Longitudinal Hierarchy Co3O4 Mesocrystals with High-dense Exposure Facets and Anisotropic Interfaces for Direct-Ethanol Fuel Cells

Novel electrodes are needed for direct ethanol fuel cells with improved quality. Hierarchical engineering can produce catalysts composed of mesocrystals with many exposed active planes and multi-diffused voids. Here we report a simple, one-pot, hydrothermal method for fabricating Co3O4/carbon/substrate electrodes that provides control over the catalyst mesocrystal morphology (i.e., corn tubercle pellets or banana clusters oriented along nanotube domains, or layered lamina or multiple cantilevered sheets). These morphologies afforded catalysts with a high density of exposed active facets, a diverse range of mesopores in the cage interior, a window architecture, and vertical alignment to the substrate, which improved efficiency in an ethanol electrooxidation reaction compared with a conventional platinum/carbon electrode. On the atomic scale, the longitudinally aligned architecture of the Co3O4 mesocrystals resulted in exposed low- and high-index single and interface surfaces that had improved electron transport and diffusion compared with currently used electrodes.


Purification and functionalization of MWCNTs
The raw MWCNTs were initially mixed with HNO 3 /H 2 SO 4 solutions at a 1:3 ratio. The mixture was then ultrasonicated at 50 °C for 5 h and then refluxed at 110 °C for 3 h. Afterward, the resulting mixture was diluted at pH 7 using deionized water. The solid (i.e., oxidized MWCNTs) was collected by centrifugation and dried at 65 °C overnight for later use.

Preparation of graphene oxide
Graphene oxide was successfully synthesized from graphite powder in accordance with the slightly modified Hummers method 1 . Typically, 2 g of graphite powder (~40 μm) was reacted with a strong oxidizing solution of concentrated H 2 SO 4 and HNO 3 (100 mL, 1:1 v/v) by vigorous stirring for 1 h at 25 °C. Afterward, the solution was placed in an ice-water bath, and 7 g of KMnO 4 was slowly added to the solution under stirring for 2 h. The mixture was then ultrasonicated for 6 h at 40 °C to avoid obtaining a homogeneous reaction solution. Deionized water (350 mL) was then mixed with the formed gel, and the whole mixture was stirred at 80 °C for 1 h. Subsequently, 120 mL of 30% H 2 O 2 and 50 mL of 15% HCl solutions were added to the mixture and then stirred for 10 min. The gel mixture was allowed to stand until brownish precipitation. The solid materials were washed repeatedly with double-distilled water and dried overnight at 60 °C prior to the next investigation.

Preparation of C-NT/CoO CPs/3D PNi nanostructures
A simple, one-pot method for the synthesis of C-NT/CoO CPs/3D PNi electrode was achieved through a microwave-assisted technique. In this method, 0.1 M cobalt nitrate hexahydrate and 0.35 M HMT were mixed with 37 mL of deionized water while stirring for 10 min. Then, 80 mg of oxidized C-NTs were immersed into the above solution and vigorously stirred for 3 h. The mixture was subsequently transferred into a Teflon-lined autoclave containing a 3D PNi sheet (Scheme 1) and then subjected to microwave irradiation (600 W) at 160 °C for 1 h. The collected sample was carefully washed with ethanol and deionized water, dried at 60 °C overnight, and finally calcined at 400 °C for 4 h in air.

Fabrication of C-NT or g-C/Co 3 O 4 or CoO/GC
To investigate the effect of the carrier substrate on EOR efficiency using a longitudinal electrode design, several electrochemical experiments were conducted using C-NT or g-C/Co 3 O 4 or CoO/GC electrodes at electrochemical conditions similar to those applied to the C-NT or g-C/Co 3

O 4 or
CoO/3D PNi electrode assays. The thin-film-layered C-NT or g-C/Co 3 O 4 or CoO/GC electrode was fabricated by dispersing the active catalyst powder onto a GC substrate through a heterogeneousassisted ink-deposition method at 25 °C. The homogeneous catalyst ink was prepared by mixing 5 mg of active catalyst to 50 μL of 5 wt% Nafion solution employed as a binder in 2 mL of Milli-Q water under ultrasonication for at least 30 min. The catalyst ink (4 μL) was then loaded onto the active area of a GC (Φ = 3 mm) at ~0.143 mg/cm 2 . The catalyst ink-loaded GC electrode was dried in a sealed oven at 50 °C to allow the formation of uniform catalyst layers over the GC substrate area.

