RAO et al.

Abstract

The landmark Net Zero Emissions by 2050 Scenario requires the revolution of today's energy system for realizing nonenergy-related global economy. Advanced batteries with high energy density and safety are expected to realize the shift of end-use sectors toward renewable and clean sources of electricity. Present Li-ion technologies have dominated the modern energy market but face with looming challenges of limited theoretical specific capacity and high cost. Li–air(O₂) battery, characterized by energy-rich redox chemistry of Li stripping/plating and oxygen conversion, emerges as a promising “beyond Li-ion” strategy. In view of the superior stability and inherent safety, a solid-state Li–air battery is regarded as a more practical choice compared to the liquid-state counterpart. However, there remain many challenges that retard the development of solid-state Li–air batteries. In this review, we provide an in-depth understanding of fundamental science from both thermodynamics and kinetics of solid-state Li–air batteries and give a comprehensive assessment of the main challenges. The discussion of effective strategies along with authoritative demonstrations for achieving high-performance solid-state Li–air batteries is presented, including the improvement of cathode kinetics and durability, solid electrolyte design, Li anode optimization and protection, as well as interfacial engineering.

INTRODUCTION

Energy is the essential material basis for human survival and development. The constant revolution in energy technology perpetually promotes socioeconomic development and the progress of human civilization. Currently, fossil fuels contribute to over 80% of the world's energy supply, propping up the functioning of modern society while breeding a world energy structure that lacks renewability and sustainability. Severe environmental problems such as global warming inevitably emerge and deteriorate, pushing global average temperature 1.1°C higher since the preindustrial age, with a distinct impact on climate change and the ecological system. Given that modern energy is indispensable to the livelihoods and the developing economies on their navigation of urbanization and industrialization, the revolution of today's energy system for realizing nonenergy-related greenhouse gas emissions becomes a historical mission for all human beings. The announced climate pledges from participating states at the 26th Conference of the Parties effectively move the needle on the International Energy Agency's (IEA's) landmark Net Zero Emissions (NZE) by 2050 Scenario. For electricity has taken the prominent sector of the energy system, cleaning up the electricity mix and extending the electrification of end-uses through the shift toward renewable and low-carbon-emission sources of electricity, is an acknowledged tactic for rebuilding global energy supply and realizing NZE. According to IEA, the electricity sector has been responsible for 36% of all energy-related CO₂ emissions in 2020 already, while electricity demand is projected to reach 42,000 TWh by 2050 (almost 80% above today's level), which urges secure electrical energy storage technologies to connect end-use sectors and intermittent supply from clean energy (solar power, wind, etc.).¹

With high flexibility and compatibility with different service scenarios, electrochemical energy storage technologies become a favored choice for the energy market. Li-ion batteries (LIBs) have established supremacy in consumer electronics and power batteries such as electric vehicle cells, figuring as the central pillar of modern energy storage systems. However, LIBs manifest an upper limit of the energy density of about 400 Wh/kg even with the predictable advancement of both electrode materials and concomitant technologies² that can meet the demand from light-duty vehicles, while cannot support heavy-duty vehicles requiring long driving range and/or heavy load, with a targeted energy density of more than 500 Wh/kg.³ Hence, “beyond Li-ion” batteries based on energy-rich redox chemistry such as a couple of metal dissolution/deposition and oxygen/sulfur conversion, emerge as promising candidates for next-generation energy storage strategies^2,4  in which, nonaqueous Li–air batteries (LABs) take the preponderance of ultrahigh theoretical energy density of 3600 Wh/kg (based on Li₂O₂ discharge product), outstripping other choices notably.

However, like LIBs, the utilization of flammable and volatile organic electrolytes brings safety concerns of nonaqueous liquid-state LABs, accompanied by a narrow electrochemical window which unmatched for Li metal anode and limited operating temperature range. The electrolyte volatilization also incessantly damages the integrity of the triple-phase interface at the cathode, which accommodates the key reactions of LABs, then renders irreversible performance decline and even battery explosion. Meanwhile, the strong oxidizing reactive oxygen species (¹O₂, O₂⁻, O₂²⁻, etc.) can readily decompose nonaqueous electrolytes and then form undesirable byproducts such as CO₂, Li₂CO₃, HCOOLi, and CH₃COOLi. These contaminants would inevitably deteriorate the long-term stability of LABs by blocking the gas diffusion channel and escalating overpotentials. The oxygen environment also pulls the Li anode of LABs into a more intractable condition, as the reactive oxygen species would parallelly proceed with the accumulation of insulating discharge products (Li₂O₂/Li₂O, Li₂CO₃, etc.) on the Li anode surface. Characterized by a half-open cell structure, the electrochemical environment of LABs is further challenged by complicated atmospheric species (CO₂, H₂O, etc.), and their shuttle behaviors in nonaqueous electrolytes arise severe threats such as parasitic reactions, complicated discharge routes, and anode corrosion. These species along with redox mediators (RMs; usually used as homogeneous catalysts) would greatly affect the composition and morphology of the solid electrolyte interphase (SEI) at the Li anode/electrolyte solution interface, leading to undesirable electrolyte consumption, anode passivation, and formidable dendrite growth. Our group has devoted intensive efforts to mechanism study,^5–8 materials design,^9–15 and building sealed system with Li₂O/Li₂O₂ chemistry for nonaqueous liquid-state LABs,¹⁶ and provided thorough discussions on this field in previous reviews.^17–22

Apparently, the nonaqueous electrolyte is the Achilles heel of LABs that dominates both cathode and anode electrochemistry, as well as battery stability and safety. However, an ideal combination of aprotic solvents and lithium salts that fits all criteria relevant to LABs is yet to be found.²³ The existing electrolyte solutions are only metastable for reactive oxygen species and limit the working potentials of standard LABs,²³ thus can hardly serve in practical LABs. Solid-state electrolyte (SSE) emerges as a promising substitute for liquid electrolyte, possessing superior stability (in thermal, chemical, and electrochemical aspects), wide electrochemical window, favorable mechanical strength, and cost efficiency.²⁴ These merits enable solid-state LABs to get rid of electrolyte evaporation and thus guarantee intact reaction interface, restrain the dendrite penetration and anode corrosion, eliminate the risks of battery combustion and explosion, and effectively enlarge working potentials and temperature range, then effectuate enhanced safety and life span as practical LABs expected.^25–27 However, the development of solid-state LABs is still in its infancy, facing several challenges, such as poor interfacial contact and/or low ionic conductivities of SSEs, limited triple-phase boundaries in the cathode, and questionable durability for the open-air circumstance.^27–29 Intensive work has identified the fundamental underpinnings and effective strategies to build up high-performance solid-state LABs. Some progress on materials has been overviewed in our previous reviews.^26,30,31 But the comprehensive report covering basic mechanisms, challenges, recent progress, and possible solutions of solid-state LABs is in demand.

In this review, the discussion will start with the fundamental science and the main challenges of solid-state LABs, then emphasize the strategies apropos of improving air cathode kinetics, solid electrolyte design, Li metal anode optimization and protection, and interfacial engineering. It is expected that this review would provide the interested reader with a systematic understanding and instrumental guidance in constructing safe, stable, and practical solid-state LABs.

FUNDAMENTALS AND CHALLENGES FOR SOLID-STATE LABs Cell configurations and chemistry

A typical solid-state LAB (Figure 1) is composed of a Li metal anode, a Li-ion conductive SSE, and a porous cathode with desirable gas diffusion channels, as similar to the liquid-state LABs except for the use of condensed electrolyte. Thus, the solid-state LABs also primarily operate with the stripping/plating of Li metal at the anode during discharge/charge processes (Equation 1), and corresponding oxygen reduction reaction (ORR)/oxygen evolution reaction (OER) at the cathode (Equation 2). The electrochemical reaction pathway of ORR is considered to be similar to the acknowledged mechanism for aprotic LAB proposed by Abraham et al.³² but presents some differences. It should start with the reduction of O₂ into LiO₂ via one-electron transfer first (Equation 3) as liquid-state LABs, but the following chance of chemical disproportionation of LiO₂ (Equation 4) in electrolyte would be not expected in solid-state LABs for the absence of LiO₂ ion pair generated by strong Li⁺ solvation. The electrochemical step for receipt of Li⁺ and the other electron (Equation 5) to form Li₂O₂ via the surface pathway is more reasonable.^33–35 But the solid evidence is hard to obtain in this condensed solid system, while the evolution of intermediates cannot be detected due to limits of in situ characterizations. Meanwhile, the recharge reaction route and relevant intermediates are still not ascertained and need to be explored carefully. Possible mechanisms have been proposed in liquid-state LABs, namely the direct two-electron transfer route,^32,36–39 the formation of Li-deficient Li₂O₂ with delithiation,^40–43 and LiO₂ intermediated mechanism^42,44,45 which has also been verified in solid-state LAB.⁴⁶ The overall battery reaction is depicted in Equation (6). The E⁰_a, E⁰_c, and E⁰_OCV are the thermodynamic anode potential, cathode potential, and open circuit voltage (OCV) under the standard condition of LABs based on Li₂O₂ discharge product, respectively. [Image Omitted. See PDF] [Image Omitted. See PDF] [Image Omitted. See PDF] [Image Omitted. See PDF] [Image Omitted. See PDF] [Image Omitted. See PDF]

Figure 1. The schematic of solid-state LABs and the challenges related to air cathode, solid electrolyte, and Li anode, respectively. LAB, Li–air batteries; OER, oxygen evolution reaction; ORR, oxygen reduction reaction.

Obviously, the reserve and decomposition of Li₂O₂ in the cathode determine the energy density and reversibility of solid-state LABs. The storage of Li₂O₂ can be optimized by electrode structure engineering to provide more active catalytic sites, gas diffusion channels, and available space. The most intractable problem is that the decomposition of Li₂O₂ during recharging is not thermodynamically preferred with a standard formation Gibbs energy (ΔG_f^θ) of −649.45 kJ/mol,⁴⁷ as a solid-state insulator with a wide bandgap delivering limited electronic and ionic conductivity,⁴⁸ inevitably retards the reversibility of LABs. But the fundamental study on Li₂O₂ oxidation is in the growth stage and faces great challenges due to the interference of uncontrollable byproducts formation such as LiOH, Li₂CO₃, and so forth. The reaction route of Li₂O₂ decomposition remains unclear and focuses on a controversial topic where the involved reactions take place; however, both the cathode/Li₂O₂ interface and the electrolyte/Li₂O₂ interface are considered possible. The observations from electron microscopy mostly favor that the cathode/Li₂O₂ interface supports Li₂O₂ decomposition. For example, Zhong et al.⁴⁹ provided in situ transmission electron microscopy (TEM) observation of preferable Li₂O₂ decomposition at the carbon nanotube (CNT)/Li₂O₂ interface instead of the electrolyte/Li₂O₂ interface in a solid-state LAB, reaching consensus with the phenomena presented by in situ TEM and scanning transmission electron microscopy techniques in liquid LABs.^50–52 They proposed that the electron-transport efficiency in Li₂O₂ should limit the oxidation kinetics of Li₂O₂ at high overpotentials, based on its insulating nature with a wide bandgap of >2 eV. But Zheng et al.⁵³ observed that the decomposition preferentially initiated at the surface of discharge products instead of their contact region with the CNT cathode or electrolyte, and continued along a certain direction of the bulk structure via an in situ environmental scanning electron microscope, in accordance with some TEM and SEM studies in liquid LABs.^54,55 In these surface-started observations, the higher electronic conductivity on the surface of Li₂O₂ than that in the bulk was considered as a trigger, while the favorable O₂ release occurred on the surface area under a vacuum condition stimulated Li₂O₂ decomposition.

Recently, Luo et al.⁴⁶ carried out a remarkable reaction mechanism study of the full-cycle reaction pathway for a solid-state Li–O₂ nanobattery, with RuO₂ decorated CNTs as a cathode, Li metal as an anode, and the Li₂O covered on Li metal as a solid electrolyte (Figure 2A). Using aberration-corrected environmental TEM achieves in situ analysis of the phase formed in discharge and charge processes. They found out that LiO₂ was initially produced on CNTs during discharge and then disproportionated into Li₂O₂ and O₂ (Figure 2D), which induced the hollow structure with a Li₂O outer shell and a Li₂O₂ inner surface (Figure 2B,E). While recharging, the Li₂O₂ lost one Li⁺ and one electron to form LiO₂ (Figure 2F), which then released O₂ accompanied by shell collapse until complete decomposition (Figure 2C,G). This research identifies the electrochemical pathway in the designed solid-state Li–O₂ system, providing insights into LABs chemistry. But limited by time resolution and the response to ordered structure, in situ TEM could not capture the transient intermediates and the amorphous products. The exquisite design for realizing real-time characterization of (electro)chemical evolution in solid-state LABs is urgently demanded, which could be based on advanced in situ X-ray absorption spectroscopy, surface-enhanced Raman spectroscopy, and so forth.

Figure 2. (A) Schematic illustration of the Li–O2 nanobattery in an environmental TEM chamber. The time-resolved TEM images for showing the evolution of (B) discharge products and (C) charge products. The corresponding SAED results and compositional illustrations for (D–E) discharging and (F–G) charging. Reproduced with permission: Copyright 2017, Nature Publishing Group.46 OER, oxygen evolution reaction; ORR, oxygen reduction reaction; SAED, selected area electron diffraction; TEM, transmission electron microscopy.

