Chemically anchored NiO x – carbon composite ﬁ bers for Li-ion batteries with long cycle-life and enhanced capacity

N-doped carbon ﬁ bers and their composites have drawn much attention because of their wide application in energy storage. In this paper, NiO x nanoparticles are anchored on N-doped carbon ﬁ bers by chemical bonds with controlled concentration of NiO x , and the ﬁ bers interweave into hierarchical structured networks. It is demonstrated that these NiO x nanoparticles consisted of both NiO and Ni 0 other than a single phase. As binder-free anodes for lithium-ion batteries, the NiO x – C ﬁ ber networks obtained at 650 (cid:1) C with 9.2 wt% NiO x could deliver a speci ﬁ c capacity of 676 mA h g (cid:3) 1 after 200 cycles at a current density of 500 mA g (cid:3) 1 . It is also found that the storage and rate capacities of the networks are dependent on both the content of NiO x and the annealing temperature. The improved lithium-ion storage properties can be ascribed to the intimate connection between NiO x and the highly conductive network of carbon ﬁ bers through chemical bonds.


Introduction
In spite of successful commercialization of lithium-ion batteries (LIBs) for more than two decades, intense research efforts have been devoted to develop new electrode materials for LIBs to meet the ever increasing demand for high energy and power density as well as good cyclic stability for energy and power hungry of new generation electronics and electric vehicles. [1][2][3] Compared to commercial graphitic anodes with a theoretical capacity of 372 mA h g À1 , transition metal oxides (TMOs) have attracted much attention because of their higher theoretical capacities, for example 717 mA h g À1 for NiO, 4,5 674 mA h g À1 for CuO, 6 and 924 mA h g À1 for Fe 3 O 4 . 7 As their lithium ion storage mechanism is based on conversion reactions, TMOs always suffer from the issue of electrode pulverization arising from the huge volumetric change during lithiation/delithiation processes. 8 Among the TMOs mentioned above, NiO with low cost and relatively high theoretical capacity has been investigated by many researchers. [9][10][11][12][13] A variety of strategies have been developed to improve the properties of NiO as anodes for LIBs, such as preparing nanosized and/or hollow structured NiO,4 and modifying them with carbon materials. 11,12 Since carbon materials with high electrical conductivity and excellent stability for LIBs are commercial anodes, a lot of researches have been focused on the preparation of NiO-carbon composites with various architectures in nano-, meso-and microscales, which were the research hot-pot of NiO based anodes for LIBs.
Amorphous carbon, carbon nanotubes, graphene, and carbon bers were commonly employed to modify NiO to improve their conductive and stability properties. 14,15 For example, both spherical and net-structured NiO were modied with carbon by a hydrothermal method to achieve enhanced properties. 16,17 Carbon nanotubes were also used to prepare NiO-carbon composites via a surface modifying method. 18 However, NiO in those composites were not intimately mixed and well protected by carbon materials because they were not coated by carbon. Graphene, a new kind of carbon materials with excellent electrical and mechanical properties, was used to modify NiO and other TMOs. 7,11,12,[19][20][21] Synthesis of NiO-graphene commonly began with nickel salts and graphene oxide, and the resultant NiO-graphene composites were easily to stack up to block the diffusion of lithium ions, resulting in a low capacity at a large current density. In comparison, composites based on carbon ber can protect TMOs from pulverization and improve the fast transfer of electron owing to carbon ber networks and easily diffusion of lithium ions because of large interface between electrolyte and active materials. [22][23][24][25][26] Preparing properly designed NiO-carbon ber nanocomposites may be a strategy to enhance their properties.
