Introduction

Lithium-ion batteries (LIBs) already widely used in consumer electronic equipment and grid-scale storage systems. Aiming to the desired energy storage, high power density and energy density are eager to be developed for high-performance LIBs [1,2,3,4,5]. However, the commercial graphite anode has a low theoretical capacity (372 mAh g−1), which places restrictions on the application of LIBs [6,7,8]. Transition metal oxides (TMOs, M = Mn, Zn, Cr, Ni, etc.) are attractive because of their high reversible capacities [9,10,11,12,13,14,15]. Among these, ZnO is supposed to be a promising candidate of anode materials for LIBs on account of its large theoretical capacity of 978 mAh g−1. However, the pulverization issue caused by the severe volume changes (228%) of crystal ZnO in LIBs hindered its practical application, resulting in a loss of electric contact, rapid capacity fading during cyclic process, and poor rate performance [16,17,18,19].

Many strategies have been applied to suppress the effect of crystal ZnO pulverization, such as morphological design of nanostructures [20, 21], hybridizing with other materials [6, 17, 18], and introduction of carbon materials [7, 16, 19, 22]. And one of the most effective strategies is to reduce ZnO size, which could decrease the mechanical stress produced by redox reaction with Li-ions, and thus restrained the inclination to fracture and crack [23, 24]. The one-dimensional (1D) nanostructure can shorten the lengths of Li-ion transport path to provide a superior rate capability [25, 26]. Furthermore, the presence of carbon could improve conductivity and mitigate volume changes so as to improve long-cycle stability [27,28,29]. Therefore, 1D nanostructure carbon compounded with ZnO had been widely studied because the nanostructure provides conductive network and transmission channels, which is beneficial to improve electrochemical performance [23, 30]. Electrospinning is an efficient way to prepare 1D nanostructure [12, 31, 32]. For instance, Zhao et al. prepared the ZnO/carbon nanofibers composited by petroleum asphalt–derived carbon. The composite anode possessed a reversible capacity of 702 mAh g−1 at 0.2 A−1 after 200 cycles [30]. Despite these achievements, the ZnO size in carbon is not small enough to avoid the pulverization. With the decrease of the particle size, the crystal ZnO particles are gradually transformed into amorphous particles.

Amorphization is considered to be advantageous to improve electrochemical performance. For crystal materials, the change from crystal state to amorphous state is electrochemically driven during the process of lithium ion insertion [33, 34]. In addition, the contact between crystals becomes worse, and then the conductive property becomes worse, resulting in a bad reaction performance with lithium ions. For amorphous materials, the volume expansion is effectively alleviated. Therefore, minimizing the pulverization process could stabilize the cycling faster [35,36,37]. Moreover, amorphous materials have superior performance during discharge/charge processes because of active sites on their surface [26, 36]. Therefore, it is an efficacious pathway to enhance electrochemical properties by improving amorphous level of ZnO particles.

Herein, we proposed an approach to make the electrospinning of nitrogen-doped carbon nanofibers generate both mesopores in carbon nanofibers and uniform distributed amorphous ZnO particles by introducing urea as pore former. The porous nitrogen-doped carbon nanofibers encapsulated ZnO (ZnO@PN-CNFs) were successfully prepared by electrospinning process and heat treatment. It was demonstrated that the amorphous ZnO particles that disperse uniformly in the porous carbon nanofibers could obtain a high discharge capacity of 1073.2 mAh g−1 after 100 cycles at 0.1 A−1. And the ZnO@PN-CNF electrode showed an excellent rate performance and high specific capacity compared with crystallite ZnO encapsulated in carbon nanofibers (ZnO@N-CNFs).

Experimental section

Preparation of the ZnO@PN-CNFs

First, 0.45 g polyacrylonitrile (PAN, AR, Mw 150 000, Sigma-Aldrich) was dissolved into 3 mL N, N-dimethylformamide (DMF, AR, Sinopharm), subsequently stirred (at 60 °C) for 1 h to form solution A. A total of 0.4 g zinc acetate dihydrate (Zn(CH3COO)2·2H2O, AR, Sinopharm) and 0.07 g urea (AR, Sinopharm) were dissolved into 3 mL DMF and upon 1 h stirring to get solution B. Next, solution B was gradually added to solution A, stirred for 24 h at 25 °C to form a uniform solution and then it was transferred into a plastic injector. During electrospinning, the distance between the aluminum collector and the injector needle was 20 cm. A positive voltage of 10–12 kV was conducted between the aluminum collector and the injector needle. The obtained precursor nanofibers were dried at 60 °C for 12 h and subsequently stabilized at 220 °C in air for 3 h. After cooling to room temperature, the precursor nanofibers were calcinated at 650 °C in Ar for 2 h. And finally, the ZnO@PN-CNFs were synthesized. For comparison, the same procedures were applied to synthesize ZnO@N-CNFs without urea.

