Advances in Air-Stable Silicon-Based Anodes and Their Application in Li–Air Batteries

Abstract

In recent years, silicon-based anodes have become a model of commercial success among various high-capacity electrode materials. They also offer a promising substitute for the lithium metal anode (LMA) in lithium–air batteries (LABs), which have the highest specific energy. However, the poor air stability of lithiated silicon-based anodes makes pre-lithiation considerably more difficult and costly in mass production to improve their initial Coulombic efficiency and cyclability, which complicates their material design and electrode manufacturing. To address this issue, intensified efforts have been devoted in recent years, mainly by constructing encapsulation structures, such as core–shell, pomegranate-like or peapod-like architectures. These designs have achieved significantly boosted stability in dry air and, in some cases, even under prolonged exposure to ambient humidity. On the other hand, it was found that silicon-based anodes often provide better cyclic stability than LMAs in LABs and lithium–oxygen batteries (LOBs); however, in most cases, the silicon-based anodes were not optimized for air stability. This review summarizes the relevant works on improving the air stability of silicon-based anodes and LABs/LOBs that used a silicon-based anode, intending to shed light on future development of air-stable silicon-based anodes and bridge the gap between the electrodes’ air-stability and their application in LABs/LOBs.

1. Introduction

The fast development of electric vehicles and portable electronics has stimulated intensive research on next-generation energy storage systems with higher energy density than modern lithium-ion batteries (LIBs). Lithium–air batteries (LABs), with their highest theoretical specific energy (gravimetric energy density) of 11,430 Wh kg⁻¹ (excluding O₂ mass), are regarded as the “holy grail” of energy storage systems because they can extract oxygen from ambient air to the cathode without pre-storage and use the lithium metal anode (LMA), which possesses both the lowest redox potential (−3.04 V vs. standard hydrogen electrode) and highest specific capacity (3861 mAh g⁻¹) among all anode materials [1,2,3,4,5,6,7,8]. When operating in pure oxygen, LABs are simplified into lithium–oxygen batteries (LOBs) at the cost of diminished specific energy (3458 Wh kg⁻¹ in theory, including O₂ mass), and Li₂O₂ is precipitated as the discharge product in the porous air cathode during the oxygen reduction reaction (ORR, discharge reaction), whereas in real air, a large portion of ORR intermediates (O₂⁻/LiO₂) and/or products (Li₂O₂) will react with other air components (H₂O, CO₂, N₂, SO_x, NO_x, CO, etc.) to form side products (LiOH, Li₂CO₃, etc.), complicating the cell chemistry [4,5,6,9,10,11,12].

Despite the appealing ultrahigh energy density, today’s LABs/LOBs still suffer from poor cyclability and rate capability, rendering them inapplicable for commercial use. At the cathode, the solid-state products cause passivation of cathode surface and clogging of gas diffusion channels during discharge and then impede their own decomposition during the following recharge (oxygen evolution reaction, OER), jeopardizing reaction kinetics and reversibility. In addition, the highly reactive oxygen species (ROS, e.g., ¹O₂, O₂⁻, LiO₂, Li₂O₂, etc.) are prone to degrading cell components (cathode substrates, binders, solvents, lithium salts, etc.) [9,10,11,13,14,15,16,17], and these issues drastically obstruct LABs/LOBs’ cyclability and rate capability. To address these issues, most studies in the LAB/LOB field focus on the air cathode side, including insights into the cell chemistry [11,17,18,19,20,21,22,23], stable cell components [24,25,26,27,28,29,30], novel cathode structures [31,32,33,34,35,36,37,38], solid-state catalysts [39,40,41,42,43,44,45,46,47,48], redox mediators (RMs) [49,50,51,52,53,54,55,56,57,58,59], ¹O₂ quenchers [60,61,62,63], etc.

At the anode, two intrinsic issues of LMA have been plaguing its cyclability and rate capability since its first introduction in lithium metal batteries (LMBs): high chemical reactivity and theoretically infinite volume variation [7,64,65,66,67,68,69,70]. Through modern characterization techniques, particularly in situ methods, the failure mechanisms of LMAs have been visually and quantitatively elucidated. For instance, cryogenic transmission electron microscopy (cryo-TEM) [71] has resolved the atomic-scale crystalline structure of Li dendrites and the nanostructure of the solid-electrolyte interface (SEI), while in situ atomic force microscopy (AFM) coupled with environmental TEM [72] has enabled the direct measurement of the growth stress (up to 130 MPa) and yield strength of individual Li whiskers, revealing the formidable mechanical power behind dendrite penetration. Furthermore, stimulated Raman scattering (SRS) microscopy [73] has been employed to map the spatiotemporal ion concentration gradients that drive dendrite initiation. Together, these observations confirm that the growth of lithium dendrites, the formation of “dead lithium” and the repeated rupture and reconstruction of the SEI are the primary culprits behind LMA failure.

In LABs/LOBs, these issues are further compounded by the involvement of air components (N₂, O₂, H₂O, CO₂, etc.) and crossover of soluble species (e.g., RMs, O₂⁻ radical, etc.). These factors exacerbate the instability of the SEI, triggering mutual consumption of electrolyte and active lithium as well as impeded mass transfer, leading to rapid cell failure [74,75,76,77]. Though most LAB/LOB studies used bare lithium foils or those with simple surface treatments (e.g., immersion in propylene carbonate) with huge excess lithium (e.g., hundreds of times higher areal capacity than the cathode), neglecting the anode influence, in recent years, it has been found that the parasitic reactions at the anode not only consume available lithium but also play an important role in increasing charge polarization to deteriorate LABs/LOBs’ round-trip efficiency and cyclability [74,75,76,77,78,79]. To pursue higher specific energy and lower the manufacturing cost, excess lithium should be strictly limited in future practical LAB/LOB devices, which would make the anode challenges even more pressing. Therefore, it is crucial to develop air-stable and high-capacity anodes for developing practical LABs/LOBs.

In the development of LMAs, three major strategies have been implemented to improve their cyclic stability, including: (1) constructing robust artificial SEIs to persistently block the electrolyte while maintaining good contact with metallic lithium underneath, (2) designing 3D host (current collector) structures to mitigate dimension variation of the electrode and reduce local current density and (3) using electrolyte additives to regulate lithium deposition behavior. In recent years, many efforts have also been devoted to improving LMA’s air stability, especially by constructing solid-state electrolyte (SSE) membranes as the artificial SEI to block air and water [74,78,79,80]. However, to the authors’ knowledge, there have been only two examples of commercial success to date, namely the lithium aluminum titanium/germanium phosphate (LATP/LAGP) ceramic disks and their derivatives from various manufacturers (represented by Ohara Inc.) and the proprietary protected lithium electrodes (PLEs) from PolyPlus Battery Company.

In the early years of LABs/LOBs (2010s), those with aqueous and hybrid electrolytes had been a popular topic. In these systems, the discharge product (OH⁻ or H₂O) is soluble as in other aqueous metal–air batteries; thus, the issue of interface instability at the air cathode (as in aprotic LABs) is intrinsically avoided [81,82,83]. To protect the LMA, ceramic SSEs, especially LATP/LAGP and their derivatives (such as Ohara Li_1+x+yAl_x(Ti,Ge)₂Si_yP_3−yO₁₂) [84,85,86], were commonly used in these systems because of their excellent stability with air/water, high room-temperature ionic conductivity (10⁻⁴~10⁻³ S cm⁻¹) and relatively low cost. However, stand-alone LATP/LAGP SSEs make instable point contact with the LMA at the solid–solid interface, resulting in high interfacial impedance. Furthermore, LATP and LAGP are thermodynamically instable against metallic lithium, and their Ti⁴⁺/Ge⁴⁺ ions spontaneously reduce to lower valence states, creating an ionic–electronic mixed conductive interface layer to induce the reaction further inwards, causing dendrite growth through the SSE and stress concentration [87,88], leading to cracks and air/water ingress, as often observed after long-term operation of aqueous and hybrid-electrolyte LABs/LOBs.