Characterization
The morphologies of the annealed samples were investigated by FE-SEM (JEOL Model 6500) at 15 kV. The C-NT or g-C/Co 3 O 4 or CoO/3D PNi electrodes were fixed onto the FE-SEM stage using carbon tape before insertion into the FE-SEM chamber. The ion sputter (Hitachi E-1030) was used to deposit thin-layered Pt films on electrodes at 25 °C.
A focused ion beam (FIB) system (JEM-9320FIB) operated at accelerating voltages from 5-30 kV with variable steps of 5 kV and magnification ranging from 150× to 300000×. The orientation axis (X and Y) of the powder samples containing C-NT or g-C/Co 3 O 4 or CoO catalysts can be changed within ±1.2 mm through a tilt angle of ± 60°. The samples were inserted inside the FIB machine using a bulk-sample holder (8 × 8 mm 2 ) after deposition by a carbon protection layer. Before FIB investigation, the powder samples of the C-NT or g-C/Co 3 O 4 or CoO catalysts were mixed with small amounts of epoxy (Gatan, Inc.) onto a small silicon wafer using a fine eyelash probe to form very thin films on the silicon substrate (Figures 2 and S7). Each thin film was baked on a hot plate at 130 °C for 10 min and subsequently coated with a uniformly thin carbon layer of about 30 nm. The samples were inserted into the FIB microscope operated at 30 kV and then roughly milled on both sides until a final thickness of 2 μm using −1.5° and +1.5° tilts. Afterward, the C-NT or g-C/Co 3 O 4 or CoO sample was cut and removed from the FIB system for subsequent HAADF-STEM microscopy.
HAADF-STEM was employed to perform (i) TEM and (ii) STEM, (iii) EDS for elemental mapping, and (iv) electron diffraction (ED). The HAADF-STEM micrographs were recorded using a JEM-ARM200F-G instrument supplied with aberration correctors at the illumination and imaging lens systems to observe TEM/STEM images at high resolution. The HAADF-STEM microscope was also equipped with a monochromated electron gun and supported by electron energy-loss spectroscopy at a high-energy resolution. Specifically, the cross-section specimens for HAADF-STEM was prepared by FIB system milling. The fine trapped probes typically sharpened the sample in the parallel direction of the longitudinal c-axis. The well-prepared FIB samples were attached to a silver grid by epoxy materials using a pick-up system. The C-NT or g-C/Co 3 O 4 or CoO attached to the silver grid was inserted again into the FIB system to produce a 100 nm-thick layer. The sample was thinned from both sides by using alternate beams with variable intensity until the final thickness of 100 nm. The 100 nm sample was viewed under the HAADF-STEM microscope to record the cross-sectional images.
The surface properties of the material involving the pore structure distribution and surface area were estimated by N 2 adsorption-desorption isotherms at 77 K using a BELSORP36 analyzer (JP. BEL Co., Ltd.). The samples were thermally treated at 200 °C for at least 6 h under N 2 atmosphere. The specific surface area (S BET ) was calculated using the Brunauer-Emmett-Teller (BET) method with multipoint adsorption data from the linear section of the N 2 adsorption isotherm. The pore size distribution was determined using nonlocal DFT (NLDFT).
The structural geometry of the catalysts was further examined by WA-XRD. The WA-XRD patterns were recorded using a 18 kW diffractometer (Bruker D8 Advance) at scan rate of 10°/min with monochromated Cu Kα -X-radiation (λ = 1.54178 Å). The DIFRAC plus Evaluation Package (EVA) software with the PDF-2 Release 2009 databases provided by Bruker AXS was used to analyze the diffraction and structure analysis diffraction data. The TOPAS package program was applied to integrate various types of X-ray diffraction (XRD) analyses.
XPS analysis was conducted on a PHI Quantera SXM (ULVAC-PHI) instrument (Perkin-Elmer Co., USA) equipped with Al Kα as an X-ray source for excitation (1.5 mm × 0.1 mm, 15 kV, 50 W) under a pressure of 4 × 10 −8 Pa. A thin film of the sample was deposited on a Si slide before the start of analysis.
Raman spectroscopy (HR Micro Raman spectrometer, Horiba, Jobin Yvon) was conducted using an Ar ion laser at 633 nm. A CCD (charge coupled device) camera detection system and the LabSpec-3.01C software package were used for data acquisition and analysis, respectively. To ensure the accuracy and precision of the Raman spectra, 10 scans of 5 s from 300 cm −1 to 1,600 cm −1 were recorded.
TG and DTA were achieved using a simultaneous DTA-TG Apparatus TG-60 (Shimadzu, Japan).