Thermodynamics and kinetics

The thermodynamics and kinetics of LABs dominate their intrinsic chemical/electrochemical behaviors and provide guidelines for improving output performance. In liquid-state LABs, some in-depth studies focused on discharge products or solvent-driven principles lead ground-breaking research and design strategies for practical devices. For instance, the highest occupied molecular orbital (HOMO) level^56,57 and acid dissociation constant (pK_a)^58,59 have been proposed as the descriptor of thermodynamic driving forces for the oxidative stability and chemical stability of solvents, respectively. The Gibbs free energy of LiO₂ solubility in aprotic solvents has been tightly related to Li₂O₂ formation routes (i.e., surface pathway in low-donor number [DN] solvents and solution pathway in high-DN solvents),⁴³ while a trade-off is required because the high-DN solvents are preferred to deliver higher capacities but have higher susceptibilities to nucleophilic attack.⁶⁰ These work offer valuable references for the thermodynamics and kinetics study targeted solid-state LABs which is still in the initial stage. Characterized by the solvent-free property of solid-state LABs, these research mainly direct on ion migration in solid electrolytes and then could share some findings in solid-state LIBs. Meanwhile, a previous study on ORR/OER that has been established in liquid-state LABs and is irrelevant to solvent should also be applicable. But considering the working condition is in an ambient atmosphere instead of pure oxygen, the development of thermodynamics and kinetics research for solid-state LABs are exceedingly challenging.

Free energy levels of adsorbed oxygen intermediate as LiO₂* and Li₂O₂* on the cathode are usually computed to identify the thermodynamically favored Li₂O₂ nucleation pathways and deduce the rate-determining step (RDS).^61–66 A balanced adsorption energy toward LiO₂* is essential to drive both O₂ activation and Li₂O₂ nucleation at low overpotentials. The weak LiO₂* bind renders the electrochemical activation of O₂ into LiO₂* as the limiting step, while the strong LiO₂* bind induces the limitation from the reduction of LiO₂* to Li₂O₂, similar to ORR and OER in aqueous systems that involve multistep electron transfer between oxygen intermediates. Theoretically, the overpotential for RDS can be used as a thermodynamic descriptor for Li₂O₂ formation and decomposition. But an accurate computational model is the prerequisite, which should be carefully designed in a not-well-defined system (nonspecific surface structure, defects, under complicated gas and temperature conditions, etc.) that makes the computed overpotentials hard to be a tool for predicting the cathode performance in LABs. And it needs to be pointed out that the direct reverse reaction steps of ORR have often been used as models for OER while theoretical calculation, even though researchers reach a consensus that the charging process of LABs does not follow the reversed route of the discharge process. The simplification could bring unreliable results and mislead readers. On the premise of the kinetics of the discharge process is limited by LiO₂* formation, Viswanathan et al.⁶⁷ proposed that the energy barrier could be predicted as demonstrated by Tafel kinetics (Equation 7): [Image Omitted. See PDF]where $$[{\text{Li}}^{+}]$$ and $${{{\rm{O}}}_{2}}^{* }\,$$are the reduction reactant; $${<mpadded >\unicode{x02206}G</mpadded>}_{0}^{\pm }$$ represents the kinetic barrier to the RDS at the equilibrium potential; $$\alpha $$ is the symmetry factor; $$e$$ is the electron charge; $${U}_{\text{dis}}$$ and $${U}_{0}$$ stand for the discharge plateau voltage and the equilibrium potential, respectively.

Undoubtedly, the growth/dissolution thermodynamics of Li₂O₂ on the Li₂O₂ itself is pivotal for the outset of discharge/end of charge. Hummelshøj et al.⁶⁸ developed a thermodynamic analysis of Li₂O₂ growth and dissolution on different stable facets, terminations, and charge-transfer sites (terrace, steps, and kinks) of Li₂O₂ surface. The surface energies of these nonstoichiometric surfaces are potentially dependent, thus enabling the prediction of dissolution barriers following different paths. In theory, low thermodynamic overpotentials (<0.2 V) existed for the charge at many Li₂O₂ sites on the facets studied. This explained some low experimental overpotentials with forming conformally deposited Li₂O₂ at early charge stages and provided a design strategy for building up Li₂O₂ electrochemistry with preferable thermodynamics. In addition, the catalyst-participated dissolution of Li₂O₂ is usual while not dependent on the unified standard. A general activity descriptor for Li₂O₂ decomposition activated by 4d transition metal catalysts was proposed by Zhao et al.⁶⁹ They described Li₂O₂ decomposition as stepwise Li⁺ desorption and O₂ evolution, that, Li⁺ → Li⁺ → O₂. The first-principles thermodynamic calculations revealed that the surface electron affinity (V_SEA) and surface ionic potential (V_SIP) of these catalysts should determine the activation of Li–O₂ bonds and the reduction of desorption barriers of Li⁺ and O₂, respectively. A balanced activity descriptor defined as surface electronegativity (V_SE, V_SE = (V_SEA + V_SIP)/2) was developed with a volcano correlation to reduced charge overpotential, identifying those catalysts with a V_SE value of 1.7–2.2 V possess high OER activity (Figure 3A). This descriptor provides a new guide for predicting OER catalysts with certain surface structures.

Figure 3. (A) The reduced overpotential with VSE. Reproduced with permission: Copyright 2020, American Chemical Society.69 (B) Cation migration mechanisms and (C, D) the energy profiles related to direct vacancy or interstitial hopping. Reproduced with permission: Copyright 2019, Springer Nature.70

The kinetics of ion migration in solid electrolytes underpins the intrinsic mass transport of solid-state LABs. Unlike in a liquid battery, there is only one type of ion (Li⁺) with mobility in SSEs.⁷¹ The diffusion can be depicted as ion hops between stable ground-state sites and/or metastable anion sites such as O²⁻, S²⁻, or polyanionic moieties.^70,72,73 The bonding environment of these sites determines the migration pathway of ions, as a result of the availability and connectivity of arranged anions. In an inorganic crystalline electrolyte, there are three typical migration mechanisms (Figure 3B–D): (1) vacancy diffusion realized by the ion migration to adjacent vacant sites, (2) direct interstitial transfer between incompletely occupied sites, and (3) concerted or correlated interstitial mechanism that the migration of interstitial ions permute neighboring lattice ions into the adjacent sites.⁷⁰ In a polymer electrolyte, the successive coordination of mobile ions and polar groups on the chain segments enables ion migration.⁷⁴ The key descriptor for ion transport in SSEs is ionic conductivity ($$\sigma )$$. For inorganic crystalline electrolytes, $$\sigma $$ is defined by a modified Arrhenius relationship related to the product of charge ($$q$$), concentration ($$n$$), and mobility of charge carriers ($$u$$): [Image Omitted. See PDF]where $${\sigma }_{0}$$ presents the pre-exponential factor for intrinsic carrier density; $$m$$ typically equals to −1; $${k}_{{\rm{B}}}$$ is the Boltzmann constant; $$T$$ is the temperature; and $${E}_{{\rm{a}}}$$ indicates characteristic activation energy for ion conduction comprising the formation energy of mobile defects ($${E}_{{\rm{f}}}$$) and the energy barrier for their migration ($${E}_{{\rm{m}}}$$). For polymer electrolytes, the description of $$\sigma $$ gives some discrepancy as shown in the following equation: [Image Omitted. See PDF]where $$R$$ stands for the ideal gas constant; and $$T$$ is the reference temperature while a positive difference compared to the glass transition temperature of a polymer benefits the inside irregular movement of Li⁺.

It is noteworthy that the ion migration in SSEs is a multiscale process and the obtained impedance in a routine electrochemical measurement is the result of all these factors. But the comprehensive understanding of mechanisms from the atomic scale to the device scale can guide the way for realizing fast ion transport. In a recent review, Famprikis et al.⁷⁰ provide detailed summarization and discussion of the mechanism study on multiscale ion transport accompanied by some progressed techniques, contributing to an in-depth and multidimensional inspiration for the development of practical SSEs.

With the utilization of SSEs, the thermodynamics and kinetics study that reside in the solid–solid interface are needed reasonably. At the cathode side, the premise for an ideal SSE is that its HOMO level should be below the Fermi energy level of the discharge product to prevent SSE oxidation and unpredictable cathode electrolyte interphase (CEI),⁷⁵ which is similar to the research results in liquid-state LABs.^56,57 Analogously, its lowest unoccupied molecular orbital should be higher than that of the Fermi energy level of the Li anode, or the reduction of SSE will deteriorate its stability and generate uncontrollable SEI. The presence of CEI and SEI is likely to increase interfacial resistance to Li⁺ migration and charge transfer, as a drag on electrochemical kinetics while the expected high-efficiency kinetics at the solid interfaces demands qualified conductivity for both electron and ion. Interfacial kinetics is a common problem for all solid-state batteries. However, the case for solid-state LABs is more complicated, as a combination of electron transfer, ion migration, O₂ circulation, and parasitic reactions. Thus, constructing a tight and integral “air–solid–solid” triple-phase contact at the cathode/electrolyte interface is a significant challenge.^76–78 Characterized by the electron-insulated property of Li₂O₂, the charge–discharge process is severely hindered at the triple-phase sites. The limited electrochemical reversibility of solid-state Li₂O₂ inevitably results in escalating recession of reaction activity and collapse of triple-phase boundaries, making the physical contact shift from a connected configuration to a “point to point” model and then exceedingly decrease the interfacial kinetics.

Overall, the thermodynamics and kinetics of solid-state LABs are the foundation for understanding the origin of discharge/charge overpotentials and decayed stability. But the unrevealed mechanisms such as the electrochemical/chemical pathway at the cathode, intermediates behaviors in solid electrolytes, and interfacial reactions at solid–solid(–gas) contact regions, highly restrict the development of instructional theories for practical devices. We urgently need ingenious computational work and in situ characterization methods to identify the RDS indicators and qualify the optimal thermodynamical/kinetics factors for battery performance from the atomic scale to device scale. Then we can build up desirable electrochemical reactions and structural engineering for the specific bottleneck.

Main challenges

As we have mentioned, the preponderance of intrinsic safety and stability enables solid-state LABs to become a practical option for next-generation energy storage techniques. However, coupled with complicated electrochemical–chemical reactions that require high integration of interfacial behaviors in multiscale, the solid-state LABs are confronted with several challenges (Figure 1).

• (i)

  Mechanisms and kinetics of air cathode. The “air–solid–solid” triple-phase reaction accommodated at the cathode is the origin of high energy density for solid-state LABs. And its reversibility and consistency dominate the overall performance such as energy efficiency, rechargeability, stability, and life span. As mentioned above, even though some remarkable in situ electron microscopy studies have provided insights into the chemical evolution of solid products,^46,49,53 the types of intermediates and their evolution involved in discharge/charge processes are poorly understood. The insufficient knowledge about cathode mechanisms induces scarce research on thermodynamics and kinetics, lack of effective criteria, and presently, intensive studies on cathode targeting for building up highly active bifunctional catalysts and favorable porous structures that can perform well in electrochemical tests. But it is hard to reach the essence of cathode electrochemistry/chemistry underneath the demonstrated properties without theoretical guidance, like operating in a black box. Parasitic chemistry of contaminates in the air such as H₂O and CO₂, arouses additional interferences for the theoretical study of cathodes working in real air conditions. Thus, the well-designed in situ characterization methods and computational models with prudent variates control are highly recommended but extremely challenging.

• (ii)

  Ionic conductivity and stability of electrolytes. The realization of applicable SSEs is a critical precondition for intrinsically safe solid-state LABs. Key properties of SSEs include high ionic conductivity that has been associated with the critical current density initiating Li dendrite at the Li anode/electrolyte interface,^79,80 negligible electronic conductivity which prevents short circuits arising from the growth of Li dendrite in SSEs, chemical compatibility with both cathode and Li anode, and wide electrochemical stability window.⁷⁸ Characterized by the sluggish ionic mobility, the Li⁺ conductivity (σ_Li) of SSEs usually dwells in the range of 10⁻⁶–10⁻⁴ S/cm, 2–4 order of magnitude below that for a liquid electrolyte (10⁻² S/cm).⁸¹ Among these SSEs, inorganic sulfides with higher ion conductivity and better compatibility with Li anode have been a favored choice in solid-state LIBs, while their instability for air atmosphere and tendency to generate toxic H₂S restrain the application in LABs. The most studied NASICON-type SSEs can supply decent σ_Li while the cations prone to be reduced by Li anode at low potentials, like Ti⁴⁺ in Li_1+xAl_yTi_2−y(PO₄)₃ (LATP) and Ge⁴⁺ in Li_1+xAl_yGe_2−y(PO₄)₃ (LAGP), rendering chemical stability concerns at the Li anode/electrolyte interface. Other stringent requirements for air tolerance, thermal stability, adequate mechanical properties, and capability of being processed into desirable thin electrolytes exert more challenges for SSEs.