To address this issue, adding an artificial interlayer has been regarded as the most effective solution, using inorganic layers such as metal oxides/nitrides/halides (Al₂O₃ [89], ZnO [90], SnO₂ [91], ZnF₂ [92], CuF₂ [88], Li_3-xPO_4−yN_y (LiPON) [93], etc.) and organic layers such as poly(ethylene oxide)/lithium salt [94], poly(ethylene glycol) methyl ether acrylate (PEGMEA) [95], polyvinylidene fluoride-hexafluoropropylene (PVDF-HFP)/lithium bis(trifluoromethanesulfonyl)imide (LiTFSI) [96], and polyvinylidene fluoride-trifluoroethylene (PVDF-TrFE) [97], as well as organic–inorganic composites [98,99]. In light of this, PolyPlus Battery Company developed its proprietary PLEs (Figure 1a) [100,101,102,103]. Their basic strategy is to sputter an ultrathin Li-compatible inner layer (such as Li₃N, Li₃P, LiPON, Li-halides, etc., or their composites with polymers, with a typical thickness of ~0.2 μm) onto the pinhole-free air- and water-stable outer layer (such as LATP, LAGP, etc., with a typical thickness of 20~30 μm) and then laminate the composite membrane with the lithium metal sheet. The inner layer emphasizes ionic conductivity and chemical compatibility with metallic lithium, while the outer layer needs to be highly impervious, ionically conductive and chemically compatible with the inner layer as well as air/water. This strategy led to the development of lithium–seawater primary batteries (essentially LABs utilizing dissolved dioxygen) with their proprietary PLEs [104]. These batteries have demonstrated ultrahigh specific energy of 2000 Wh kg⁻¹ and 69 days of stable underwater operation (down to 400 m) [105]. They also show excellent scalability: a 560 Ah stack was built with gravimetric and volumetric energy densities of 1600 Wh kg⁻¹ and 1050 Wh L⁻¹, respectively [106]. And the pilot production line of PLEs became operational in July 2022. In their design, the thick PLEs for primary batteries have a thickness up to 3.35 mm (650 mAh cm⁻²). For secondary (rechargeable) batteries, the areal capacity shrank to 3~5 mAh cm⁻² (15~25 μm) to reduce volumetric variation. Such areal capacity could match that of the air cathode in some LAB/LOB pouch-cell prototypes, e.g., the one Samsung developed with an areal capacity of 3.55 mAh cm⁻² and ultrahigh specific energy of 1214 Wh kg⁻¹ [34].

However, in these thin PLEs, the areal mass of protective layers exceeds that of metallic lithium, diminishing the core value of the LMA: the ultrahigh specific capacity. For example, in primary batteries, the PLE can provide a near-theoretical specific capacity of 3736 mAh g⁻¹ (calculated based on 3350 μm thick Li, 0.2 μm thick LiPON and 20 μm thick LATP layer), whereas in secondary batteries, the specific capacity plummets to 709 mAh g⁻¹ with 25 μm thick Li (assuming the thicknesses of LiPON and LATP are unchanged). Considering the areal mass of other auxiliary components, such as current collectors, packaging materials and oxygen selective membranes (OSMs) [109] (as illustrated in Figure 1b), higher areal loading and capacity are desirable to further improve the energy density once the oxygen mass transfer issue is addressed. For example, recent advanced air cathodes with aligned microchannels and/or hierarchical pores (Figure 1c) have achieved ultrahigh areal capacities of up to 56 mAh cm⁻² [32,33]. To meet such high areal capacity, lithiophilic 3D hosts are necessary to reduce local current density and dimension variation. However, to date, there is still a huge gap between laboratory-made lightweight (e.g., carbon-based) 3D hosts and industry-level production.

In contrast, silicon-based anode materials, with their specific capacity second only to LMA (3579 mAh g⁻¹_Si or 1858 mAh g⁻¹_Li15Si4 for Li₁₅Si₄), such as Si/C composites and SiO_x, have already been successfully commercialized in LIBs [110]. Unlike intercalation/deintercalation in graphite or plating/stripping on the LMA, in Si-based anodes, Li atoms are alloyed with Si atoms to form various stoichiometries (such as Li₁₂Si₇, Li₇Si₃, Li₁₃Si₄, Li₁₅Si₄, Li₂₁Si₅, etc.), as well as amorphous ones [111]. Currently, they are only used as capacity-enhancing additives (with the content of a few percents) in the graphite substrate; however, with the advances in tackling their volume expansion and low initial Coulombic efficiency (ICE) issues (as discussed below), their content is progressively increasing to become the main active material of the anode [112,113,114]. They not only offer an alternative for LMAs in the midway to the “holy grail”—LMBs (especially LABs)—but can also function as the lithiophilic 3D host in high-areal-capacity LMAs. In these thick electrodes, lithium metal in the bulk is bound to the lattice, and its slow self-diffusion restrains the surface delithiation rate, whereas a small amount of silicon-based anode material can act not only as a lithiophilic deposition host but also as ion transport busways in the LMA to enhance lithium ion conductivity to 10⁻⁴~10⁻³ S/cm [111], help homogenize lithium consumption/deposition through the depth of the LMA and significantly extend the Sand’s time while not significantly decreasing the whole electrode’s specific capacity, as illustrated in Figure 1c (the proposed PLE-like thick LMA with Li_xSi additive).

As widely perceived, the main challenges of silicon-based anodes are their huge volume expansion after lithiation (3.8-fold for Li₁₅Si₄) and low ICE (50~80%) [115]. The former issue is mainly tackled by nanosizing, compositing with buffering substrate (e.g., carbon), building internal voids (e.g., yolk–shell silicon–carbon composites [116] and nanoporous silicon structures [117,118,119]), screening stronger adhesives (e.g., poly(acrylic acid) (PAA) [120,121] and alginate salts [122]) and using SEI-forming electrolyte additives (e.g., fluoroethylene carbonate (FEC) and vinylene carbonate (VC) [123]). To improve the ICE, aside from the abovementioned measures, pre-lithiation is also an important strategy [124] and is regarded as the “last piece of the puzzle” for high-content utilization of silicon in LIB anodes. However, the poor air stability of lithiated silicon-based anodes makes their mass production more difficult and costly. In recent years, increasing efforts have been devoted to enhancing the air stability of silicon-based anodes, but to our best knowledge, there has not been a dedicated review on this topic. On the other hand, several previous works have reported that Li-ion–air (or O₂) batteries with silicon-based anodes (as illustrated in Figure 1c) provide better cyclic stability than LMAs in LABs/LOBs; however, in most cases, the silicon-based anodes were not optimized for air stability. Under these circumstances, we think it is necessary to summarize the advances in improving the air stability of silicon-based anodes and their application in LABs/LOBs to bridge the gap and shed light on the development of ambient-operable LABs, as well as more convenient production of pre-lithiated silicon-based anodes without the need for dry rooms.

2. Advances in Air-Stable Silicon-Based Anodes

When exposed to ambient environments, the self-limiting oxidation reaction would produce a thin (1~3 nm) and dense native oxide (SiO₂) layer on the surface of elemental silicon and SiO_x, preventing their further degradation. However, once lithiated, the formed Li_xSi would readily react with air components, especially water vapor. This high reactivity not only affects the use of silicon-based anodes in LABs/LOBs but, to a broader extent, imposes a big challenge for the production of all pre-lithiated silicon-based anodes, which are desirable to mitigate volume expansion and improve the ICE. In this section, we mainly focus on the long-term air stability of silicon-based anodes, with the prospect to use them in LABs/LOBs; nevertheless, a couple of works focusing on short-term stability were also discussed, which may help eliminate the need of dry rooms for the production of LIBs with pre-lithiated silicon-based anodes.

2.1. Core–Shell Structures

A core–shell is the most straightforward protection structure to design and manufacture. In 2014, Cui’s group [125] reported Li_xSi-Li₂O core–shell nanoparticles (NPs) with improved dry air stability. The Li_xSi NPs (100~200 nm in diameter) were obtained by mechanical stirring of Si NPs (~50 nm in diameter) and molten Li at 200 °C in argon. Due to the trace oxygen (<3 ppm), a dense Li₂O passivation layer (~10 nm thick) was formed on the surface of Li_xSi NPs, as illustrated in Figure 2a. They found that 1,3-dioxolane (DOL) and toluene have good compatibility with Li_xSi-Li₂O NPs for electrode casting, showing a Li extraction capacity of 1200~1400 mAh g⁻¹_Si, while most of the capacity was lost after treatment with the normally used N-methyl-2-pyrrolidinone (NMP) or diethyl carbonate (EC), as shown in Figure 2b. The Li_xSi-Li₂O NPs exhibited almost unchanged capacity (~1600 mAh g⁻¹_Si) in a 20-cycle operation; using it as the pre-lithiation reagent for Si NPs, the ICE improved from 76% to 94%, and the cyclability was also slightly improved (Figure 2c). As shown in Figure 2a, after 3 days of exposure to dry air (dew point −50 °C), the Li₂O layer was thickened, while the X-ray diffraction (XRD) pattern almost unchanged. Although the capacity retention rate gradually decreased as the exposure was prolonged (Figure 2e), the exposure time is sufficient for battery manufacturing. However, when the dew point increased to above −10 °C, the capacity quickly decayed as Li_xSi and Li₂O converted to LiOH, as shown in Figure 2f. The key metrics are summarized and compared in

Table 1. Summary of works on improving the air stability of silicon-based anodes.