Material characterization

X-ray diffraction (XRD, Rigaku Dmax/2550VB+18KW) was used to characterize the crystal structure of the samples. The morphology of the samples was observed by scanning electron microscopy (SEM, FEI Nova Nano SEM230) and transmission electron microscopy (TEM, JEOL JEM-2010). X-ray photoelectron spectroscopy (XPS, Thermo ESCALAB 250XI) was employed with Al Kα (hυ = 1486.8 eV) as the excitation source. The Quadrasorb SI surface characterization analyzer with nitrogen adsorption-desorption isotherms was used for evaluating the specific surface areas of samples as well as pore size distribution. The thermogravimetric analysis of ZnO@PN-CNFs was conducted with a Netzsch STA 409EP TGA-DSC apparatus. The Raman spectrum was obtained by using a Raman spectrometer (Horiba, LabRAM Aramis).

Electrochemical measurements

To obtain the electrodes, mix prepared nanofibers, acetylene carbon black, and polyvinylidene fluoride at mass ratios of 7:2:1, dissolved in N-methyl-2-pyrrolidinone to form slurry. After that, the slurry was casted onto Cu foil and dried at 60 °C overnight under vacuum then cut into circular pieces. The loading active material density on Cu foil was 0.8–1.0 mg cm−2. Thereafter, the CR2032 coin-type half-cells were assembled in an argon-filled glove box with polypropylene film (Celgard 2400) as separator, Li metal as the reference electrode, and LiPF6 (1M, EC: DMC = 1:1) as electrolyte. The electrode capability was measured by a battery testing system (LANDCT2001A) within a voltage area of 0.005–3.0 V. The electrochemical workstation (CHI760E) was employed on the tests of cyclic voltammetry (CV) within 0.005–3.0 V and the EIS measurements with the frequency range of 100 kHz to 0.01 Hz.

Results and discussion

The fabrication process of ZnO@PN-CNFs is illustrated in Fig. 1. The samples were divided to two groups: one was ZnO@N-CNFs as the contrast samples and the other was ZnO@PN-CNFs with urea addition in the precursor solution. The electrospinning precursor nanofibers were transformed into ZnO@N-CNFs and ZnO@PN-CNFs after the subsequent stabilization and carbonization. It was believed that the presence of carbon can alleviate the volume change of zinc oxide during Li+ insertion and extraction processes [38]. Although compounded with carbon, crystalline ZnO in ZnO@N-CNFs will pulverize after repeated Li+ insertion and extraction, and then the crystal state was transformed into an amorphous state. However, amorphous ZnO in ZnO@PN-CNFs could still maintain amorphous state even after many cycles.

Fig. 1
figure 1

Schematic illustration of the synthesis process and changes during cycling of ZnO@PN-CNFs and ZnO@N-CNFs

Full size image

The PAN-derived nitrogen-doped carbon fibers could improve the electron conductivity, provide ion transport channels, and shorten the ion diffusion path during lithiation/delithiation processes [39, 40]. Figure 2 a and b show the SEM images of the ZnO@PN-CNFs precursor and ZnO@PN-CNFs, respectively. The precursor nanofibers had uniform nanostructure with smooth surface. However, the addition of urea significantly enlarged the diameters of the ZnO@PN-CNF precursor. After stabilization and carbonization, the ZnO@PN-CNFs became a twisted and crooked structure like noodles without crystalline ZnO on the surface in Fig. 2b. And Fig. 2 c and d show the SEM images of the ZnO@N-CNF precursor and ZnO@N-CNFs, respectively. As the contrast sample, many crystalline ZnO particles were mounted on the surface of the ZnO@N-CNFs, as revealed in Fig. 2d. The average diameter of the ZnO@PN-CNFs was about 130 nm while that of ZnO@N-CNFs is about 100 nm. The thicker diameters of ZnO@PN-CNFs are attributed to the effect of urea pore former.