• (iii)

  Lithium anode issues. Compared with liquid electrolytes, the SSEs possess a superior mechanical advantage over protecting Li anode by restraining the dendrite penetration, which leads to low Coulombic efficiency and safety concerns for all Li-metal batteries. Specifically, the solid-state inorganic ceramics were opined to be unaffected by Li dendrite propagation and benefited from their high Li transfer number,⁷⁸ high shear modulus with the magnitude of tens to hundreds of GPa,^82–84 and low fracture toughness.⁸⁵ However, the observations that Li dendrites preferably penetrate the SSEs through or along grain boundaries, surface defects, and coterminous pores, are challenging the conviction of impervious dendrite penetration in SSEs.^86–88 Except for the dendrite issue, the Li anode corrosion could be triggered by the shuttle behaviors of O₂, H₂O, and CO₂ into the solid matrix under air environment, along with volume change during Li stripping/plating, inevitably retrograding Coulombic efficiency and cycle performance. Apparently, the Li anode assembled in a solid-state LAB faces more challenges than other Li-metal batteries operated as sealed systems, requiring more scrupulous study to impel secure utilization of Li anode.

• (iv)

  Interfaces in solid-state LABs. As shown in Figure 1, except for the triple-phase interface for ORR/OER, the solid-state LABs accommodate the other four types of interfaces as Li anode/electrolyte, electrolyte/electrolyte, addictive/electrolyte, and cathode/electrolyte interfaces. The quantity of triple-phase interfaces determines the overall activity of ORR/OER, which need consistent accessibility and integrity but is challenged by the coverage of solid products and structural collapse of the cathode. Other interfaces can be categorized as follows: (1) voids, formed via unideal close packing of electrode and SSEs, as well as electrode pulverization during cell operation, could induce high contact resistance and dendrite growth; (2) grain boundaries, presenting a Li-deficient space-charge layer generated by Li⁺ transfer between two adjacent particles with different electrochemical potentials, could greatly inhibit ion conductivity at the interface; (3) those produced by chemical/electrochemical reactions, coming from the mismatch chemical potential of electrode and SSEs, or the instability of SSEs under low/high voltages, could contribute to the formation of undesirable interfaces and exacerbate the charge transfer resistance.⁸⁹

STRATEGIES FOR HIGH-PERFORMANCE SOLID-STATE LABs

Characterized by a hierarchical solid structure, half-open operating environment, and the utilization of Li metal anode, the solid-state LABs face miscellaneous challenges such as intractable interface physics and interface chemistry, parasitic reactions induced by air contaminants, anode dendrite, corrosion, and so forth. An in-depth study of thermodynamics and kinetics would provide basic strategies for constructing high-performance solid-state LABs, but comprehensive guiding criteria are absent yet. Extensive efforts have been devoted to researching practical tactics for optimizing specific components and/or enhancing overall battery performance, usually accompanied by specific theoretical studies to find out how these tactics work. In general, there are mainly four effective routes to reaching high-performance solid-state LABs: (1) the improvement of cathode kinetics, with highly active bifunctional catalyst and desirable gas diffusion channels, has been acknowledged as an efficient method to obtain high energy efficiency and extended cycle life; (2) rational electrolyte design to deliver high σ_Li and chemical/electrochemical stability, as well as competent mechanical properties which are pivotal for realizing an applicable solid-state LAB; (3) the optimization and protection of Li metal anode via microstructural design, surface functionalization, and introducing functional layer to build up a roust anode; (4) interfacial engineering targeted for low interfacial impedance and stable chemical/electrochemical environment, supports the fast kinetics and intrinsic stability of solid-state LABs.

Improvement of cathode kinetics and durability

The reversible oxygen conversion reaction makes solid-state LABs distinct from traditional rocking-chair batteries based on ion intercalation, enabling ultrahigh energy density along with unremitting oxygen supply from the external air environment. Theoretically, the capacity of the cathode is no longer a limit for output energy while the available capacity from the Li anode is. However, in a practical solid-state LAB, the limited space for accommodating solid-state discharge products and their sluggish kinetics of formation/decomposition at the cathode severely restrict the overall battery performance. The catalytic behaviors and physical properties of the cathode determine its intrinsic electrochemical/chemical kinetics for ORR and OER, the efficiency of oxygen diffusion, the storage capability of discharge products, and durability for long-term operation. Thus, an ideal cathode for solid-state LABs should contain a highly active catalyst driving reversible oxygen conversion under low overpotentials, a rationally designed porous structure for preferred oxygen diffusion and accessibility to catalytic sites, superior conductivity for sufficient electron supply, and adequate void volume for reserving discharge products.

Carbon materials (such as super P, Ketjen black, CNTs, graphene nanosheets, and reduced graphene oxide)^90–93 possessing desirable electronic conductivity, large surface area, light mass, and low cost, are widely employed in solid-state LABs as electronic conductive networks and catalysts. Moreover, the desirable interface with low interfacial impedance between carbon and SSE particles can be constructed by simple mechanical milling or heat treatment which is imperative in the fabrication of solid-state LABs.⁹⁴ We constructed a two-dimensional (2D) graphene–graphite composite layer on a LISICON electrolyte sheet via simple pencil drawing, which functioned as an air electrode for semi-solid-state LAB, which gave a discharge capacity of 950 mAh/g at 100 mA/g.⁹⁵ And we also proposed a kind of cathode composed of CNTs and LAGP (or its analog) (Figure 4A,B),^96,97 which was sintered at high temperature to form Li ion conducting networks, affording relatively low charge transfer resistance. Compared to multiwall CNTs (MWCNTs), single-wall CNTs (SWCNTs) were perceived with higher electrical conductance and fewer defects for reducing side reactions.⁹⁶ The assembled solid-state LAB with a cathode of SWCNTs/LAGP showed a larger capacity of ∼2800 mAh/g for the first cycle than the battery with MWCNTs/LAGP cathode (1700 mAh/g) (Figure 4C).⁹⁶

Figure 4. (A) Schematic illustration of the solid-state LAB composed of Li metal anode, LAGP SSE, and a cathode of SWCNTs and LAGP particles. Reproduced with permission: Copyright 2015, American Chemical Society.96 (B) The schematic of the solid-state LAB composed of Li metal anode, polymer SSE, and a cathode of MWCNTs/LAGP. Reproduced with permission: Copyright 2012, John Wiley & Sons Inc.97 (C) The comparison of discharge–charge curve for the solid-state LAB with SWCNTs/LAGP or MWCNTs/LAGP at a current density of 200 mA/g. Reproduced with permission: Copyright 2015, American Chemical Society.96 (D) Schematic demonstration of fabricating PI@IrOx NFs. (E) Cycling performance of the solid-state LAB working in dry-air (O2/CO2/N2) atmospheres at 500 mA/g with a limited capacity of 500 mAh/g for 600 h. Reproduced with permission: Copyright 2021, Elsevier Ltd.98 (F) The schematic of a solid-state LOB with carbon-free RBC cathode and LiOH reaction chemistry. (G) Discharge–charge profile of the proposed LOB in humidified O2, and (H) the corresponding partial electron yield spectra of cathodes at different stages during discharge/charge as the points shown in (G). Reproduced with permission: Copyright 2022, AAAS.99 LAB, Li–air battery; LAGP, Li1+xAlyGe2−y(PO4)3; LOB, Li–O2 battery; MWCNT, multiwall CNT; RBC, Ru-based composite; SSE, solid-state electrolyte; SWCNT, single-wall CNT.

Despite the advantages of carbon materials, air electrodes with carbon inevitably encounter carbon decomposition arising from the side reactions between carbon and discharge intermediates under high voltages. The accumulation of Li₂CO₃ from the reaction of carbon with Li₂O₂ could readily elevate charging overpotentials and irreversibly consume Li ions, bringing an unexpected decline in the energy efficiency and cycling performance of solid-state LABs. Thus, carbon-free catalysts such as Pt, Au, Ru, RuO₂, and transition metal oxides have been developed to eliminate this concern.^100–103 Liu et al.¹⁰⁴ synthesized a Co₃O₄ cathode with a porous urchin-like microstructure enabling adequate active sites for reactions and storage space for discharge products. The integration with a hybrid solid electrolyte (HSE) membrane composed of inorganic Li₇La₃Zr₂O₁₂ particles and poly(vinylidene fluoride-co-hexafluoropropylene) (PVDF-HFP), empowered a solid-state Li–O₂ battery (LOB) to achieve a high initial specific capacity of 8000 mAh/g, close to that of the battery with liquid electrolyte.¹⁰⁴ Kim's⁹⁸ group designed a hybrid cathode by depositing an electronically conductive and catalytic layer (Ir and outermost IrO_x) on porous polyimide nanofibers (PI@IrO_x NFs) (Figure 4D). The PI@IrO_x NFs supported the reversible conversion of discharge products as Li₂O₂ and Li₂CO₃ during cycling for more than 600 h in dry-air (O₂/CO₂/N₂) atmospheres (Figure 4E), and decent shape deformability for the application in flexible LABs.¹⁰³

An all-in-one electrocatalytic design for meeting the needs of multifunctional cathodes is essential for solid-state LABs but is a challenge. Recently, Kim et al.⁹⁹ reported an ingenious strategy to simultaneously enhance the capacity and reversibility of solid-state LOBs (Figure 4F). The carbon-free ceramic cathode built up on both electronically and ionically conductive Ru-based composite, and LiOH-based reaction chemistry triggered by the addition of water vapor (Figure 4G,H), rendered remarkable promotion of specific capacity and battery stability of 200 mAh/g over 665 discharge–charge cycles, while other reported cathodes achieved only ∼50 mAh/g and ∼100 cycles.¹⁰⁴ Sun et al.¹⁰⁵ realized the dual modulation of electronic and ionic microenvironment on a novel Li-decorated RuO₂ (Li–RuO₂) cathode with an amorphous structure, creating plenty of open frameworks with unsaturated sites and defects for promoting electronic–ionic transport, which benefited fast kinetics at the gas–solid interface for solid-state oxygen electrolysis (Figure 5G) and homogeneous distribution of discharge products on the cathode (Figure 5A–F). The assembled LOB with A–Li–RuO₂ exhibits a high specific capacity of 15,219 mAh/g at 100 mA/g and low polarization overpotential between discharge and charge (1.2 V), far superior to these values for C–Li–RuO₂ (11,900 mAh/g, 1.4 V) and rutile RuO₂ (7896 mAh/g, 1.6 V) (Figure 5H), as well as most reported solid-state LOBs (Figure 5I). These studies provide instructive exemplification of integrated design strategy for solid-state electrolysis in solid-state LABs. However, the LiOH chemistry with fast kinetics or other desirable reactions is highly relevant to interfacial mass transport and electrochemical environment, which are contingent on the properties of solid electrolytes.

Figure 5. SEM images of (A) A–Li–RuO2 and (B) C–Li–RuO2 cathodes after discharging. Synchrotron X-ray tomography reconstruction with 3D volume rendering images of (C) A–Li–RuO2 and (D) C–Li–RuO2 cathodes after discharging. The active 3D area of (E) A–Li–RuO2 and (F) C–Li–RuO2 cathodes. (G) Influence mechanism of mass transfer discrepancy for A–Li–RuO2 and C–Li–RuO2. (H) The discharge–charge profiles of solid-state LOBs with A–Li–RuO2, C–Li–RuO2, and rutile RuO2 cathodes, respectively. (I) Comparison of the discharge capacity and overpotential of the solid-state LOB with A–Li–RuO2 with some previously reported systems. Reproduced with permission: Copyright 2022, AAAS.105 3D, three-dimensional; LOB, Li–O2 battery; SEM, scanning electron microscopy.

Solid electrolyte design

An ideal solid electrolyte for LABs is characterized by high total ionic conductivity (10⁻²–10⁻³ S/cm), negligible electronic conductivity, wide electrochemical window (>5 V vs. Li/Li⁺), favorable compatibility with electrode materials, good thermal stability under operation temperature, as well as desirable cost efficiency and environmental benignity. For the case of solid-state LABs, indispensable durability to oxidative intermediates and air contaminants such as H₂O, CO₂, hydrocarbons, and so forth. inevitably leaves us limited options. Categorized by the migration mechanisms of Li⁺, there are three common solid electrolytes applicable for solid-state LABs, such as inorganic electrolytes, polymer electrolytes, and composite electrolytes.

Solid-state inorganic electrolytes

The developed solid-state inorganic electrolytes include sodium superionic conductor (NASICON),^106–108 garnet,^109,110 perovskite^111–113 and antiperovskite,^114,115 zeolites,^116,117 LiPON,¹¹⁸ as well as sulfides like Li₂S–P₂S₅ binary system^119,120 and thio-LISICON.^121,122 The advantages and disadvantages of these inorganic SSEs have been summarized in Table 1. The inorganic sulfides possess a notable preponderance of high ionic conductivity, favorable compatibility with Li anode, and wide electrochemical window, qualifying as a competitive choice in Li⁺ conductive solid electrolytes.^77,130 However, the high sensitivity to humidity accompanied by risky leakage of toxic H₂S, makes inorganic sulfides hard to be processed and utilized in the open air. Thus, inorganic oxides present as a more eligible choice for solid-state LABs with established supremacy of chemical stability in the air, along with decent ionic conductivity and electrochemical window, wherein NASICON, garnet, perovskite, antiperovskite, and zeolite are typical representatives.