Electrochemical measurements
Electrochemical measurements were obtained in a home-made electrochemical cell using mercury/mercury oxide (Hg/HgO, 1 M NaOH) and platinum wire (Φ = 0.1 mm) as the reference and counter electrodes, respectively. The active catalyst C-NT or g-C/Co 3 O 4 or CoO grown on 3D PNi with surface structures of CPs, LSs, MCSs, and BCs and with active loading of l.5 mg served as the working electrodes for electrochemical investigation. The data were recorded using a Zennium/ZAHNER electrochemical work station (Elektrik GmbH & Co. KG) controlled by the Thales Z 2.0 software. Initially, all of the working electrodes were cycled at least 10 times at a scan rate of 50 mV s −1 until the signals were stabilized. Then, the CV data were collected. Current density refers to the geometrical surface area of the investigated working electrodes (1 cm 2 ). The measured potentials were reported with respect to the Hg/HgO reference electrode. The freshly prepared electrolyte (0.5 M NaOH) was de-aerated by bubbling a slow stream of purified N 2 above the electrolyte in the electrochemical glass cell. The N 2 flow was maintained during the electrochemical measurements to ensure an N 2 -saturated electrolyte. To guarantee the reproducibility of the recorded results, freshly prepared electrolyte solutions were used for every electrochemical measurement.

Mathematical modelling
DFT is a promising approach to effectively illustrate the electronic correlation effects. In this study, all calculations investigated by DFT were performed in accordance with the DMol3 of BIOVIA Dassault systems 2,3 . The exchange-correlation energy function was represented by the Perdew-Burke-Ernzerhof (PBE) formalism 4 . The Kohn-Sham equation was expanded in a double numeric quality basis set (DNP) with polarization functions. To consider the relativistic effect, the DFT Semicore Pseudo-potentials 5 were used for the treatment of the core electrons of the doped clusters. The orbital cutoff range and Fermi smearing were selected as 5.0 Å and 0.001 Ha, respectively. The selfconsistent-field (SCF) procedures were performed to obtain well-converged geometrical and electronic structures at a convergence criterion of 10 −6 a.u. The energy, maximum force, and maximum displacement convergence were set to 10 −6 Ha, 0.002 Ha/Å, and 0.005 Å, respectively.
Meanwhile, the electrostatic site potential is a measure of the Coulomb interaction per unit charge experienced by an ion at a given position in space. DFT was also used to calculate the electrostatic potential (EP) distribution. Modeling was performed to show a physical quantitative survey at each point on the isosurfaces using a feature of the surface-charging map. Typically, the isosurfaces of the electron densities were colored on the basis of EP intensities (EPI) using a lattice representation in which the charges are mapped on the cubic lattice in the so called contour where the EP is calculated.
The slab model was constructed with nine atomic layers of each catalyst ( Figure 6). To compare the active center within the structure, oxygen atoms at the surface and subsurface layers were involved in the stoichiometric mode. EP was investigated over the range of −0.06 eV to +0.6 eV as shown in the optimized model

