Direct Growth of MoS2 Nanowalls on Carbon Nanofibers for Use in Supercapacitor

Direct growth of MoS2 nanowalls on vapor grown carbon nanofibers (VGCNFs) has been achieved using a microwave-assisted hydrothermal (MAH) method under an acidic condition. The acidic condition was obtained through the addition of an HCl aqueous solution. We demonstrate that the HCl not only modifies the pH value for limiting the growth rate but also leads to the formation of NaCl, which is the key for the direct and unique growth of MoS2 on the VGCNF surface. A growth mechanism is therefore proposed. The growth of MoS2 onto the high electrically conducting VGCNF creates a unique structure that not only reduces the aggregation of MoS2 but also improves the electrical conductivity of the resulting composite electrode. The MoS2 nanowall/VGCNF composite shows Csp as high as 248 F g−1 at 5 mV s−1 and excellent electrochemical stability with a retention of 96% after 1,000 cycles at a high charge rate of 200 mV s−1. The ease of composite fabrication and electrochemical stability suggest that the MoS2 nanowall/VGCNF composite is a promising candidate electrode material for supercapacitor.

a template to grow MoS 2 15 . It was found that more RGO than MoS 2 is needed in order to form few-layered but not multiple stacked-layered MoS 2 . This allows the composite to give enhanced inter-and intra-sheet charge storages and better retention (92% after 1,000 cycles). MoS 2 /N-doped graphene (NDG) composite has been synthesized using a hydrothermal process in which the growth of MoS 2 nanosheets and the formation of NDG occur in the same pot 11 . The NDG contains N-and O-functional groups which improve the wettability of the electrode surface and allow additional faradaic reactions, hence leading to enhanced pseudocapacitance. However, graphene or its derivatives are easily to stack or re-stack and aggregate. Therefore, there is a need of surface modification to prevent these phenomena. On the other hand, the addition of a conducting polymer, especially polypyrrole (PPy) and polyaniline (PANI), into MoS 2 for enhanced electrical conductivity is also reported 10,17 . However, the major drawbacks of the resulting composites include poor cyclic life due to the irreversible redox reactions and highly undesirable degradation of the composite during the charge-discharging process.
As mentioned above, MoS 2 has been mixed with graphene or its derivatives, however, the growth of MoS 2 on a carbon material is not direct. The contact between a carbon material and MoS 2 is limited, leading to large interface resistance. Carbon is normally negatively charged 18 and so is MoS 2 in acid condition 19 . Under such as circumstance, the use of surfactant or the existence of some functional groups on the carbon surface is required to improve the interfacial contact between MoS 2 precursor and carbon. Up to now, several approaches have been taken. Glucose was used to act as a binder to help the interaction between MoS 2 precursor and carbon [20][21][22] . Cationic surfactant (cetyltrimethylammonium bromide) modifies graphene surface through electrostatic interaction, thus reducing the charge incompatibility between the graphene and the MoS 2 precursor 23 . Also, gemini surfactant homogeneously functionalizes the surface of GO so that the wettability of GO is improved 24 .
In this work, we demonstrate the synthesis of MoS 2 nanowalls on a carbon material, namely, vapour grown carbon nanofiber (VGCNF), without the use of any binder, surfactant, or surface modification. MoS 2 nanostructures especially nanoflower-like MoS 2 have been synthesized using the conventional hydrothermal technique 5, 25-28 . However, this approach requires a long reaction time ranging from 15 to 48 hours. Here we have used a facile microwave-assisted hydrothermal (MAH) method for the synthesis of MoS 2 nanowalls in a just few minutes under an acidic condition. The pH value plays an important role in controlling the growth of MoS 2 , for example, few-layered MoS 2 exposing plenty active sites was obtained under an acid condition that reduced the MoS 2 growth rate 29 . The acidic condition was achieved through the addition of an HCl solution. However, we show that the HCl in fact not only modifies the pH value but also leads to the formation of NaCl in the solution. The formation of NaCl is desirable since it improves the hydrophilicity of the carbon nanofiber by providing additional O-functional groups and hence makes the surface charge to become more positive 30,31 . Therefore, we demonstrate that the presence NaCl on the surface of the VGCNF allows the growth of MoS 2 on the carbon nanofiber. Furthermore, to our knowledge, MoS 2 /VGCNF composite for use as a supercapacitor electrode has not been reported. VGCNF is a highly graphitic carbon, exhibiting excellent thermal and electrical conductivity [32][33][34][35] . As a result, the growth of MoS 2 on VGCNF creates a unique structure that not only reduces the aggregation of MoS 2 but also improves the electrical conductivity of the resulting electrode.