Table 1 Properties of several typical oxide and sulfide inorganic SSEs.

Inorganic solid electrolytes	σLi (S/cm)	Advantages	Challenges	Ref.
NASICON	10−4‒10−3	Eligible σLi, stable in air, and good mechanical strength.	Unstable with Li anode.	[106–108, 123, 124]
Garnet	10−4‒10−3	Eligible σLi, stable with Li metal, and good mechanical strength.	Sensitive to humidity and CO2 high interfacial resistance.	[109, 110, 125, 126]
Perovskite LLTO	10−4‒10−3	Stable in air, has good mechanical strength, and low cost.	Unstable with Li anode, high grain boundary resistance.	[111–113]
Antiperovskites	10−4‒10−2	Stable with Li anode Li-rich and is light weight.	Poor structural and cycling stabilities.	[114, 115, 127–129]
Zeolites	10−6‒10−4	Stable in air, good interfacial compatibility.	Insufficient σLi.	[116, 117]
Amorphous Li2S–P2S5	10−4‒10−2	Stable with Li metal, high conductivity, and low grain boundary resistance	Sensitive to humidity and O2.	[119, 120]
Thio-LISICON	10−2	High σLi.	Unstable with Li anode, sensitive to humidity and O2.	[121, 122]
LiPON	10−6	Stable with Li metal, good mechanical strength.	Low σLi, expensive.	[118]

Abbreviation: LLTO, Li_3xLa_2/3−x□_1/3−2xTiO₃; SSE, solid-state electrolyte.

NASICON-type solid electrolytes origin from the discovery of fast Na⁺ transport in Na_1+xZr₂Si_xP_3−xO₁₂ (0 ≤ x ≤ 3) demonstrated by Goodenough et al.¹³¹ in 1976, in which the Na⁺ conductivity at 300°C reached up to 0.25 (Ω cm)⁻¹. As shown in Figure 6A, NASICON-type solid electrolyte displays a 3D skeleton with a rhombic unit cell composed of MO₆ octahedron and PO₄ tetrahedron by sharing vertex oxygen ions, giving an $$R\bar{3}c$$ space group with a general formula of AM(PO₄)₃ (A = Li, Na, or K, while M = Ge, Zr, or Ti). Depending on the distortion of the structure and the radius of the alkali metal ions, the alkali metal ions would occupy different positions and form ion migration channels between them, which constitute the ion conduction network of the NASICON structure. When the ionic conductor structure matches the size of the migrating ion, the ionic conductor can obtain the maximum diffusion coefficient and the lowest activation energy for delivering high ionic conductivity. Thus, in a Li⁺-conductive NASICON electrolyte, the mobility of Li⁺ is mainly controlled by the narrowest point located in the conduction path known as the “bottleneck.” LiTi₂(PO₄)₃ and LiGe₂(PO₄)₃ represent two excellent host structures with suitable structures for Li⁺ migration, while the σ_Li can be further improved via partial substitution of Ti⁴⁺/Ge⁴⁺ ions by divalent or trivalent cations such as Al³⁺, Ga³⁺, In³⁺, Fe³⁺, Cr³⁺, Zn²⁺, and Ca²⁺ to effectively widen the bottleneck.^84,132–134 The LATP ceramic commonly used in solid-state LABs has a high ionic conductivity of over 10⁻⁴ S/cm at room temperature.¹³² And even a maximum conductivity of 10⁻³ S/cm in polycrystalline Li_1.3Al_0.3Ti_1.7(PO₄)₃ prepared by field-assisted sintering was observed.¹²³ Similarly, the Li_1.5Al_0.5Ge_1.5(PO₄)₃ demonstrated high ionic conductivity of 2.4 × 10⁻⁴ S/cm at room temperature. Whereas, under the intimate contact of LATP or LAGP with Li anode, Al^3+, and Ge⁴⁺ in the electrolyte are prone to be reduced by Li metal to form Li–Al alloy or Li–Ge alloy while Ti⁴⁺ tends to be reduced into Ti³⁺, inducing an unstable structure and even electrolyte cracking with the change of internal stress.^106,124 In addition, the electrolyte decomposition could produce oxygen, which further reacts with Li metal and then renders intense thermal runaway.¹³⁵ For this reason, NASICON electrolytes should not be placed at excessive temperatures when used in Li-metal batteries. To resolve the contact problem, introducing an interlayer like liquid/polymer electrolyte and SEI has been certified as a practical strategy for stabilizing the electrolyte/Li anode interface. The details will be discussed in Sections 3.2.3 and 3.4.2.

Figure 6. Crystal structures of (A) NASICON-type, (B) garnet-type, (C) perovskite-type, and (D) zeolite-type SSEs. SSE, solid-state electrolyte.

The garnet structure is characterized by an $$Ia\bar{3}d$$ space group with the general formula of A₃B₂(XO₄)₃ (A = Ca, Mg, Y, or La, while B = Al, Fe, Ge, or Mn), wherein the cubic unit cell accommodating AO₈ distorted cube, BO₆ octahedron, and XO₄ tetrahedron (Figure 6B). When X sites are occupied by Li atoms, the garnet-type Li⁺ conductor as Li₃A₂B₂O₁₂ could form. To obtain high σ_Li, the valence of A³⁺ and B⁶⁺ could be intentionally modulated with excess Li⁺ stuffing for charge neutrality, providing Li₃ (Li₃Ln₂Te₂O₁₂, Ln = Y, Pr, Nd, Sm-Lu), Li₅ (Li₅La₂M₂O₁₂, M = Nb, Ta, Sb), Li₆ (Li₆Ala₂M₂O₁₂, A = Mg, Ca, Sr, Ba, while M = Nb, Ta), and Li₇ (Li₇La₃M₂O₁₂, M = Zr, Sn) types.¹²⁵ Li⁺ preferentially occupies the fourfold coordinated cation sites until the coordination number reaches three. But the tetrahedrally coordinated Li hardly participates in ion migration, resulting in low ionic conductivity of Li₃-type garnets, such as Li₃Nd₃Te₂O₁₂ delivering 1 × 10⁻⁵ S/cm at 600°C merely and high activation energy of 1.22 eV.¹²⁵ In 2003, Thangadurai et al.¹²⁶ reported the Li₅-type garnet as a novel family of fast Li⁺ conductor, namely Li₅La₂M₂O₁₂ (M = Nb, Ta), in which three 4-fold coordinated Li⁺ and two 8-fold coordinated Li⁺ form 3D ion migration channels, presenting decent σ_Li of ≈1 × 10⁻⁶ S/cm at room temperature. However, a practical SSE for Li batteries works on the premise of qualifying a σ_Li of no less than 10⁻⁴ S/cm. In 2007, Murugan et al.¹³⁶ explored a sheet-like Li₇-type garnet, that is, Li₇La₃Zr₂O₁₂ (LLZO), in which the Li atoms located at the tetrahedral site and distorted octahedral site, respectively, formed a loop structure accommodating the 3D network of Li⁺ migration pathway and then achieved superior σ_Li of 7.74 × 10⁻⁴ S/cm at room temperature and low activation energy of 0.34 eV. High-valence dopants have been intensively studied to increase Li vacancy concentrations and reduce short-range ordering for enhancing σ_Li. The highly improved σ_Li close to 1–2 mS/cm for Ta/Te/Ga-doped LLZO like Li_6.4La₃Zr_1.4Ta_0.6O₁₂ (1.0 × 10⁻³ S/cm),¹³⁷ Li_6.5La₃Zr_1.75Te_0.25O₁₂ (1.02 × 10⁻³ S/cm),¹³⁸ and Li_6.55Ga_0.15La₃Zr_2.5O₁₂ (2.06 × 10⁻³ S/cm)¹³⁹ has been observed. Despite the desirable σ_Li, the susceptibility to humidity and CO₂ of garnet-type Li⁺ conductors inevitably produces a carbonaceous layer and then increases the interfacial resistance, and unexpected formation of Li dendrites originating from uneven interfacial contact. The introduction of a transition layer with strong chemical bonding energy to Li metal can remarkably alleviate the interfacial problem as demonstrated in Sections 3.2.3 and 3.4.2.

Perovskite-type ionic conductors, ideally presenting a $$Pm\bar{3}m$$ space group with the general formula of ABO₃ (A = Ca, Sr, La; B = Al, Ti), comprise cubic unit cells in which A ions and B ions take 12-fold and 6-fold coordinated sites, respectively (Figure 6C). The elements doping brings Li⁺ into perovskite and achieves a type of Li⁺ conductor, namely Li_3xLa_{2/3−x□1/3−2x}TiO₃ (LLTO). The migration of Li⁺ proceeds along the A-site vacancies and goes through the bottleneck containing four oxygen-adjacent ions. Thus, the type of introduced A-site cations, Li⁺ and/or vacancy concentration as well as the interaction between them, have a prominent impact on the σ_Li of LLTO.²⁷ The enhanced σ_Li of LLTO can be obtained via controlling the concentrations of vacancies and Li⁺ ions, with enlarged bottleneck when A sites are partially occupied by rare earth ions or alkaline earth ions possessing large radius. For example, Li et al.¹⁴⁰ applied a codoping strategy of Ta⁵⁺ and Sr²⁺, realizing four-fold enhancement of σ_Li from 3.7 × 10⁻⁵–1.4 × 10⁻⁴ S/cm in LLTO. Although elemental doping can improve the crystal conductivity of LLTO, its grain boundary problem is basically unresolved and becomes the most intractable issue, which not only deteriorates total σ_Li but also becomes the inducement for Li-dendritic propagation.^84,141 However, it is hard to distinguish the origin of the grain-boundary core as the discrepancy in defect segregation energy between the bulk and grain boundaries, or the formation of an insulating phase at the grain boundaries since its extremely small width of 2–3 unit cells.^142,143 Effective tools for ascertaining the origin are urgently needed to design practical LLTO keeping from the poor grain-boundary transport. In addition, like LATP, the instability of LLTO toward the Li anode stemming from the reduction of Ti⁴⁺ into Ti³⁺, further limits the application of LLTO in solid-state LABs. In 2012, halogen-based Li-rich antiperovskites with a general formula of Li₃OA (A = Cl or Br) were proposed as high-performance superionic Li⁺ conductors, showing a high σ_Li of more than 1 × 10⁻³ S/cm at room temperature and low activation energy of 0.2–0.3 eV.¹²⁸ The simulation results of ab initio molecular dynamics indicated that antiperovskites with perfect crystal structures were not desirable Li⁺ conductors, while Li vacancies and structural disorders facilitated Li⁺ migration by lowering the enthalpy barriers along favorable pathways, rendering high σ_Li.¹²⁹ This implies that the σ_Li of antiperovskites can be rationally tuned by Li vacancy concentration and structural disorder. Li et al.¹⁴⁴ transformed Li₂OHCl antiperovskites from an orthorhombic phase to a cubic phase via partially substituting OH⁻ by F⁻, which increased the tolerance factor for favoring disordered OH⁻ orientation. In this way, the rotations of relevant hydrogen bonds get promoted and then decrease the activation energy for a Li⁺ transfer to an adjacent vacancy site.¹⁴⁴ Whereas, the Li₃OA antiperovskites also face stability problems due to their thermodynamical metastability,¹²⁹ as Li₃OCl could decompose into Li₂O₂, LiCl, and LiClO₄ when the applied voltage surpasses 2.5 V.¹¹⁴ Thus, effective strategies for optimizing the stability of antiperovskites is essential to bring their superiority of superionic conductivity into full play.

Zeolites emerge as a novel type of SSEs for solid-state LABs as first explored by Chi et al.^27,116 As shown in Figure 6D, the structure of zeolites is characterized by corner-sharing tetrahedra of Al, Si, and P joined by oxygen bridge bonds as the primary building units, giving an open framework with well-defined micropores.¹⁴⁵ Benefiting from the superiority of ordered micropores and incessant ion-migration pathway, as well as intrinsic stability toward Li metal and air atmosphere, zeolites demonstrate fascinating compatibility in solid-state LABs.^116,146 The Li⁺ exchanged zeolite membrane (LiXZM) presents a high σ_Li of 2.7 × 10⁻⁴ S/cm and a low electronic conductivity of 1.5 × 10⁻¹⁰ S/cm, accompanied by the high compatibility to Li anode and air, provided an all-in-one strategy for solving interfacial issues, Li dendrites, and stability concerns in solid-state LABs.¹¹⁶ A solid-state LAB was integrated by in situ assembly of LiXZM with cast lithium and CNTs (Figure 7A) and delivered an ultrahigh capacity of 12,020 mAh/g (Figure 7B) and an extended cycle life of 149 cycles at 500 mAh/g with a fixed capacity of 1000 mAh/g in ambient air, much outperformed LAGP-based solid-state LABs of 13 cycles (Figure 7C). Owing to favorable flexibility and electrochemical performance, the solid-state LAB with integrated cathode and LiXZM (C-LiXZM) showed a promising prospect for practical energy storage devices (Figure 7D).

