Ji Young
Kim
,
Dong Jae
Chung
,
Tae Rim
Lee
,
Donghan
Youn
and
Hansu
Kim
*
Department of Energy Engineering, Hanyang University, 222 Wangsimni-ro, Seongdong-gu, Seoul 04763, Republic of Korea. E-mail: khansu@hanyang.ac.kr
First published on 13th February 2025
Silicon monoxide (SiO) is one of the most promising post-graphite anode materials for lithium-ion batteries. Prelithiation of SiO has been proposed as a potential method to improve its initial coulombic efficiency (ICE), which is a persistent challenge for SiO. However, pre-lithiation can decrease the capacity of SiO due to the formation of lithium silicate phases. To address this issue, we developed a strategy to improve the ICE and reversible capacity of SiO through lithium and oxygen engineering. Prelithiated Si-enriched SiO0.5 prepared by high-energy mechanical milling of Si and SiO with lithiation followed by LiH treatment, exhibited a capacity of 2093 mA h g−1 with an ICE of 88.1%, significantly surpassing the performance of both pristine and prelithiated SiO. While increasing the Si content may typically result in poor capacity retention, the unique porous structure formed by the Si and lithium silicate phases in this study mitigated this effect, ensuring capacity retention over 300 cycles by alleviating the expansion of Si during the lithiation/delithiation process.
Fig. 2a shows the X-ray diffraction (XRD) patterns of the synthesized SSO composites, along with the micron-sized Si particles and SiO particles used in the composites. The XRD patterns indicate that the Si particles exhibit crystallinity, while the SiO particles are amorphous. The Bragg peaks corresponding to the crystalline Si phase in the SSO composites broadened after high-energy mechanical milling due to the inhomogeneous strain induced during the milling process.23 The broad peak at 23° in the XRD patterns of the SSO composites corresponds to amorphous SiO. After prelithiation of the SSO composites with LiH, the XRD pattern of the LSSO material showed the disappearance of the Bragg peak for amorphous SiO, and the emergence of new Bragg peaks corresponding to the Li2Si2O5 and Li2SiO3 phases. These lithium silicate phases were formed by the prelithiation of SiO with LiH. 16,17Fig. 2b shows the Raman spectra of the SSO-1, LSSO, micro-Si, and amorphous SiO particles. Two different Raman bands are observed in the Raman spectrum of SSO-1: a broad band from 400 to 480 cm−1, attributed to the amorphous Si phase (a-Si) from the SiO particles, and a sharp band at 520 cm−1 corresponding to the crystalline Si phase (c-Si) from the Si particles. According to the random mixture model of SiO, SiO consists of a-Si and amorphous SiO2 (a-SiO2) in an equal molar ratio.16,17,24 These results indicate that SSO-1 particles maintain both the physical properties of both Si and SiO even after high-energy ball milling. By comparing Raman spectra of SSO-1 and LSSO, we observed the phase transition before and after prelithiation. The Raman bands corresponding to a-SiO2 and a-Si disappeared, while only the distinctive Raman band of the transverse optical mode of c-Si was detected at 520 cm−1. This indicates that the amorphous SiO reacted with Li released from dehydrogenated LiH, forming c-Si and lithium silicate phases during the heat treatment.16,17,25 These results are consistent with previous studies on the prelithiation of SiO using LiH. To further understand the phase transformation of SSO-1 during prelithiation, 29Si magic angle spinning nuclear magnetic resonance (29Si-MAS-NMR) spectroscopy was performed, as shown in Fig. 2c. The 29Si NMR spectrum of SSO-1 exhibits two chemical shifts at −75 ppm and −108 ppm corresponding to Si (−70 ppm for a-Si and −81 ppm for micro-Si) and a-SiO2, respectively. These results are in good agreement with the Raman spectra of the SSO-1 material. After prelithiation, the chemical shifts associated with a-Si and a-SiO2 disappeared, and the distinctive chemical shifts corresponding to lithium silicates (−92 ppm for Li2Si2O5 and −75 ppm for Li2SiO3) and c-Si (−83 ppm) appeared. These results align well with the XRD pattern of the LSSO material. Notably, the silicon phase did not react with lithium from LiH to form a Li–Si alloy phase. These results clearly show that the Si particles in SSO-1 remained chemically unreacted throughout the process, whereas the amorphous SiO particles reacted with Li to form lithium silicate phases and c-Si.