Results and Discussion
Structure and morphology of MoS 2 nanowall/VGCNF composites. The synthesis of MoS 2 /VGCNF composites was conducted through a simple MAH method. Prior to the MAH treatment, the solution was found to be milky. After the solution was dried, the obtained solid precursor sample was subjected to XRD analysis. The resulting XRD diffraction pattern shows that the solid precursor consists of NaCl, Mo 4 O 11 , and MoO 3 (Fig. 1). The milky appearance is therefore due to the formation of the Mo oxide particles as shown in Figure S1 in the Supporting Information. Before the MAH treatment, the HCl in the precursor solution triggers the hydrolysis of Na 2 36,37 . It is noted that, based on the XRD peak areas, the amount of the Mo 4 O 11 was found to be about 40% that of the MoO 3 . Also, the CH 4 N 2 S reacts with H 2 O and form H 2 S gas 10 . H 2 S gas is known to react Scientific RepoRts | 7: 5999 | DOI:10.1038/s41598-017-05805-z with MoO 3 to form MoS 2 according to Eq. 1A 7 . In the meantime, we believe that the Mo 4 O 11 also reacts with H 2 S in a similar fashion as shown in Eq. 1B.
Although the above reactions explain the formation of MoS 2 itself, as mentioned above, the growth of MoS 2 on a carbon material is not straight forward. The formed NaCl prior to the MAH process plays a crucial role that allows the direct growth of MoS 2 on VGCNFs. It was found that the zeta potentials of individual VGCNFs and MoS 2 solid precursors are −56.6 and −15.7 mV, respectively. The negatively charged species hence repel each other. However, during the MAH synthesis, the presence of NaCl in the solution modifies the surface of VGCNF by giving various O-functional groups 30 . As to be shown later, O-functional groups such as -OH and -COOH pair with the carbon lattice on the VGCNFs surface such that the wettability the Mo radicals on the carbon nanofibers is increased, hence allowing the growth of MoS 2 on the VGCNFs. Furthermore, a recent report shows that NaCl changes the surface charge of graphene oxide from negative to positive 31 . In this study, the critical role of NaCl will be discussed later. Also, with a higher molar ratio of Mo precursor, more MoS 2 was formed not only heterogeneously on the VGCNFs but also homogeneously by the VGCNFs. The detailed morphology will be explained with the SEM analysis given below.
The crystalline phase was determined using XRD. All the composites show the hexagonal phase of MoS 2 (JCPDS 37-1492) with several major diffraction peaks at 12.9°, 33.3°, 37.6°, and 58.2°, corresponding to the (002), (100), (103), and (110) planes of MoS 2 , respectively (Fig. 2a). There is no other peak observed in the XRD patterns, indicating the phase purity of the obtained MoS 2 . No carbon peak was detected due to the small amounts of VGCNFs, except that a carbon peak at 26.4° was detected for MPR25-3 which has the highest amount of VGCNFs. It is noted that there is no peak at 14.4° indicating the formation of partially exfoliated MoS 2 38 . We can infer that the MoS 2 in the composites consists of a few layers which can be observed obviously using TEM analysis to be shown later. However, it is seen that the intensities of peaks are low and the peaks are broadened, indicating poor crystallinity of the MoS 2 . With increasing VGCNFs, the peak at 12.9°, corresponding to the MoS 2 (002) plane diffraction, becomes more broadened. The full width half maximum (FWHM) are 3.7, 3.2, 5.2, and 7.3 for MoS 2 , MPR25-1, -2, and -3, respectively. It is believed that the presence of VGCNFs restrains the growth of layered MoS 2 crystal during MAH synthesis especially the (002) plane 11,16 , hence resulting in reduced crystal size of MoS 2 . Furthermore, the (002) peak shifts to the smaller angle side after the addition of VGCNFs. Such a peak shift is more