$ is shown in Figure 4C. There were slight differences in linearity depending on the base structures; however, $$\unicode{x02206}{G}_{{\rm{H}}}$$ increased with a decrease in the reorganization energy. Therefore, we confirmed that the facile structural reorganization of defective and distorted carbon sites present in the ultramicropore region in the 3D-TCE-800 material is the origin of the good HER activity.

To confirm the feasibility of the 3D-TCE-based anode in Z-RABs, Zn/3D-TCE-1600//V₂O₅ and Zn//V₂O₅ full cells were prepared, and their electrochemical performances were compared in a voltage window of 0.2–1.4 V versus Zn²⁺/Zn at a current rate of 0.1 A g⁻¹ (Figure S16). In the Zn/3D-TCE-1600 and Zn anodes, 1000 mAh cm⁻² of Zn metals were electrochemically deposited on 3D-TCE-1600 and stainless steel substrates, respectively. Also, the V₂O₅ cathode was fabricated by using the previously reported method.⁶⁶ In the second galvanostatic charge/discharge curves of the Zn/3D-TCE-1600//V₂O₅ cells, a specific capacity of ~308 mAh g⁻¹ was obtained and the capacity was sharply decreased with cycles (Figure S16A). After 30 cycles, approximately 20% of the initial capacity was retained (Figure S16). In contrast, the Zn/3D-TCE-1600//V₂O₅ cells revealed better capacity retention, where approximately 65% of the initial capacity (~200 mAh g⁻¹) was maintained after 30 cycles (Figure S16C,D). This result clearly supports the positive effect of 3D-TCE-1600 in ZMA for Z-RABs.

CONCLUSION

In summary, 3D-TCEs with different local graphitic orderings, pore structures, and surface properties were prepared, and their electrochemical performance as an electrode for ZMA was investigated. The 3D-TCE-800 sample, which has a high ultramicropore volume and oxygen functional groups, suffered from the high activity for the HER and corresponding large amounts of Zn(OH)₂ byproduct formation, showing poor CE values and insufficient cycling stabilities. In contrast, 3D-TCE-1600, which has a graphitic microstructure similar to that of 3D-TCE-800, exhibited highly stable cycling behaviors over 2000 cycles with high average CE values of ~99.28%, and the cycling stability and average CE values were slightly increased in 3D-TCE-2400 with highly ordered graphitic structures. These results indicate that a local graphitic microstructure is not a key factor in the electrochemical performance of CEMs for ZMA. In addition, the increased oxygen content in 3D-TCE-1600-O did not affect its electrochemical performance compared with that of 3D-TCE-1600, demonstrating that the oxygen groups of CEMs are not essential factors for the HER. However, ultramicropore-reduced 3D-TCE-800 (3D-TCE-800-M) revealed highly reduced HER and much less Zn(OH)₂ formation, leading to improved cycling performance and significantly increased average CE values. DFT calculation data also showed higher HER activity in ultramicropores of CEMs, substantiating our claim that ultramicropore-free CEMs with well-developed graphitic ordering and high surface area can lead to high-performance ZMA for Z-RABs.

ACKNOWLEDGMENTS

This study was supported by the Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education (NRF-2019R1A2C1084836, NRF-2021R1A4A2001403, and NRF-2022R1C1C1011484).

CONFLICTS OF INTEREST

The authors declare no conflicts of interest.