Figure 7. (A) Schematic illustration of the assembled solid-state LABs with LiXZM SSE and the conduction mechanism. (B) Specific capacities and (C) cycling performance of the solid-state LAB with integrated C-LiXZM, nonintegrated cathode, and LiXZM (C|LiXZM) or LAGP (C|LAGP) at 500 mAh/g with a fixed capacity of 1000 mAh/g. (D) Safety, abuse tolerance, and flexibility of the solid-state LABs with C-LiXZM. Reproduced with permission: Copyright 2021, Springer Nature.116 CNT, carbon nanotube; LAB, Li–air battery; LAGP, Li1+xAlyGe2−y(PO4)3; LiXZM, Li+ exchanged zeolite membrane; SSE, solid-state electrolyte.

To sum up, inorganic oxides including NASICON, garnet, perovskite, antiperovskite, and zeolite are typical SSEs for solid-state LABs. Compared with perovskite and antiperovskite, NASICON and garnet SSEs normally possess higher σ_Li, while the instability toward Li metal and/or air arouses more interfacial problems to be solved. Structural modification through optimizing bottlenecks underpinned by element doping, building up 3D ion migration channels, as well as modulating Li⁺ and/or vacancy concentrations is an acknowledged strategy to obtain enhanced σ_Li and structural stability in inorganic SSEs. However, the guiding criteria are still absent that require extensive studies on different crystal systems from both theoretical and experimental points. The successful utilization of zeolites paves a new way to construct practical SSEs with high σ_Li and intrinsic stability. But the in-depth research on conductive behaviors related to different cage sites, and the balance between mechanical strength and conductivity of zeolite SSEs are indispensable.

Polymer electrolytes

Compared with inorganic SSEs, polymer electrolytes exhibit the superiority of better interfacial contact, high processing ability, cost efficiency, and mechanical flexibility, thus gain intensive research for various batteries. In 1996, Abraham and Jiang¹⁴⁷ first proposed a LOB with polyacrylonitrile (PAN)-based solid polymer electrolyte (SPE) integrated with a Li foil anode and a carbon composite cathode, realizing a battery package with stacked band-aid-type strips showing a specific energy of 250‒350 Wh/kg. Typically, the SPEs can be fabricated by dissolving Li salts such as Li bisfluorosulfonimide (LiFSI), Li bis(trifluoromethanesulfonyl)imide (LiTFSI), or Li trifluoropotassium sulfonate (LiCF₃SO₃) into polymer hosts, such as PAN, poly(ethylene oxide) (PEO), poly(vinylidene fluoride) (PVDF), PVDF-HFP, poly(methyl methacrylate) (PMMA), poly(vinyl chloride) (PVC), poly(vinylpyrrolidone) (PVP), poly(tetrafluoroethylene) (PTFE), and so forth.

As the first discovered SPE with conductive behavior of alkali ions,¹⁴⁸ PEO has shown several advantages such as strong Li⁺ solvating ability, preferred dimensional stability, and mechanical properties, thus has been widely explored in solid-state LABs (Table 2). For instance, Balaish et al.¹⁶⁶ reported a PEO-based LOB displaying a higher discharge voltage (≈80 mV) and a lower charge voltage (≈400 mV) at 80°C than the liquid-state LOB. However, PEO is with a semi-crystalline structure, wherein the amorphous phase with activated chain segments above the glass transition temperature (T_g) supports Li⁺ itineration through continuous Li‒O bonds formation and breaking. Therefore, typical linear PEO exhibits insufficient σ_Li based on the high crystallinity, which hardly provides free volume for ionic transition with a stiff structure, especially at low temperatures.¹⁶⁷ Apparently, decreasing the crystallinity of PEO is an effective strategy to construct a desirable SPE with enhanced σ_Li. A series of approaches such as the introduction of plasticizers^168–171 or nanofillers^172,173 containing large anionic groups, polymer blending,^174–176 and designing block copolymers^177,178 have been implemented to improve the σ_Li of PEO-based SSEs in LIBs while not gotten studied in LABs yet. Meanwhile, PEO-based SPEs are readily oxidized by the discharge products of LABs, inevitably generating stability concerns. The auto-oxidation of PEO in an oxygenated environment arising from the accelerated radical formation at the applied potential higher than OCV has been validated by Harding et al.¹⁵⁰ They also studied the chemical stability of some common polymers utilized in LABs in the presence of Li₂O₂.¹⁷⁹ Of the polymers researched, PEO may suffer from some cross-linking, while PAN containing electrophilic nitrile group allowed nucleophilic attack, and the halogenated polymers (PVC, PVDF, and PVDF-HFP) underwent dehydrohalogenation reactions with the existence of Li₂O₂, showing a reactivity tendency of PAN>>PVC≈PVDF>PVDF-HFP>>PVP. Meanwhile, PMMA with methyl and methoxy functionalities contributed to the reduction of potential reaction pathways, while PTFE and perfluorosulfonic acid resin (Nafion) with a lack of α-H or β-H adjacent to the electron-withdrawing fluoride groups, allowed the superior stability toward the nucleophilic environment of LABs. This study provides a guideline for the bottom-up design of stable polymers for LABs characterized by the inhibition of general reactivity toward functional groups with nucleophiles. But it is more complicated when the SPEs work in a practical condition instead of Li₂O₂ alone, like stable PMMA is challenged by accelerated decomposition by any impurity such as water in the cell, with nucleophilic substitution by strong bases like O₂^•− at the carbonyl center.^179–181 Thus, the stability mechanisms of these polymers under working conditions are expected to be adequately understood, which requires valid in situ evidence but remains challenging.

Table 2 Summary of recent progress in polymer electrolytes for solid-state LABs.

Electrolyte type	Polymer matrix	Year	Ionic conductivity (S/cm)	Electrochemical window	Cycle life	Ref.
SPE	PEO	2015	10−3	4.9 V	20 Cycles/500 mAh/g	[149]
SPE	PEO	2015	–	–	5 Cycles/1000 mAh/g	[150]
SPE	PEO	2018	10−3	4.5 V	40 Cycles/42 mAh/g	[151]
SPE	Li-Nafion	2021	–	–	182 Cycles/500 mAh/g	[152]
GPE	PVDF-HFP	2017	–	4.8 V	400 Cycles/1000 mAh/g	[153]
GPE	ETPTA	2017	1.4 × 10−3	4.5 V	55 Cycles/500 mAh/g	[154]
GPE	PVDF-HFP	2018	1.01 × 10−4	–	100 Cycles/500 mAh/g	[155]
GPE	PVDF-HFP	2018	4.2 × 10−4	–	553 Cycles/100 mAh/g	[156]
GPE	PVFM	2018	10−3	5.0 V	150 Cycles/1000 mAh/g	[157]
GPE	PVDF-HFP	2019	–	–	95 Cycles/1000 mAh/g	[158]
GPE	DES-PE	2021	1.19 × 10−3	5.0 V	300 Cycles/500 mAh/g	[159]
GPE	PVDF-HFP	2021	10−7–10−4	4.9 V	16 Cycles/100 mAh/g	[160]
GPE	BMA-PEGDA	2021	–	–	35 Cycles/100 mAh/g	[161]
GPE	PVDF-HFP	2022	1.06 × 10−3	–	194 Cycles/1250 mAh/g	[162]
HSE	PE	2016	–	5.2 V	100 Cycle/1000 mAh/g	[163]
HSE	PVDF-HFP	2020	–	–	146 Cycles/1000 mAh/g	[164]
HSE	PMMA	2020	1.06 × 10−3	–	194 Cycles/1250 mAh/g	[165]

Abbreviations: BMA, buthyl methacrylate; DES-PE, deep eutectic solvent-based polymer electrolyte; ETPTA, ethoxylated trimethylolpropane triacrylate; GPE, gel polymer electrolyte; HSE, hybrid solid electrolyte; LAB, Li–air battery; PE, polyethylene; PEGDA, polyethylene glycol diacrylate; PEO, poly(ethylene oxide); PMMA, poly(methyl methacrylate); PVDF-HFP, poly(vinylidene fluoride-co-hexafluoropropylene); SPE, solid polymer electrolyte.

With a low to moderate cross-linking density, SPE can be swollen by a plasticizer in abundance presenting as a quasi-solid-state gel polymer electrolyte (GPE). Balanced with the mechanical preponderance of a polymer network and the high ionic conductivity of liquid electrolytes, these GPEs usually provide high σ_Li, low interfacial resistance, and preferred flexible properties while avoiding the evaporation and leakage of liquid solvents, as well as effectively protect Li anode from air contaminants in an open system. In previous research studies, many host polymers as PAN, PMMA, PVDF, PVDF-HFP have been explored in GPE-based LABs (Table 2).^182–185 Among them, PVDF-HFP possesses lower crystallinity and T_g, superior chemical and mechanical stability, and thus gets intensively studied. Of the developed plasticizers, ionic liquid (IL) electrolytes can induce a more stable and conductive SEI on the surface of the Li anode accompanied with high hydrophobicity against water crossover, thus promising more eligible GPEs to be used in open environment.¹⁸² For example, Zhang et al.¹⁸⁶ reported a combination of IL (PMMITFSI), silica, and PVDF-HFP polymer matrix, enabling a GPE with high σ_Li (1.83 × 10⁻³ S/cm) and good Li anode protection. Amanchukwu et al.¹⁸³ presented a pioneering work on the control of oxygen reduction chemistry (from 2e to 1e mechanism) via the formation of IL-superoxide complexes in a IL (PYR₁₄TFSI)-LiTFSI-PMMA GPE. According to Pearson's hard soft acid base theory, they proposed that Li₂O₂ would form if Li⁺ presented when O₂ was reduced to O₂^•− (hard acid–hard base), or the IL-superoxide complexes would govern the consequent process (soft acid–soft base) (Figure 8A). This study indicates an energy-intense system for LABs and expands the future for Na–air batteries and K–air batteries. Recently, Hoffknecht et al.¹⁸⁷ pointed out that the Li⁺ transport in GPE with TFSI-based ILs still followed a sluggish chain pathway as solid PEO, while the ILs with more coordinating anions like TFSAM competing with PEO for Li⁺ solvation, could achieve higher Li⁺ transport efficiency and had been applied in LIBs. This research provides a new sight for design high-performance GPEs, and may inspire future advances in the field of LABs. In addition, the RMs are also applied to increase the energy efficiency and prolong the cycle life of GPE-based LABs. Fu et al.¹⁸⁸ constructed a GPE with gradient distributed RMs (LiI, TMPD, or DMPZ) through  in situ cross-linking of liquid glycol dimethyl ether (TEGDME, G4) containing LiClO₄ and RMs with lithium ethylenediamine modified Li anode. The RM-free part tightly adhered to the Li anode and the RM-containing part enriched at the cathode, effectively inhibited the direct contact of Li and RMs while assured their high mobility for participating in cathode reaction. Ren et al.¹⁸⁹ got inspired by the catalytic behavior of superoxide dismutase (SOD) on the dismutation of O₂^•− in biological systems, then applied oxidized activated carbon (OAC) serving as a SOD mimetic. As shown in Figure 8C, the redox cycle of OAC enabled efficient reduction of soluble LiO₂ into LiOH instead of Li₂O₂ (Figure 8B), which could mitigate side reactions triggered by soluble LiO₂, and thereby achieved prolonged life span. As a promising power resource for wearable devices, flexible LABs based on bendable GPEs provide alluring advantages of high energy density, favorable safety, and deformation ability.^190–198 Wang et al.¹⁹² integrated a rippled air electrode composed of aligned CNT sheets stacked on a GPE, and a Li array electrode, endowing a highly flexible fiber-type LAB with high electrochemical performance under stretching, bending, or twisting condition (Figure 8D–H).

Figure 8. (A) The oxygen reduction process leaded by IL-superoxide complexes (1e process) or Li+ (2e process). Reproduced with permission: Copyright 2020, Oxford University Press.183 (B) Discharge process of O2 to Li2O2 in LOBs following solution and surface mechanisms. (C) Discharge process of O2 to LiOH in LOBs with OAC-involved GPEs. Reproduced with permission: Copyright 2021, American Chemical Society.189 Discharge curves of the fiber-type flexible LAB under (D) stretching with increasing strains, (E) bending or (F) twisting with increasing angles, respectively. (G) The discharge curve of the flexible LAB under the dynamic stretching/releasing condition with increasing speeds. (H) Dependence of discharge voltage plateau before and after deforming (V0 and V, respectively) when the LAB is stretched at a strain of 75%, bended at an angle of 90° and twisted at an angle of 180°, respectively. Reproduced with permission: Copyright 2016, Royal Society of Chemistry.192 GPE, gel polymer electrolyte; IL, ionic liquid; LAB, Li-air battery; LOB, Li–O2 battery; OAC, oxidized activated carbon.

Despite the attractive merits of polymer electrolytes on low cost, high processing ability, and favorable tolerance to volume change, their application is retarded by low σ_Li, matrix decomposition, and so forth. The polymer-in-salt¹⁹⁹ concept could be a feasible strategy to increase the σ_Li of polymer electrolytes with T_g well below the ambient temperature but has not been explored in LABs yet. The most studied solution for a practical solid electrolyte is constructing the hybrid inorganic/polymer system, which will be overviewed below.