Fig. 2
figure 2

SEM image of a ZnO@PN-CNFs precursor, b ZnO@PN-CNFs, c ZnO@N-CNFs precursor, d ZnO@N-CNFs

Full size image

To further investigate the microstructure of ZnO@PN-CNFs, TEM was employed. As shown in Fig. 3, ZnO crystallite was observed neither on the nanofibers surface nor inside the nanofibers. The crystal lattice fringe of ZnO was not found in Fig. 3b of the high-resolution TEM (HRTEM) images of ZnO@PN-CNFs either, which indicates the amorphous state of ZnO. Amorphous ZnO helps to achieve higher insertion and extraction rates of Li ions and accelerates to equilibrium as soon as possible without the pulverization of the ZnO in the oxidation reduction reactions (ORR), which simultaneously improves the electrical conductivity, cycling stability, and interfacial contact property [25, 36]. Thereafter, element mapping identified the elemental distribution of ZnO@PN-CNFs. Figure 3 c–f present elemental mapping results revealing the distribution of C, Zn, O, and N elements. The images of Zn and O elements indicated that ZnO was homogeneously distributed in the nanofibers, which was conducive to avoid the pulverization and alleviate the volume expansion. The presence of C and N indicated the formation of nitrogen-doped carbon from carbonized PAN, which was confirmative to enhance the conductivity as well as provide many active sites [23, 41]. The TEM and HRTEM of ZnO@N-CNFs are shown in Fig. S1.

Fig. 3
figure 3

a TEM image of ZnO@PN-CNFs. b HRTEM image of ZnO@PN-CNFs. c–f Corresponding elemental mapping images of ZnO@PN-CNFs

Full size image

The composition and crystallographic structure of ZnO@PN-CNFs had been further investigated. Figure 4 a shows the XRD patterns of ZnO@PN-CNFs and ZnO@CNFs, respectively. For the XRD patterns of ZnO@N-CNFs, all the characteristic peaks of ZnO@N-CNFs can be well corresponding to ZnO (JCPDS 36-1451) [42]. The XRD patterns of both materials showed a broad diffraction peak at about 23°, attributed to the (002) lattice planes of carbon [23]. The XRD patterns of ZnO@PN-CNFs only showed a broad diffraction peak at 23°, and no obvious peaks related to ZnO were detected while that of ZnO@N-CNFs indicated the ZnO was well crystallized corresponding to the SEM image of Fig. 2d. This can be attributed to the coexistence of urea (as a porogen) and zinc acetate suppressing the crystallization of ZnO, forming an amorphous state or transforming into smaller nanosized particles [23, 43]. The XRD result also confirmed the amorphous state of ZnO in ZnO@PN-CNFs, which was consistent with TEM observation.

Fig. 4
figure 4

a XRD patterns of ZnO@PN-CNFs and ZnO@N-CNFs. b TG analysis of ZnO@PN-CNFs. c, d N2 isotherms and pore size distribution ZnO@PN-CNFs and ZnO@N-CNFs

Full size image

To determine the content of ZnO in ZnO@PN-CNFs, TGA analysis was conducted from 25 to 1000 °C in air (Fig. 4b). The slight weight loss occurred in 25–200 °C, which is corresponding to evaporation of absorbed water. And an apparent weight loss occurred in 200–900 °C due to the combustion of carbon. The calculated ZnO content of ZnO@PN-CNFs from the TGA data was 35.54 wt%.

The BET-specific surface area and related pore-size distribution were studied by N2 adsorption-desorption isotherms (Fig. 4c, d). The BET-specific surface area of the ZnO@PN-CNFs and ZnO@N-CNFs is 30.72 m2 g−1 and 29.28 m2 g−1, respectively. Both isotherms display noticeable uptake at a relative pressure of P/P0 < 0.01, revealing the existence of microporous structure [44]. The hysteresis loop of isotherms suggested the existence of mesopores [39, 44]. The pore size distribution was obtained by DFT method, as displayed in Fig. 4 c and d. Compared to ZnO@N-CNFs, the specific surface area of ZnO@PN-CNFs was not significantly improved, which occurred because urea plays a role during the stabilization and carbonization in making difference for nanofiber structure; the mesopores were increased and micropores were decreased. The presence of mesopores is beneficial for the transmission of Li ions [45]. Proper pore size can shorten the transport pathway, increase the transport rate of lithium ions, and mitigate the volume change of ZnO during the processes of lithiation and delithiation [44, 46, 47].