Ni (or NiO)-carbon ber composites have attracted much interest in the elds of LIBs, 27-32 supercapacitors, 33 and catalysts. 34 Those composites commonly fabricated by means of electrospinning using polyacrylonitrile and nickel salts as precursors, and followed with annealing/pyrolysis treatment in an inert gas at high temperatures. [27][28][29] It should be noted that the standard electrode potential of Ni 2+ /Ni (4 Q A ¼ À0.257 V) is higher than that of Co 2+ /Co (4 Q A ¼ À0.277 V), Ni 2+ would be reduced by carbon (or CO) more easily than that of Co 2+ . 35 Cobalt compounds in electrospun CoO x -C bers have been reported to be Co 0 . 36 Similarly, most of the nal product of Ni(or NiO)-carbon ber composites were Ni-carbon ber composites. [27][28][29] Considering the fact that TMOs storing lithium ions is based on the conversion reaction, transition metal do not have direct contribution on the storage of lithium ions. 37 However, some Ni-carbon ber composites have been reported to show improved specic capacity as compared to pure carbon bers. 27 Similar results were found in Co-carbon ber composites as advanced anodes for LIBs. 36 Therefore, it is of great signicance to gure out the valence of nickel and the mechanism of NiO x improving the properties of carbon bers.
In this study, NiO x -carbon ber nanocomposites were synthesized by electrospinning starting from polyacrylonitrile and nickel acetate tetrahydrate (NiAc 2 ). Aer heat treated in N 2 at a high temperature (about 650 C), polyacrylonitrile was pyrolyzed to carbon, and NiAc 2 was converted to nickel oxide and partially to nickel (Ni 0 ), which was conrmed by means of XPS analysis. The composites showed improved excellent cyclic stability and specic capacities especially at large current densities. The improvement may be ascribed to the carbon ber networks with high conductivity and fast diffusion rate of lithium ions, the present of nickel nanoparticles in the composites, and the chemical bonds between nickel and carbon.

Experimental sections
Polyacrylonitrile (PAN, M w ¼ 150 000, Sigma-Aldrich Co., Ltd., USA), NiAc 2 (Alfa Aesar Co., Ltd., USA), and N,N-dimethylformamide (DMF, J. T. Baker Co., Ltd. USA) were purchased from the companies and used without any purication. To prepare the precursor for electrospinning, PAN and NiAc 2 were dissolved in DMF to form solutions with concentrations of 6.5 wt% and 1.6-2.3 wt%. Then they were transferred to a 3 mL syringes with a stainless steel needle whose inner diameter was 0.6 mm. The ow rate of the precursor solution was set to about 0.4 mL h À1 by a syringe pump. A piece of aluminum foil, which was vertically positioned about 15 cm away from the needle, was used as the collector for bers. The potential between the needle and the collector was controlled to be 13-17 kV by a high voltage DC power. Finally, a piece of PAN-NiAc 2 ber network was obtained from the aluminum foil. Aer being treated in air at 225 C for 6 h, the resulting brown networks were annealed at 550-700 C in N 2 for 2 h to carbonize the PAN and decompose the NiAc 2 . When the weight ratio of NiAc 2 in precursor solutions was 1.95%, the nal products obtained at different temperature were A-2-X (X means the nal annealed temperature). When NiAc 2 was 1.6 wt% and 2.3 wt% in precursor solutions, the nal products were marked as B-2-X and C-2-X, respectively. Pure carbon bers were synthesized using PAN solution without any NiAc 2 . Carbon bers synthesized at 650 C were marked as E650.
The microstructure and morphology of composites were characterized using a JEOL JSM-7000F scanning electron microscope (SEM), and a FEI Tecnai G2 F20 transmission electron microscope (TEM) operating at 200 kV accelerating voltage. Elemental analysis of samples was achieved using energy dispersive spectroscopy (EDS) on SEM. Thermogravimetric analysis (TGA) data were collected on a Netzsch STA449C thermal analyzer. The valence of the atoms in samples was detected by X-ray photoelectron spectroscopy (XPS, Surface Science Instruments S-probe spectrometer).
The binding energy scale was calibrated by assigning the lowest binding energy of the C 1s peak to a binding energy of 285.0 eV.
