The synthesis route of the PSi@C(N) anode material and the characterization of key intermediates are summarized in Fig. 1a. The Mg2Si particles were prepared at 650°C by mixing Mg, microsilica and NaCl. After the removal of Mg by HCl washing, the porous silicon particles were coated with a layer of RF resin polymer by sol-gel method, and then carbonized under nitrogen atmosphere. The XRD patterns (Fig. 1b) confirmed the success of each synthesis step. The disappearance of SiO2 peaks and the appearance of Mg2Si peak (JCPDS card No. 36–0773) ( M. Chen et al., 2018), indicated that the precursors of the powders were completely converted into Mg2Si. The pattern of the final material PSi@C(N) suggested the successful removal of MgO and NaCl, as well as the existence of silicon (JCPDS card No. 27-1402) (W. Wu et al., 2016). Noted the peak narrowing of PSi@C(N) was attributed to lattice deformation due to stress generated in the surrounding the Si atom network after nitrogen doping (M. Ashuri et al., 2020). To further prove the existence of carbon-coating, Raman spectra were employed (Fig. 1c). The peak centered at ~ 510 cm− 1 stood for the silicon Raman phonon vibration. In addition, the two peaks located at 1342 cm− 1 and 1589 cm− 1, corresponding to the D and G bands of carbon materials, respectively, which indicated the formation of the carbon coating layer (Y. Cheng et al., 2016).
The porous microstructure of microsilica and PSi@C(N) were studied by nitrogen adsorption-desorption analysis. The specific surface area values were calculated from the Barrett-Emmett-Teller (BET) method, while pore volume and pore size distribution were computed by the Barrett-Joyner-Halenda (BJH) method (Fig. S2 and Table S2). Microsilica particles displayed relatively low specific surface area (22.225 m2 g− 1) and low pore volume (0.080 cm3 g− 1). In contrast, PSi@C(N) particles exhibited higher surface area (63.942 m2 g− 1) with larger pore volume (0.265 cm3 g− 1). The tight carbon-coating structure on the outer layer of PSi@C(N) resulted in a decrease in the pore size from 4.256 nm to 3.385 nm. The developed porous microstructure with high specific surface area is favorable for Li+ ions migration and can effectively relieve structural stress roots from cycling (J.R. Matos et al., 2003).
XPS is employed to analyze the individual components of composites (Fig. 1d-1e). There are three main peaks in the high-resolution XPS spectra of Si 2p, of which the peaks at 101.5 eV and 103.5 eV are attributed to the Si 2p1/2 and Si 2p3/2 of Si, respectively, while the characteristic peaks of Si-O-Si bond of SiOx at 100.8 eV correspond to Si-O-Si (Y. Xiao et al., 2014). This has demonstrated that, SiOx is present on the surface of silicon particles. The C 1s spectrum has been deconvoluted into three distinct peaks, the peaks at 284.5 eV, 286.2 eV and 282.5 eV, which corresponds to the sp2 carbon atom, C-O bond and C-Si bond, respectively. This indicates the presence of graphitic carbon (M. Xia rt al., 2020), and there is a small amount of SiC is formed at the interface between Si and C during the process of high temperature carbonization (H. Tang et al., 2015) .that. The N 1s spectrum (Fig. 1f) contains two sets of peaks. The peaks at 395.5.8 eV and 398.5 eV, which stands for the N2 and Si3N4 phases. In conclusion, the XPS analysis results have further confirmed the successful synthesis of PSi@C(N) nanocomposites. The TGA measurement (Fig. S2) revealed a mass loss of 9.62 wt%, caused by the oxidation of the carbon layer.
The microstructure and morphology of the products were examined by SEM and TEM. The microsilica particles exhibit spherical morphology with particle diameter ranging between 100 to 500 nm (Fig. 2a). After magnesiothermic reduction, the forming Mg2Si particles does not show any particular morphology (Fig. 2b). Oxidative acid etching results in a formation of a porous Si nanoparticles with fluffy morphology, as shown in Fig. 2c. Silicon nanoparticles with diameter ranging from 100 to 200 nm are beneficial for fast diffusion of Li+ ions (Fig. 2d), while a thin SiOx layer (~ 5 nm), which is formed around the porous Si nanoparticles, can effectively alleviate the stress caused by the volume change of silicon particles during the cycling (Fig. 2e) (D. Chen et al., 2012, X F.-F Cao et al.,2011, S. Zhang et al., 2010). After carbon coating, PSi@C(N) particles with primary nano building blocks and many void spaces are formed (Fig. 2f). Elemental mapping confirms the presence of N-doped carbon layer (Figs. 2g-2j). This carbon layer improves the electronic conductivity of material and prolongs the cycle life of electrode (N. Liu et al., 2012).