![]() | ||
Fig. 2 (a) XRD pattern of Si, SiO, SSO-4, SSO-2.03, SSO-1, SSO-0.25 and LSSO, (b) Raman spectra of SSO-1 and LSSO and (c) 29Si MAS NMR spectra of Si, SiO, prelithiated SiO, SSO-1, and LSSO. |
Fig. 3a and b show the electrochemical performance of the prepared SSO composite electrodes for the first cycle. The SSO composite electrodes exhibited initial reversible capacities of 3157 mA h g−1, 2753 mA h g−1, 2129 mA h g−1, and 1840 mA h g−1 for Si/SiO ratios of 4, 2.03, 1, and 0.25, respectively. The ICEs of the SSO composite electrodes were 80.5% (SSO-4), 78.4% (SSO-2.03), 69.7% (SSO-1), and 59.4% (SSO-0.25). Both the discharge capacity and the ICE of the SSO composite electrodes were higher than those of the pristine SiO electrode, which delivered an initial reversible capacity of 1076 mA h g−1 with an ICE of 44.8% (Fig. S1†). These results also show that the ICE and discharge capacity of the SSO composite electrodes increase with higher Si content. Among the SSO composites, SSO-1 demonstrated optimal performance with a discharge capacity exceeding 2000 mA h g−1 while maintaining stable cycle performance. Based on these results, SSO-1 was chosen for prelithiation to further improve the electrochemical performance. The LSSO, the prelithiated form of SSO-1, electrode delivered a discharge (lithiation) capacity of 2378 mA h g−1 and a charge (delithiation) capacity of 2093 mA h g−1 corresponding to an ICE of 88.1%. This represents a substantial 19.8% improvement compared to the ICE of the SSO electrode with the same Si content (69.7%). The increase in ICE is attributed to the pre-emptive formation of irreversible phases during prelithiation with LiH, which reduces lithium consumption during the initial cycle. The differential capacity plots (DCP) of the SSO-4, SSO-2.03, SSO-1, SSO-0.25, and LSSO electrodes for the initial cycle show a DCP peak at approximately 200 mV (vs. Li/Li+) and a peak below 100 mV (vs. Li/Li+) upon lithiation. (Fig. S2†) The peak at 200 mV is associated with the irreversible reaction of a-SiO2 with Li, forming lithium silicates, while the peak below 100 mV corresponds to the alloying reaction of the Si phase. As the portion of SiO in the SSO composite increases, the peaks between 460 and 200 mV (vs. Li/Li+) become more pronounced, showing the maximum intensity in the SSO-0.25 composite. The disappearance of these peaks in the DCPs during the second cycle suggests that these peaks reflect the irreversible electrochemical reaction of SiO in the first cycle. In contrast, the LSSO electrode shows a more negative and sharper DCP peak at 70 mV (vs. Li/Li+), suggesting that the a-SiO2 in SSO-1 effectively reacted with LiH during prelithiation, eliminating the significant peak at approximately 200 mV (vs. Li/Li+). Upon delithiation, the SSO electrodes showed a similar tendency. The peak intensity around 450 mV, corresponding to the de-alloying reaction of crystalline Li3.75Si (c-Li3.75Si), increased with an increase of Si content in the composite. Notable differences were observed between the SSO-1 and LSSO electrode. The SSO electrode shows two peaks at approximately 300 and 450 mV (vs. Li/Li+) corresponding to lithium removal from the amorphous LixSi alloy and c-Li3.75Si (the richest Li–Si phase), respectively. These results show that the SSO-1 electrode has the electrochemical features of both micro-Si and SiO materials; however, after prelithiation, the lithium storage mechanism of the LSSO electrode changes due to the phase transition of SSO-1 into Li-active (Si phase) and Li-inactive phases (lithium silicate phases). Fig. 3c shows the cycle performances of the SSO-1 and LSSO electrodes over 300 cycles at a rate of 500 mA g−1, following three formation cycles at a rate of 100 mA g−1. The SSO composite electrode with the highest Si content (SSO-4) showed the highest specific capacity but experienced the fastest capacity fading within the initial 50 cycles, whereas the SSO and SSO-0.25 electrodes showed stable cycle performance with lower discharge capacities. These results indicate a trade-off relationship between specific capacity and cycle performance of the SSO-1 composite