Fiber networks (including NiO x -C, carbon bers) were directly used as binder-free anodes for LIBs. A Celgard 2400 microporous polypropylene membrane was used as a separator. A solution of 1 M LiPF 6 in ethylene carbonatedimethyl carbonate (1 : 1 by volume) was used as electrolyte to assemble coin cells (CR-2016). Pure lithium foils were used as both counter and reference electrodes. All of the coin-cells were assembled in an argon-lled glove-box with the moisture and oxygen levels less than 1 ppm. Discharge and charge measurements were carried out using Arbin BT2000 and LAND battery testing systems with the cut off potentials being 0.005 V for discharge and 3 V for charge. The cyclic voltammetry curves were collected on an electrochemical workstation (CHI660B).  Fig. 1b conrms the NiO x -C bers (B-2-650) with very few nanoparticles when the weight ratio of NiAc 2 in the precursor solution was 1.6 wt%. As the NiAc 2 weight ratio in precursor increased to 2.3 wt%, NiO x -C bers (C-2-650) with inhomogeneous distribution of nanoparticles are observed in Fig. 1c. If the NiAc 2 concentration was adapted to be a suitable level of 1.95 wt%, the resulted bers (A-2-650) with homogenous distribution of nanoparticles in a local eld could be achieved, as shown in Fig. 1d. Fig. 1e is a low-resolution SEM image of the sample A-2-650, showing those bers with uniform diameters and lengths more than 20 mm forming a network. This phenomenon is similar to our previous study about CoOcarbon ber mats, 23 indicating that the framework of carbon bers are not changed by neither CoO nor NiO x . The A-2-650 sample was characterized by EDS further. The EDS results shown in Fig. 1f reveal the presence of Ni, C, O, and N, while the N is the residual nitrogen from PAN. 23 Fig . 2 compares the TGA thermograms of NiO x -C ber samples. There is little weight loss at the temperature below 300 C which could be attributed to the desorption of adsorbed water. The sharp weight loss in B-2-650 (lowest concentration of NiO x among three samples) is higher than those of both A-2-650 and C-2-650, showing a negative effect of NiO x on the thermal stability of NiO x -C bers. A possible reason is that Ni or NiO may be a catalyst for the oxidation of carbon. 27 When the temperature increased higher than 550 C, the weight of those samples were changeless at 9.2%, 3.2%, and 17.1% for A-2-650, B-2-650, and C-2-650, showing good stability of the nal products. According to the literature, they could be indexed to NiO. 28 Therefore, the NiO x in the samples A-2-650, B-2-650, and C-2-650 could be roughly estimated to be 9.2 wt%, 3.2 wt%, and 17.1 wt%, respectively.

Results and discussion
The electrochemical properties of above NiO x -C bers with regards to the storage of lithium ions were estimated by assembling and measuring coin cells, and the results are shown in Fig. 3. The initial discharge capacities of A-2-650, B-2-650, and C-2-650 are 1203, 1022, and 914 mA h g À1 , corresponding to Coulombic efficiencies of 80%, 67%, and 57%, respectively. Owing to the irreversible formation of solid electrolyte interphase (SEI) lms in the rst lithiation procedure, their discharge capacities decrease to 938, 635, and 506 mA h g À1 in the second cycle, respectively. Aer lithiation/delithiation processing at a current density of 100 mA g À1 for ten times, the discharge capacity of A-2-650 is as high as 865 mA h g À1 while they are only 582 and 524 mA h g À1 for B-2-650 and C-2-650, indicating that the weight ratio of NiO x has great effect on the electrochemical properties of NiO x -C bers. Aer the current   The specific capacities, cyclic properties and Coulombic efficiencies of three NiO x -C fibers at different current density: 100 mA g À1 in the range of 1 to 10 cycles and 500 mA g À1 in subsequent cycles. density increases to 500 mA g À1 at the 11 th cycle, the A-2-650 could deliver a discharge capacity of 676 mA h g À1 , which is 1.4 and 1.8 times that of B-2-650 and C-2-650. In addition, the Coulombic efficiency of A-2-650 is xed at approximate 100% at a current density of 500 mA g À1 , showing a better cyclic stability of A-2-650 as compare to either B-2-650 or C-2-650. Therefore, the most suitable concentration of NiO x in NiO x -C bers may be around 9.2 wt%, which is very close to a published paper about Ni/C composites as anodes for LIBs with highest capacity. 