1. Introduction
In recent years, the rapid spread and scale of power-consuming devices, new smart cars, and large-scale power grids have created an incredible demand for high-energy density energy storage devices. Lithium-sulfur batteries (Li-S batteries) have an important development prospect in new energy storage systems, which is owed to their theoretical mass energy density (2600 Wh kg
−1) and bulk energy density (2800 Wh L
−1) being much higher than that of lithium-ion batteries (Li-ion batteries) [
1]. Apart from this, the advantages of natural abundance and non-toxicity of sulfur make Li-S batteries more advantageous in terms of cost and environment [
2]. Despite considerable efforts by researchers to commercialize Li-S batteries, they have several critical issues that remain unresolved to date [
3]. During Li-S batteries cycling, intermediates of higher-order liquefied lithium polysulfides (LiPSs) are produced at the sulfur cathode and transferred to the Li anode side, where they are eventually reduced to form the lower-order solidified Li
2S
2/Li
2S by the highly electrochemical activity Li anode, (i.e., the “shuttle effect”). Furthermore, in the process of discharge, the slow kinetics of sulfur reduction and the volume change (about 80%) due to the formation of LiPSs also led to performance degradation [
4,
5,
6]. These phenomena result in a significant decrease in active materials (S and Li) and the short cycle lifespan of the batteries, affecting the actual application of Li-S batteries [
4].
In order to address these notorious issues, a range of specific carbon structures for LiPSs capture have been designed and fabricated over the last few decades, including porous carbon spheres, graphene, nanofibers, and porous carbon [
7,
8,
9,
10,
11,
12]. These carbon materials have higher conductivity, a large specific area, and are physically adsorbed to sulfur, which is more conducive to increased sulfur loading and limited volume expansion and physically slows down the “shuttle effect” of LiPSs [
13,
14]. However, as the affinity of non-polar carbon materials for polar LiPSs is weak, which essentially exhibits physical adsorption and cannot effectively limit the “shuttle effect”, and also cannot solve the electrode kinetics for reaction of sluggish Li-S batteries, it will eventually lead to battery performance degradation as well [
15]. Therefore, the challenge remains in using carbon materials as cathode carriers to sustain the long-term stability of Li-S batteries. Recently, a series of mainstream approaches to address the “shuttle effect” show the incorporation of polar metallic materials into porous or hollow carbon frameworks, which not only boost the conversion kinetics of LiPSs but effectively inhibit the “shuttle effect” [
16,
17,
18]. Polar heteroatom doping by chemical modification of the surface (such as N, Pt, Co, etc.) can change the hybridized orbital patterns of carbon and effectively raise the chemisorption ability of carbon materials to LiPSs [
19,
20,
21,
22].
Herein, we propose a simple electrostatic spray synthesis method for the in-situ preparation of cobalt (Co) doped N-containing porous carbon spheres (Co/N-PCSs) has a high specific area of the surface (212 m2 g−1) based on polyvinylpyrrolidone (PVP) as the carbon precursor and polytetrafluoroethylene (PTFE) emulsion as the template by doping polar Co metal nanoclusters into N-containing porous carbon spheres (N/-PCSs). This synthesized Co/N-PCSs sulfur host material is not only able to adsorb LiPSs to alleviate the “shuttle effect,” but also accelerate the conversion kinetics of LiPSs due to the presence of Co nanoclusters. The Co/N-PCSs sample exhibits good cycling stability and rate performance with a capacity of 650 mAh g−1 after 300 cycles at a current density of 1 C, and a discharge specific capacity of 576 and 451 mAh g−1 at current densities of 3 C and 5 C.