Hybrid solid electrolytes

As previously discussed, the inorganic ceramic electrolytes are characterized by good thermal stability, desirable Young's modulus, and superior ionic conductivity, but their rigidity results in untight contact with electrodes rendering high interfacial resistance. Meanwhile, the GPEs offer excellent flexibility and good wettability toward electrodes, but inferior ionic conductivity and low mechanical strength to stop Li dendrite penetration. Obviously, both an inorganic ceramic electrolyte and a GPE alone cannot meet all expectancies for an ideal SSE while the two are complementary. Thus, integrating the HSE with an inorganic SSE and a GPE is a promising strategy for taking advantage of both sides while remedying their shortages mutually.^{163–165,200–202}

An eligible HSE should be qualified with (i) indispensable stability in O₂ or air,²⁰³ (ii) a rigid inorganic component with 3D ion transport channels covering the surface of Li anode evenly to enable uniform Li⁺ distribution and high mechanical strength,^204,205 (iii) adequate flexibility and smoothness to ensure tight contact with electrodes,¹⁶⁴ and (iv) homogeneous distribution of inorganic electrolyte and GPE to avert local disorder of Li⁺.^206–208 Here, some paradigm designs based on typical SSEs such as NASICON-type LAGP, garnet-type LLZO, perovskite-type LLTO, and amorphous SSE are illustrated for in-depth understanding.

Wang et al.¹⁶⁴ fabricated a novel HSE with a rigid nanosized LAGP core and ultrathin flexible PVDF-HFP shell. The force-separation curve certified that the surface of LAGP solid particles was homogeneously wrapped by a soft PVDF-HFP polymer with a thickness of around 5 nm (Figure 9A–C). Apparently, the core-shell interface strongly hinged on the ratio of LAGP to PVDF-HFP polymer as well as the size of LAGP. To understand the interface mechanism of LAGP@PVDF-HFP, three control groups characterized by the high-content and micro-sized LAGP (HSE-I), low-content and nano-sized LAGP (HSE-II), and GPE in the absence of LAGP were employed. The symmetric Li/Li battery with LAGP@PVDF-HFP exhibited the longest cycle life (Figure 9F) without distinct Li dendrites formation (Figure 9E). The mechanism study suggested that the nanocrystallization of LATP contributes to a full-time uniform distribution of Li⁺ while HSE-I suffers from the later formation of sphere dendrites; and the homogeneous coverage of PVDF-HFP on LAGP ensures tight and stable Li anode/HSE interface, while the insufficient density of HSE would induce local Li⁺ disorder rendering more severe dendrite problem than GPE (Figure 9D,E). As a demonstration, the quasi-solid-state LOB with LAGP@PVDF-HFP afforded a long cycle life (146 cycles) at 300 mA/g with a fixed capacity of 1000 mAh/g (Figure 9G), and markedly outperformed the control groups.

Figure 9. (A) The force-separation curve of LAGP@PVDF-HFP HSE. (B) Atomic force microscopy topography of LAGP@PVDF-HFP HSE and (C) the corresponding force-separation curves of six points. (D) Schematic diagram of Li deposition, (E) SEM images of Li anode after (F) the cycling tests derived from the symmetric Li/Li batteries, and (G) the cycle life of LOBs with different electrolytes as LAGP@PVDF-HFP HSE, HSE-I, HSE-II, and GPE, respectively. Reproduced with permission: Copyright 2020, Oxford University Press.164 GPE, gel polymer electrolyte; HSE, hybrid solid electrolyte; LAGP, Li1+xAlyGe2−y(PO4)3; LOB, Li–O2 battery; PVDF-HFP, poly(vinylidene fluoride-co-hexafluoropropylene); SEM, scanning electron microscopy.

Zhao et al.¹⁶⁵ constructed an HSE comprising Li_6.4La₃Zr_1.4Ta_0.6O₁₂ (LLZTO) SSE with 3D porous garnet microstructure (PSSE) infused with PMMA-based GPE (denoted as PSSE/GPE) (Figure 10E). The PSSE/GPE accommodated a rigid backbone for suppressing Li dendrites, and the consecutive GPE in pores contributed a high σ_Li (1.06 × 10⁻³ S/cm) and sufficient compactness for blocking O₂ crossover. Meanwhile, the flooded GPE on the surface of PSSE/GPE endowed small interfacial resistance and high compatibility with the Li anode. Benefiting from these advantages, the assembled LOB with PSSE/GPE delivered a long cycle life of 194 cycles, which was far superior to that with GPE (Figure 10F). Le et al.²⁰¹ proposed an assembly of PVDF-HFP polymer with Al-doped LLTO (A-LLTO) particles covered with a modified SiO₂ (m-SiO₂) layer (Figure 10D), which effectively protected A-LLTO from chemical decomposition while enhancing its adhesion to PVDF-HFP. The obtained HSE achieved a high σ_Li of 1.22 × 10⁻³ S/cm and remarkable suppression of the growth of Li dendrites compared to GPE (Figure 10B,C). Compared to highly crystallized inorganic SSEs, the amorphous SSE with low grain boundary resistance is preferred to provide higher Li⁺ diffusivity. Thus, our group¹⁶³ combined the amorphous LiNbO₃ with poly(methylmethacrylate-styrene) on a polyethylene (PE) substrate, realizing a quasi-solid-state HSE with desirable thermal stability keeping from shrinkage or degradation even after heating at 120°C for 1 h (Figure 10A). The stable interfacial resistance supported by this HSE effectuated a stable cycle performance of the assembled LOB over 100 cycles.

Figure 10. (A) Shrinkage of PE and prepared LiNbO3/PMS/PE HSE when heated at 120°C for 1 h. Reproduced with permission: Copyright 2016, John Wiley & Sons Inc.163 The growth process of Li dendrites in a symmetric Li/Li battery with (B) pure GPE or (C) A-LLTO/PVDF-HFP HSE at 0.5 mA/cm2. (D) Schematic illustration of the LOB with A-LLTO/PVDF-HFP HSE. Reproduced with permission: Copyright 2016, American Chemical Society.201 (E) Schematic illustration of the processing of LLZTO/PMMA HSE (PSSE/GPE). (F) Cycling stability of the LOBs with PMMA-based GPE and PSSE/GPE, respectively, with a limited capacity of 1250 mAh/g at 312.5 mA/g. Reproduced with permission: Copyright 2020, American Chemical Society.165 GPE, gel polymer electrolyte; HSE, hybrid solid electrolyte; LLTO, Li3xLa2/3−x□1/3−2xTiO3; LLZTO, Li6.4La3Zr1.4Ta0.6O12; LOB, Li–O2 battery; PMMA, poly(methyl methacrylate); PMS, poly(methylmethacrylate-styrene); PSSE, porous garnet microstructure; PVDF-HFP, poly(vinylidene fluoride-co-hexafluoropropylene).

Very recently, Kondori et al.²⁰⁹ presented an intriguing study on an HSE that Li₁₀GeP₂S₁₂ (LGPS) nanoparticles embedded in a PEO matrix, realizing a four-electron reaction with Li₂O formation under room temperature. In this HSE, LGPS chemically bonded with a silane-coupling agent, mPEO-TMS, which protected LGPS from decomposition at the electrode/SSE interfaces. The Li₂O discharge product was detected with increasing peak intensity over 1 h of discharging in the in situ Raman spectroscopy (Figure 11A,B). This four-electron reaction was confirmed by acid titration coupled with ultraviolet–visible spectroscopy and ex situ differential electrochemical mass spectroscopy (DEMS) (Figure 11C), and its reversibility was further elucidated by in situ DEMS result (Figure 11D). The X-ray diffraction patterns also indicated the highly reversible Li₂O electrochemistry while discharge and charge, and the small amount or amorphous nature of LiO₂ and Li₂O₂ products (Figure 11E). The authors proposed that the four-electron reactions were driven by the abundance of O₂ at the interface, excess Li⁺ originating from high σ_Li of HSE, and mixed electron-conductor behavior at the cathode, but the lack of detailed mechanism study. The battery was rechargeable for 1000 cycles (Figure 11F) with a low polarization gap that increased from 50 mV initially to ≈430 mV at the end, while the energy efficiency gradually dropped from 92.7% to 87.7% (Figure 11G).

Figure 11. (A) In situ Raman spectroscopy experiments and (B) relative Raman peak intensities at different time intervals during discharge with a limited capacity of 125 mAh/g at 1 A/g. Calculated electron transfer number using (C) acid titration coupled with UV–vis spectroscopy (inset: ex situ DEMS result) during discharge with a limited capacity of 1 Ah/g at 1 A/g, and (D) in situ DEMS during charge with a limited capacity of 5 Ah/g at 1 A/g. (E) XRD patterns of the discharged/charged cathodes at different cycle numbers during (F) the galvanostatic cycling over 1000 cycles with a limited capacity of 1 Ah/g at 1 A/g. (G) The Coulombic efficiency, energy efficiency, and polarization gap during cycling. Reproduced with permission: Copyright 2023, AAAS.209 DEMS, differential electrochemical mass spectroscopy; UV–vis, ultraviolet–visible; XRD, X-ray diffraction.

These pioneering research studies validate the advanced nature of HSEs over individual inorganic SSEs and GPEs. However, a practical HSE is not a simple mixture with uncertain structure, but a well-designed integration with a homogeneous distribution of the hybrid structure for stable and efficient interfacial behaviors, adequate compactness for blocking O₂ crossover and avoiding local disorder of Li⁺, as well as sufficient rigidity for suppressing Li dendrites. As a result, the construction of HSEs is challenged by structural control, processing difficulty, and a balance between the merits of inorganic SSEs (high σ_Li, thermal stability, and Young's modulus) and GPEs (preferred flexibility and wettability). When faced with future commercialization, the overall cost and air compatibility should also be vital concerning factors.

Li anode optimization and protection

As one kind of Li metal battery, undoubtedly, solid-state LAB dictates the need for a robust and chemically stable Li anode that is qualified for curbing Li dendrites formation and bearing the volume change while sequent Li plating/stripping. Meanwhile, the half-open structure of solid-state LAB introduces concerns about Li anode corrosion and mechanical strength decay, induced by the permeation of air contaminants such as H₂O and CO₂ into the free volume of SSE matrices. Specifically, a small amount of H₂O can trigger a fast N₂-attacking reaction (Equations 10 and 11) to form Li₃N and LiOH.²¹⁰ [Image Omitted. See PDF] [Image Omitted. See PDF]

Several strategies have shown alluring prospects for Li anode optimization and protection, such as Li alloying for elevating anode stability, fabrication of super-hydrophobic SSEs or introducing a functional layer to avoid anode corrosion, and construction of preferred interface environment for realizing low interfacial resistance and inhibited dendrite growth (see details in Section 3.4.2).

Introducing alloying elements was found to effectively diminish the discrepancy in surface energy of Li anode and SSEs, contributing to higher wettability and enhanced (electro)chemical stability at the anode side (Figure 12A).²¹¹ The Li alloys such as Li–In,^212–216 Li–Si,^217–221 Li–Al,^222–224 and Li–Sn^{213,214,225–227} have been widely studied as advanced substitutes for Li metal anodes in solid-state batteries, while Li–Si²²⁸ and Li–Al²¹⁰ also have shown preponderance over Li metal in liquid-state LABs with suppressed O₂/air-attacking-induced polarization. However, the alloying with inactive elements would inevitably sacrifice the specific capacity of the Li anode, exacerbating the limitation of anode capacity in LABs. Herein, Ma et al.²²⁹ proposed a Li–Na alloy serving for a novel aprotic bimetal Li–Na–O₂ battery with enhanced cycling stability (Figure 12B). The electrostatic shield effect that the adsorbed Li⁺ ions on the tips were prone to exclude incoming Na⁺ and force Na⁺ to deposit away from the tips, successfully endowing the suppression of anode dendrites and cracks. This research provides a paradigm shift in designing Li alloy anodes but needs to be further explored in solid-state LABs.

Figure 12. (A) Schematic illustration of the wettability between Li/Li alloy and substrates. Reproduced with permission: Copyright 2017, John Wiley & Sons Inc.211 (B) Cycling stability of the bimetal Li–Na–O2 battery with a limited capacity of 1000 mAh/g at 200 mA/g. Reproduced with permission: Copyright 2018, Springer Nature.229 (C) The schematic of the working mechanism for the LAB with or without LDPE film, and (D) the corresponding XRD patterns of Li2O2 under different conditions. Reproduced with permission: Copyright 2017, John Wiley & Sons Inc.190 (E) Schematic illustration of the solid-state LOBs in a humid atmosphere with common LiSICON film or well-designed super-hydrophobic SSE. Reproduced with permission: Copyright 2016, John Wiley & Sons Inc.230 (F) Digital photograph and (G) the Fourier-transform infrared spectroscopy plots presenting the corrosion resistance of Li metal with PS-QSE before and after exposure to air for 30 min. Reproduced with permission: Copyright 2019, John Wiley & Sons Inc.194 LAB, Li–air battery; LDPE, low-density polyethylene; LOB, Li–O2 battery; QSE, quasi-solid electrolyte; SSE, solid-state electrolyte; XRD, X-ray diffraction.

