In-situ growth of MnO2 crystals under nanopore-constraint in carbon nanofibers and their electrochemical performance

Growing MnO2 nanocrystals in the bulk of porous carbon nanofibers is conducted in a KMnO4 aqueous solution aimed to enhance the electrochemical performance of MnO2. The rate of redox reaction between KMnO4 and carbon was controlled by the concentration of KMnO4 in a neutral solution. The MnO2 nanoparticles grow along with (211) crystal faces when the redox reaction happens on the surface of fibers under 1D constraint, while the nanoparticles grow along with (200) crystal faces when the redox reaction happens in the bulk of fibers under 3D constraint. The composite, where MnO2 nanoparticles are formed in the bulk under a constraint, yields an electrode material for supercapacitors showing good electron transport, rapid ion penetration, fast and reversible Faradaic reaction, and excellent rate performance. The capacitance of the composite electrode could be 1282 F g−1 under a current density of 0.2 A g−1 in 1 M Na2SO4 electrolyte. A symmetric supercapacitor delivers energy density of 36 Wh kg−1 with power density of 39 W kg−1, and can maintain 7.5 Wh kg−1 at 10.3 kW kg−1. It exhibits an excellent electrochemical cycling stability with 101% initial capacitance and 95% columbic efficiency even after 1000 cycles of charge/discharge.

MnO 2 nanocomposite electrode. Lang 27 et al. showed that a supercapacitors made of nanoporous gold and nanocrystalline MnO 2 resulted in a specific capacitance of 1145 F g −1 at 50 mV s −1 . Fei group 28 reported the reverse micro emulsion method to prepare 3-20 nm MnO 2 /3D RGO composites. The 3D RGO could provide channels for rapid ionic and electronic transport. The maximum specific capacitance of 709.8 F g −1 at 0.2 A g −1 of the MnO 2 (66.4 wt%)/RGO composite. After 1000 cycles, 97.6% of the initial capacitance value can be maintained. Chen 29 et al. prepared nanostructured MnO 2 /CNTs-sponge composite electrodes, which could give a high specific capacitance as 1230 F g −1 at 1 mV s −1 . Based on the comparison in Table S1, the cost of the methods above is relatively high and the flexibility of the electrode is limited in most of the cases.
To develop a flexible and cost-effective method, Wang 30  However, their attention was paid mainly on the influences of the length of reaction time 17,30 and pH value of the solution 31,32 . The size of MnO 2 obtained were submicrometers to micrometers. To improve the performance of MnO 2 /carbon fibers, Gao group 18 and Qu group 19 designed in-situ growth of MnO 2 on the surface of activated carbon fibers. Gao 18 reported that the composite displayed a capacitance of 117 F g −1 . 116 F g −1 was kept after 3000 cycles. Qu 19 et al. reported that the MnO 2 /PCNFs exhibited a specific capacitance of 520 F g −1 at 0.5 A g −1 and 92.3% retention of the initial capacitance after 4000 cycles in a 6 M KOH aqueous solution. It should be indicated that although porous carbon materials were used, both of them pointed out the growth of MnO 2 on the surface of the fiber.
To further improve the capacitance of the MnO 2 /CNFs electrode, nano-MnO 2 in the bulk of porous carbon nanofibers are needed. Amorphous carbon materials have strong reactivity with KMnO 4 . As the electrospun carbon fibers are amorphous [33][34][35][36] , there should be a competition of KMnO 4 diffusion into the bulk and reaction of KMnO 4 with carbon on the surface of the carbon fibers. If the diffusion time is shorter than the reaction time, the KMnO 4 can diffuse into the bulk. Otherwise, the MnO 2 crystals cover the surface of the fibers and block the diffusion route of KMnO 4 into the bulk. Developing MnO 2 with much smaller sizes and high dispersion in carbon scaffold is urgently needed.