Protection Architecture	Preparation Method and Protective Component	Electrolyte Composition	Initial Li Extraction Capacity and ICE	Cyclability	Air Exposure Condition and CRR 1 (of 1st Cycle Unless Specified)	Year and Reference
Core–shell	Oxidation by trace O2 during lithiation with molten Li; Li2O	1M LiPF6 in EC/DEC (1:1, w/w) + 1 v% VC and 10 v% FEC 2	~1600 mAh g−1Si (@C/20); ICE: 94% 3	Almost unchanged ~1600 mAh g−1Si in 20 cycles @C/20	Td 4 −50 °C, 1 day; 91%	2014 [125]
					Td −50 °C, 3 days; 73%
					Td −50 °C, 5 days; 67%
	Reaction of LixSi with aluminum isopropoxide (in cyclohexane); LixAlySiOz/ Li2O	1M LiPF6 in EC/DEC (1:1, v/v) + 2% VC	1005 mAh g−1A-LixSi; ICE: N/A	Decays from 430 to 170 mAh g−1 in graphite/A-LixSi composite after 20 cycles	30 ± 2% RH, 5 1 h; 84%	2022 [126]
					30 ± 2% RH, 3 h; 70%
					30 ± 2% RH, 5 h; 45%
	Reaction of LixSi with 1-fluorodecane (in cyclohexane); LiF and lithium alkyl carbonate	1M LiPF6 in EC/DEC (1:1, w/w)	2078 mAh g−1Si; ICE: 96.8% 3	Undiminished ~1500 mAh g−1Si in 70 cycles @C/2 after first 3 cycles @C/20 (~2200 mAh g−1Si)	Td −50 °C, 5 days; 92%	2015 [127]
					10% RH, 6 h; 77%
					20% RH, 6 h; 36%
					20% RH, 6 h; 12%
	Heating CYTOP polymer to generate F2 gas to react with LixSi; LiF	1M LiPF6 in EC/DEC (1:1)	2504 mAh g−1Si in NMP, 2879 mAh g−1Si in DOL (@C/20); ICE: ~95%	~1300 mAh g−1Si after 650 cycles @C/2 with 87% CRR and 99.92% average Coulombic efficiency (CE)	40% RH, 6 h; ≥99%	2017 [128]
					40% RH, 24 h; 85.9%
	Coating with resorcinol–formaldehyde resin-derived carbon, followed by electrochemical pre-lithiation; carbon and Li2O	1M LiPF6 in EC/DEC (1:1, v/v) + 5 v% FEC for half-cell and 1M LiPF6 in EC/DMC 6/DEC (1:1:1, v/v) for ASP-Hp-SiOx@C-LFP full cell	1179 mAh g−1 (@1.0 A g−1); ICE: 99.2%	1059 mAh g−1 after 1000 cycles @1.0 A g−1 with 89.8% CRR and 550 mAh g−1 after 1000 cycles @10 A g−1 with 66.2% CRR	10~20% RH, 48 h; N/A (all battery tests were conducted after air exposure)	2022 [129]
	Coating with amorphous TiO2, followed by pre-lithiation with molten lithium; LixTiO2 and Li2O	1M LiPF6 in EC/DEC/VC/FEC (44.5:44.5:1:10, v/v)	2326 mAh g−1Si (@C/20); ICE: 238%	2000 mAh g−1Si after 40 cycles @C/20 with negligible capacity fading and ~1300 mAh g−1Si after 500 cycles @C/2 with 77% CRR	10% RH, 30 days; 87%	2018 [130]
Pomegranate-like	Mixing SiO or SiO2 NPs with molten lithium; Li2O	1M LiPF6 in EC/DEC (1:1, w/w)	2059 mAh g−1SiO2 (equivalent to 3236 mAh g−1Si) @C/50; ICE: ~95.5%	Undiminished ~961 mAh g−1SiO after 400 cycles @C/2, with 99.87% average CE from 200th to 400th cycle	Td −50 °C, 5 days; 91%	2016 [131]
					10% RH, 6 h; ~91%
					40% RH, 6 h; 60%
Peapod-like	Mixing LixSi NPs with graphene sheets and SBS 7 binder in toluene, followed by pressing under 40 MPa; few-layer graphene	1M LiPF6 in EC/DEC/VC/FEC (44.5:44.5:1:10, v/v)	1600 mAh g−1LixSi; ICE: 98.8%	98% CRR after 400 cycles @1 mA cm−2 and ~2.4 mAh cm−2 with 99.92% average CE	Td −50 °C, 14 days; 94.3%	2017 [132]
					20~60% RH, 3 days; ~80%
Film	Drop casting PMMA 8 (in DOL) onto the lithium-plated Cu foil; PMMA	1 M LiPF6 in EC/DEC	2961 mAh g−1; ICE: 116%	1340 mAh g−1 after 100 cycles with 58% CRR after air exposure, similar to the unexposed (1456 mAh g−1 and 63% CRR)	10% RH, 0.5 h; 92% (remaining capacity after 100 cycles and air exposure divided by that without exposure)	2016 [133]
Silicon–graphite hybrid	Spray drying a mixture of a lithium-enriched graphite, Si NPs and pitch in tetrahydrofuran followed by calcination; a novel O-Li-Si bonding composite	Commercial electrolyte for silicon–graphite	502 mAh g−1; ICE: 116%	~520 mAh g−1 after 100 cycles	~71% RH, 7 days; almost unaffected initial specific capacity (502 mAh g−1) and ICE (113.6%)	2024 [134]

¹ CRR: capacity retention rate. ² EC: ethylene carbonate; DEC: diethyl carbonate; VC: vinylene carbonate; FEC: fluoroethylene carbonate. ³ Si NPs/protected Li_xSi composite. ⁴ Dew point. ⁵ Relative humidity. ⁶ Dimethyl carbonate. ⁷ Poly(styrene-butadiene-styrene). ⁸ Poly(methyl methacrylate).

.

In 2022, Li’s group [126] used aluminum isopropoxide (dissolved in cyclohexane) to react with Li_xSi to construct a thin but dense Li_xAl_ySiO_z/Li₂O protective layer on the surface of Li_xSi NPs. The protected Li_xSi (denoted by A-Li_xSi) displayed a Li extraction capacity of 1005 mAh g⁻¹, slightly lower than the bare Li_xSi (1137 mAh g⁻¹) due to the surface passivation consumption. However, the remaining capacities of A-Li_xSi after 1, 3 or 5 h exposure to 30 ± 2% RH air were 840, 705 and 454 mAh g⁻¹, respectively, considerably higher than the bare Li_xSi, which only retained 326 and 189 mAh g⁻¹ after 3 and 5 h exposure, respectively.

Since Li₂O can readily react with H₂O, these protection structures are still intrinsically susceptible to humid air. In contrast, LiF is a highly stable SEI component in various environments and can be intentionally utilized to construct air- and water-stable protection structures. In light of that, in 2015, Cui’s group [127] developed artificial-SEI-protected Li_xSi NPs with 1-fluorodecane as the precursor. The prepared Li_xSi NPs were added to 1-fluorodecane dissolved in anhydrous cyclohexane and reacted for 2 h, as illustrated in Figure 3a. 1-fluorodecane was chosen for its excellent processability in nonpolar solvents (e.g., cyclohexane) so the capacity loss of Li_xSi could be eliminated. The dissolved 1-fluorodecane was reduced on the surface of Li_xSi NPs, generating a conformal, continuous and dense coating (as shown in Figure 3b,c), which consists of LiF and lithium alkyl carbonate with long hydrophobic carbon chains to mitigate susceptibility to air and water. The extraction capacity of coated Li_xSi was 2078 mAh g⁻¹_Si, 10% lower than the bare Li_xSi, indicating ~10% of active lithium consumed in the coating formation. After 5 days in dry air (dew point −50 °C), the coated Li_xSi remained 92% of the initial capacity (as shown in Figure 3d), considerably higher than bare Li_xSi (67%) and their previously reported Li_xSi-Li₂O NPs [125]. However, as the relative humidity (RH) further rose to 10%, 20% and 40%, the extraction capacity decreased to 1604, ~740 and ~250 mAh g⁻¹_Si after 6 h exposure, respectively, as shown in Figure 3e.