pronounced for Sample MGO, whose (002) peak shifts to an even lower angle of 10.4° (Fig. 2a). The peak shift indicates both the expansion of the d-spacing and the existence of strains 25 . The expansion is obviously as shown in the HR-TEM images given in Fig. 2b. Accordingly, the interlayer distances of bare MoS 2 , MPR25-1, and MPR25-3 are 0.62, 0.64, and 0.74 nm, respectively. We also believe that strains exist in the obtained MoS 2 on the VGCNFs. This accounts for the different in the d-spacing determined from the XRD and TEM (Table S1). The d-spacing determined from the XRD are larger than that determined from the HRTEM images. This is attributed to the existence of strains. Such strains can be estimated using the Wagner-Agua method 39 . The value of the strain increases with the amount of fibers (Fig. 2c). Furthermore, the addition of VGCNFs results in significant decrease of the number of MoS 2 layers as shown in the HR-TEM images given in Fig. 2b. For example, the number of layers significantly decreases from ~14 to ~6 layers for MoS 2 and MPR25-1, respectively. Figure 3 shows the morphologies of the bare MoS 2 and its composites. The MoS 2 exhibits a flower-like morphology having MoS 2 nanowall petals ( Fig. 3a,b). It is known that the formation of MoS 2 nanoflowers begins with the occurrence of MoS 2 nanowalls which then aggregate, due to the van der Waals interaction, and then form a flower-like morphology 40 . However, the growth of MoS 2 on a VGCNFs surface involves more effort as mentioned above. In the composite samples, MoS 2 nanowalls and nanoflowers were found on the fibers surfaces ( Fig. 3c) and by the fibers (Fig. 3d), respectively. The amount of MoS 2 nanoflowers by the VGCNFs in general increases with the Mo precursor. The growth of MoS 2 nanoflowers by the fibers follows the mechanism describe above. On the other hand, the growth of MoS 2 nanowalls on the fibers surface in this study is not straightforward but due to the addition of NaCl which changes the polarity of VGCNFs surface and gives O-functional groups. Hence, under such a circumstance, heterogeneous growth of MoS 2 nanowalls on the fibers surface occurs. When the amount of Mo precursor exceeds a certain amount, the homogeneous growth of MoS 2 nanoflowers by the fibers also occurs. It appears that the nanowalls are vertical to the fiber surface, as shown in Fig. 3e. Figure 3e also shows that each nanowall consists of a few stacked layers of MoS 2 . The contact between the MoS 2 nanowall and fiber surface is intimate. Furthermore, with increasing fiber, the nanowalls become thicker and their packing becomes denser, as shown in Figure S2. It shows that compared to MPR-3, MPR25-1 nanowalls are thinner, approximately 22 nm, and loosely packed. MPR25-2 ( Fig. 3c) and -3 ( Figure S2) show denser packing of the nanowalls whose thicknesses are thicker too, approximately 25 nm and 30 nm, respectively. Thinner nanowalls and less packing density give more empty spaces in between the nanowalls which is beneficial for ion intercalation. For comparison, MoS 2 was also grown on GO, which was reduced to RGO during synthesis. Layers of MoS 2 were found to grow on the RGO surface, laying parallel to the RGO surface (Fig. 3f), while MoS 2 nanoflowers are seen beside the RGO ( Figure S3). The fact that MoS 2 grows on either the fiber or RGO surface demonstrate the critical role of NaCl, which facilitates the nucleation and growth of MoS 2 , as to be discussed more later. To further investigate the growth mechanism, a control experiment has been done following the condition for MPR25-1, except without the sonication. The sonication helps to disperse the precursors such that the nucleation is more homogeneous and therefore the MoS 2 nanowalls also distribute more homogeneously, as shown in Figure S4.