Herein, we have developed a controllable method for preparing high performance supercapacitor electrode with the nanocrystalline-MnO 2 in the bulk of porous CNFs by controlling the kinetics of an in-situ redox reaction of KMnO 4 under nanopore-constraint of CNFs (Eq. 1). The scaffold of porous CNFs, which was prepared from a mixture of polyimide (PI) and PVP via electrospinning 35,36 , acts as a matrix for electric double layer formation and a good electronic conductor to enhance the pseudocapacitive behavior of the nanocrystalline MnO 2 . More importantly, the nanopores provide volume for nanocrystalline MnO 2 to grow. The pore volume limits the finial size of MnO 2 . Such kind of structure provides the enhanced utilization of MnO 2 particles by easy diffusion of the electrolyte to the interfaces. In this work, reactants concentration and reaction temperature were considered to control the rate of chemical reaction mentioned above. A wide range of the concentration of KMnO 4 was employed, i.e., 0.76 × 10 −4 , 1.52 × 10 −4 and 7.60 × 10 −4 M for the deposition of MnO 2 nanoparticles onto the pristine CNFs. Since the relative concentration of KMnO 4 in the solution can be expressed by 1:2:10, the MnO 2 /CNFs composites synthesized in these solutions are denoted as MC1, MC2 and MC10, respectively. For the comparison, the pristine CNFs without MnO 2 loading was also used by denoting as MC0. Their structures, particularly those of deposited MnO 2 on CNFs, and capacitive performance in Na 2 SO 4 were characterized.

Experimental results
Structure. Figure 1 shows the SEM images of the MnO 2 /CNFs composites. It is seen that the external surface of the pristine CNFs (MC0) is smooth and the diameters of fibers are in the range of 300-600 nm (Fig. 1a). After loading MnO 2 by a redox reaction with KMnO 4 , the appearance of CNFs is in sharp contrast to that of MC0, particularly that of MC10. For those CNFs reacted with KMnO 4 in a low concentration, MC1 and MC2, the surfaces of CNFs became rough and nanoparticles of MnO 2 can be recognized, as shown in Fig. 1b and c, respectively. For MC10, which is prepared in a high concentration of KMnO 4 , large-sized needle-like crystals of MnO 2 are formed on the fiber surface (Fig. 1d).
Detailed transmission electron microscope (TEM) observations were carried out by using thin slices of the fibers. Figure 2a shows TEM image of the cross-section of a MC2 fiber. High resolution TEM (HRTEM) images of an external part and an inner part of the cross-section are shown in Fig. 2b and c, respectively. In these HRTEM images of MC2, lattice fringes are clearly seen, spacing of fringes being ca. 0.24 nm in the external part and ca. 0.50 nm in the inner part (Fig. 2d). In the HRTEM images of MC10 (Fig. 2e,f), lattice fringes are clearly seen, spacing of fringes being ca. 0.24 nm in the external part and no clear MnO 2 were formed in the inner part. N 2 adsorption/desorption measurements were performed at 77 K to characterize the surface areas, pore volumes and the pore-size distributions of MC1, MC2, MC10 as well as pristine CNFs. All isotherms shown in Fig. 3a are typical IUPAC type I, suggesting the microporous structure 37 . As can be seen in Fig. 3b, MC10 and MC0 share a similar micropore distribution with peaks positioned below 0.4 nm and at around 1.2 nm in the pore size range of 0.5-2 nm, while a broad peak appeared in the range of 0.8-1.1 nm on MC2. As shown in Fig. 3c, the mesopores are detected in the range of 2-8 nm for MC0, but MC2 composites with MnO 2 have only negligibly small amount of mesopores. When the concentration of KMnO 4 increased to 0.12 g L −1 (MC10), pore-size distribution in micropore range becomes very similar to the pristine fibers (MC0), but the mesopores at 2-8 nm in MC0 disappear completely. On MC2 composite prepared by using the concentration of KMnO 4 of 0.024 g L −1 , very different distribution of micropore-size is observed.
In Table 1, surface areas and pore volumes are shown by dividing them into three regions, < 0.7 nm, 0.7-2 nm and > 2 nm, together with BET surface area S BET and total pore volume V total . With increasing KMnO 4 concentration for the MnO 2 loading onto MC0 fibers, both S BET and V total increase slightly. For MC2 composite, however, surface area and pore volume contributed by the pores with the sizes less than 0.7 nm, S <0.7 nm and V <0.7 nm , decrease markedly and those by 0.7-2 nm increase markedly, although only slight increases in these parameters are detected for MC10.