In 2017, Cui’s group [128] also used a commercial fluorinated thermoplastic polymer CYTOP (developed by AGC Chemicals; its structure is shown in Figure 4a) as the precursor to generate fluorine gas at 350 °C and in situ form a uniform and continuous LiF coating layer on the Li_xSi surface, as illustrated in Figure 4a. This dense and crystalline LiF coating (shown in Figure 4b) effectively isolates the reactive Li_xSi from air and slurry dispersants. Unlike the bare Li_xSi, which suffers from great capacity loss when the slurry dispersant shifts from DOL to NMP, the LiF-coated Li_xSi retained an extraction capacity of 2504 mAh g⁻¹_Si using NMP, close to that with DOL (2879 mAh g⁻¹_Si), as shown in Figure 4c. After 1 day of exposure to 40% RH air, the LiF-coated Li_xSi retained 85.9% of its initial capacity, much higher than the Li₂O-Li_xSi NPs, which only remained at 20% of their initial capacity after 6 h exposure (shown in Figure 4d). The protective LiF coating also improved the average Coulombic efficiency (CE) from 99.79% to 99.92%, with a capacity retention rate of 87% after 650 cycles (Figure 4e).

2.2. Pomegranate-like and Peapod-like Compartmented Protection Structures

Despite the fruitful progress in core–shell-structured protected Si-based anodes, they are not fail-safe designs in nature; because it is difficult to realize perfect encapsulation, any pinhole will compromise the whole “defense line”, providing a pathway for moisture to react with inner Li_xSi. In contrast, compartmented designs can theoretically provide better protection; even if some compartments were compromised by pinholes, other independent compartments are still protected. In 2016, Cui’s group [131] developed a Li_xSi/Li₂O composite with excellent ambient air compatibility through a one-pot metallurgical process, using SiO or SiO₂ as the starting material to alloy thermally with molten Li metal, as illustrated in Figure 5a. The synthesized Li_xSi/Li₂O composite had a pomegranate-like structure (Figure 5a,b); it exhibited an initial Li extraction capacity of 2059 mAh g⁻¹_SiO2 (equivalent to 3236 mAh g⁻¹_Si) at C/50 and lost only 9% of initial capacity after 5 days of exposure to dry air (dew point −50 °C), much lower than core–shell-structured Li_xSi-Li₂O NPs (30%), as shown in Figure 5c. Furthermore, the pomegranate-structured Li_xSi/Li₂O composite maintained a specific capacity of 1240 mAh g⁻¹_SiO2 after 6 h exposure to humid air with 40% RH (Figure 5d,e). It also exhibited excellent cyclability, achieving an average CE of 99.87% from the 200th to 400th cycle, as shown in Figure 5f. The improved air stability was attributed not only to the highly dense and crystalline protective Li₂O matrix formed at high temperature (250 °C) and prolonged reaction time (5 days) but also to the compartmented Li_xSi nanodomains, such that each Li_xSi nanodomain has localized Li₂O protection.

In 2017, Cui’s group [132] reported a freestanding peapod-like Li_xM/graphene foil (M = Si, Al, Sn and other materials that can form alloys with Li), which is composed of few-layer (<10 layers) graphene sheets and impregnated Li_xM NPs. Compared to Li₂O in their previous Li_xSi/Li₂O composite [131], the graphene sheets not only have better chemical stability against humidity to protect the impregnated Li_xM NPs but also provide higher electronic conductivity, achieving excellent air/water and cycling stability. The Li_xM alloy NPs were prepared by mechanically stirring a stoichiometric mixture of vacuum-dried M NPs and molten Li in argon at a temperature between the melting points of M and Li. The Li_xM alloy was then mixed with graphene sheets and poly(styrene-butadiene-styrene) (SBS) binder in toluene, and the mixture was cast onto a polyethylene terephthalate (PET) release film and left to dry. Then, thin layers of graphene sheets were coated on the double sides of the foil to protect the Li_xM NPs on the outer surface, forming the freestanding Li_xM/graphene foil. Finally, the Li_xM/graphene foil was pressed under 40 MPa. Since the Li_xM alloys were already in their fully expanded state, no extra void space was needed to accommodate the volume expansion. This pressing step not only increased the tap density to improve the volumetric capacity to 1800~2000 mAh cm⁻³ (corresponding to a specific capacity of 1600 mAh g⁻¹_LixSi), close to the theoretical value of Li metal (2061 mAh cm⁻³), but also eliminated the void space that allows air/water/electrolyte ingress. With a foil thickness of 19 μm, the Li_xSi/graphene exhibited an initial delithiation areal capacity of 3.8 mAh cm⁻². Because the Li_xSi NPs are confined in the graphene matrix, the volume variation is buffered, and dendrite growth is mitigated effectively, enabling stable operation of 400 cycles with 98% capacity retention rate at an areal capacity of ~2.4 mAh cm⁻² and current density of 1 mA cm⁻² in a half-cell, corresponding to an excellent average CE of 99.92% after the initial two cycles. Furthermore, the Li_xSi/graphene foil exhibited excellent air/water stability, retaining 94.3% initial capacity after 2 weeks storage in dry air (dew point −50 °C) and ~80% after 3 days in ambient air (with 20~60% RH). Importantly, this preparation method makes the foil thickness highly tunable from 12 to 42 μm (corresponding to initial delithiation capacities of 2.0~8.3 mAh cm⁻²), as demonstrated in their work.

2.3. Protection Before Lithiation

As mentioned at the beginning of this section, when not lithiated, Si and SiO_x are quite stable in ambient conditions. If lithiation could be carried out after building the protection structure rather than before, it would be much more convenient for production. In 2022, Wang et al. [129] reported a hollow porous SiO_x@C sphere (Hp-SiO_x@C) with excellent cyclability and air stability. The SiO_x spheres were coated with resorcinol–formaldehyde (RF) resin, then incubated in hot water to form the hollow structure and finally calcinated to obtain Hp-SiO_x@C. It exhibited an initial specific capacity of 1475.6 mAh g⁻¹ at 0.1 A g⁻¹ with 72.2% ICE. After activation at 0.1 A g⁻¹, Hp-SiO_x@C displayed a specific capacity of 1126.9 mAh g⁻¹ at 1.0 A g⁻¹ and a high-capacity retention rate of 91.2% after 1000 cycles, much better than the solid SiO_x@C (42.4% capacity retention rate after 700 cycles). To improve the ICE and air stability, Hp-SiO_x@C was electrochemically pre-lithiated in electrolyte and subsequently heated to 200 °C to evaporate electrolyte and form a robust Li₂O-rich passivation layer on the outer surface, resulting in the air-stable pre-lithiated Hp-SiO_x@C (ASP-Hp-SiO_x@C). Even after 48 h of exposure to 10~20% RH air, it showed an initial specific capacity of 1179 mAh g⁻¹ with a very high ICE of 99.2% and a capacity retention rate of 89.8% after 1000 cycles at 1.0 A g⁻¹.

In the above work, pre-lithiation was performed electrochemically by assembling a Si||Li half-cell, as in numerous previous works, which is cumbersome with the need for cell assembling and disassembling, as well as removing residual pre-lithiation electrolytes. In contrast, both metallurgical and chemical lithiation are more desirable for industrial productions. In 2018, Lu’s group [130] developed a thermal lithiated TiO₂ as a robust and electron-conducting protection coating for Li_xSi. As illustrated in Figure 6a, the amorphous TiO₂ layer (5~15 nm thick) was first coated on the Si NPs with a facile sol–gel process, and then the Si@TiO₂ NPs (with ~75 wt% Si) were alloyed with molten lithium to obtain the Li_xSi-Li₂O/Ti_yO_z product. After lithiation, the surface amorphous TiO₂ converted to Li_xTiO₂, which is stable against air and water (as shown in Figure 6d) and has a greatly improved electron conductivity of ~1 × 10⁻⁶ S cm⁻¹. As shown in Figure 6e, the Li_xSi-Li₂O/Ti_yO_z composite exhibited an initial Li extraction capacity of 2326 mAh g⁻¹_Si at C/20 (1C = 4.2 A g⁻¹_Si) and maintained 2000 mAh g⁻¹_Si after 40 cycles. At a higher rate of C/2, the capacity was maintained at 1692 mAh g⁻¹_Si at the fourth cycle with a 99.4% CE and ~1300 mAh g⁻¹_Si after 500 cycles. As shown in Figure 6g, after 30 days of exposure to 10% RH air, the stored capacity that preloaded in the alloy (determined by subtracting the first lithiation capacity from the delithiation capacity) only decayed by 13% from 1560 mAh g⁻¹_Si to 1357 mAh g⁻¹_Si. In comparison, the bare Li_xSi NPs experienced a 79% decay (Figure 6h). In this work, the protective layer (amorphous TiO₂) was constructed prior to alloying; thus, it can be carried out in an ambient environment, and after alloying with molten lithium (the only step requiring inert atmosphere in the crucible), the lithiated TiO₂ is already in place to protect Li_xSi underneath from air and water, making the production much more convenient.