Figure 5 a displays the Raman spectra of ZnO@PN-CNFs and ZnO@N-CNFs; two dominant peaks located at 1327 cm−1 and 1575 cm−1 were related to D-band and G-band, respectively [3, 48]. The D-band was assigned to disordered (sp3), while G-band was associated with crystalline graphite (sp2). The intensity ratio (ID/IG) of ZnO@PN-CNFs and ZnO@N-CNFs is 1.41 and 1.07, respectively. The ID/IG represented the carbon of composites with disordered structures [49]. ZnO@PN-CNFs showed a more disordered structure.

Fig. 5
figure 5

a Raman spectra of ZnO@PN-CNFs and ZnO@N-CNFs. XPS spectrum of ZnO@PN-CNFs: b survey, c C 1s, d O 1s, e N 1s, f Zn 2p

Full size image

To further identify the surface chemical states of ZnO@PN-CNFs, XPS was carried out. As revealed in Fig. 5b, the characteristic peaks in the XPS spectrum indicated the presence of C, N, O, and Zn elements in ZnO@PN-CNFs. As shown in Fig. 5c, there were three characteristic peaks in the C 1s spectrum fitted at 284.1, 284.8, 286.3, and 288.7 eV that corresponded to C–C, C–N, C=N, and C=O bonds, respectively [23, 44, 50, 51]. According to the O 1s spectrum in Fig. 5d, the peaks at 530.9 eV and 533.8 eV were indicative of the Zn–O bond, and the peak at 532.1 eV corresponded to C–O bond [50, 52, 53]. The high-resolution spectrum of N 1s displayed three characteristic peaks (Fig. 5e). The peak located at 398.4 eV was corresponding to pyridinic N and the peak at 400.0 eV evidenced the presence of the pyrrolic N, while the peak of 400.5 eV corresponded to graphitic N [53, 54]. Pyridinic N and graphitic N were found to be the most highly active sites for the ORR [55]. As shown in Fig. 5f, the Zn 2p spectrum of ZnO@PN-CNFs contained a peak at 1022 eV (Zn 2p3/2) and another peak at 1045.1 eV (Zn 2p1/2) [23, 54].

The electrochemical properties of the ZnO@PN-CNFs and ZnO@N-CNFs were evaluated in coin cells (CR2032). The CV curves of ZnO@PN-CNFs electrode were tested at a sweep rate of 0.2 mV s−1 in Fig. 6a. In the initial cathodic process, the well-defined peak at around 0.76 V, which is ascribed to the conversion reaction of ZnO to Zn and Li2O, and the formation of a solid-electrolyte interface (SEI) film due to electrolyte decomposition [30, 56]. The weak peak located at 0.26 V resulted from the alloying reaction of the Zn-Li alloy [23]. The peak near 0 V corresponded to the Li+ intercalate reaction into carbon [10]. In the anodic process, the peaks appear at 0.53, 0.8, 1.76, and 2.1 V. The peaks at 0.53 and 0.8 V are corresponding to the multistep dealloying reaction of the Zn-Li alloy [56, 57]. The peaks appearing at 1.76 V and 2.1 V corresponded to the regeneration ZnO (the reaction of Zn and Li2O) [9, 58]. After one cycle, the CV curve almost overlaps without significant change, indicating the ZnO@PN-CNF electrode possesses excellent reversibility. The electrochemical reactions can be described as follows [23]:

$$ \mathrm{ZnO}+2\ {\mathrm{Li}}^{+}+2{\mathrm{e}}^{-}\leftrightarrow {\mathrm{Zn}}^0+2\ {\mathrm{Li}}_2\mathrm{O} $$
(1)
$$ \mathrm{Zn}+x{\mathrm{Li}}^{+}+x{\mathrm{e}}^{-}\leftrightarrow {\mathrm{Li}}_x\mathrm{Zn} $$
(2)
$$ 6\mathrm{C}+\mathrm{Li}\leftrightarrow {\mathrm{LiC}}_6 $$
(3)
Fig. 6
figure 6