X-ray diffraction patterns are shown for the composites, including MC0, in Fig. 4a. For MC0, there are two broad peaks at 2θ of around 26 and 44°, which can be ascribed to 002 and 10 diffraction of amorphous carbon 36 . Raman spectra of the composites are shown in Fig. 4b. The D-and G-bands of amorphous carbon 38 are observed for the composites as well as the pristine MC0, of which intensity ratio I D /I G was calculated to be 2.7, indicating the turbostratic structure of the CNFs. For the composites, one additional band from MnO 2 is observed at 646 cm −1, which can be assigned to the symmetric stretching vibration of Mn-O in the MnO 6 groups 25,39 . It should be noted that the bands belonging to MnO 2 become more and more distinct with the increase of KMnO 4 concentration.
To investigate the chemical composition and chemical states of various elements in the composites, XPS analysis was performed. The spectrum in Fig. 5a reveals the existence of O, Mn and C in the MC2 composite. From the binding energy separation of 11.7 eV between the peaks at 653.9 and 642.2 eV attributed to Mn2p1/2 and Mn2p3/2, respectively 40   Sample S BET (m 2 /g) S <0.7 nm (m 2 /g) S 0.7~2 nm (m 2 /g ) S >2 nm (m 2 /g) V total (cm 3 /g) V < 0.7 nm (cm 3 /g) V 0.7~2 nm (cm 3 /g) V >2 nm (cm 3  Electrochemical properties. Cyclic voltammetry (CV) curves in 1 M Na 2 SO 4 electrolyte with different scan rates from 2 to 100 mV s −1 are shown for the composites in Fig. 6a and Supplementary Fig. S1. The composite MC2 is expected to have the highest capacitance. Galvanostatic charge/discharge curves measured with the current density of 0.2 and 10 A g −1 are summarized for the composites in Fig. 6b and Supplementary Fig. S1, respectively. For the composites MC2 and MC10, the curves are not straight during both charging and discharging, revealing certain contribution of pseudocapacitance due to MnO 2 . The specific capacitance calculated from these curves with 0.2 A g −1 was 1282 F g −1 for MC2, while 151 and 515 F g −1 for MC0 and MC10, respectively. When increasing the charging/discharging current density to  10 A g−1, all curves present representative relatively linear voltage-time relationship with quasi-symmetric triangular shapes compare to the charging/discharging curve at 2 A g−1, revealing a remarkable reversibility, which indicate the capacity is mainly contributed by physically adsorption under larger current density.
The dependences of specific capacitance on current density are plotted for the composites in Fig. 6c. It is seen that the specific capacitance decreases with the increase of current density, abruptly up to 10 A g −1 and then gradually. For MC2, the specific capacitance remains as high as 335 F g −1 at a current density of 50 A g −1 , the capacity retention ratio being 26%. The capacitance retention with the current density of 1 A g −1 is shown for MC2 in Fig. 6d, the capacitance fluctuating during the first 200 cycles and then being stabilized at about 100%. The capacitance fluctuation in the beginning is probably due to the low utilization of inner MnO 2 as the limitation of electrolyte diffusion into the bulk of the fibers. After 200 cycles, as the electrodes gradually soaked by electrolyte, more MnO 2 are involved in the charge storage process. The retention ratio start to increase, then slowly exceed its initial value. Electrodes display an ideal cycling stability with full permeation of electrolyte after approximate 500 cycles, reaching a remarkable capacitance retention of 101%. 95% Coulombic efficiency remained after 1000 continuous cycles measured using the galvanostatic charge-discharge technique. The electrochemical impedance spectroscopy (EIS) of the composite is displayed in Fig. 6e, the equivalent circuit diagram being in the inset. The equivalent circuit includes bulk solution resistance Rs, charge transfer resistance Rct, pseudo-capacitance Cp due to the redox reactions of MnO 2 and Warburg impedance Zw. It is seen that all composite electrodes exhibit the nearly vertical line along the imaginary axis in the low-frequency region, revealing an ideally capacitive behavior of the electrode materials due to the fast and reversible Faradic reaction of nano-MnO 2 . It should be noted that the slope is larger with MC10 than MC2. The Rs for composite electrodes remained almost the same because this parameter is insensitive to the electrode surface. In the high frequency region, the charge transfer resistance (Rct) was 11.5 and 2.5 Ω for MC2 and MC10, respectively, which seems to depend strongly on the concentration of KMnO4 during loading of MnO 2 .