In order to evaluate the electrochemical performance of the as-prepared PSi@C(N) particles as active material, half cells with Li chip as reference electrode were fabricated. The specific capacity was normalized to the total weight of PSi@C(N) material. The PSi@C(N) electrode was pre-lithium after 10 initial cycles at 200 mAh g− 1, and its performance was tested in a long-term cycle at 1000 mA g− 1 (Fig. 3a). The PSi@C(N) electrode delivers 803.92 mAh g− 1 after 300 cycles, which is ~ 60% of the initial capacity. This is much higher than both microsilica and porous silicon (Fig. 3b). The capacity of microsilica electrode is below 300 mAh g− 1 because silica is a poor conductor. Porous silicon has delivered higher capacity, 1024.21 mAh g− 1 after 50 cycles. This is attributed to nanoscale size effect and porous structure of PSi, which leaves plenty of room for the silicon to expand in volume during charging. The cycling stability of the PSi anodes is further improved by carbon coating. The reversible capacity of PSi@C(N) anodes is 1304.6 mAh g− 1 after 50 cycles. Figure 3c shows the voltage distribution of PSi@C(N) electrode to Li+/Li between 0.01 and 2 V at a current density of 1000 mA g− 1. The initial discharge and charge capacities are 1302.4 and 1075.6 mAh g− 1, respectively. The corresponding initial Coulomb efficiency (ICE) is 76.92%. After 5 cycles, the C.E. gradually increases to 98%, indicating the formation of stable SEI layer and reversible Li+ storage (S. Weckend et al.,2016, J. Cui et al., 2017). Rate capability testing results of PSi@C(N) electrode are presented in Fig. 3d. By step-wise increasing the current density from 200 mA g− 1 to 4000 mA g− 1, the average capacity was increased from 1957.2 to 1209.3 mAh g− 1. When the current rate is switched back to 200 mA g− 1, the reversible capacity is restored to 1892.9 mAh g− 1. Cyclic voltammetry (CV) measurements at the PSi@C(N) electrode were made in the range of 0.01 to 2.0 V at a scanning rate of 0.1 mV s− 1, as shown in Fig. 4e. In the first cycle, a wide cathode peak at ~ 0.6V appears, which is not detected in the following cycles. We attribute this peak to SEI membrane formation, which is an irreversible reaction. The peak observed at ~ 0.2V is related to the alloying reaction between lithium and silicon, which is consistent with the charge-discharge curve (L. Su et al., 2015, M. Gu et al., 2015). The anodic peaks of LixSi at 0.35 V and 0.5 V belong to the dealloying process respectively (J. Kong et al.,2013).
Figure 4 shows the electrode kinetics of the preparation of key materials for each component in the synthesis process. Figure 4(a, b, c, S4) shows the EIS spectra of microsilicon, porous silicon and PSi@C(N) electrodes as well as the fitting lithium-ion diffusion coefficient, which proves that the composite electrode has high lithium-ion diffusion ability and low resistance. Figure 4(d, e, f) shows the CV curves of PSi@C(N) electrodes at different scanning rates and the calculation of battery properties. The results indicate that capacitive behavior plays a dominant role in the storage and release of lithium ions at high scanning rates. The introduction of N doping layer enhances the connection between electrolyte and active material and improves the pseudocapacitive behavior (V. Augustyn et al., 2014). Figure 4(g, h, i) illustrates the diffusion behavior of lithium ions by Galvanostatic Intermittent Titration Technique (GITT). The three dynamics fully explain the effective contribution of each component of the composite to the overall click. The electrode with PSi@C(N) as active material has better dynamic behavior.
To further investigate the effect of the materials design on the performance of the electrode, coin cells were disassembled after 100 cycles and PSi@C(N) electrodes were extracted and examined under SEM (Figs. S5b-S5c). The measured thickness of the coated material on the fresh electrode is ~ 43 µm. After 100 charge/discharge cycles, the thickness is increased to ~ 51 µm (18.6% increase). Considering the huge volume change of silicon in lithiation/delithiation process, as depicted in Fig. S5a, we can relate the small increase in the electrode thickness to the unique design of PSi@C(N) composites. The uniform distribution of nanocrystalline silicon inside the matrix has inhibited the inhomogeneous expansion, and thus the structural porosities reserve sufficient space for the silicon expansion and maintain the integrity of the electrode for several cycles. The stringent SiOx layer formed around the core, acts like a cushion, which could harness volume expansion of silicon and preserve the structural integrity of the electrode. Furthermore, the N-doped carbon shell promotes the electronic conductivity of the electrode and shortens the diffusion path of Li+ ions. In addition, the carbon layer prevents the vigorous side reactions between electrode and electrolyte at the interface. Therefore, it is concluded that, a steady SEI layer is formed around the active materials (F. Xi et al., 2021).