28 The bers fabricated from the precursor solution with 1.9 wt% NiAc 2 were investigated in detail in subsequent research. Fig. 4a shows a TEM image of A-2-650 at a low magnication. Uniform carbon bers with some nanoparticles on their surface are clearly visible. In an amplied TEM image (Fig. 4b), lots of ne nanoparticles are found inside the carbon bers. The nanoparticles attached on the surface of carbon bers are relative larger than those inside the carbon bers, indicating that carbon matrix can suppress the growth of nanoparticles. A high-resolution TEM image in Fig. 4c shows four small nanoparticles on the edge of the ber and a large one outside the ber. Fig. 4d shows an amplied image of the marked area in  (JCPDS 4-850). These results reveal that most of nanoparticles on the surface of the ber are Ni 0 . Thus the microstructure of NiO x -C bers is schematically displayed in Fig. 4f that most of nanoparticles on the surface of carbon bers are Ni 0 nanoparticles. XPS analysis was conducted to determine the bond state of nickel in NiO x -C bers. The experiments were carried out on a Surface Science Instruments S-probe spectrometer, which had a monochromatized Al K a X-ray source and a low energy electron ood gun for charge neutralization of nonconducting samples. The XPS spectrum of A-2-650 NiO x -C bers is shown in Fig. 5a, which reveals the present of carbon, nitrogen, oxygen, and nickel. This result is highly consistent with the conclusion of EDS that carbon bers are doped by nitrogen. Fig. 5b shows the Ni 2p XPS spectrum of A-2-650, which can be deconvoluted to three peaks with a Gaussian t. The main peak at 855.1 eV associated a satellite peak at 860.9 eV is attributed to the Ni 2p 3/2 spin-orbit levels of NiO. 11,38 A weak peak is found at 852.4 eV, which can be ascribed to Ni 0 . 39 So there are two kinds of nickel atoms in A-2-650 NiO x -C bers, including NiO and Ni 0 with a molar ratio of 6.1 to 93.9, which agree with the results of TEM. The ne XPS spectrum of O 1s is displayed in Fig. 5c, which can be deconvoluted to ve peaks using a Gaussian t. The peak at 529.7 eV is of low intensity, and it can be probably due to the O 1s in NiO. 4,11 The relative low intensity may be ascribed to the encapsulation of NiO in carbon bers. The peaks at 532.1, 533.1, and 534.1 eV correspond to C]O, C-OH (and/or C-O-C), and chemical sorbed oxygen (and perhaps adsorbed waters). 40,41 The extra peak at 530.9 eV may be attributed to the possible present of Ni-O-C binds because Ni-C bonds are approximately located at either 853 or 285 eV. 11 The XPS spectrum of C 1s shown in Fig. 5d can be deconvoluted to ve peaks at 290.4, 289.2, 287.9, 286.5, and 285 eV. Those data are consistent with our previous results about PAN-based carbon bers. 23 Thus NiO x nanoparticles, including NiO and Ni 0 , are anchored on carbon bers by chemical bonds other than mechanical force and Van der Waals' force. Fig. 6 shows the cyclic voltammetry (CV) curves of pure carbon bers (E650) and A-2-650 NiO x -C bers. As shown in Fig. 6a, the CV curve of E650 in the rst cycle exhibits a clear cathodic peak at approximately 0.3 V, resulting from the irreversible formation of SEI lms. 28 It shows two anodic peaks at about 0.2 and 1.3 V, corresponding to the delithiation of graphite-like carbon and lithium extraction from the defective sites (and/or micropores) of carbon bers. [42][43][44] The CV curve in the second cycle nearly overlaps with the fourth one, indicating a good cyclic stability of E650. The initial lithiation potential of NiO x -C bers is approximate 0.5 V, similar to that of a previous report about nickel-carbon bers. 27 A board anodic peak in the  range of 0.5 to 1.3 V is found in the rst delithiation procedure, which is almost unchanged in the subsequent cycles. The typical cathodic and anodic peaks of NiO are located at approximately 1.1 and 2.2 V. 10,45,46 Therefore, the board anodic peak cannot be indexed to the delithiation of Ni 0 and Li 2 O. In some of reports about NiO or NiO-based composites as anode for LIBs, the anodic peak at about 1 V was attributed to the partial decomposition of SEI lms. [14][15][16]28 Whereas, the anodic peak at about 1 V only is observed in the rst cycle in those studies. In present investigation, the result is different from above papers that the peak potential and relative intensities of the anodic peak in the