Compared to metallurgical pre-lithiation, which requires high-temperature treatment in an inert atmosphere, chemical pre-lithiation in ambient conditions is more desirable. To date, lithium foil and stabilized lithium metal powder (SLMP) are still the two major anode pre-lithiation reagents. SLMP was developed by FMC Corporation in the 2000s. It is surface passivated by a Li₂CO₃ coating (<3 wt%), providing significantly improved stability in dry air than regular lithium foil; however, it is still incompatible with aqueous slurry, which is usually used for anode production; furthermore, this proprietary product costs 1~2 times more than regular lithium foil. In 2016, Yang’s group [133] developed an anode pre-lithiation method with lithium foil that was protected by a soluble polymer coating. In their work, metallic lithium was electrochemically deposited onto the copper current collector, and then poly(methyl methacrylate) (PMMA) was dissolved in DOL and drop cast onto the lithium-plated copper foil. After evaporation, the uniform PMMA protective coating was formed. Then, this PMMA-coated lithium–copper film can be used as substrate to prepare artificial graphite (AG) or silicon electrode sheets, as illustrated in Figure 7a. The PMMA-coated lithium–copper film showed good compatibility with humid air (up to 10% RH, as shown in Figure 7b,c) and even aqueous slurry, with the metallic luster unchanged for 5 min (with 20 μm thick PMMA) or over 30 min (with 100 μm thick PMMA) in contact with water. This good protecting effect against air/water provides sufficient time for battery manufacturing. Nevertheless, in their battery tests, ethanol was used as the slurry dispersant for AG/silicon casting to minimize the loss of available lithium in the microcracks. After battery assembly, the PMMA coating was dissolved in electrolyte, and the metallic lithium underneath was exposed to contact AG/silicon to accomplish pre-lithiation even without pressure. With 30 min of exposure to 10% RH air, the PMMA-protected Si electrode retained a capacity of 1340 mAh g⁻¹ (58% retention rate) after 100 cycles, similar to that without air exposure (1456 mAh g⁻¹), considerably higher than the unprotected one (809 mAh g⁻¹), as shown in Figure 7d. However, when the exposure time exceeded 1 h in 10% RH air or when the RH reached 30%, the improvement in electrochemical performance quickly vanished, unlike that with the AG/PMMA/Li electrode, possibly attributed to silicon’s more hydrophilic property and higher surface area than graphite, so the moisture penetrates more easily. Although this work showed good short-term air stability of the PMMA-protected Li/Si composite anode, it cannot be directly transplanted to LABs/LOBs because of the dissolution of PMMA coating.

In 2020, Qu’s group [135] developed an air-stable chemical lithiation compound, Li−9,9-dimethyl-9H-fluorene-tetrahydrofuran (Li-Flr-THF). It was facilely prepared by dissolving 1 M 9,9-dimethyl-9H-fluorene (Flr) in tetrahydrofuran (THF) and then adding excess Li metal in the solution and reacted for 2 h. Although the prepared pre-lithiation reagents are susceptible to the moisture in the air, they can remain in the dried ambient air for several days without degrading. The lithiation was carried out by simply immersing the pristine graphene-coated SiO (SiO/G) electrode (SiO:graphene = 45:55, w/w) for a predetermined duration, followed by washing with THF and drying. Flr exhibited a lower (0.18 V vs. Li/Li⁺) redox potential compared to other lithium radical anions of polycyclic aromatic hydrocarbons (PAHs), such as biphenylide (Bp, 0.32 V) and naphthalenide (Nap, 0.25 V); this led to the formation of a relatively thin and complete SEI layer with Flr. In contrast, the more positive redox potentials of Nap and Bp not only took longer time (>1 h vs. ~10 min) to form SEI but also resulted in incomplete SEI coverage. Consequently, in half-cells, the SiO/G electrode with Flr treatment exhibited an initial delithiation capacity of 130 mAh g⁻¹, much higher than those with Bp (40 mAh g⁻¹) or Nap (10 mAh g⁻¹). The cell with Flr treatment also presented higher ICE (87.1%) than that with Bp (85.8%) or Nap (82.9%), confirming that Li-Flr-THF was the optimal pre-lithiation reagent. In SiO/G||NCM622 full-cell tests, the Flr-treated sample gave a much higher ICE (87.1%) than the untreated one (61.1%).

Similarly, in the same year, Lee’s group [136] developed a lithium–arene complex (LAC) as the chemical pre-lithiation reagents for SiO_x (x ≈ 0.5). The LAC was facilely prepared by dissolving lithium metal into methyl-substituted biphenyls (e.g., 4,4′-dimethylbiphenyl, abbreviated as 4,4′-DMBP, shown in Figure 8a). When in contact with SiO_x, the LAC (e.g., Li-4,4′-DMBP) was immediately oxidized because of its lower redox potential (~0.2 V vs. Li/Li⁺) than Si (~0.3 V vs. Li/Li⁺), donating lithium ions to SiO_x to complete the pre-lithiation and formation of the SEI. The pre-lithiated SiO_x exhibited similar extraction capacity around 1600 mAh g⁻¹ before and after 1 h of exposure to dry air, as shown in Figure 8b, much higher than that with Li-Flr-THF treatment (130 mAh g⁻¹) [135]. However, after exposure to 35~40% RH air, the extraction capacity plummeted to ~300 mAh g⁻¹.

2.4. Silicon–Graphite Hybrid Electrode

In today’s commercial LIBs, Si/C and SiO_x are usually composited with other anode materials (graphite, soft/hard carbon, etc.) to buffer Si’s volume variation and improve electronic conductivity. Although they compete in the allocation of Li⁺ and electric current, they work independently from a particle-level viewpoint. However, in 2024, Zhang’s group [134] reported a lithium-enriched silicon/graphite (LESG) anode with unexpected excellent air and water stability. It was prepared by spray drying a mixture of a lithium-enriched graphite, nano-silicon powders and pitch in THF and subsequent calcination. The prepared LESG (with 12.44 wt% Si) presented a specific capacity of 620 mAh g⁻¹ and excellent stability in air and water. It can be immersed in water without detectable H₂ emission using gas chromatography. Furthermore, after 1 week of exposure to humid air (~71% RH), LESG displayed an initial discharge specific capacity of 505 mAh g⁻¹, almost the same as that without air exposure, and the cell with exposed LESG also showed negligible capacity loss after 50 cycles. Through molecular dynamics simulation, they found the formation of a novel O-Li-Si structure, and the dissociation of all Li-adjacent bonds in LESG requires a much higher energy of 4.72 eV than that in Li_xSi (3.38 eV for equidistant site and 3.03 eV for top site), which may be the reason for LESG’s higher air and water stability.

In summary, there have been several silicon-based anode materials displaying excellent air stability for prolonged exposure (1~30 days) to ambient air (10~60% RH) along with high specific capacity, such as the core–shell-structured Li_xSi@LiF with CYTOP-derived LiF-based SEI [128] (2879 mAh g⁻¹_Si, 85.9% capacity retention rate after 1 day of exposure to 40% RH air), pomegranate- or peapod-structured Li_xSi/graphene composite foil [132] (1600 mAh g⁻¹_LixSi, ~80% capacity retention rate after 3 days of exposure to 20~60% RH air) and core–shell-structured Li_xSi-Li₂O/Ti_yO_z composite [130] (1560 mAh g⁻¹_Si, 87% capacity retention rate after 30 days of exposure to 10% RH air). Furthermore, due to the high-quality (artificial) SEI formed with the protective components and structures, in some works, long cyclability could be achieved without fluorine-containing electrolyte additives (such as VC and FEC [127,128,131,133], as listed in Table 1). These are promising anode candidates for Li-ion–air batteries.

3. Lithium-Ion–Air (Or O₂) Batteries with Silicon-Based Anodes

In 2012, a joint study by the Scrosati, Sun and Amine groups [137] reported the first effort to replace LMAs with a silicon-based anode in LABs/LOBs. They synthesized a spherical nano-Si-C composite by heating a mixture of nano-Si, graphite and petroleum pitch; then, the obtained Si-C composite (with a Si-C mass ratio of 3:7) was lithiated by direct contact with a lithium foil in an electrolyte solution to form Li_2.6Si, which displayed a specific capacity of 650~780 mAh g⁻¹_Si-C in the first 20 cycles of a half-cell with tetraethylene glycol dimethyl ether (TEGDME) as the electrolyte solvent. Ex situ XRD patterns of the cathode and anode after discharge and recharge of the Li-ion–O₂ full cell revealed the formation and disappearance of Li₂O₂ at the cathode, as well as the conversion between Si and Li_xSi. The Li-ion–O₂ full cell stably operated 15 cycles under a fixed capacity of 1000 mAh g⁻¹_carbon, even with a small N/P ratio (2.34), demonstrating the feasibility of this concept. The corresponding data are listed in

Table 2. Summary of LABs/LOBs with silicon-based anodes.