2. Results and Discussions
The preparation procedure of the material Co/N-PCSs is shown in
Figure 1, the porous carbon spheres samples were prepared using PVP-Co (NO
3)
2-PTFE mixture by a sample electrostatic spray method in which PVP formed a good cross-linked network by electrostatic adsorption with PTFE. The experimental section provides details of the material preparation. Due to the fact that PVP melts at approximately 130 °C and decomposes at 400 °C to 500 °C, the carbon structure was first stabilized by pre-oxidation at 260 °C for 1 h to avoid excessive decomposition of PVP during high-temperature heat treatment. It is then heated at a temperature of 1200 °C for 3 h under Ar atmosphere. During the heating process, PVP gradually thermally decomposed and finally formed an N-containing carbon skeleton, while the microscopic and macroscopic phase separation brought about a uniform pore distribution. As examined by thermogravimetric analysis (TGA,
Supplementary Materials, Figure S1), the pre-oxidation enhanced the carbon yield. Both PVP and PTFE could be converted to porous carbon spheres with a total carbon yield of ~6.1%. The pyrolysis of small molecules (PVP) produced mesopores and micropores, while the decomposition of PTFE bulk nanoparticles produces a continuous series of macropores. It is noteworthy that the morphology of porous carbon spheres is strongly influenced by the mass ratio of PVP, PTFE, and deionized water (
Supplementary Materials, Figure S2a–c). With a fixed ratio of PVP to PTFE, the maximum ratio of deionized water is 60% to obtain a porous carbon sphere structure. The viscosity of the precursor solution directly affects the morphology of the microspheres obtained by electrostatic spraying. If the percentage of deionized water is increased to more than 60%, porous carbon spheres would be formed after pyrolysis; while reducing the percentage to 55% and 50% would result in increased viscosity and a fibrous structure. Interestingly, the large pore sizes of these different porous carbon structures are similar, as seen in the scanning electron microscopy (SEM) images, which also indicates that the continuous macropores are decomposed from PTFE. The SEM image (
Supplementary Materials, Figure S2c) shows that the prepared sample Co/N-PCSs has the morphology of porous microspheres with an average diameter of about 2 µm and a large number of pore structures on the surface at the time when the solvent is 60% of the solution mass. The high porosity is conducive to the increase of sulfur loading and, thus, higher energy density.
To research the microstructure of pores and elemental profiles of the synthesized samples, transmission electron microscopy (TEM) and SEM tests were performed on Co/N-PCSs and N-PCSs samples.
Figure 2a shows the SEM characterization of Co/N-PCSs, showing its porous morphology. The surface morphology of Co/N-PCSs samples was further characterized by TEM (
Figure 2b); interconnected macropores running through the entire interior of the porous carbon sphere reveal the highly porous characteristics of Co/N-PCSs. The interconnected pores form a 3D skeleton structure inside, which can not only effectively raise the S fixation capacity but alleviate the volume expansion [
23]. The Co nanoclusters are uniformly distributed inside the carbon sphere.
Figure 2c show the elemental mapping of the Co/N-PCSs sample, where the elements C, O, and N are closely mixed and spread throughout the porous carbon spheres, and surprisingly the Co nanoclusters are uniformly spread on the porous carbon sphere surface. The HRTEM image of Co/N-PCSs further shows the clear lattice edges of the Co nanoclusters with 0.21 nm crystal plane spacing, which corresponds to the Co (JCPDS: 15-0806) in the (111) crystal plane diagram (
Figure 2d,e). Clear carbon lattice fringes are also observed, indicating the presence of graphitic carbon and amorphous carbon within the porous carbon spheres.
Figure 2f shows the lattice diffraction pattern of Co/N-PCSs, illustrating its polycrystalline structure. This indicates that the Co nanoclusters are successfully introduced into the N-containing porous carbon spheres. The synthesized N-PCSs sample also shows highly porous morphological properties in TEM images (
Figure 2g). The carbon spheres have a diameter of about 2.7 μm, which is in agreement with the results observed in the SEM images (
Supplementary Materials, Figure S2). Based on the elemental mapping of the elemental distribution inside the carbon spheres, it clearly shows that the C, O, and N elements in the porous carbon spheres are uniformly spread throughout the carbon spheres (
Figure 2h).
To further determine the crystal structure in these different samples, The X-ray diffraction (XRD) pattern is shown in
Figure 3a. The unheated carbon spheres (PVA+ PTFE+ Co (NO
3)
2) are first characterized, as shown by the blue line in
Figure 3a, not showing the characteristic diffraction peak of Co. The black curve in
Figure 3a shows the XRD pattern of N-PCSs, which shows a broad peak around 42° compared to the curve of the untreated sample, which is the (110) crystalline plane of graphitic carbon. The XRD pattern of Co/N-PCSs obtained by adding Co (NO
3)
2 has a clear and subtle difference from the other samples (red curve in
Figure 3a), with a more pronounced diffraction peak located at 42°, which proves that a higher degree of graphitization has occurred in the carbon material, and two other diffraction peaks which are sited at 44° and 51° that correspond well to the Co (JCPDS: 15-0806) standards of the card. Compared with N-PCSs, Co/N-PCSs also shows better electrical conductivity, probably due to the higher graphitization catalyzed by Co nanoclusters at high temperatures. The degree of graphitization and defects in the N-PCSs and Co/N-PCSs samples are compared more convincingly by Raman analysis. The feature peaks of the two samples can be seen in
Figure 3b in the D-band shown at 1338 cm
−1 and the G-band shown at 1580 cm
−1, which match disordered carbon and ordered graphitic carbon structures, respectively. The I
D/I
G ratio (the ratio of the relative intensity of the D-peak to the G-peak) is usually commonly employed to indicate the extent of graphitization of carbon materials, and the I
D/I
G ratios of N-PCSs and Co/N-PCSs are 1.01 and 0.97, respectively. It is obvious that the graphitization degree of Co/N-PCSs is higher. The above analysis indicates that the doping of Co nanoparticles is beneficial to improve the graphitization degree of the material, and such an ordered structure can improve the electrical conductivity of Co/N-PCSs.