second and fourth cycle are almost the same. So, the board peak may not arise from the partial decomposition of SEI lms. According to the mechanism that lithium ions could be stored in micropores and defective sites of carbon with high capacities, the board peak of A-2-650 at 1 V is attributed to the storage of lithium ions in micropores and disordered carbon. In addition, the present of NiO x results in the shi of this peak to a lower potential, showing the activation effect of NiO x . Fig. 7a shows the voltage prole of A-2-650 NiO x -C bers in the 1 st , 2 nd , 15 th , and 200 th cycle. In the 1 st cycle, A-2-650 NiO x -C bers delivers a discharge capacity of 1203 mA h g À1 at a current density of 100 mA g À1 , corresponding to a Coulombic efficiency of 80%. During the discharge procedure, an extended voltage plateau is found around 1 V, which may be attributed to the formation of SEI lms. 11,28 A more inclined discharge slope in the range of 0.8 to 0.005 V is observed, which could be ascribed to the partial capacity contribution from carbon bers. 28 In the subsequent discharge process, the plateau becomes higher and more sloping than those in the 1 st cycle, showing that the lithiation reaction becomes easier. When the current density increases to 500 mA g À1 , the charge curve in the 15 th cycle almost overlaps with that in the 200 th cycle, indicating the excellent cyclic properties of A-2-650 NiO x -C bers. Fig. 7b compares the specic capacities of NiO x -C bers obtained at different temperatures, and corresponding Coulombic efficiencies are shown in Fig. 7c. The current density in the rst ten cycles is 100 mA g À1 , which increases to 500 mA g À1 in the subsequent cycles. NiO x -C bers synthesized at 550 C (A-2-550) could deliver specic capacities of 487 and 345 mA h g À1 in the 10 th and 200 th cycles. As the annealing temperature increasing to 600 C, the corresponding capacities of NiO x -C bers (A-2-600) enhance to 578 and 387 mA h g À1 , showing positive effects of improving annealing temperature on the electrochemical properties of NiO x -C bers. 47 The capacity increase of both A-2-550 and A-2-600 can be ascribed to the activation of NiO x and the electrolyte solvent soakage into composites. 22,36,48 When the annealing temperature was set at 650 C, the resulting NiO x -C bers (A-2-650) could store lithium ions with capacities of 865 and 676 mA h g À1 in the 10 th and 200 th cycles. The samples (A-2-700) prepared at 700 C can deliver capacities of 887 and 658 mA h g À1 in corresponding cycles, whose properties are very close to those of A-2-650. In addition, the ratio of the specic capacities in the 200 th cycle to that in the 10 th cycle are 0.71, 0.67, 0.78, and 0.74 along with the increase of annealing temperature from 550 to 700 C, indicating A-2-650 with the highest retention ratio. Besides, the Coulombic efficiencies of A-2-650 almost maintain at 99.5% in all the cycles except the 1 st and 11 th cycles, showing an excellent cyclic stability. Because a high annealing temperature can result in carbon bers with high graphitization degree and conductivity and may cause the transformation of NiO to Li-inactive Ni 0 , the A-2-650 NiO x -C bers with best properties may be attributed to the annealing temperature of 650 C which can make a balance between the high electrical conductivity of carbon bers and avoiding the conversion of NiO to Ni 0 and the reduction of nitrogen concentration in carbon bers. 18,22 The electrochemical properties of A-2-650 NiO x -C bers and E650 pure carbon bers are compared in Fig. 7d. They display nearly unchanged Coulombic efficiencies and specic capacities at constant current densities, indicating good cyclic stability of carbon bers-based composites. Whereas, A-2-650 could deliver a specic capacity of 865 mA h g À1 at a current density of 100 mA g À1 , which is 1.7 times that of E650. In addition, the specic capacity of A-2-650 at 500 mA g À1 is 676 mA h g À1 , also 280 mA h g À1 higher than that of E650 at the corresponding cycle. Therefore, the NiO x nanoparticles even with very low concentration have great effects on improving the properties of carbon bers regards the storage of lithium ions. The improvement may be ascribed to the micropores and/or defect sites arising from the present of NiO x because the possible present of Ni-O-C chemical bonds according to XPS results.