Anode Material and Loading	Cathode Material and Loading	Other Components, Treatments and Conditions	Anode’s Initial Li Extraction Capacity, ICE and N/P Ratio	Cyclability	Cyclability of Control	Year and Reference
Electrochemically pre-lithiated homemade Si-C composite (Si:C = 3:7, w/w); binder: CMC-SBR 1 loading: 3 mg cm−2	Super P; loading: 1.0 ± 0.1 mgcarbon cm−2	Electrolyte: LiCF3SO3-TEGDME (1:4, molar)	780 mAh g−1Si-C @100 mA g−1Si-C; ICE: ~98.3%; N/P ratio 2: 2.34	Stable ~680 mAh g−1Si-C in 20 cycles in half-cell	N/A	2012 [137]
				15 cycles @200 mA g−1carbon and 1000 mAh g−1carbon in LOB	N/A
Electrochemically pre-lithiated commercial Si NPs; binder: PAA 3; loading: 0.65 or 1.0 mg cm−2	Ru/KB (1:4, w/w); loading: 0.5 mg cm−2	Half-cell electrolyte: 1 M LiTFSI-FEC/TEGDME (1:4, v/v)	2083 mAh g−1Si @50 mA g−1Si; ICE: 87%; N/P ratio 2: 3.9	1550 mAh g−1Si after 50 cycles @100 mA g−1 with 78% CRR in half-cell	900 mAh g−1Si after 50 cycles with 45% CRR in FEC-free half-cell	2016 [138]
				~1850 mAh g−1Si after 50 cycles @100 mA g−1 with ~68% CRR in half-cell in O2 atmosphere	~290 mAh g−1Si after 50 cycles @100 mA g−1 with ~11% CRR in FEC-free half-cell in O2 atmosphere
		LOB electrolyte: 50 μL 0.2 M LiTFSI + 0.8 M LiNO3 in TEGDME		100 cycles @500 mA g−1Ru/KB and 1000 mAh g−1Ru/KB in LOB	20 cycles @500 mA g−1Ru/KB and 1000 mAh g−1Ru/KB in LOB without prior FEC treatment
			N/P ratio 2: 2.5	60 cycles @500 mA g−1Ru/KB and 1000 mAh g−1Ru/KB in LOB	N/A
Commercial Si NPs; binder: PAA; loading: 3 mg cm−2	Li3N–super P–CNT composite (90:5:5, w/w); loading: 8 mgLi3N cm−2	50 μL 1 M LiTFSI in TEGDME and 2% FEC (v/v)	~3000 mAh g−1Si @400 mA g−1Si; ICE: N/A; N/P ratio 4: 23.2	~2480 mAh g−1Si after 50 cycles @400 mA g−1Si with ~83% CRR in half-cell	~300 mAh g−1Si after 50 cycles @400 mA g−1Si with ~12% CRR in half-cell without FEC	2016 [139]
				129 cycles @250 mA g−1carbon and 500 mAh g−1carbon in LOB (anode capacity ~11 mAh)	65 cycles @250 mA g−1carbon and 500 mAh g−1carbon in LOB with Li foil anode (~14 mAh)
				40 cycles @200 mA g−1carbon and 800 mAh g−1carbon in LOB	N/A
Electrochemically pre-lithiated homemade Si-C composite; binder: CMC; loading: ~1.13 mg cm−2	KB; loading: 0.156 mg cm−2	140 μL 1 M LiTFSI in TEGDME and 2.5 mM bi-CoPc	1541 mAh g−1SCC @100 mA g−1SCC; ICE: 88.0%; N/P ratio 2: 48.6	944 mAh g−1SCC after 100 cycles @100 mA g−1SCC with 54% CRR in half-cell	N/A	2025 [140]
				86 cycles @100 mA g−1KB and 300 mAh g−1KB in LAB (anode capacity 3.5 mAh, artificial air N2-O2, 78:22, v/v)	60 cycles @100 mA g−1KB and 300 mAh g−1KB in LAB with Li foil anode (capacity 83 mAh)
					43 cycles @100 mA g−1KB and 300 mAh g−1KB in LAB with Cu@Li anode (capacity 24 mAh)
Electrochemically pre-lithiated homemade Si–CNT hybrid fiber (83.84 wt% Si); binder: none; loading: 0.023 mg cm−1 (longitudinal)	Aligned CNT sheet; loading: 0.02 mg cm−1 (longitudinal)	UV-cured GPE 5 based on LiTFSI, LiNO3, TEGDME and other components	1250 mAh g−1Si/CNT @0.1 mA; ICE: ~96.7%; N/P ratio 2: 2.88	~84% CRR after 100 cycles @0.1 mA in half-cell	N/A	2017 [141]
				100 cycles @500 mAh g−1cathode in LAB (5% RH air)	N/A
Chemically pre-lithiated, freestanding and flexible Si/C film (51 wt% Si) by electrospinning a dispersion containing PAN 6 and Si NPs and subsequent calcination; binder: none; loading: 5.4~6.1 mg cm−2	MWCNTs 7; loading: ~0.3 mg cm−2	GPE with 1M LiFSI 8 in TEGDME and 50 mM LiI	1300 mAh g−1Si-C @100 mA g−1; N/P ratio: 26.7~30	140 cycles @500 mA g−1MWCNT and 1000 mAh g−1MWCNT in LOB	80 cycles @500 mA g−1MWCNT and 1000 mAh g−1MWCNT with GPE and LMA in LOB	2019 [142]
					90 cycles @500 mA g−1MWCNT and 1000 mAh g−1MWCNT with liquid electrolyte and Si/C film anode in LOB
					44 cycles @500 mA g−1MWCNT and 1000 mAh g−1MWCNT with liquid electrolyte and LMA in LOB

¹ Carboxymethyl cellulose and styrene butadiene rubber. ² Based on the given fixed specific capacity of the cathode. ³ Polyacrylic acid. ⁴ Based on the first charge capacity for the anode and the given fixed specific capacity for the cathode ⁵ Gel polymer electrolyte. ⁶ Polyacrylonitrile. ⁷ Multiwall carbon nanotubes. ⁸ Lithium bis(fluorosulfonyl)imide.

. They also noted a small but progressive drop of both the discharge and charge voltage, which could be attributed to gradual degradation of Li_xSi due to oxygen attack. Since this foundational work, several other Li-ion–air (or O₂) batteries with Si-based anodes have also been reported [84,138,139,140,141,142,143,144], mainly focusing on the electrolyte formulation to improve SEI quality and combination with (quasi-)SSE membranes to suppress the crossover of air and electrolyte components, as discussed below.

3.1. Electrolyte Formulation for Improving SEI Quality

In 2016, Zhou’s group [138] introduced fluoroethylene carbonate (FEC) into the electrolyte to construct a robust SEI on the surface of commercial Si NPs during electrochemical pre-lithiation in a half-cell. The addition of FEC in the TEGDME-based electrolyte did not affect the initial Li extraction capacity (2090 mAh g⁻¹_Si without FEC and 2083 mAh g⁻¹_Si with FEC) but significantly improved the capacity retention rate from ~45% to ~78% after 50 cycles at 100 mA g⁻¹_Si, as shown in Figure 9a,d. Then, they used the pre-lithiated Si anodes in Li-ion–O₂ full cells with an N/P ratio of 3.9 to test their cyclability with a fixed-capacity limit of 1000 mAh g⁻¹_Ru/KB at 500 mA g⁻¹_Ru/KB, where Ru-loaded Ketjenblack (Ru/KB) was used as the cathode substrate. As shown in Figure 9b,c,e,f, the cell with the FEC-treated Si anode sustained 100 cycles without capacity fading, significantly better than that without FEC, which started to show capacity fading within 20 cycles. Even when the N/P ratio was reduced to 2.5, the cell with FEC-treated Si still showed a good lifespan of 60 cycles. Furthermore, ex situ XRD tests revealed that after first recharge in the Li-ion–O₂ full cell, the diffraction peaks corresponding to Li_xSi were similar to the pristine electrode with the FEC-treated Si anode, whereas the signal was much weaker after recharge for the cell with untreated Si. These results confirm the critical role of anode protection for the longevity of Li-ion–O₂ batteries with Si-based anodes. They also conducted ex situ X-ray photoelectron spectroscopy (XPS) tests for the pre-lithiated Si anodes, revealing LiF and Li₂CO₃ as the main components of the SEI for the untreated Si anode; in contrast, the FEC-treated Si anode possibly contains a new polyfluorocarbon component, though some LiF and Li₂CO₃ were also present. The stabilization effect of FEC treatment was further evidenced by the greatly improved cyclability of Si-Li half-cells in an O₂ atmosphere, where the cell with FEC retained ~78% of initial capacity after 50 cycles and that without FEC retained only ~11%. In addition, the cell with FEC-treated Si maintained a stable open-circuit voltage (OCV) around 2.42 V and stable impedance during 64 days of storage in O₂; in contrast, the cell with untreated Si experienced a continuous OCV drop (from ~2.36 V to ~2.07 V) and impedance increase. These results demonstrate the significant effect of FEC treatment in improving the SEI robustness to resist oxygen permeation and SEI rupture by the volumetric variation of Si.