The difference in the morphology results in different SSA, as shown in Table 1. It shows that the BET SSA of MPR25-1 > bare MoS 2 > MPR25-2 > MPR25-3 > MPR25-1-2. Initial addition of fiber to the MoS 2 increases the SSA of the resulting MPR25-1 composite from 26.98 to 32.26 m 2 g −1 . However, further increasing the fiber amount (MPR25-2 and -3) leads to decreased SSA due to the thicker and more densely packed nanowalls, as mentioned above. This also indicates that the MoS 2 has a higher SSA than the VGCNF. Likewise, MPR25-1-2 has smallest SSA among the composites due to the vigorous aggregation of the nanowalls. The result strongly suggests that the introduction of an appropriate amount of VGCNFs can inhibit the aggregation of MoS 2 and will be further explained later in the mechanism part. Compared to MPR25-1, MGO has a much smaller SSA because the RGO layers are easy to stack or restack. The nitrogen adsorption-desorption isotherms of bare MoS 2 and its composites exhibit type IV characteristics which indicates the existence of large pores or mesopores (Fig. 4). Using the BJH method the pore size distributions were characterized, as shown in insets of Fig. 4. It can be clearly seen that MPR25-1 has the largest pore volume (Table 1). It is known that the flower-like morphology has larger pore size than the wrinkle layered morphology. Therefore, as shown in Table 1, bare MoS 2 has the highest pore size and the composites have similar pore sizes too. However, among all of the samples, MGO has the highest pore size. This is attributed to the existence of MoS 2 by the RGO that exhibits a fully 3D flower-like structure, as shown in Figure S3.
As mention above, the presence of NaCl on the VGCNF surface is the key for the growth of MoS 2 on the fiber. It is known that in H 2 O, NaCl is completely dissolved due to the polar nature of H 2 O. As a result, the electrostatic force between Na + and Cl − breaks, leading to the formation of their solvated ions. Furthermore, it is known that the interaction between Na + and π-electrons is stronger than that between Cl − and π-electrons, therefore the solvated Na + is attracted to the fiber surface by the π-electrons on the carbon fiber 41 . Under such a circumstance, the fiber surface becomes positively charged due to the existence of the Na + and the solution becomes negatively charged due to the presence of Cl − . In the meantime, during the hydrothermal process, the solvated Na + that are attracted to the fiber surface react with the C-C/C=C and O-functional groups on the fiber surface to form  Pore Volume (cm 2 g −1 ) Pore  Fig. 5a. XPS C1s spectra of the rest of the composites are shown in Figure S5. Three common peaks are seen in both spectra. The dominant peak at 284.7 eV is attributed to graphitic carbon C-C/ C=C, while the other two peaks at 286.3 and 288.8 eV are related C-OH and C(O)OH respectively 26,42 . However, there is an additional peak of C=O at 287.7 eV in the C1sp spectrum of MPR25-1 and the rest of composites as shown in Figure S5. This indicates the aforementioned Na + adsorption to and reaction on the fiber surfaced have indeed occurred. The Mo3d and S2p high resolution spectra for Sample MPR25-1 were also obtained, as shown in Fig. 5b,c, respectively. There are three Mo orbital peaks at 229.5, 232.6, and 235.9 eV, representing Mo 4+ 3d 5/2 , Mo 4+ 3d 3/2 , and Mo 6+ 3d 5/2 , respectively, and a S2s peak at 225.6 eV in the high resolution Mo3d spectrum. The dominant peak of Mo 4+ is from the MoS 2 bonding and the weak peak of Mo 6+ is likely due to the presence of small amount of Mo-O bonding which commonly occurs due to either the bonding between Mo atom and O atom from the functional groups on the fiber or partially oxidation 11 . The S2s peak represent the Mo-S bonding.
On the other hand, there are various bonding at 161.4, 162.6, and 163.7 eV, representing S2p 3/2 and S2p 1/2 for Mo-S and C-S bonding, respectively, in the high resolution S2p spectrum 11, 43 . Direct growth mechanism of MoS 2 nanowalls on VGCNFs. Based on the XRD, SEM, TEM, and XPS analysis, the growth mechanism of MoS 2 on the VGCNF is proposed as shown schematically in Fig. 6. During the synthesis, there are both heterogeneous growth of MoS 2 on the fiber surface and homogeneous growth of MoS 2 by the fiber. As mentioned previously that the growth of MoS 2 on the fiber surface is due to the use of NaCl. As shown in Fig. 6a, fibers in the solution are surrounded by the solvated Na + and Cl − , MoO 3 , and Mo 4 O 11 . Due to the interaction with the π-electron from the fibers, the solvated Na + are attracted to the fiber surfaces (Fig. 6b), hence making the fiber surface positive as supported by the Zeta potential measurements ( Figure S8 in the Supplementary information). As a result, due to the existence of the solvated the Na + on the fiber surface, C=O bonding forms during the MAH according to Eq. 2.