Ragone plot is presented to characterize the MnO 2 /CNFs composites in Fig. 6f together with the data published. It is noteworthy to mention that energy density of MC2 is 36 Wh kg −1 with power density of 39 W kg −1 , and can maintain 7.5 Wh kg −1 at 10.3 kW kg −1 . Comparing the present composite MC2 with those published on electrospun carbon fibers and activated carbons, the performance of MC2 surpasses most of MnO 2 /carbon composites reported [12][13][14]20,[44][45][46][47][48][49][50][51][52][53] . With regard to active materials, the energy density can be calculated through the following equations: Two strategies are commonly used to develop electrode with high energy density under high power density. (1) One is to extend the potential window in aqueous electrolytes via designing an asymmetric configuration 50,54-55 or using special electrolyte with high potential window 14,49,56,57 . By using Na 2 SO 4 aqueous electrolyte, the potential window of the supercapacitor can be 1.6-2.2 V 14,49,57 . The achieved performance curve is overlapped with our results, although the MnO 2 /C composite performance is about 292 F g −1 (2.5 A g −1 ). (2) The other is designing electrode with high specific capacitance. Many efforts have been done to design the host material micro-architecture for loading nanoscopic MnO 2 to balance electron and ion transport inside the composites, as well as mass loading. Except the energy and power density shown in Fig. 6(f), more details for comparison are shown in Table S3. Several groups reported that, shorter redox reaction time of KMnO 4 and carbon in neutral solution can help to achieve thinner MnO 2 with higher specific capacitance and high energy density [12][13]30 . Larger mass loading accompanies with lower specific capacitance in most of the cases. With porous structure, more nano-sized MnO 2 can be loaded 50 . High utilization of active materials can be achieved by combining the homogeneous deposition of MnO 2 nanocrystals in the bulk because the rapid intercalation/deintercalation of Na + on the surface layer of MnO 2 has been widely accepted as the charge-storage mechanism in mild electrolytes. The combination of MnO 2 and porous CNFs can achieve excellent capacitive performance. More interface between the MnO 2 /conductive material can improve the MnO 2 utilization mass ratio. In most of the cases shown in Fig. 6d, The MnO 2 crystal size is tens of nanometers. Although the MnO 2 is exposed to the electrolyte, the interface between the MnO 2 and carbon is limited. The interface between 3D contrained nano-MnO 2 and porous CNFs has been enhanced dramatically while the resistance of the matrix increases with more loading of nano-MnO 2 in the bulk of CNFs. So it is a trade-off between the energy density and power density. Furthermore, comparing CNFs with activated carbons and porous carbon spheres, the pores in CNFs are shallow ones since the fiber diameter is only sub-micros while the pores in activated carbons or carbon spheres are deep and narrow. The electrolyte penetration resistance of CNFs should be smaller than that of activated carbons. So, the electron transport loops and the diffusion loops of cations in most of the cases are longer than the structure shown in this work, which is related to the power performance of an electrode.