The introduction of super-hydrophobic SSEs or functional layer is with a direct effect on repelling the permeation of H₂O. For example, our group²³⁰ constructed a super-hydrophobic SSE effectuated by the combination of super-hydrophobic SiO₂ matrix and Li⁺ conducting ILs (Figure 12E). The quasi-solid electrolyte (QSE) showed a decent σ_Li of 0.91 × 10⁻³ S/cm, excellent thermal stability below 230°C, and super hydrophobicity with a contact angle over 150°, underpinning a safe and long-life solid-state LOB in a humid atmosphere. Similarly, Shu et al.¹⁹⁴ developed a QSE (PS-QSE) with SiO₂ filler and PVDF-HFP matrix, enabling desirable corrosion resistance of Li metal (Figure 12F,G) and improved anodic reversibility for over 200 cycles. Wang et al.¹⁹⁰ designed a low-density polyethylene (LDPE) film (Figure 12C) with low H₂O permeability (0.825 g/(m² d)) while high O₂ permeability (40.3 Barrer). The favorable selectivity could effectively restrain Li corrosion induced by H₂O, and impede the Li₂CO₃ formation in ambient air (Figure 12D).

Although the protective strategies mentioned above markedly retard the Li corrosion caused by humidity and undesirable gas crossover, they arouse some side effects as well, such as the sacrifice of energy density and high interfacial resistance. Constructing a stable Li metal/SSE interface with homogeneous Li⁺ distribution for inhibiting dendrite growth, and sufficient compactness for blocking contaminants is a more promising way, which will be discussed in Section 3.4.2.

Interfacial engineering

Except for the essential components for LABs, the chemical, electronic, and mechanical properties of the involved interfaces in a solid-state LAB are of critical importance for determining its long-term electrochemical performance and viability. As previously mentioned, the solid-state LABs accommodate four kinds of interfaces as (i) triple-phase interfaces for ORR and OER, (ii) voids forming via the untight contact of electrodes and SSEs, (iii) grain boundaries in SSEs, and (iv) the electrode/SSE interfaces. Due to the poor wettability, excessive stiffness, and/or inferior Li⁺ conductive behavior of SSEs, the solid-state LABs face with more intractable interface problems than liquid-state LABs, albeit decreased influence from the crossover of discharge products and air contaminants. Especially, the high interfacial resistance between cathode/SSE and Li anode/SSE, as well as the instability of SSE toward oxidative products or Li anode toward O₂, H₂O, CO₂, and so forth requires rational strategies for interface optimization. Here, these strategies will be categorized and discussed for cathode/SSE interfaces and Li anode/SSE interfaces.

Interface optimization between cathode and electrolyte

The quantity, availability, and stability of triple-phase interfaces dominate the overall performance of air cathodes, while challenged by inadequate contact with SSE, sluggish Li⁺ transfer, and structural collapse of cathode/SSE interface. Except for constructing cathodes with high electrocatalytic performance and desirable physical structure as discussed in Section 3.1, improving the interfacial conductivity and contact of cathode/SSE interface through constructing abundant triple-phase boundaries, adding the electrolyte with high σ_Li, cathode coating, and so forth is an effective tactic for accelerating cathodic kinetics.

For example, Wang et al.²³¹ developed a cathode with in situ introduced porous plastic crystal electrolyte (composed of succinonitrile, LiTFSI, PVDF-HFP, etc. denoted as SLPB) (Figure 13A). The high softness and firm adhesion of SLPB (Figure 13B) enabled tight interfacial contact between SSE and electrode, then formed continuous and abundant triple-phase boundaries at the cathode side. The electrolyte resistance and interface resistance of SLPB-based LAB were effectively decreased compared to the traditional solid-state LAB with LATP or PEO (Figure 13C), presenting a high specific capacity (5963 mAh/g) and long cycle life of up to 130 cycles. Zhao et al.²³² applied an ionically superconductive Li₃InCl₆ electrolyte as the interface modifier for N-doped CNTs (NCNTs) cathode and LAGP SSE (Figure 13D). The high ionic conductivity (1.3 × 10⁻³ S/cm) and solution-based preparation method of Li₃InCl₆ enabled its uniform distribution within the cathode and continuous contact with LAGP. The interfacial resistance of cathode/SSE was significantly decreased from 2056 to 569 Ω (Figure 13E), realizing fast cathode kinetics and prompting reversibility of discharge products (Figure 13F). Meanwhile, the point-to-point triple-phase junction of SSEs, cathode, and Li₂O₂ particles usually leads to high charge polarization and quick failure of solid-state LABs. To conduct optimization, they developed a transformation of the cathode/SSE interface from three-phase contact to two-phase contact by building a hybrid cathode as LiTaO₃-coated NCNT (NCNT@LiTaO₃), wherein the NCNT core acted as an electronic conductor while the LiTaO₃ coating layer performed as an ionic conductor (Figure 14A).²³³ The in situ environmental TEM results visually verified the continuous growth of Li₂O₂ film in NCNT@LiTaO₃ enabling highly reversible decomposition from a two-phase process (Figure 14B), while pure NCNT suffered from particle-shaped Li₂O₂ and its residual followed by a three-phase process (Figure 14C). This optimized cathode/SSE interface achieved steady charge transfer and intact catalytic regions, endowing a remarkable increase in Coulombic efficiency from 38.6% to 80.8%.

Figure 13. (A) The schematic of conventional solid-state LAB and that with adjustable-porosity SLPB. (B) The Young's modulus and adhesion of PEO and SLPB. (C) The electrolyte resistance and interface resistance of symmetric LiF–Li/LiF–Li batteries with LATP, PEO, or SLPB as electrolytes. Reproduced with permission: Copyright 2020, John Wiley & Sons Inc.231 (D) The schematic of fabricating the LAGP-NCNT-Li3InCl6 hybrid air electrode and the Li3InCl6-modified decomposition process of discharge products. (E) Nyquist plots and (F) the cycling performance (with a limited capacity of 500 mAh/g at 100 mA/g) of the LOBs with LAGP-NCNT-Li3InCl6, LAGP-NCNT, and LAGP-NCNT-liquid air electrodes, respectively. Reproduced with permission: Copyright 2020, Elsevier.232 LAB, Li–air battery; LAGP, Li1+xAlyGe2−y(PO4)3; LATP, Li1+xAlyTi2−y(PO4)3; LOB, Li–O2 battery; NCNT, N-doped CNT; PEO, poly(ethylene oxide); SLPB, a plastic crystal electrolyte composed of succinonitrile, lithium bis(trifluoromethanesulphonyl)imide, poly(vinylidene fluoride-hexafluoropropylene), and 2,6-di-tert-butyl-4-methylphenol.

Figure 14. (A) Schematic illustration and the corresponding SEM images for displaying the discharge/charge processes at the interface between the LAGP electrolyte and a CNF-based cathode characterized with mixed ionic and electronic conductors (point-to-point contact with Li2O2) or hybrid ionic and electronic conductor electrode (facet-to-facet contact with Li2O2). In situ observations of the time-resolved discharge/charge processes with (B) NCNT@LiTaO3 or (C) NCNT in an environmental TEM chamber. Reproduced with permission: Copyright 2020, John Wiley & Sons Inc.233 CNF, carbon nanofiber; LAGP, Li1+xAlyGe2−y(PO4)3; NCNT, N-doped CNT, carbon nanotube; SEM, scanning electron microscopy; TEM, transmission electron microscopy.

Besides, an all-in-one design with the integration of catalysts and SSE to serve as both cathode and electrolyte emerges as a promising strategy, with intrinsically uniform distribution of cathode/SSE interfaces and decreased overall thickness to deliver dramatically suppressed interfacial resistance. Zhu et al.²³⁴ first proposed this battery configuration, in which the LATP scaffold supported efficient Li⁺ transport, while a carbon coating onto it ensured electron transfer and catalytic activity for ORR and OER at the same time (Figure 15A,B). This design achieved a thin electrolyte layer with a 90% decrease from that in conventional solid-state batteries and a highly porous cathode with a porosity of 78%. The LOB based on this integrated structure delivered a cycle life of 100 cycles at 0.15 mA/cm² with a fixed capacity of 1000 mAh/g_carbon (Figure 15C). Similarly, they proposed an ultrathin integrated structure (≈19 μm) with porous LATP and carbon, accompanied by the Si-oil film coating to repel H₂O and CO₂ from reaching reaction sites (Figure 15D,E), contributing to a LAB with high practicability in ambient air when cycled at 0.3 mA/cm² with a fixed capacity of 5000 mAh/g_carbon for 50 cycles (125 days).²³⁵

Figure 15. The SEM images for (A) an integrated structure of LATP and carbon coating, and (B) one pore in the porous cathode (the inset shows a thin carbon coating layer on LATP). (C) The cycling performance of a LOB with LATP@cabon integration at 0.15 mA/cm2 with a fixed capacity of 1000 mAh/gcarbon. Reproduced with permission: Copyright 2015, Royal Society of Chemistry.234 (D) High-resolution observations of the outer part of LATP grain and surface layer, and the contact angle of Si-oil on the LATP surface or the carbon-coated LATP surface. (E) Schematic illustration of the LAB with the integration of LATP membrane, carbon-coated LATP cathode, and a Si-oil film onto the cathode. Reproduced with permission: Copyright 2015, Royal Society of Chemistry.235 LATP, Li1+xAlyTi2−y(PO4)3; LOB, Li–O2 battery; SEM, scanning electron microscopy.

Although these strategies markedly enhance the interfacial contact and stability between cathode and SSEs. But considering the ORR/OER is characterized by a surface reaction and requires space for accessing O₂ and reserving solid products, the balance between interfacial compactness and catalyst availability is of great challenge. The trade-off strategies have not been studied yet.

Interface optimization between anode and electrolyte

It is acknowledged that Li metal is highly electropositive and reactive, rendering the spontaneous reaction with most SSEs at room temperature and the formation of SEI. Most binary ionic conductors show intrinsic chemical stability with Li metal as the involved anions present a fully reduced state. For widely used ternary and quaternary SSEs, the stability of Li metal is contingent on the formation energy of decomposition products. As per the previous statement, the LATP and LAGP react with the Li anode at room temperature, resulting in the formation of Li–Al or Li–Ge alloy and the reduction of Ti⁴⁺ into Ti³⁺.^106,124 The binary ionic conductors usually enable a chemically stable interface with Li anode while no SEI layer, where Li⁺ could efficiently transfer from Li metal to SSEs. The SSEs (LiPON, Li₃PS₄, etc.) with some electronically insulating decomposition products but at least one ionically conductive component, produce a stable SEI once formed. For SSEs (like LLZO, LATP, and LAGP) that are most used in solid-state LABs, with a thermodynamically favorable reductive decomposition, result in a mixed SEI accommodating both electronically and ionically conductive components which are unstable and tend to grow continuously during cycling then generate increasing interfacial resistance.⁸⁹ Meanwhile, the development of most exceptional SSEs with high σ_Li is hindered by the large interfacial resistance between Li anode and SSE, due to the rigid nature of SSEs. Although some simple mechanical methods have been proposed to optimize the wettability of Li anode/SSE interface, such as simple nanopolishing²³⁶ and ultrasonic-assisted fusion welding²³⁷ we reported previously, the long-lasting interfacial stability required by LABs is challenging.

There are mainly two ways to avoid such an SEI and the unexpected contact problem to bring these SSEs with the advantage of high σ_Li for practical battery applications. One way is to utilize the Li alloy to realize stable Li stripping and plating under kinetically stabilized SEI, but remains a controversial failure under large current density.^88,89,238 The other way is to construct a thin buffer layer between Li anode and SSEs but without a distinct sacrifice of ionic conductivity, such as ultrathin metal/metal oxide coating, Li⁺-conductive solid interlayer, gel polymer interlayer, and so forth.

For example, our group²³⁹ sputtered an amorphous Ge thin film on the surface of LAGP, then suppressed the reduction of Ge⁴⁺ by Li and realized greatly dropped interfacial resistance from 2506 to 147 Ω cm² in symmetric Li/Li batteries (Figure 16B). A quasi-solid-state LAB further demonstrated this advantage and realized a stable cycle life of 30 cycles in the air. Similar enhancement was achieved in the garnet system by Luo et al.,²⁴⁰ through depositing a thin Ge layer (20 nm) on an LLZO garnet, endowing intimate contact of Li metal and SSE (Figure 16A). With the formation of Li–Ge alloy, the decreased interfacial resistance from ≈900 to ≈115 Ω cm² and stable Li stripping/plating was achieved. Similarly, they engineered a Ca, Nb-substituted LLZO (LLCZNO) garnet with Al coating, which induced enhanced wettability of the LLCZNO surface (Figure 16C–H) via forming Li–Al alloy layer (Figure 16I).²⁴¹ The computation result indicated the high stability between Li–Al alloys and garnet SSE with mutual reaction energies of −60 to −40 meV/atom (Figure 16J), while these minor interfacial reactions supported kinetic stabilization and improved wettability at the interface. Then a markedly reduced interfacial area-specific resistance from 950 Ω cm² for the Li|garnet SSE to 75 Ω cm² for the Li|Al-coated garnet SSE was produced.