where R is aromatic of carbon. Therefore, the wettability of the fibers is improved as shown in Fig. 6b. Under such a circumstance, both the negatively charged MoO 3 and Mo 4 O 11 are attracted to the positively charged fiber surface, as shown in Fig. 6c. During the hydrothermal process, these sub-oxides are reduced to MoS 2 by H 2 S released from thiourea as shown in Eq. 1A and B. The single-and then few-layered MoS 2 grow on the fiber surface due to the lowest surface energy of (002) planes of MoS 2 as shown in Fig. 6d. The layered growth phenomena at the initial stage can be clearly seen in MGO sample as shown in Fig. 3f. As the MoS 2 grows layer by layer to the [001] direction, more edges are show pore size distributions obtained using the BJH method.
exposed in the few-layered structure. The edges have different local stoichiometric as compared to the basal plane, i.e., lower coordination numbers. The unsaturated S atoms on the edge sites from either their atomic structure of H phase or structural defect represent the microstructure having low coordination numbers, hence making the edges more reactive than basal plane 44,45 . Furthermore, the edge sites also show metallic phase and therefore accommodate ions, making them more reactive too 44 . Consequently, thermodynamic driving force makes the MoS 2 to grow vertically to the fiber or RGO, as shown in Fig. 6e, to minimize the exposures of the highly energetic and atomically undercoordinated edges. On the other hand, a larger amount of fibers gives a lower pH, meaning more positively charged surfaces. The fibers then attract more MoO 3 and Mo 4 O 11 , i.e., forming more nucleation sites for the growth of MoS 2 . As a result, due to the increased nucleation sites, the MoS 2 grows to smaller sizes in the basal planes on the fiber surfaces, as shown in Fig. 6f. Also, as discussed above, strains were found in the composite samples and the strain increases with the fibers (Fig. 2c). In the meantime, a larger amount of fibers, giving a larger strain, leads to a larger d-spacing (Fig. 2b,c). The strain is interfacial strain that occurs due to the lattice mismatch between the fiber and MoS 2 . Such interfacial strain leads to uniaxial tensile strain along [001] direction that causes the expansion of the lattice along the [001] direction 46 . As a result, a larger amount of fibers, meaning a higher strain, leads to a larger d-spacing.
Electrochemical performance of MoS 2 nanowall/VGCNFs composites. The electrochemical performance of bare MoS 2 and its composites were evaluated using CV and EIS tests. Figure 7a shows the CV curves of bare MoS 2 and its composites. The curves show quasi-rectangular shapes without any obvious redox peak, seemly suggesting the charge storage of all the samples are likely from EDLC. However, the specific capacitance determined from these CV curves is considered to be more than that can be obtained based on the consideration of only the moderate specific surface areas of the electrode materials (Table 1). Therefore, it is believed that pseudocapacitance also exists. In other words, the charge storage of MoS 2 in this study comes from both EDLC and pseudocapacitance, which is contributed by diffusive-intercalation and surface redox reactions. Comparing all of the CV curves, MPR25-1 shows more distortion at the positive potential side. This occurs due to its higher pseudocapacitance as explained later. The Csp of all the samples were calculated by integrating the area under the CV curve. As shown in Fig. 7a, the MPR25-1 has a wider area as compared with other samples followed by bare MoS 2 and MPR25-2 then MPR25-3. Therefore, the Csp (at 5 m V s −1 ) are in the same order: MPR25-1 (248 F g −1 ) > bare MoS 2 (212 F g −1 ) > MPR25-2 (169 F g −1 ) > MPR25-3 (130 F g −1 ) > MPR25-1-2 (138 F g −1 ) > VGCNF (17 F g −1 ) (The CV curve of VGCNF is given in Figure S6). MPR25-1 has the highest Csp due to the highest SSA and thinnest nanowalls. Noteworthy, the distance between the layers of MoS 2 is very important in charge transport. A sufficient distance between adjacent nanowalls prevents the electrostatic repulsion between adsorbed ions 12 . It also shows in Table 1 that MPR25-1 exhibits large pore volumes which favour ion diffusion. It indicates the , it is seen that the VGCNF is more favourable for the electron transport than GO. The reason is that the GO exhibit a lower electrical conductivity. However, there is limit for the addition of fiber, as shown in Table 1. The SSA and pore volume decrease with the amount of the VGCNF. Therefore, the Csp of MPR25-2 and MPR25-3 are lower than MPR25-1. Thicker and aggregated MoS 2 nanowalls were observed for MPR25-2. Both observations are linked to less favourable electron transport. For the bare MoS 2 , it has the second highest Csp due to the high SSA and appropriate pore sizes for ion diffusion. Therefore, from the above observation, a clear relationship between SSA and Csp can be obtained, as shown in Fig. 7b for bare MoS 2 and MoS 2 /VGCNF composites.