In the above work, the electrochemical pre-lithiation requires the assembly of a dedicated Si||Li half-cell with SEI-forming additives, followed by disassembly of the half-cell to retrieve the pre-lithiated Si anode and removal of residual SEI-forming additives to prevent undesirable parasitic reactions on the surface of air cathode. In 2016, Yang, Wang and co-workers [139] developed an in situ lithiation method with Li₃N preloaded in the air cathode (Li₃N–super P–CNT composite, with a mass ratio of 90:5:5, as illustrated in Figure 9g), eliminating the need for a dedicated pre-lithiation half-cell. They compared the cyclability performance of commercial Si NPs with and without FEC in TEGDME-based electrolyte, as well as that with FEC in carbonate electrolyte, finding that the addition of 2% in TEGDME-based electrolyte is optimal for a Li-ion–O₂ full cell. As shown in Figure 9h, on the one hand, it provides similar anode specific capacity (~3000 mAh g⁻¹_Si) and cyclability performance (~83% capacity retention rate after 50 cycles in half-cell) to the carbonate electrolyte with 10% FEC (which cannot be used in LABs/LOBs due to carbonate degradation by superoxides); on the other hand, the 2% FEC can be completely consumed in 2 cycles to prevent its further parasitic reactions at the air cathode. Ex situ XRD tests after the first charge (~10.3 mAh with a cutoff voltage of 3.6 V to ensure complete decomposition of Li₃N) indicated the disappearance of Li₃N and Si, along with the formation of Li_xSi, demonstrating the feasibility of this in situ pre-lithiation method. In the full-cell cyclability tests with Li₃N-preloaded cathode at a fixed capacity limit (500 mAh g⁻¹_carbon, equivalent to 0.44 mAh cm⁻²), the cell with Si anode and preloaded Li₃N survived 129 cycles, much longer than that with LMA (65 cycles), as shown in Figure 9i,j, though the active Li reserve in the in situ-formed Li_xSi was smaller (11 mAh) than the LMA (14 mAh).

3.2. Suppressing Crossover of Air and Electrolyte Components

As mentioned at the beginning of this review, in recent years, it has been recognized that the crossover of air components and other soluble species in the electrolyte has critical impacts on the LMA’s polarization and cyclability; however, in 2025, Yan et al. [140] found that silicon-based anodes are much less affected by the crossover of RMs. They synthesized a Si-C composite (SCC) using Si NPs, phenolic resin, sodium alginate and bitumen as the precursors. The synthesized SCC exhibited an initial discharge capacity of 1750 mAh g⁻¹_SCC with 88.04% ICE in an SCC||Li half-cell. The capacity retention rate was 54% after 100 cycles at 0.1C, as shown in Figure 10f. Then, the electrochemically pre-lithiated SCC was assembled into Li-ion–air full cells (in N₂-O₂, 78:22, v/v) for cycling tests with a fixed capacity limit of 0.072 mAh cm⁻². As shown in Figure 10g,h, although the cell with SCC has a much lower areal capacity (1.74 mAh cm⁻²) than the Li foil (41.27 mAh cm⁻²) and Li-deposited Cu foil (abbreviated as Cu@Li, 11.93 mAh cm⁻²), it sustained significantly more (86) cycles than that with Li foil (60 cycles) or Cu@Li (43 cycles). Furthermore, ex-situ XPS and energy-dispersive X-ray spectroscopy (EDS) mapping after cycling reveals no deposition of RM (binuclear cobalt phthalocyanine, bi-CoPc) on the SCC (Figure 10d), despite the absence of an SSE membrane to prevent crossover. In contrast, the Li foil after cycling displayed clear Co signal in the EDS mapping, and the whole electrode turn green (Figure 10e, characteristic of bi-CoPc). These results demonstrate the higher compatibility of silicon-based anodes with RMs than LMAs.

To isolate silicon-based anodes from ambient air components, using (quasi-)SSE membranes has been a straightforward and effective approach. In 2017, Peng’s group [141] reported the first device-level prototype of a Li-ion–air battery with a GPE-protected (where GPE stands for gel polymer electrolyte) Si anode. They designed a coaxial fibrous structure where the electrochemically pre-lithiated silicon/carbon nanotube (CNT) hybrid fiber served as the inner anode, GPE as the interlayer and a bare CNT sheet as the outer cathode. As illustrated in Figure 11a, Si NPs were first coated onto two stacked CNT sheets, which was then twisted into a Si/CNT hybrid fiber (with 83.84 wt% Si, shown in Figure 11b–d). Electrochemical pre-lithiation was performed by galvanostatic cycling of the hybrid fiber with a Li sheet in a coin cell. It exhibited a reversible specific capacity of 1250 mAh g⁻¹_Si/CNT with a capacity retention rate over 80% after 100 cycles (Figure 11h), which was attributed to the spiral structure that enabled firm combination of Si NPs on the CNTs’ surfaces, as well as the numerous voids that provided space for volume expansion. The resulting Li_xSi/CNT fiber anode was then coated with a layer of UV-cured GPE, consisting of LiTFSI, LiNO₃, TEGDME, PVDF-HFP, NMP, 2-hydroxy-2-methyl-1-phenyl-1-propanone (HMPP) and trimethylolpropane ethoxylate triacrylate (TMPET). After that, an aligned CNT sheet was wrapped around the hybrid fiber to serve as the air cathode, with an anode–cathode mass ratio of 1.15:1, as shown in Figure 11e–g, and finally, a punched heat shrinkable tube was used to seal the fibrous cell. With a fixed capacity limit of 500 mAh g⁻¹_cathode (corresponding to an N/P ratio of 2.88), the full cell stably operated for 100 cycles in 5% RH air, with a specific energy of 512 Wh kg⁻¹ (based on the mass of the cathode and anode, excluding the mass of GPE and packaging), as shown in Figure 11i,j. These results strongly suggest that silicon-based anodes can be effectively protected by GPE against O₂ and moisture.

Later in 2019, Yang’s group [142] further improved the Li-ion–O₂ battery by optimizing the electrolyte salt. They synthesized a freestanding and flexible Si/C film (with 51 wt% Si) by electrospinning a dispersion containing polyacrylonitrile (PAN) and nano-silicon powder and subsequent calcination. When assembling the LAB/LOB, a Li foil was placed between the Cu-foil current collector and freestanding Si/C film, so the Si/C was chemically pre-lithiated once in contact with the electrolyte. Although the Si/C film exhibited a similar initial Li extraction capacity around 1000~1400 mAh g⁻¹_Si/C with different electrolyte formulations, the Si/C||Li half-cell with 1M LiTFSI TEGDME showed rapid capacity decay, which was ascribed to the SEI’s instability. However, by replacing LiTFSI with lithium bis(fluorosulfonyl)imide (LiFSI), a dramatically improved capacity retention rate (from 15% to 72% after 40 cycles) was achieved even without the FEC additive, close to that with carbonate electrolyte and FEC (83%). This improvement was attributed to the formation of a robust SEI, which contains Li₄SiO₄ and SiO₂ but without fluoride on the anode surface. The Li-ion–O₂ full cell was then assembled with a GPE based on PVDF-HFP, LiFSI and LiI, and multiwall carbon nanotubes (MWCNTs) were used as the air cathode with a loading of ~0.3 mg cm⁻². In addition, a composite Cu@Li foil served as both the anode current collector and pre-lithiation reagent. The cell with Si/C (with an areal capacity of 8~9 mAh cm⁻²) and GPE stably operated for 140 cycles with a fixed capacity of 1000 mAh g⁻¹_MWCNT (equivalent to 0.3 mAh cm⁻²), significantly higher than that with an LMA (80 cycles). In contrast, the cyclability with liquid electrolyte (1 M LiFSI TEGDME + 50 mM LiI in TEGDME) was much poorer, with only 90 cycles with Si/C and 44 cycles with the LMA. The improved cyclability with Si/C anode and GPE was attributed to the higher stability of Li_xSi than the LMA and GPE’s suppression effect on the crossover of O₂ and other soluble intermediates, respectively. They also tested the cyclability of Li-ion–air cells in humid air (40% RH): the cell with Si/C survived 20 cycles under a fixed capacity of 1000 mAh g⁻¹_carbon, while the cell with the LMA survived only 9 cycles.