Hydrothermal-assisted formation of hierarchal Co 3 O 4 nanocomposites
Scheme 1 shows the morphological evolution of the designed electrodes. The C-NT or g-C/Co 3 O 4 nanohybrid structures derived from the hierarchical metal framework grown directly on the 3D PNi substrate with robust mechanical adhesion were obtained after adequate pyrolysis of the assynthesized C-NT or g-C/ Co(OH) x (CO 3 ) 0.5 •0.11H 2 O/3D PNi electrodes at 400 °C for 4 h (additional details are found in the Experimental section). We adopted scalable and flexible methods to fabricate morphology-controlled nanohybrids along the longitudinal direction vertically oriented toward the 3D PNi skeleton. In these processes, the surface morphologies of the CPs and BCs in the NR architecture and MCSs and LSs in nanosheet dominates were engineered. GO sheets (g-C) or functionalized multi-walled carbon nanotube (C-NT) counterparts acted as carbon supports.
Basically, the unique structures of the fabricated electrodes were contributed by the directing basic salt (urea and HMT) and cobalt precursor. To well understand our unique and reasonable electrode building along the longitudinal scales, Table S1 summarizes some of the reported materials based sheets, nanowires, and rod structures by different routes. Moreover, such a unique structure provides well-defined pathways for the electrolyte solution to pass through the graphene layer and along the carbon tubes, leading to fast electron transport.
Interestingly, the controlled longitudinal growth of the C-NT or g-C/Co 3 O 4 or CoO/3D PNi electrodes generates more favorably exposed surfaces and interfaces containing an extensive domain of Co 3+ active sites. This effect enhances the kinetics of the EOR, as evidenced by the HR-HAAF-STEM micrographs (Figures S10).

Supplementary S1
Addition of urea and hexamethylenetetramine (HMT, C 6 H 12 N 4 ) agents is essential to achieve NR-, LS-and MCS-like morphologies, determine the orientation of growth of Co 3 O 4 or CoO mesocrystals, and construct the atomic structure along with the active exposure of low-and high-index single and interface plane surfaces ( Figures S1-S4). The dose of urea species significantly affects the formation of CP-(high urea concentration) and BC-(low urea concentration) NR Co 3 O 4 structures with specific crystal planes and, consequently, the catalytic performance in EOR. In addition, the proposed growth mechanism of Co 3 O 4 or CoO mesocrystals in hybrid or electrode fabrics most likely involves two stages: (i) homogeneous nucleation of cobalt seeds potentially forming thermodynamically stable and actively centered nuclei sites of Co(CO 3 ) 0.5 (OH) x ·0.11H 2 O composition domains and (ii) time-and temperature-dependent controlled growth of active and stable-centered seeds to achieve longitudinal growth around the c-axes in the final structures.

Supplementary S5
To further illustrate the thermal conversion from the precursors Co(OH) x (CO 3 ) 0.5 •0.11H 2 O to the final product Co 3 O 4 , thermal gravimetric analysis (TGA) was conducted. Through this approach, the thermal properties and chemical composition of the Co(OH) x (CO 3 ) 0.5 .0.11H 2 O CP precursor powder were initially determined ( Figure S5). A DSC plot of the sample shows two exothermal peaks located in the gravimetric gain region centered at 228 and 320.5 °C. At around 395 °C, the total weight loss of the sample is about 17.45%, which was ascribed to the successive release of CO 2 during the thermal decomposition of the Co(OH) x (CO 3 ) 0.5 •0.11H 2 O to Co 3 O 4 and the evaporation of adsorbed water molecules. As revealed by thermogravimetry (TG)-differential thermal analysis (DTA), no weight loss was observed for the precursor after 400 °C, indicating complete decomposition from the precursor to Co 3 O 4 . On the basis of the TGA data, we performed the thermal heat treatment of the sample at 400 °C.  double enhancement of the surface-area-to-volume ratios of exposed catalytic surfaces, and (iii) controlled uniformity of the mesopore distribution ( Figures S7A-S7F). These surface structural features enable the multi-accessible windows of ethanol fluids to bind with interior active Co 3+ sites during the EOR process.

Supplementary S8
To investigate the atomic core-level organization of the geometrical C-NT or g-C/Co 3

Supplementary S10
The crystallographic nature and morphological shape of the individual mesocrystals of the hierarchical C-NT/Co(OH) x (CO 3 ) 0.5 •0.11H 2 O BC-NRs were evident from the HAADF-STEM and ED images (inset), as shown in Figures S10A and S10B. With high-temperature treatment, no change was observed in the morphological shape of the BC-NRs (Figures S8C-S8E).

