Rice husk-originating silicon–graphite composites for advanced lithium ion battery anodes

High capacity silicon (Si) anodes have received much attention from the battery community because their superior specific capacities can increase the energy densities of lithium ion batteries (LIBs) substantially [1–4]. Nonetheless, the cycle lives of Si anodes do not usually meet the commercial standards because of the immense volume change of Si during repeated charge–discharge cycles [5–7]. The volume change causes the pulverization of active material, delamination of the electrode, and unstable solid–electrolyte-interphase (SEI) layer.

In an effort to address the aforementioned issues related to the volume expansion, a variety of nanostructured Si materials have been introduced, as the nanometer dimensions can effectively release the stress built during the volume expansion of Si. Si nanomaterials along this direction include diverse morphologies, such as nanoparticles (Si NP) [8–12], nanowires [13–16], two-dimensional (2D) nanosheets [17–19], and etc. Although these nanostructured Si demonstrated improved cycling performance to great extents, they suffer from low tap densities and are more liable to unwanted surface reactions due to the large surface-to-volume ratios [20, 21]. Thus, it is desired to develop Si microparticles in which internal nanostructures are embedded. Meso- or microporous Si materials are well-aligned to this design consideration because their particle sizes are in the micrometer range while the wall thicknesses of pores are in the nanometer range.

In developing meso- or microporous Si materials, one critical aspect is scalability and batch-to-batch consistency in the synthesis. In most syntheses of Si nanomaterials, it is difficult to avoid a certain level of deviation in particle size and morphology. In fact, the generation of Si microparticles is nontrivial to completely exclude. In this regard, the magnesiothermic or magnesio-milling reduction of rice husk-originating SiO₂ (RH-SiO₂) is advantageous, as rice husks have highly consistent pore structures and are produced in massive quantity worldwide (96 million tons in 2016) [22, 23]. SiO₂ occupies ~20 wt% of rice husk, and mesoporous SiO₂ can be easily obtained by a thermal treatment (>600 °C) of rice husks. Whereas mesoporous Si originating from rice husk SiO₂, namely Si_RH, has been recently demonstrated as an excellent LIB anode material, exclusive use of Si_RH leaves a gap before immediate commercial adoption. Since inevitable volume expansion of Si_RH is still large in such a way that the electrode swelling and charge–discharge reversibility are not as controllable as those of current graphite counterparts. Si_RH also suffers from low electrical conductivity. These limitations suggest Si_RH–graphite composite as a more viable solution for the very next generations of practical cells.

Based on the given motivation, herein, we provide a systematic investigation on Si_RH–graphite composites as practical LIB anodes. By varying the mass ratio between Si_RH and graphite, we endeavored to find an optimal condition and also see the effect of Si_RH in terms of cycle life focusing on Coulombic efficiency (CE). Also, in order to overcome the relatively low initial Coulombic efficiency (ICE) of Si, a controllable pre-lithiation scheme was employed. The present study is expected to serve as a useful ground in developing high capacity LIB anodes integrating Si materials, with great potential towards commercialization from the viewpoints of volumetric energy density and resource scalability.

Si_RH was produced by a magnesio-milling process (Fig. 1a) by considering the scalability of the given process [24, 25]. In this synthetic scheme, Si and MgO were first generated after the magnesio-milling process. MgO was then removed by an acid treatment to yield bare Si_RH. For electrochemical testing based on enhanced electrical conductivity, Si_RH was carbon-coated using a chemical vapor deposition (CVD) process. The carbon-coated Si_RH is denoted as c-Si_RH.

Figure 1b displays the first charge–discharge profiles of four different Si electrodes: Si NP, c-Si NP, Si_RH, and c-Si_RH. Whereas all electrodes showed consistent lithiation and delithiation plateaus at around 0.1 and 0.4 V vs. Li/Li⁺, they delivered different gravimetric capacities. It is first noted that the carbon-coated derivatives exhibited larger specific capacities than those of the bare Si counterparts, indicating that the low conductivity of intrinsic Si is needed to be compensated for efficient Li storage. Also, both types of Si NPs showed clearly higher specific capacities than those of the Si_RH counterparts, which can be explained by the fact that the porous structure of Si_RH does not allow for as much volume expansion as Si NP, resulting in a smaller amount of Li storage in the given voltage range. The inevitable residue of silica in Si_RH is another reason for the smaller specific capacities. Considering the specific capacities of both Si_RH and c-Si_RH electrodes are far higher than those of typical commercial cathodes, the smaller specific capacities of Si_RH and c-Si_RH electrodes compared with those of the Si NP counterparts are unlikely to be a critical issue in practical cell design.