In general, the use of silicon-based anodes in LABs/LOBs is still in its infancy but has shown good prospects. Electrolyte optimizations and combination with GPE have resulted in significant cycling performance improvements, even with low N/P ratios. However, most reported batteries used commercial Si NPs or homemade Si-C composites that are not optimized for long-term air and water stability. In addition, in most works, the air stability of these silicon-based anodes was not specifically tested in half-cells. These factors should be considered in the future development of Li-ion–air batteries with silicon-based anodes.

4. Conclusions and Prospects

According to the above discussions, pre-lithiated silicon-based anodes can achieve remarkable air stability, even under prolonged exposure (up to 30 days) to ambient humidity (up to 71% RH) through proper designs of protective layers (e.g., LiF, Li_xTiO₂ and graphene). These architectures allow the pre-lithiated anodes to retain most of their initial capacities, reaching up to 1600 mAh g⁻¹_LixSi. Furthermore, these protective architectures can serve as artificial SEIs to significantly enhance ICE. While pre-lithiation is essential for improving ICE and cyclability and can potentially alleviate the need for excessive buffering substrates or complex internal void structures (e.g., yolk–shell designs), its industrial application has been historically bottlenecked by the poor air stability of lithiated silicon. Given these advancements in air stabilization, the mass production of pre-lithiated silicon materials and their increased integration into graphite-based anodes are expected in the near future.

With further advancements in ICE and cyclability, silicon-based materials are poised to transition from additives to the dominant active components in anodes, where they would rival LMAs in specific capacity. Although PolyPlus PLEs offer impeccable air and water stability alongside near-theoretical specific capacity, their effective specific capacity in secondary batteries (e.g., 709 mAh g⁻¹ with 25 μm thick Li, as calculated in Section 1) can be surpassed by certain air-stable silicon-based anodes (such as Li_xSi/graphene composite foils). This performance gap arises because maintaining robust interfacial contact between the lithium metal and the protective layers (e.g., LiPON/LATP) during cycling necessitates the use of thin lithium foils, which inadvertently leads to a disproportionately high mass percentage of these auxiliary components. To pursue extreme specific and areal capacities, ultrathick LMAs (e.g., PLEs with 3350 μm thick Li) remain the ultimate pathway; in this scenario, silicon-based materials can still function as critical structural scaffolds, providing lithiophilic deposition hosts and high-flux lithium transport pathways, as illustrated in Figure 1c. Furthermore, drawing inspiration from the concept of “reserve lithium-ion batteries” [145], air-stable pre-lithiated silicon can act as a robust “lithium reservoir”. This setup not only enables in situ lithiation of the anode to compensate for initial capacity loss but also facilitates the recovery of lost capacity in both electrodes during long-term cycling. By integrating air-stable protective layers, these silicon-based reservoirs can mitigate the risks associated with pure lithium metal while ensuring a continuous lithium supply, thereby significantly extending the lifespan of LABs/LOBs.

As for the Li-ion–air (or O₂) batteries, although most utilized silicon-based anodes were not optimized for air stability, they have still shown a significant advantage in cyclability over unprotected LMAs even with a much smaller lithium reserve, indicating that silicon-based hosts can effectively mitigate the catastrophic side reactions between active lithium and permeated oxygen species (e.g., O₂ or O₂⁻ radical). From the perspective of hard and soft acids and bases (HSAB) theory, while metallic Li⁰ acts as a highly reactive “soft base” that reacts indiscriminately with “hard” oxygen-containing species, the lithium in Li_xSi alloys possesses a more localized electron density, shifting its character toward a harder acid. This increased “hardness” promotes a more chemically robust interface and thermodynamic compatibility within the lithium–air (or O₂) environment. Consequently, unlike the formation of thick LiOH-rich passivation layers typical of unprotected LMAs in LABs/LOBs, the lithiated silicon framework provides a more stable kinetic landscape for lithium storage. With the integration of air-stable protective layers (e.g., those mentioned in Section 2), pre-lithiated silicon anodes could not only rival LMAs in capacity but also surpass them in practical robustness for long-term LAB operations.

Despite these promising advancements, several critical challenges must be addressed to fully realize the potential of air-stable pre-lithiated silicon anodes in next-generation batteries.

First, there is a prominent “bottleneck” in bridging material-level stabilization with system-level application. While Section 2 highlights significant breakthroughs in air-stable pre-lithiated silicon anodes, the practical LOB/LAB studies summarized in Section 3 still predominantly rely on conventional or unprotected silicon materials, and the superior durability of these newly developed protective architectures (e.g., Li_xTiO₂ and graphene-based scaffolds) has yet to be validated under the aggressive oxidative environments and complex radical chemistries of functioning LABs/LOBs. Future research should prioritize the integration of these air-stable anodes into full-cell configurations. This would not only verify their effectiveness in suppressing parasitic reactions (such as LiOH and/or Li₂CO₃ accumulation) but also provide a realistic assessment of their energy density advantages over lithium metal-based systems under ambient conditions.

Second, further insights into the reaction mechanisms between the pre-lithiated species and electrolyte/air components are required. Although silicon-based anodes have been extensively investigated for decades, the chemical compatibility of their pre-lithiated states with air components (O₂, H₂O, CO₂, N₂, etc.), as well as with specific LAB/LOB electrolyte components (e.g., ethereal solvents, lithium salts, RMs and crossover intermediates), has received far less attention. Establishing such a fundamental understanding is vital for further enhancing the ambient stability of silicon-based anodes and ensuring their successful integration into future practical, long-cyclable LABs/LOBs. Furthermore, such advancements could relax the stringent dry-room requirements for battery assembly and potentially lead to their eventual elimination in silicon-based LIB/LAB production. In the investigations of such reaction mechanisms, modern analytical tools, such as in situ/operando characterization techniques (e.g., cryo-TEM, near-ambient pressure TEM, surface-enhanced Raman spectroscopy (SERS) and isotope tracing), first-principles calculation tools (e.g., Vienna Ab initio Simulation Package (VASP), Gaussian and ORCA) and machine learning/artificial intelligence (ML/AI) tools, will play critical roles.

Third, identifying optimal electrolyte formulations tailored for silicon-based LOBs/LABs is paramount. While glymes (e.g., TEGDME) remain the conventional choice due to their balanced stability at the cathode and anode, they are still vulnerable to nucleophilic attack by superoxide radicals. Emerging candidates such as fully methylated ethers [27,30], amides [28,29] and particularly localized high-concentration electrolytes (LHCEs) [146,147,148] have shown great promise in stability improvements at the air cathode; some of them (notably LHCEs) can also enhance stability against singlet oxygen and LMAs while maintaining/enhancing favorable oxygen solubility/diffusivity. Furthermore, the transition toward gel polymer electrolytes (GPEs) could be a decisive strategy to physically block the crossover of detrimental air components to the Si-based anode. Regarding additives, while SEI-forming additives (e.g., VC and FEC) are essential for Si-based anodes, their dosage must be carefully balanced with cathode-centric additives, such as RMs (e.g., LiBr [149] and bi-CoPc [55]) and singlet oxygen quenchers (e.g., TPA [60]) to simultaneously ensure anode protection and prevent cathode passivation; if high-quality protection structures are successfully formed, the need for these SEI-forming additives could even be eliminated, as demonstrated in Cui’s series of studies [127,128,131]. Integrating these advanced electrolyte systems with air-stable silicon anodes could be a critical frontier for achieving high-performance LOBs/LABs.

Fourth, a comprehensive techno-economical analysis (TEA) is indispensable for evaluating the commercial viability of these high-capacity anodes. While the sophisticated design and manufacturing of protective architectures (e.g., atomic layer deposition or specialized carbon coating) may increase initial material synthesis costs, these expenses can be strategically offset by substantial savings at the system and manufacturing levels. Specifically, the enhanced ambient stability of pre-lithiated silicon significantly relaxes the stringent dew-point requirements for dry rooms, one of the most capital-intensive and energy-consuming components of battery production facilities, thereby drastically reducing both capital expenditure (CAPEX) and long-term operational expenditure (OPEX). In addition, the simplified battery assembly process, which bypasses the need for external lithium-compensation equipment or sacrificial additives, could further streamline production lines. Future TEA should therefore adopt a holistic “cradle-to-gate” approach to balance the incremental costs of advanced material processing against the significant gains in manufacturing efficiency and device-level energy density, ensuring that these anodes remain economically competitive with state-of-the-art graphite and lithium metal technologies.

We envision that the integrated design of air-stable, high-capacity silicon-based anodes coupled with (quasi-)SSE membranes, OSMs and/or pre-lithiation reagents will yield significant improvements in energy density and cyclability. Such holistic strategies, supported by modern analytical tools, are expected to advance the practical application of rechargeable LABs/LOBs, especially those targeting high areal capacities.