electrodes. Prelithiation was expected to address this trade-off relationship in SSO composite anode materials and improve the ICE of the SSO composite electrodes. After 300 cycles, the SSO-1 electrode maintained a capacity of 940 mA h g−1 (44.2% retention), while the LSSO electrode delivered 1070 mA h g−1 with a capacity retention of 51.1%. The superior cycle performance of the LSSO electrode compared with the SSO-1 electrode can be attributed to the Li-inactive lithium silicate phase, which suppressed the expansion of the electrode and maintained its integrity during the charge/discharge process. Although the SSO-1 electrode did not perform as well as LSSO, it still showed improved cycle performance compared to pristine Si and SiO electrodes, which exhibited rapid capacity fading with capacity retention of 16.5% and 33.6% of the initial capacity, respectively, after 300 cycles (Fig. S1, ESI†). These results revealed that the Li–inactive lithium silicate buffer phase was not the only factor contributing to the improved cycling performance of the LSSO electrode. The free voids in the SSO composite particles created through high-energy ball milling also played a role in relieving the mechanical stresses generated in the electrode during cycling. It is widely recognized that a porous microstructure with large surface areas can accommodate volume changes in the Si phase during cycling and shorten the diffusion path for Li ions, thereby facilitating Li ion transport in Si-based anode materials.26–28Fig. 3d shows the rate capability of SSO-1 and LSSO electrodes at current densities ranging from 0.1 to 5 A g−1 after three formation cycles at 100 mA g−1. At a high current density of 5.0 A g−1, the SSO and LSSO electrodes maintained 65.7% and 77.6% of the capacity retention obtained at a rate of 0.1 A g−1, respectively. The materials showed excellent rate capabilities, largely due to their porous structures, which facilitate Li ion transport within the electrode.29
Fig. 4a and c show scanning electron microscopy (SEM) images of the SSO-1 and LSSO materials. The top-view SEM images show that both SSO-1 and LSSO have rough surfaces compared to the starting materials, Si and SiO particles (Fig. 4a and c; see also S3†). Cross-sectional observation by focused ion beam-scanning electron microscopy (FIB-SEM) revealed the presence of internal pores within the particles (Fig. 4b–d and see also S4†). These pores were generated during the high-energy ball milling process, where mechanical impact causes the fracture of primary particles, leading to particle refinement.30,31 These refined particles then aggregate into secondary particles through cold welding during the continuous milling process, simultaneously creating free voids within the secondary structures due to imperfect particle packing. More importantly, FIB-SEM revealed two distinct domains, differentiated by brightness levels: Si and SiO. To investigate the microstructure of SSO-1 and LSSO particles in more detail, high-angle annular dark-field scanning transmission microscopy (HAADF-STEM), energy dispersive X-ray spectroscopy (EDS), selected area electron diffraction (SAED), and high-resolution transmission electron microscopy (HR-TEM) were conducted. Fig. 4e and g show HAADF-STEM images of the SSO-1 and LSSO particles, both of which exhibit porous structures, with voids appearing as black areas, similar to the FIB-SEM images. Combined with EDS analysis, the particles can be classified into two regions: Si and O coexisting domains and Si-rich domains, both of which have reduced sizes compared to the starting materials (Fig. 4e–f for SSO-1 and 4g–h for LSSO). In the Si and O coexisting domains, a microstructural difference was observed between SSO-1 and LSSO. There is no Z-contrast observed in SSO-1, whereas there is a distinguishable Z-contrast in LSSO. To further investigate the microstructural difference, the coexisting domains were magnified up to 800 K (Fig. 4i and k). The HAADF-STEM images of SSO-1 and LSSO at 800 K magnification, along with the corresponding element mapping from EDS analysis revealed that there are two distinct Si-rich and O-rich domains in