Here in, the charge storage mechanism is further discussed. A quantitative analysis has been done to determine the percentage of diffusion-controlled and capacitive capacitance following the Trassatti method 47 . The result shows that diffusion-controlled capacitance is the major storage mechanism for all the samples, as shown in Fig. 7c. The diffusion-controlled capacitance is believed to come from the intercalation of K + into the MoS 2 interlayers, as shown in Eq. 3. Scientific RepoRts | 7: 5999 | DOI:10.1038/s41598-017-05805-z Meanwhile, the capacitive capacitance is from double layer charges and the surface redox reaction as shown in Eq. 4.
Furthermore, the kinetic charge storages at different scan rates determined by the Dunn's method 48 also show that the diffusion-controlled is dominant, as shown in Fig. 7d for MPR25-1. The capacitive contribution increases with the scan rate. Likewise, the other samples exhibit the same behavior as shown in Figure S7. On the other hand, the addition of fibers to the MoS 2 improves the Csp retention that favours the high power density as shown in Fig. 7e. It shows that a composite with a higher amount of VGCNF has better retention than the bare MoS 2 .
Especially, MPR25-2 shows an enhancement of 25% over the bare MoS 2 at 100 mV s −1 , indicating the fibers help the electron transport during the charge-discharge process. These observations indicate that, for both the electrochemical characteristic of bare MoS 2 and composites, (i) the Csp value depends on the morphology and SSA and (ii) the Csp retention is controlled by the VGCNF. On the other hand, supercapacitor exhibiting long term cyclic stability is also highly desirable. The evaluation of cyclic stability was performed at a high charge transfer of 200 mV s −1 as shown in Fig. 7f. The MPR25-1 electrode shows only 4% reduction after 1,000 cycles, demonstrating good long term stability which is much better than the bare MoS 2 , which shows 24% reduction. This is achieved due to the high charge storage and high electrochemical stability of MPR25-1 after the addition of VGCNF. On the other hand, MGO has a smaller Csp than MPR25-1 due to the lower SSA and conductivity. Compared to the published results listed in Table 2, we clearly demonstrate the potential application of these composites for supercapacitor application.
To analyse the resistivity of supercapacitor cell, EIS test was performed and the resulting Nyquist plots are shown in Fig. 8. At the low frequency region, the slope indicates idealness of capacitor behaviour. It shows that the composites are closer to an ideal capacitor due to the presence of VGCNFs that improve the charge transfer. At the high frequency region, the first point at the X-axis is the equivalent series resistance (ESR/Rs) which is associated with the intrinsic resistance of electrode material and the bulk ionic resistance in electrolyte solution 42 . As the same electrolyte was used, the difference is from the intrinsic resistance of the electrode material. As shown in Table 3, the addition of the conductive VGCNF decreases the Rs, except MPR25-2. The electrical conductivity increases with the fiber amount. The unique structure of MoS 2 nanowall-VGCNF composite with intimate interface contact between MoS 2 layers and VGCNF provides a diffusion path for enhanced charge transfer, i.e., improved electrical conductivity of composites. For MPR25-2, the higher Rs is attributed to the fact that MPR25-2 has the highest strain among the composites, leading to the formation of such defects that reduce the electrical conductivity. The MGO also shows higher Rs than the bare MoS 2 due to the abundant oxygen atoms and O-functional groups on the GO. Furthermore, VGCNF also decreases the contact resistance (Rc) which is the intercept of quasi-semicircle with the X-axis. Rc is associated with the contact among the constituents in the MoS 2 /VGCNF composite electrode, and between the electrode and current collector. For the contact between the electrode and the current collector, the presence of VGCNFs bridges the transfer of electrons at electrode/ current collector interface. However, MPR25-3 has a larger Rc because its thickest MoS 2 layers make the transfer of electrons through interface less straightforward. Similarly, MGO