NiO x -C bers were further characterized by rate capacity tests, and the results are shown in Fig. 8a. It can be found that A-2-650 maintains reversible capacities of 687, 601, 500, 423, and 319 mA h g À1 at current densities of 0.1, 0.2, 0.5, 1, and 2 A g À1 , respectively. Aer the current density decreases to 0.1 A g À1 , their capacities quickly increase to 790 mA h g À1 . The improved capacity as compare to the original value can be attributed to the activation of the embedded NiO in carbon ber matrix and electrolyte solvent soakage, and similar phenomena have been found in previous reports about electrospun composites for LIBs. 22,36,48 Comparing the rate capacities of A-2-650 (with 9.2 wt% NiO x ), C-2-650 (with 17.1 wt% NiO x ), and E650 (pure carbon bers), it can be concluded that the concentration of NiO x in carbon bers should be controlled at approximate 9.2 wt% because the rare property of A-2-650 is the best one among those three samples. The property of A-2-650 is also better than that of A-2-600. A possible reason is that the relative higher temperature would result in carbon ber with high conductivity which is of benet for the electron transfer and the reduction of polarization. 22,47 AC impedance experiments were carried out aer rate capacity tests to investigate the reason why A-2-650 NiO x -C bers were of excellent properties. The Nyquist plots of these samples in Fig. 8b are composed of two partially overlapped semicircles at high-and medium-frequency ranges, and an inclined line in the low frequency range which could be considered as Warburg impedance. [49][50][51] The AC impedance results were tted using an equivalent circuit, as shown in Fig. 8c. The intercepts of four curves on the Z real axis are almost the same values, suggesting the uniform electrolyte resistance (R e ). C f and R f are the capacitance and resistance of the SEI lms (high-frequency). C dl and R ct are the double-layer capacitance and charge-transfer resistance (mediumfrequency). According to the equivalent circuit, the tting values of R f are 161, 170, and 576 U for A-2-600, C-2-650, and E-650 while it is 136 U for A-2-650, showing a lowest R f of A-2-650. The R ct of A-2-600,A-2-650, and C-2-650 are 394, 178, and 161 U which are lower than that of E650, indicating that nickel may enhance the graphitization of carbon resulting in an improved conductivity. 9 Besides, it can be found by comparing the R ct of A-2-650 with A-2-600 that an improved heat treatment temperature results in an increased graphitization and conductivity of carbon. Although the R ct of C-2-650 is lower than that of A-2-650, the sum of R ct and R f is 314 U for A-2-650 which is lower than that of C-2-650 (331 U). Considering the differences of heat treatment temperatures, and NiO x concentrations and distribution in those NiO x -C bers, the excellent properties of A-2-650 could be ascribed to their high conductivity arising from improved graphitization carbon bers and the possible Ni-O-C bonds. In addition, the network composed of NiO x -C bers improves the contact area between the active materials and electrolyte, shorten the transfer distance of lithium ions, and are favor for the fast diffusion of lithium ions. 23,48

Conclusions
NiO x -C bers with controllable content of NiO x can be synthesized via a simple electrospinning approach and following thermal treatment. NiO x is demonstrated to be complex including both NiO and Ni 0 other than a single component. NiO x -C bers obtained at 650 C with the homogenous distribution of 9.2 wt% NiO x nanoparticles could deliver a reversible capacity of 865 and 676 mA h g À1 at a current density of 100 and 500 mA g À1 in the 11 th and 200 th cycles, respectively. Those properties are better than those of both pure carbon bers and NiO x -C bers with different either NiO x concentrations or annealing temperatures. Besides, the rate capacities of NiO x -C bers synthesized at 650 C with 9.2 wt% NiO x are also better than other anodes. The enhanced properties of NiO x -C bers could be ascribed to the micropores and/or defective sites arising from the present of NiO x nanoparticles, and the network of carbon bers to accelerate the transfer of electrons and the diffusion of lithium ions. It is believed that this study may provide a new insight to synthesize carbon-transition metal (oxide) with high properties for energy storage.