a CV curves of the ZnO@PN-CNFs electrode at a scan rate of 0.2 mV s−1. b Discharge/charge profiles of the ZnO@PN-CNFs electrode at a current density of 0.1 A g−1. c Cycling performance of the ZnO@PN-CNFs and ZnO@N-CNFs electrode at 0.1 A g−1. d Rate capability of the ZnO@PN-CNFs and ZnO@N-CNFs electrode. e Cycling performance of the ZnO@PN-CNFs and ZnO@N-CNFs electrode at 1 A g−1

Full size image

Figure 6 b displays the charge/discharge profiles during 1st, 2nd, 5th, 50th, and 100th cycles at 0.1 A−1. It can be observed that there is a long slope region below the 1V voltage in the initial cycle, which is attributed to the conversion reaction of ZnO, the generation of SEI film, and the alloying reaction of Zn-Li alloy. The initial discharge and charge capacities of the ZnO@PN-CNF electrode are 1836 mAh g−1 and 1448.1 mAh g−1, respectively. The capacity loss of 397.1 mAh g−1 in first cycle is ascribed to the irreversible loss of electrolyte and the generation of SEI film [27, 28].

Figure 6 c shows the cycling properties of the ZnO@PN-CNF and ZnO@N-CNF electrodes at 0.1 A−1. The ZnO@PN-CNF electrode exhibits higher initial charge and discharge capacity than the ZnO@N-CNF electrode. The initial coulombic efficiency of the ZnO@PN-CNF electrode is 78.87% and quickly stabilizes above 98% after the third cycle, indicating its excellent cycle stability. In the 100th cycle, the discharge capacities of the ZnO@PN-CNF and ZnO@N-CNF electrode maintain at 1073.2 mAh g−1 and 816.3 mAh g−1, respectively. The capacity of ZnO@PN-CNF electrode is higher than that of ZnO@N-CNF electrode, which is attributed to amorphous ZnO and the special structure. The amorphous phase has more Li ions storage sites than Crystal phase. Moreover, amorphous ZnO particles that disperse uniformly in carbon nanofibers can provide more interface, which can lead to additional capacity.

The rate performance of the ZnO@PN-CNF and ZnO@N-CNF electrodes was tested in the range of 0.1 to 3 A−1 every 10 cycles, as shown in Fig. 6d. The reversible discharge capacities of ZnO@PN-CNFs are 1047, 955, 804.3, 653.7, 485.6, and 374.2 mAh g−1 at 0.1, 0.2, 0.5, 1, 2, and 3 A−1, respectively. Upon returning the current density to 0.1 A−1, the reversible discharge capacity maintains 1029.2 mAh g−1 while that of ZnO@CNFs electrode is 722 mAh g−1, which indicates that the ZnO@PN-CNF electrode has a higher stability and more excellent reversibility. The mesoporous structure increases the contact area between electrolyte and the ZnO@PN-CNF electrode, resulting in increased transport rate of lithium ions and therefore enhances the electrochemical performance.

Figure 6 e displays the cycling characteristics of the ZnO@PN-CNF and ZnO@N-CNF electrodes at 1 A−1. After 400 cycles, the ZnO@PN-CNF electrode delivers a reversible capacity of 703.1 mAh g−1, and its coulombic efficiency is nearly 100%. The discharge capacity of the ZnO@N-CNF electrode only maintains at 550.4 mAh g−1. At first 200 cycles, the ZnO@N-CNF electrode has a capacity increase process. This phenomenon is due to the amorphization process during the charge and discharge could provide more interface between ZnO and carbon, resulting in increased capacity. As for ZnO@PN-CNF electrode, there is no capacity increase process, because amorphous ZnO in the electrode could maintain amorphous state even after many cycles, which will not bring more interface. Moreover, the flat and straight cycling curve of ZnO@PN-CNF electrode in Fig. 6e indicates that the ZnO@PN-CNF electrode has an outstanding cyclic stability and excellent reversibility. Uniformly distributed amorphous ZnO effectively alleviates the volume expansion and avoids pulverization effect, giving the ZnO@PN-CNF electrode more excellent cyclic stability compared to ZnO@N-CNF electrode.