has the largest Rc due its abundant oxygen atoms and O-functional groups that hinder the transfer of electrons at the interface. For the ion diffusion within the pores (Rd), bare MoS 2 has higher Rd than the composites. This is due to that the bare MoS 2 has thick mesoporous wall (determined from its pore volume). The mesoporous walls affect the diffusion distance in a manner that a thinner and thicker mesoporous wall leads to smaller and longer diffusion distance, respectively 49 . For the composites, the Rd values are lower, implying their mesoporous walls are thinner than that of the bare MoS 2 . Therefore, the diffusion distance is lower, favouring the electron transport. Moreover, the unique composite structure exhibiting low degrees of aggregation is highly desirable for ions intercalation.   1). For comparison, GO was also used in this study using the same experimental condition, except without any VGCNF, and the resulting sample is designed as MGO. GO was prepared following a previous study using a modified Hummers' method 42,50 . Another set of experiments has been done following the condition for preparing MPR25-1 except that VGCNF were dispersed in DI-water following by the dissolution of Mo and S precursors under stirring without sonication. The obtained sample is designed as MPR25-1-2. Meanwhile, without being subject to the MAH, the aforementioned Mo precursor/VGCNF mixture was also dropped on a glass substrate and then dried in an oven at 60 °C for several minutes. The dried products were analyzed for the phase and morphology.
Characterization. The crystalline phase was examined using X-ray diffraction (XRD, Rigaku X-Ray Diffractometer) with Cu-Kα radiation (λ = 0.15406 nm). The surface chemistry and bonding were investigated using X-Ray photoelectron spectroscopy (XPS, Electron Spectroscopy for Chemical Analysis, PHI 5000). The surface morphology and microstructure were investigated using scanning electron microscopy (SEM, JSM-6701F) with an acceleration voltage of 3 kV and transmission electron microscopy (TEM, JEOL-2100F CS STEM) with an acceleration voltage of 200 kV. The zeta potential was determined using zeta potential analyser (Malvern   Table 3. EIS analysis of bare MoS 2 and composites. Nano zs MRK791-02). The specific surface area (SSA) and pore size distribution were determined using the Brunauer-Emmett-Teller (BET) and the Barrett-Joyner-Halenda (BJH) method, respectively.
Fabrication of working electrode. To prepare the working electrode, 70 wt% of the active material and 30% of carbon black were first dispersed in 1 mL ethanol which was then deposited onto a 1 × 1 cm 2 Advantech filter paper using a filtration method. The ethanol was removed by drying the resulting sample at 60 °C. The active material side of the sample was attached to a stainless steel current collector to form the working electrode. The cell also had a filtration membrane (Critical process) and a 1 M KCl aqueous solution electrolyte.
Electrochemical measurements. All the electrochemical measurements were carried out at room temperature. The supercapacitor performance was evaluated in a three electrode cell for cyclic voltammetry (CV) measurement and a two electrode cell for electrochemical impedance spectroscopy (EIS) measurement using an Autolab PGSTAT. A three electrode cell consisting of a working electrode, a platinum counter electrode, and a saturated calomel electrode (SCE) reference electrode. CV measurement was performed within −0.9 to 0.4 V at scan rates from 5 to 100 mVs −1 . The preparation of the electrode for the two-electrode cell or symmetric supercapacitor was same as that of the three-electrode cell. The specific capacitance (Csp) was calculated by first integrating the area under the CV curve to obtain the charge (Q in Coulombs), which was then divided by the mass of the active material (m in g), the scan rate (v in V s −1 ), and the potential window (V = Va − Vc in V) according to Eq. (5): where I(V) is the current as a function of voltage, and Va and Vc are anodic and cathodic potentials, respectively. EIS was used to evaluate the electrical resistivity of the cell. It was conducted in the frequency range between 100 kHz and 2 mHz with a perturbation amplitude of 5 mV versus the open-circuit potential.