Discussion
At the interface of electrode and Na 2 SO 4 electrolyte, two charge storage mechanisms of MnO 2 are proposed, including surface adsorption/desorption and bulk insertion/extraction of Na + , which are showed below 42 : To get more insights into the advantages of the MC2 electrode, a high resolution TEM has been used. To figure out the affection of porous structure in CNFs on in-situ reaction between KMnO 4 and carbon to form MnO 2 , the electrospun CNFs prepared from PI without PVP were used as scaffold for MnO 2 by the same condition as MC2 (KMnO 4 concentration of 0.024 g L −1 ). Pore-size distributions for the pristine (MnO 2 -unloaded) and MnO 2 -loaded CNFs are shown in Fig. 7(a,b), displaying an increase of pores having the size of 1.0-1.5 nm by sacrificing the mesopores having 2-8 nm sizes. The change in pore structure in the CNFs with MnO 2 -loading is very similar to that happened in the composite MC2 (Fig. 3). Comparing to the pristine CNFs, the pore volume contributed from pores smaller than 0.7 nm disappear, a big increase of pore volume from pores in the range of 0.7-2 nm appeared and a decrease of pore volume happens with the pore volume contributed from pores larger Scientific RepoRts | 6:37368 | DOI: 10.1038/srep37368 than 2 nm for MC2 (Fig. 3) and also another CNFs (MnO2-loaded in Fig. 7). This should be related to the PVP induced pores. Porous CNFs has a larger pore volume and more defects. The defects in the bulk can prove more active-edges for the reaction with KMnO 4 to produce MnO 2 in the 3D constrains.
Based on the pore distribution and TEM images, we could conclude that KMnO 4 penetrated to the pores situated inside of the porous CNFs. The MnO 2 nanoparticles formed inside of the pores covered the inner surface of the pores. So, the pores smaller than 0.7 are blocked by MnO 2 either deposited at the entrances of pores or formed in the pores, resulting in the disappearance of the peaks for the pores smaller than 0.7 nm. The increase of pore volume in the range of 0.7-1.2 nm should be related to the MnO 2 nanoparticles formed in the pores larger than 2 nm.
Based on the CV test and galvanostatic charge/discharge results, it could be seen that the performances of MnO 2 /PVP induced porous CNFs electrode are all better than MnO 2 /PVP free CNFs electrode. The PVP free samples MnO 2 @CNFs shows better capacity at smaller current density and the lower capacity at larger current density than MC10 (Fig. 7c). The results indicate that MnO 2 with low crystalline was produced in low concentration KMnO 4 , which bring us a higher utilization rate of MnO 2 in the supercapacitor performances.
As shown in Fig. 6a, Supplementary Fig. S1 and Fig. 6e of the CV test and EIS results, MC2 shows larger capacitance than MC10 while the resistance of MC10 is much lower than MC2 and MC1. This should be related to formation of MnO 2 in the bulk of the electrode.
With a high concentration of KMnO 4 , the redox reaction is reasonably supposed to occur very fast with amorphous carbons, and so MnO 2 nanoparticles are formed on the surface of the fibers, which seems to prevent the penetration of KMnO 4 into the bulk of the fiber. The morphology of the electrode is quite different with that in a low concentration KMnO 4 , even in the form of flaks, as shown in Fig. 1d for MC10. The crystallinity of MnO 2 produced in a high concentration KMnO 4 is lower than that of MnO 2 produced in low concentration KMnO 4 , which is proved by XRD as shown in Fig. 4a. In MC10, the pore volume does not change too much. Such a pore-size distribution suggested that MnO2 deposition occurs only on the surface of the CNFs. The resistance of such kind of structure should be low which is also approved by EIS data.
When the time for the diffusion of KMnO 4 into pores is the same order of magnitude with the time for the redox reaction of KMnO 4 with carbon, MnO 2 nanocrystals can be precipitated in the pores of the fibers under a strong constraint. As MnO 2 is an insulator, the bulk resistance should increase. Such kind of MnO 2 /C interface connection is very well. Therefore, the performance under larger current is also good. Conclusion A 3D MnO 2 /C nano-composite with remarkable electrochemical behaviors can be on-site synthesized with porous carbon nanofibers as a scaffold. For MC2, the capacitance of the composite electrode could be 1282 F g −1 under a certain current density of 0.2 A g −1 in 1 M Na 2 SO 4 electrolyte. It is noteworthy to mention that energy density of MC2 is 36 Wh kg −1 with power density of 39 W kg −1 , and can maintain 7.5 Wh kg −1 at 10.3 kW kg −1 .