LSSO, whereas SSO-1 showed a uniform distribution of Si and O (Fig. 4i and j for SSO-1 and Fig. 4k and l for LSSO). SAED patterns showed the presence of only the crystalline c-Si phase (d022 = 1.90 Å) in SSO-1, while LSSO showed three different crystalline phases: Si (d111 = 3.11 Å), Li2Si2O5 (d110 = 5.39 Å) and Li2SiO3 (d111 = 3.30 Å) (Fig. 4m for SSO-1 and n for LSSO; see also Fig. S5†). HR-TEM observations are in good agreement with the XRD and NMR analyses. Based on these phase analysis results, the areas with uniform Si and O distributions were assigned to SiO, and the Si- and O-rich domains were assigned to a mixture of c-Si and lithium silicate (Li2SiO3 and Li2Si2O5) phases (Fig. 4j and l). To further understand the pore structure of SSO-1 and LSSO observed through FIB-SEM and STEM analysis, Brunauer–Emmett–Teller (BET) analysis was performed (Fig. 4o). The SSO-1 material possessed a BET surface area of 16.73 m2 g−1, which is larger than those of the starting materials Si (8.05 m2 g−1) and SiO (2.62 m2 g−1). The pore volume of SSO-1, estimated at 0.06699 cm3 g−1 using the BJH method, was also larger than those of the starting materials (0.02616 cm3 g−1 for Si particles and 0.00718 cm3 g−1 for SiO particles) (Fig. S5, ESI†). This suggests that ball milling increases both the surface area and pore volume of the material.25,32 The surface area and pore volume of the LSSO material after prelithiation of SSO-1 were 9.52 m2 g−1 and 0.02564 cm3 g−1, respectively. This reduction in surface area and pore volume in LSSO materials might be closely related to the prelithiation process. Similar results were reported by Alanoina et al., where prelithiation using lithium stearate filled pores and repaired surface cracks, as verified by TEM and BET analyses.33
Fig. 5a shows the thicknesses of the SSO-1 and LSSO electrodes in the pristine state after lithiation (discharging to 10 mV vs. Li/Li+), after 1 cycle, and after 300 cycles. The thickness change after the lithiation of the LSSO electrode was 95%, significantly lower than the expansion observed in the SSO-1 electrode (130%) under the same conditions (Fig. 5b). During the initial lithiation, the presence of pre-emptive phases accounts for the differences in electrode expansion. The greater expansion of the SSO-1 electrode is attributed to the formation of irreversible phases such as Li4SiO4 and Li2O. The thickness changes of these electrodes after 1 cycle followed a similar trend: 22% for SSO-1 and 13% for LSSO. After 300 cycles, the LSSO electrode expanded by 145%, which is much less than the expansion observed in the SSO-1 electrode (292%). The dimensional stability of the electrodes is directly related to their cycling performance, as shown in Fig. 3c.
To further test the viability of the full cell, a coin type full cell was assembled using SSO-1 and LSSO blended with graphite as the negative electrode (NE) and LiNi0.88Co0.06Mn0.06O2 as the positive electrode. Fig. 6a shows first cycle voltage profiles of full cells employing SSO-1 and LSSO. The first cycle discharge areal capacities of full cells with LSSO as the NE were higher than those of full cells with SSO as the NE (2.81 mA h cm−2 for SSO-1 and 3.29 mA h cm−2 for LSSO), which is mainly due to the improved ICE of the LSSO NE (81.7% for LSSO and 70.7% for SSO-1). The full cell employing LSSO as the NE showed greater discharge areal capacity for 100 cycles after two formation cycles compared to the full cell with SSO-1 as the NE (Fig. 6b). Fig. 6c shows the capacity retention and coulombic efficiency of the full cells. The capacity retention of the LSSO electrode was superior to that of the SSO-1 electrode. Upon prolonged cycling up to 100 cycles, the full cell with LSSO (99.3%) showed higher average coulombic efficiency than the full cell with SSO-1 (98.9%). These results suggest that microstructural modifications, such as incorporating free voids and pre-emptive phases, can enhance both the ICE and cycle performance while also increasing the specific capacity of SiO-based anode materials.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d4ta08234f |
This journal is © The Royal Society of Chemistry 2025 |