The surface-specific and porosity distributions of Co/N-PCSs and N-PCSs were tested by N
2 sorption. The N
2 sorption analysis curves illustrated in
Figure 3c and
Figure S3 in the Supplementary Materials are calculated according to the Brunner Emmet Teller (BET) method. Due to the intense correlation between the adsorbent and the surface, the adsorption amount of both samples increased rapidly at low relative pressures, and the adsorption/desorption analysis curves showed an upward convexity. The curve shows strong adsorption at equal pressure, forming a plateau and then the adsorption isotherm curve continues to rise, forming the typical characteristics of a type II isotherm due to the coexistence of macro- and mesopores [
24,
25]. The pore size distribution curves in the inset also confirm the presence of main mesopores and macropores in the synthesized samples, with the mesopore pore size concentrated around 3 nm and the macropore pore size at 60 and 110 nm.
Table S1 shows the specific surface areas of the two samples using different calculation methods, which are calculated using the BET multipoint method and are 194.2 m
2 g
−1 and 212 m
2 g
−1 for N-PCSs and Co/N-PCSs, respectively. Carbon-based materials with large specific surface areas doped with Co nanoclusters will provide more storage space for the active substance sulfur while possibly inhibiting the “shuttle effect” and accelerating the transformation between LiPSs.
To further reveal the atomic ratios and chemical states of Co/N-PCSs, X-ray photoelectron spectroscopy (XPS) characterization of N-PCSs and Co/N-PCSs samples are performed, respectively, and the results are shown in
Figure 3d–f. The high-resolution C 1s spectra showed that the most abundant C elements in Co/N-PCSs and N-PCSs samples are mainly in the presence of C-C/C=C, C-O/C-N and C=O forms, with peaks at the binding energies of 284.58 eV, 285.58 eV and 289.08 eV, respectively (
Figure 3d) [
26]. The absence of C-N bonds indicates that N doping is achieved in both samples. The high-resolution N 1s profiles of N-PCSs and Co/N-PCSs are shown in
Figure 3e. It is shown from the figure that three N elemental binding states are present in both samples, namely Pyridinic-N, Pyrrolic-N, and Graphitic-N [
27]. Among them, the incorporation of Pyrrolic-N and Pyridinic-N can improve the energy of binding to soluble LiPSs. In particular, the Li bond between pyridinic-N and Li is a dipole-dipole interaction and, thus, has a strong adsorption effect on Li
+ in LiPSs [
27,
28]. In addition, the strong electron-giving ability of Graphitic-N is also beneficial to boost the electrical conductivity and rate performance of the material. Calculation of the percentage of bound states of the three N elements based on the percentage of N 1s split peak area shows that the content of Graphitic-N in Co/N-PCSs is 44.5%, which is higher than that of N-PCSs, indicating that it possesses higher electrical conductivity [
29]. The high-resolution O 1s spectra in
Figure S4 in Supplementary Materials show that the porous carbon spheres are rich in O-containing functional groups, which can provide more electrochemical sites. The presence of metallic Co is further indicated by the Co (0) peak with a bonding energy of 779.68 eV in the high-resolution Co 2p profile obtained for the partitioning of elemental Co in Co/N-PCSs in
Figure 3f. The characteristic peaks of Co 2p
3/2 and Co 2p
1/2 appear at bonding energies of 781.68 eV and 796.28 eV, respectively, while those of 786.08 eV and 802.88 eV are considered satellite peaks [
30]. Based on the XPS analysis, it was verified that the Co metal exists as metal nanoclusters on the porous carbon spheres, which will help to improve the electrochemical performance by increasing the adsorption capacity and accelerating the transformation of LiPSs.