To analyze the reaction kinetics of ZnO@PN-CNF electrode, the CV curves in a scan range of 0.2–1 mV s−1 were conducted (Fig. 7a). The relation of the peak current (i) to the scan rate (v) is shown in the following formulas [59]:

$$ I=a\ {v}^b $$
(4)
$$ \log (i)=b\ \log (v)+\log (a) $$
(5)
Fig. 7
figure 7

a CV curves of the ZnO@PN-CNFs electrode at different scan rates. b Corresponding log(i) versus log(v) plots at each redox peak (peak current: i, scan rate: v) of the ZnO@PN-CNFs electrode. c Capacitance contributions at a scan rate of 0.8 mV s −1. d Contribution ratio of capacitance and diffusion at different scan rate

Full size image

Here, a is a variable, and b value is a parameter calculated from the linear relation between log(i) and log(v). The calculated results of three peaks are shown in Fig.7b. Generally, if b = 0.5, electrochemical storage process is diffusion-controlled process, while b = 1 is capacitance-controlled process [60]. The b-values of peaks 1, 2, and 3 are 0.777, 0.848, and 0.988, respectively. The b-values of all peaks are close to 1, indicating the coexistence of the capacitance and the diffusion processes [61, 62].

According to the following formula, the contribution of capacitance to the ZnO@PN-CNF electrode can be determined [53]:

$$ i\kern0.5em ={\mathrm{k}}_1v+{\mathrm{k}}_2{v}^{1/2} $$
(6)

In Eq. (5), the relationship can be separated into k1v and k2v1/2, where k1v is corresponding to capacitive effect and k2v1/2 is corresponding to diffusion-controlled effect. Figure 7 c shows that the contribution ratio of the capacitance effect is around 68% for the total charge storage at the scan rate of 0.8 mV s−1. Figure 7 d indicates the contribution value of capacitance and diffusion at different scanning rates (0.2–1 mV s−1). As the scan rate gradually increases, the capacitance contribution value increases progressively. The fast reaction kinetic results in excellent rate performance.

To further investigate the transport kinetics of materials, the electrochemical impedance spectra were measured. Figure S2 shows the equivalent circuit diagram, the Nyquist plots and the corresponding fitting line of ZnO@PN-CNF and ZnO@N-CNF electrode [38, 63]. The electrolyte resistance (Re) of ZnO@PN-CNF and ZnO@N-CNF electrode is 2.73 Ω and 1.38 Ω, respectively. The SEI layer resistance (Rf) of ZnO@PN-CNF and ZnO@N-CNF electrode is 76.8 Ω and 85.4 Ω, respectively. The charge-transfer resistance (Rct) of the ZnO@PN-CNF electrode (55.6 Ω) is lower than the ZnO@N-CNF electrode (Rct = 61 Ω), because nitrogen-doping carbon and active sites of amorphous ZnO improve the electronic conductivity of ZnO@PN-CNF electrode.

Unique 1D porous nanostructure is the reason why ZnO@PN-CNFs show an excellent electrochemical performance. The structure shortens the time to reach stability and does not need to undergo the process of cracking and pulverization to form an amorphous state during the reaction of ZnO crystals and Li ions. The transport interfaces of nanoparticles are almost unchanged before and after the reaction under the coating of carbon. The existence of mesopores leads to further shortening the distance of the insertion and extraction of Li ions, which is conducive to large current charge and discharge. This is why the capacity of ZnO@PN-CNFs is higher than the comparative sample and the reversible performance is excellent.

Conclusion

In summary, the ZnO@PN-CNFs were successfully prepared by facile urea-assisted electrospinning followed by stabilization and carbonization. The addition of urea changes the morphology of the nanofibers, resulting in the formation of amorphous ZnO. The resulting special structure provides good electron and ion transport capabilities and alleviates the volume change of ZnO@PN-CNFs. The ZnO@PN-CNF electrode delivered a superior reversible capacity of 1073.2 mAh g−1 at 0.1 A−1 for 100 cycles. Besides, the ZnO@PN-CNF electrode displayed a discharge capacity of 703.1 mAh g−1 at 1 A−1 after 400 cycles. The synthesis of zinc oxide is facile and scalable and can be used as a reference for the design of superior-performance anode materials for LIBs.