The key point of the work is to make the diffusion time of KMnO 4 into the bulk of the porous carbon fibers and the redox reaction time of KMnO 4 with carbon on the same order of magnitude by adjusting the concentration of KMnO 4 . The results indicate that the MnO 2 growth along with (211) crystal face when KMnO 4 and carbon redox reaction happens with an un-limited 1D constrain; while the MnO 2 growth along with (200) crystal face when KMnO 4 and carbon redox reaction happens with a comparable 3D constrain. The optimized KMnO 4 concentration in our case should be 0.024 g L −1 . The good capacitance performance of the MnO 2 /CNFs composite electrode is clearly attributed to its unique nanostructure. The porous CNFs serves as a highly conductive matrix for fast ion and electron transport, while the ultrathin MnO 2 nanocrystals on the inner surface of the pores enable a short diffusion path of electrolyte and provide a more electrochemically active surface area for pseudocapacitance through fast and reversible Faradic reaction. Fabrication of carbon nanofibers. The electrospinning solution was prepared with PMDA, ODA and PVP in DMF with molar ratio of PMDA and ODA 1:1. The PVP was dissolved in the solution with mass ratio 20% (to total masses of PMDA and ODA). The blended solution was stirred at 0 °C for 24 h. Electrospinning parameters were set as follows: applied voltage of 20 kV, tip-to-collector distance of 20 cm and flow rate of 0.8 ml h −1 . As-spun fibers were imidized and carbonized in a horizontal tubular furnace. Temperature of the furnace was first increased from room temperature to 300 °C at the rate of 3 °C min −1 and was maintained for 30 min in a flow of air for imidization. Then the temperature was increased to 900 °C at the same heating rate and was maintained for 1 h under N 2 atmosphere for carbonization.
The carbon fiber electrode was soaked into the KMnO 4 solution and the weight ratio of CNFs to KMnO 4 was set to 4:1. The CNFs was taken out until the purple color of KMnO 4 solution had fade into golden brown. The mass loading remains to 10-11% for all samples. The MnO 2 /CNFs composites were rinsed by deionized water for several times, and finally dried at 80 °C for 10 h under vacuum condition.
Characterization. The morphology of CNFs was investigated using a scanning electron microscope (SEM, LEO 1530). A high-resolution transmission electron microscopy (HRTEM) was employed to characterize the distribution and crystallization of the composites. The CNFs, pristine and MnO 2 -loaded, embedded in epoxy resin was cut into 20 nm slices using a microtome (EM UC6, Leica, Germany) with a diamond blade (DiATOME) and then placed onto a 300 mesh Cu grid for examination by TEM.
X-ray diffraction (XRD) pattern of the composites was examined using D/MAX-RM 2000 at the scanning rate of 2° min −1 in a range of diffraction angle 2θ from 5° to 70°. Raman spectrum was recorded by HR800 (HORIBA spectrometer) with a wavelength of 633 nm. X-ray photoelectron spectroscopy (XPS) measurement was performed with a thermo ESCALAB 250 spectrometer. XPS data analysis was performed using Thermo Avantage software. Sorption/desorption isotherms of N 2 at 77 K were measured by a Belsorp Max apparatus (Japan) and analyzed to evaluate specific surface areas of the composites by using the Brunauer-Emmett-Teller (BET) method, S BET , and pore-size distribution of the composites by density functional theory (DFT) method.
A symmetrical two-electrode supercapacitor was assembled using two pieces of the composite and a separator to investigate the electrochemical performance as an electrochemical capacitor. To achieve sufficient saturation of the electrodes and separator by the electrolyte, they were immersed in 1 M Na 2 SO 4 solution under vacuum for 24 hrs before being assembled. Cyclic voltammetry (CV) and galvanostatic charge/discharge analysis were carried out to evaluate electrochemical performance of the electrodes with a potential window ranging from 0 to 1 V in 1 M Na 2 SO 4 electrolyte using an electrochemical workstation (CHI 600, Shanghai Chen Hua Instrument Company). The electrochemical impedance spectroscopy (EIS) was conducted in the frequency range of 0.01 Hz ~100 kHz with perturbation amplitude of 5 mV versus the open circuit potential. The average specific capacitance was calculated according to the charge/discharge tests based on the following equation: where I, t, m and U are the charge/discharge current (A), the discharge time (s), the total mass of active materials in the two electrodes (g) and the potential after the deduction of IR drop (V), respectively.