To investigate whether the electrochemical performances of the Co nanocluster doped porous carbon spheres are enhanced, the active sulfur material is loaded into the synthesized N-PCSs and Co/N-PCSs samples by the melt-diffusion method; consult the experimental section for specific experimental details. The sulfur-containing cathodes are denoted as Co/N-PCSs@S and N/-PCSs@S, respectively.
Figure S5a in the Supplementary Materials shows the SEM image of Co/N-PCSs@S. As illustrated in the figure, the loading of sulfur resulted in the disappearance of the multi-porous structure on the surface of the carbon sphere and no large blocks of sulfur monomers appeared, indicating that the sulfur monomers entered the interior of the porous carbon sphere through the surface pore structure, and the Co/N-PCSs@S sample is perfectly synthesized. The specific content of sulfur as an active substance in the samples is tested by TGA.
Figure S5b, Supplementary Materials shows that the mass of both samples does not decrease significantly until 160 °C under Ar atmosphere. When the temperature exceeds 160 °C, the liquid sulfur keeps evaporating, so the masses of Co/N-PCSs@S and N-PCSs@S decrease significantly. When the temperature exceeds 400 °C, with the mass change stabilized, it indicated that the sulfur is basically completely evaporated, from which it could be concluded that the sulfur contents of N-PCSs@S and Co/N-PCSs@S are 69% and 70%, respectively. The sulfur content of both is almost the same, and the active mass content ratio in the electrochemical performance test is obtained from the TGA test data. The performance of these cathodes is evaluated in specific Li metal cells.
Figure 4a presents the cyclic voltammetry (CV) profiles of Co/N-PCSs@S for an initial three cycles at a scan rate of 0.1 mV s
−1 and a test voltage of 1.7~2.8 V. It can be noticed that the CV profile is similar to that of a typical Li-S cell, consisting of two significant reduction peaks and one for oxidation [
31]. Two peaks of reduction at 2.2 V and 1.9 V can be noticed in the CV profile of the first cycling, representing the electrode reactions of S
8+Li
++e
−→Li
2S
n (4 ≤
n ≤ 8) and Li
2S
n+Li
++e
−→Li
2S+Li
2S
2, respectively. During the anodic scan, the peak of oxidation at 2.4 V represents the conversion of Li
2S and Li
2S
2 to the final S
8 process [
32]. After the first cycle, the position of the peak of reduction is slightly shifted to the right and gradually stabilizes in the subsequent reactions which shows the higher reversibility of the Co/N-PCSs@S sample.
Figure 4b displays the initial cycle charge/discharge profile of the electrode materials at a current density of 0.1 C (1 C = 1675 mAh g
−1). The initial discharge-specific capacity of Co/N-PCSs@S is as high as 1420 mAh g
−1, which is significantly higher than N-PCSs@S (1277 mAh g
−1). It is worth noting that the polarization voltage of Co/N-PCSs@S is only 0.16 V, which is much lower than that of N-PCSs@S (0.19 V). It shows that the polarization voltage of the porous carbon after Co nanocluster doping is smaller, owing to the faster electron transport rate and the better catalytic conversion performance.
As seen in
Figure 4c, the electrochemical impedance spectroscopy (EIS) pattern is made up of a semi-circle in the high-frequency area and the slope in the low-frequency area, and the intersection of the high-frequency region with the Z’ axis represents the inherent internal resistance R
S inside the cell The semi-circle in the mid-high-frequency area is corresponding to the charge transfer resistance R
ct at the electrode-electrolyte interface and the slope in the high-frequency area corresponds to the ion diffusion resistance Z
ω [
33]. The inset from
Figure 4c shows that the internal resistance R
S and R
ct is smaller for Co/N-PCSs@S compared to N-PCS@S, indicating that the charge transfer at the electrolyte interface is accelerated due to the Co nanoclusters encapsulated within the porous carbon spheres as conductors. The linear slope magnitude in the low frequency area is negatively correlated with the ion diffusion resistance Z
ω. The slope of the straight-line part of Co/N-PCSs@S is larger, indicating that the ions in the electrolyte can quickly reach the electrode surface for reaction.