In order to see the commercial feasibility of c-Si_RH, the electrochemical performance of c-Si_RH–graphite composites with c-Si_RH-to-graphite mass ratios from 1:9 to 4:6 were examined. The higher portions of graphite reflect its dominant occupation in current commercial cells. From morphology perspective, graphite showed smoother surfaces with more uniform particle sizes near 15 μm (Fig. 2a). In contrast, Si_RH forms secondary particles from primary particles (Fig. 2b) that correspond to the original rice husk powder.

As expected, the specific capacity increased with increasing c-Si_RH portion (Fig. 2c). The reversible capacity in the first cycle increased from 544.5 to 1173.6 mAh/g, as the composite ratio was varied from 1:9 to 4:6. The increased specific capacities of the composites with higher c-Si_RH portions were reflected in the longer plateaus at 0.1 V (lithiation) and 0.42 V (delithiation) in Fig. 2c and the enhanced peaks in differential capacity (dQ/dV) plots in Fig. 2d. In Fig. 2d, the two small peaks around 0.1 V assigned to the staging phenomenon of graphite decreased with decreasing the portion of graphite. The dQ/dV peaks also unveil the lithiation sequence in such a way that Li ions are first intercalated into graphite (stage IV and III) and then alloyed with Si. The ICE increased with increasing c-Si_RH portion, such as 80.5% of c-Si_RH–graphite (1:9) to 84.9% of c-Si_RH–graphite (4:6). This observation is attributed to the fact that the reactive edges of graphite are involved in the reaction with the electrolyte, although detailed mechanism needs to be unveiled via an additional in-depth study [26–28].

Similar to c-Si_RH alone, the composite electrodes exhibited trade-off relation between specific capacity and cycle life (Fig. 2e). While c-Si_RH–graphite (4:6) started at 1030.6 mAh/g and ended at 657.5 mAh/g after 100 cycles, c-Si_RH–graphite (1:9) started at 494.6 mAh/g and ended at 432.2 mAh/g after the same number of cycles. These capacity changes lead to capacity retentions of 63.8 and 87.4%, respectively. These results remind the challenge of Si_RH, associated with the volume expansion of Si. The more challenging feature of higher portions of c-Si_RH with regard to cycle life was reflected in the CE data over cycling (Fig. 2f). The average CEs of c-Si_RH–graphite (1:9) and c-Si_RH–graphite (4:6) were 99.2 and 98.9%, respectively.

It is known [3, 29] that increasing ICE is critical in practical cell design and the energy density of a full-cell. In an attempt to maximize the ICE, a pre-lithiation scheme was adopted [30, 31]. In particular, simple contact with Li metal foil while adjusting the circuit resistance was employed to control the degree and kinetics of lithiation more easily and accurately. Figure 3a displays the voltage profiles during pre-lithiation. With increasing the c-Si_RH portion, the voltage plateau during pre-lithiation becomes higher in reflection of the higher redox voltage of Si as compared with that of graphite. Since the given pre-lithiation uses the potential difference between both electrodes, a greater portion of graphite leads to lower cell potential and consequently to lower current density (Fig. 3b). This means that for the same pre-lithiation duration, the pre-lithiation of a higher portion of graphite may not be as efficient as that of a smaller portion of graphite. However, in practice, c-Si_RH plays a more dominant role in determining the ICE due to its high capacity, leading to smaller ICEs with increasing the c-Si_RH portions after pre-lithiation (Fig. 3c, d). The ICE decreased from 93.8 to 89.1% as the c-Si_RH–graphite ratio was changed from 1:9 to 4:6. In all composite ratios, the ICE increased conspicuously to around 90%, reconfirming the usefulness of pre-lithiation. Notably, as different c-Si_RH–graphite ratios yield the different levels of pre-lithiation, controllable pre-lithiation scheme is highly desirable in optimizing the cell conditions.

In conclusion, we have tested c-Si_RH alone and c-Si_RH–graphite composites with various compositions as LIB anodes, in order to take advantage of Si_RH in both structural and manufacturing aspects. While c-Si_RH demonstrated more sustainable operation than c-Si NP counterparts for long-term cycling, the trade-off relation between specific capacity and cycle life is to be overcome and optimized. The electrochemical performance of c-Si_RH–graphite composites with various compositions exhibited a similar trade-off phenomenon, impairing the charge–discharge reversibility with increasing the c-Si_RH portion. Further enhancement in the reversibility is an important ‘must-solve’ task for these composites to be integrated into practical cells. Diminishing side reactions by advanced surface coatings and treatments may be a feasible solution along such direction.