Supplementary S11
To control the electrochemical performance of the electrodes, the cyclic voltammetry (CV) technique was initially applied in 0.5 M NaOH using a N 2 -saturated electrolyte at a scan rate of 50 mV s −1 at room temperature (details are available in the Electrochemical Measurements segment of the Supplementary Section). Notably, the current density of well-cleaned 3D PNi was basically negligible (~0.5 mA/cm 2 ) compared with that of the proposed electrode designs ( Figure S11A).
Under pure alkaline conditions, all of the electrodes revealed two sets of redox couples produced from the reversible reactions of Co 3 O 4 and CoOOH (peaks I/IV) and between CoOOH and CoO 2 (peaks III/II) (Figures 3, S11, S12, S13, S14A-D, and S15). To study the synergetic contribution of the g-C and C-NT counterparts to the EOR electroactivity of the hybrid electrodes, we carried out a set of electrochemical experiments (see Figure S11-S14). The CVs of pure Co 3 O 4 and g-C or C/Co 3 O 4 based electrodes were recorded ( Figures 4A-B, S14E and S11D-E). Interestingly, the remarkable enhancement of the catalytic activity of the g-C or C-NTs/Co 3 O 4 electrodes is evident. This result indicates the significant role of the counterparts to provide a fast charge transport, to enhance the diffusion of active species along the electrode building 25,26 and to afford abundant of active sites for EOR. Moreover, the g-C or C-NTs/Co 3 O 4 electrodes show higher anodic current densities and lower onset potentials than that of pristine Co 3 O 4 electrodes (Figs 4A-B and S11F). Our finding indicates that the highly conductive g-C counterpart provides double current density and facile electron transport at the electrode-electrolyte interfaces of the C-NT support, which leads to the ultrafast electron transport kinetics of the electrooxidation reactions (See Supporting S11B&C) 16,19 . Consequently, the facile charge/electron transport through the tunneling C-tubular cylinders or ordered g-C sheet layers leads to the effective removal of intermediates and the continuous oxidation of organic molecules onto the electrode surfaces 41

Supplementary S13
To further illustrate the superior effect of the direct longitudinal growth of electroactive C/Co 3 O 4 catalysts into free-supporting 3D PNi electrode on electrochemical activity in terms of current density and stability of electrodes, we performed the EOR using a C-NT or g-C/Co 3 O 4 /GC electrode ( Figures S13-S15). The GC-based electrode was fabricated through a heterogeneous-assisted ink deposition method (see Experimental Section in Supplementary Information). A set of EOR catalytic experiments was also performed using both electrodes based on C-NT or g-C/Co 3

O 4 and C-NT/CoO
CPs to show the effective influence of Co 3+ site composition on EORs.
Similar to the electroactivity of the 3D PNi modified electrodes in the absence or presence of ethanol, the catalytic activity of the GC-based electrodes was improved indiscriminately by the synergetic role of the counterparts ( Figure S15). However, the disparity in catalytic activity ( Figure S13) among the electrodes can be ascribed to the interactions of the catalytically exposed sites. Particularly, the interplay of Co 3+ with the counterparts added significant advantages to the catalytic activity for the EOR. A higher change in the relative current of GC-based electrodes was observed after long-term cycling (i.e., 18,000 s) relative to that of the free-supporting 3D PNi -based electrodes. This finding indicates the efficient design and stability of the PNi-based electrode ( Figures S13-S13 Among the electrodes, the C-NT/Co 3 O 4 CP electrode exhibited the highest activity, suggesting more exposed active sites in this electrode, which indicates that numerous Co 3 O 4 mesocrystals have participated effectively in the reaction.
The lower onset potential of the carbon/Co 3 O 4 electrode with respect to those of other electrodes denotes the lower overpotential and superior kinetics for the EOR. Moreover, the relatively high oxidation current and lower onset potential demonstrate the higher catalytic activity of the electrode towards EOR. On the other hand, the C-NT/CoO electrode mainly contains a wide range of Co 2+ active sites that are inactive for EOR. The lower electroactivity and slow kinetics of the C-NT/CoO CP electrode can be explained by its lower oxidation current and higher positive onset potential.