To demonstrate that Co/N-PCSs@S possesses better electrochemical performance, two electrode materials, N-PCSs@S and Co/N-PCSs@S, are tested for rate performance and cycle performance. For the N-PCSs@S electrode, the discharge-specific capacities are 1066, 838, 714, 653, 179, and 112 mAh g
−1 at current densities of 0.1, 0.3, 0.5, 1.0, 3.0, and 5.0 C, respectively. For the Co/N-PCSs@S electrode, the average discharge capacities at current densities of 0.1, 0.3, 0.5, 1.0, 3.0, and 5.0 C are 1095, 930, 854, 736, 576, and 451 mAh g
−1, respectively (
Figure 4d). In comparison with N-PCSs@S, Co/N-PCSs@S possesses a higher discharge-specific capacity, especially at different current densities, which is associated with the doping of Co nanoclusters to promote electron transport in the electrode materials. To evaluate the cycle stability of the two electrode materials, three cycles are first activated at a lower current density of 0.1 C, followed by a subsequent charge/discharge test at a high current density of 1 C.
Figure 4e shows the discharge-specific capacity of N-PCSs@S and Co/N-PCSs@S electrode materials in the first 300 cycles. In terms of the decay rate of the single cycle capacity, Co/N-PCSs@S is only 0.051%, while N-PCSs@S is 0.086%. In addition, the Coulombic efficiency (CE) of Co/N-PCSs@S remained at almost 100% after the initial three cycles. Notably, Co/N-PCSs@S exhibits great potential as a high-performance cathode material for Li-S batteries.
In the Li-S batteries, the adsorption of LiPSs by carbon matrix materials is essential to promote the stability of the battery cycle. To verify whether the Co nanoclusters enhance the adsorption of LiPSs, adsorption experiments are performed on the synthesized samples and the adsorbed solutions are analyzed using UV-vis absorption spectroscopy [
34]. After adding N-PCSs to the initially brown 1 mM Li
2S
6 solution and leaving it for 6 h, the color of the Li
2S
6 solution showed no significant change and the solution color remained light brown (
Figure 5a). However, after the addition of Co/N-PCSs samples, the color of the Li
2S
6 solution has been colorless within 6 h. It indicates that due to the presence of Co nanoclusters, the Co/N-PCSs samples exhibited stronger adsorption of LiPSs relative to the N-PCSs samples. This fact is further confirmed by the subsequent UV-vis spectrogram (
Figure 5b), where Li
2S
6 mixed with the Co/N-PCSs sample shows the weakest peak strength of absorption in the range of 350–700 nm. This weak peak represents the minimum remaining level of Li
2S
6 [
35]. On the other hand, the peak strength of absorption of the N-PCSs sample immersed in Li
2S
6 solution is higher, which indicates the minimum remaining level of Li
2S
6 is higher, reflecting its weaker ability to absorb Li
2S
6. This is complementary to the above experimental findings. To obtain data on the catalytic performance of Co nanoclusters in Co/N-PCSs samples that are capable of speeding up the transformation of LiPSs. We assembled two electrode materials of Co/N-PCSs and N-PCSs into separate symmetric cells with an electrolyte of 1 mM Li
2S
6 dissolved in 1 M LiTFSI in a 1:1 (
v/
v) DME/DOL. CV testing of symmetric half-cells is conducted at a scan rate of 10mV s
−1 over a voltage range of −1.0 V to 1.0 V. During the CV test, the Li
2S
6 undergoes a series of reversible transition processes at the electrode-electrolyte interface, re-generating the appropriate current reactions and wave peaks [
36]. The results showed that in the Li
2S
6 electrolyte, Co/N-PCSs showed much higher peak currents and sharper peaks than N-PCSs, and the redox reaction of LiPSs has higher reversibility, indicating that Co nanoclusters doping is more favorable to catalyze the transformation of LiPSs and raise the electrochemical kinetics of redox. (
Figure 5c). Li
2S nucleation experiments are further tested and Li
2S deposition rates are calculated based on Faraday’s law,. The results are shown in
Figure 5d [
37]. By comparing the deposition area on Co/N-PCSs and N-PCSs electrodes, Co/N-PCSs has a larger deposition area and higher deposition capacity, which indicates that Co/N-PCSs provides more active sites for achieving high capacity. The electrode of Co/N-PCSs has a faster Li
2S deposition rate (average deposition rate of 7.85191 × 10
−5 mA s
−1) compared to the case of the N-PCSs electrode (average deposition rate of 1.40716 × 10
−5 mA s
−1). The above results show that the doping of Co nanoclusters provides additional activity sites, regulates the deposition of Li
2S, and effectively accelerates the rapid conversion of higher-order liquidized LiPSs and lower-order solidified Li
2S, as well as offering the robust chemoadsorption of LiPSs. Thus, the Co/N-PCSs sulfur host material can run longer with sustainable capacity and release more capacity at high currents. [
38].