EOR measurement of C-NT/Co 3 O 4 CPs/GC electrode
The electrochemical performances of the C-NT/Co 3 O 4 CPs/GC modified electrode were evaluated by CV (Figures S14-S15) in 0.5 M NaOH in the absence and presence of ethanol. By considering the same active mass of the C-NT/Co 3 O 4 CP catalyst uniformly loaded on both substrates, the C-NT/Co 3 O 4 CPs/3D PNi conductive substrate creates a cell capable of multiplying the efficiency of EOR to more than three times higher in current density than the film-casted/GC-based electrode ( Figure S14). The marked currents of the C-NT/Co 3 O 4 CPs/3D PNi electrode is mainly due to the direct contact of the active material to the current collector (3D PNi), which ensures fast electron transport. Moreover, the absence of polymeric binders can greatly improve the electrode's electrocatalytic activity.  Supplementary S16

Effects of ethanol concentration and scan rate on the performance of C-NT/Co 3 O 4 CPs/3D PNi electrode
In this electrochemical assay, several key factors, including ethanol concentration and scan rate, significantly affected the EOR activity ( Figures S16 and S15). Concentrated ethanol solution is known to be suitable for use in DEFCs to minimize their size. In this study, ethanol concentration was changed from 0.05 M to 0.5 M. Figure S16A and S15 presents the CV profiles of C-NT/ The gradual increase in cathodic peak with a slightly negative shift with increasing scan rate suggests that the EOR in the cathodic path is scan-rate dependent. This trend shows that ethanol oxidation of the higher-valence metal oxides is a rate-determining process, as reported by Fleishmann et al. 33 . The ratio of the anodic to cathodic current (Ia/Icat) ( Figure S16E) is high at low scan rates, which might be due to the rapid formation of the active CoOOH layer. The change in the Ia/Ic ratio is approximately negligible after a sweep rate of 100 mV s −1 . The catalytic rate constant was measured in accordance with the reported equations (3, 4) as follows 34 : where I C is the catalytic current of C-NT/Co 3 O 4 CPs/3D PNi in the presence of ethanol measured from the current-time spectra ( Figure 3C-a), I L is the limiting current in the absence of ethanol measured from the current-time data ( Figure S17A), γ is the error function and equal to kC o t, and C o is the ethanol concentration. As the value of γ exceeds 2, the error function will be equal to 1, and subsequently the above equation can be reduced to: where k, C 0 , and t are the catalytic rate constant (cm 3 mol −1 s −1 ), ethanol concentration (mol cm −3 ), and time duration (s), respectively. The value of k was evaluated from the slope of the I C /I L versus time 1/2 graph. It is a measure of the kinetics of EOR at the electrode. Figure S17B shows the I C /I L plot at C-NT/Co 3 O 4 CPs/3D PNi electrode. Results show that the average value of k was found to be 4.56 × 10 3 cm 3 mol −1 s −1 , indicating the high EOR kinetics.
Furthermore, the linear dependence of the net currents I C ( Figure S17C) and I L ( Figure S17D

Stability of the hierarchical structures of electrodes with reuse/cycles
To explore the long-term stability of the longitudinal electrode designs after multiple reuse/cycles, the hierarchical structures, orientational surface crystals, and morphological shapes of the g-C or C-NT/Co 3 O 4 /3D PNi electrodes were investigated by HAADF-STEM ( Figure S19). Figure S19 shows evidence that the hierarchical BC structures, for example, were well preserved without any changes  that among all the designed electrodes in this study, the C-NT/Co 3 O 4 CPs/3D PNi electrode showed smaller semicircle diameters than the other electrodes, as shown in Figure 3F. The highly exposed energy surface interfaces longitudinally aligned along NRs, LSs, and MCSs considerably reduce internal resistance, provide continued pathways for electron/ion transfer, and subsequently increase the redox reaction rate. These advantages contribute to the fast kinetics of electron transport, high catalytic activity, and durability with multiple cycles that are required for DEFCs ( Figure 3 and Supplementary Sections S13, S14 and S16-S18).

Proposed mechanism of EOR
Basically, the proposed mechanism of EOR can be interpreted and analyzed on the basis of the adsorption phenomena of ethanol molecules and the reaction kinetics on the anisotropic surfaces (Scheme 2). In this EOR process, the oxidation of ethanol occurs in multiple steps and may generate