Figure 16. (A) The schematic of the contact improvement between LLZO garnet/Li metal through coating a thin Ge layer on LLZO. Reproduced with permission: Copyright 2017, John Wiley & Sons Inc.240 (B) The impedance performances of amorphous Ge thin film coated LAGP in symmetric Li/Li batteries. Reproduced with permission: Copyright 2018, John Wiley & Sons Inc.239 (C) The digital pictures and SEM images for showing the wetting behaviors of Li|garnet SSE (D and E) and Li|Al-coated garnet SSE (F–H). (I) The phase diagram of coated ultrathin Al layer. (J) The calculation of mutual reaction energy at the interface of LLCZN garnet and Li–Al alloy. Reproduced with permission: Copyright 2017, AAAS.241 LAGP, Li1+xAlyGe2−y(PO4)3; LLZO, Li7La3Zr2O12; SEM, scanning electron microscopy; SSE, solid-state electrolyte.

The oxide coating (Al₂O₃, ZnO, etc.) known for its high stability and wettability to Li metal, can act as a desirable buffer layer between Li anode and SSEs. Han et al.²⁴² introduced an ultrathin Al₂O₃ coating on LLCZNO garnet by atomic layer deposition (ALD) and enabled tight Li anode/SSE contact leading to a remarkable decline of interfacial resistance from 1710 to 1 Ω cm² (Figure 17C,D). They also demonstrated an all-in-one strategy for simultaneously improving the interfacial property with intimate Li anode/SSE contact and a high tolerance for anode volume change.²⁴³ For this concept, an ultrathin and conformal ZnO surface coating was built up in 3D porous garnet SSE via ALD, which notably promoted the wettability between garnet SSE and the molten Li from bulk surface to internal structure by forming Li–Al alloy (Figure 17A,B), endowing a low interfacial resistance of about 20 Ω cm². In addition to oxides, Le et al.²⁴⁴ constructed a bilayer ceramic SSE comprising Li⁺-conductive LiPON and Al-substituted LLTO (LiPON/A-LLTO), relying on the advantage of LiPON interlayer to form stable SEI while contact to the Li anode. The symmetric Li/Li batteries with LiPON/A-LLTO presented enhanced cycling stability since the effective suppression of dendrite growth (Figure 17E–H). And a long cycle life of 128 cycles in the LOB with LiPON/A-LLTO was obtained.

Figure 17. (A) Schematic illustration and (B) the cross-section SEM images of the Li infiltration behavior into 3D porous garnet with or without the modification of an ALD-coated ZnO surface layer. Reproduced with permission: Copyright 2016, American Chemical Society.243 (C) SEM images and (D) EIS profiles for demonstrating the improved contact at the Li|garnet SSE interface with ALD-coated Al2O3. Reproduced with permission: Copyright 2016, Nature Publishing Group.242 The cycle performance of symmetric Li/Li batteries with (E–F) glass fiber impregnated with liquid electrolyte and (G–H) LiPON/A-LLTO, respectively. Reproduced with permission: Copyright 2016, Royal Society of Chemistry.244 3D, three-dimensional; ALD, atomic layer deposition; EIS, electrochemical impedance spectroscopy; LLTO, Li3xLa2/3−x□1/3−2xTiO3; SEM, scanning electron microscopy; SSE, solid-state electrolyte.

Except for the (electro)chemical properties, the mechanical nature is also pivotal for determining the performance of coating layers, because Li metal experiences incessant volume change during cycling that will accumulate internal stress in the coatings. The stress can induce crack formation in a stiff coating layer with high Young's modulus as aforementioned, while the cracks serve as hot spots for the nucleation and growth of Li dendrite, resulting in severe performance decay and safety concerns for solid-state LABs. Mechanically, polymer coating layers with lower stiffness can accommodate the stress evolution preferably. For instance, Zhou et al.²⁴⁵ coated a single-Li⁺ conducting polymer film composed of PEO and lithium poly(acrylamide-2-methyl-1-propane-sulfonate) (PAS) on LLZTO, realizing good adhesion of PEO–PAS to LLZTO and avoiding direct contact between Li anode and the grain boundaries residing in LLZTO pellet (Figure 18A). The PEO–PAS polymer coating with markedly improved Li⁺ transfer number is obtained to decrease interfacial resistance and suppress Li dendrite formation (Figure 18B). Liu et al.²⁴⁶ stored an apropos amount of liquid electrolyte inside a PVDF-HFP matrix as a polymer coating for LLCZNO, affording a significant decline in interfacial resistance from 1.4 × 10³ to 214 Ω cm² for the Li anode (Figure 18C,D).

Figure 18. (A) Cross-section SEM image for the symmetric Li/Li battery with PEO–PAS polymer coated LLZTO after cycling test. (B) The ionic conductivity of PEO–PAS at different PEO/PAS ratios and the corresponding Li+ transport number. Reproduced with permission: Copyright 2018, Elsevier.245 (C) EIS plot for the symmetric Li/Li battery with gel coated-LLCZNO. (D) Comparison of the interfacial resistance between LLCZNO SSE and electrode with and without the gel interlayer. Reproduced with permission: Copyright 2017, American Chemical Society.246 EIS, electrochemical impedance spectroscopy; LLZTO, Li6.5La3Zr1.5Ta0.5O12; PAS, poly(acrylamide-2-methyl-1-propane-sulfonate); PEO, poly(ethylene oxide); SEM, scanning electron microscopy; SSE, solid-state electrolyte.

To sum up, introducing an artificial SEI layer into solid-state LABs is an effective strategy for optimizing the Li anode/SSE interface. The criteria for an ideal coating layer include high thermodynamical stability with both Li metal and SSEs, electrical insulation, sufficient Li⁺ conductivity, and good mechanical ductility for tight interface contact. High chemical bonding to Li metal, metal/oxide coating, and GPE buffer layer have shown prominent improvement in the wettability and stability of the Li anode/SSE interface. The integrated HSE combining the merits of both ceramic SSE and GPE should also be regarded as an effective all-in-one demonstration for realizing overall high σ_Li and anode protection simultaneously, as discussed in Section 3.2.3. Besides, constructing 3D Li anode/SSE interfaces is an effective strategy for alleviating the Li⁺ concentration polarization at the interface and endowing homogeneous Li⁺ flux on the Li anode. The interfacial fluctuation can be well reduced by applying a 3D Li anode,^247–249 a 3D SSE,^250,251 or hosting Li in a 3D SSE framework.^243,252 Although these design strategies give promising prospect in developing practical Li metal anode for solid-state LABs, there remain great challenges in structural engineering and processing technologies.

SUMMARY AND PERSPECTIVE

As the most energy-rich “beyond Li-ion” batteries, LABs have motivated intensive research for approaching the high theoretical energy density rivaling combustion engines by clean electricity sources. Compared to liquid-state counterparts, solid-state LABs provide a more practical prospect with superior air compatibility, intrinsic safety, wide electrochemical window, favorable mechanical properties, environmental benignity, and cost efficiency. However, characterized by the complicated electrochemical–chemical reactions at the cathode, and assorted interfacial behaviors residing in the Li anode/electrolyte, electrolyte/electrolyte, addictive/electrolyte, and cathode/electrolyte interfaces, the solid-state LABs are confronted with several intractable challenges. Many prominent research groups have contributed paradigms of mechanism studies and effective strategies for propelling the development of solid-state LABs. In this review, we proposed the main challenges and related strategies of this field in detail.

• 1.

  The insufficient mechanisms and kinetics study. The energy-rich property of solid-state LABs originates from the reversible oxygen conversion reaction at the cathode, while its reaction routes and involved intermediates evolution are poorly understood. The insufficient knowledge about discharge–recharge mechanisms, leads to the absence of criteria for enhancing cathodic thermodynamics and kinetics at the atomic level. And a simplification in the theoretical calculation that modeling OER with direct reverse reaction steps of ORR in many papers could bring misleading results against a consensus in this field that OER does not follow the reversed route of ORR. Except for the cathode, the in-depth studies on mechanisms and kinetics for SSE and Li anode are hardly developed. But considering the shared properties with solid-state LIBs, tremendous work in this field would provide a valuable reference. Overall, to identify the specific bottleneck and construct targeting tactics for battery performance from the atomic scale to the device scale, we urgently need to develop ingenious computational methods and in situ characterization techniques.

• 2.

  To-be-improved ionic conductivity and stability of electrolytes. An ideal SSE for LABs should possess high σ_Li (10⁻²–10⁻³ S/cm), negligible electronic conductivity, wide electrochemical window (>5 V vs. Li/Li⁺), eligible compatibility with Li anode and cathode, durability to oxidative intermediates and air contaminants, adequate stiffness, and so forth. The inorganic ceramic electrolytes such as NASICON (LAGP, LATP, etc.), garnet, perovskite, and antiperovskite are the most developed SSEs for LABs, showing the merits of good thermal stability, desirable Young's modulus and σ_Li, while their rigidity leads to a to-be-improved interfacial resistance. Meanwhile, the chemical instability of the Li metal of LAGP/LATP, the susceptibility to humidity and CO₂ of garnet, the grain boundary problem of perovskite, and the thermodynamical metastability of antiperovskite inevitably impede the application of these SSEs. Structural modifications such as element doping, modulating Li⁺ and/or vacancy concentrations, and building up artificial interlayers could effectively enhance their σ_Li and structural stability. However, the guiding criteria are not established yet and demand systematic studies with both theoretical and experimental efforts. Zeolite emerges as a promising candidate for constructing practical SSEs with high σ_Li and intrinsic stability, which not only contributes to LABs but also brings a chance for LIBs. To prompt its development, in-depth research on mechanisms like conductivity related to different cage sites, and the balance between mechanical strength and conductivity should be comprehensively carried out. Compared to inorganic SSEs, the SPEs often show attractive merits of excellent ductility and favorable wettability toward electrodes, while suffering from low σ_Li, matrix decomposition, and Li dendrite penetration. Strategies such as decreasing their crystallinity, and being swollen by a plasticizer to form GPEs could produce applicable polymer electrolytes for LABs. With complementary properties, the integrated HSE of an inorganic SSE and a GPE is a promising strategy that meets most expectancy for an ideal SSE. It should be noted that a practical HSE is not a simple mechanical mixture but with a well-designed hybrid structure for stable and efficient interfacial behaviors, which need to be meticulously researched.

• 3.

  Lithium anode issues. Although the Li dendrite propagation should be alleviated to a large extent in solid-state LABs, with reaping the mechanical benefits of SSEs, the observations of Li dendrites penetration residing in grain boundaries, surface defects, and coterminous pores, repudiate the impervious dendrite penetration in SSEs. Meanwhile, when operated under ambient conditions, the Li anode corrosion accompanied by mechanical strength decay could be induced by the permeation of O₂, H₂O, and CO₂ into the free volume of SSE matrices. The continuous Li stripping and plating also result in undesirable volume change. These negative effects inevitably retrograde the overall Coulombic efficiency and cycle performance of solid-state LABs. Strategies for Li anode optimization and protection include substituting Li metal with Li alloys to enhance anode stability, utilizing super-hydrophobic SSEs or introducing a buffer layer to avoid anode corrosion, and constructing artificial SEI to decrease interfacial resistance and restrain dendrite growth, and so forth. Apparently, the Li anode worked in solid-state LABs with a half-open feature faces more complicated circumstances than other sealed Li-metal batteries, requiring more scrupulous study to qualify a practical Li anode.

• 4.

  Complicated interfaces in the battery. The chemical, electronic, and mechanical properties of the involved interfaces, especially that between electrolyte and cathode/anode, are crucial for determining the output energy density and long-term viability of a solid-state LAB. The poor wettability, incompatibility to Li anode/cathode products, excessive stiffness, and/or inferior Li⁺ conductive behavior of SSEs, often produce high interfacial resistance and an unstable interfacial environment. Adding auxiliary electrolytes with high σ_Li, cathode coating, integration of catalyst and SSE, and so forth are presented as effective methods for accelerating cathodic kinetics. But the tradeoff strategies for balancing interfacial compactness and catalyst availability are absent yet, which needs further in-depth investigations. Introducing an artificial SEI layer with high thermodynamical stability, electrical insulation, sufficient Li⁺ conductivity, and eligible mechanical ductility for intimate contact, has been proposed as an applicable tactic for optimizing the Li anode/SSE interface. The metal/oxide coating, GPE buffer layer, and all-in-one design of HSE have shown prominent improvement in the wettability and stability of the Li anode/SSE interface. Despite these strategies for interface engineering giving promising prospects in developing practical solid-state LABs, there still are many challenges in structural engineering and processing technologies that need to be overcome.

When it comes to practical application, the energy density of a full LAB has been predicted to be 610 Wh/kg and 680 Wh/L, with distinct preponderance over Li–S battery (450 Wh/kg and 450 Wh/L) or the LIB with Li anode and Li-rich NMC cathode (300 Wh/kg and 500 Wh/L).²³ But the maximum performance is guaranteed by operating in the air as an open battery. Characterized by the nonvolatile and (electro)chemically stable nature of SSEs, solid-state LABs show an alluring prospect in the LAB devices working with air. The activities of air contaminants such as CO₂ and H₂O in solid-state LABs are highly restricted compared to their liquid-state counterparts due to their low reservation ability in SSEs, but not totally obviated. The residual H₂O molecules at the air cathode could react with Li₂O₂ and form LiOH, which has been proposed to enable an energy-rich reaction with four electron transfers and fast discharge/charge kinetics.^99,189 But LiOH is prone to evolve into Li₂CO₃ with the existence of CO₂, then would greatly elevate polarization potential and retrograde cycling ability. For that reason, the reversible LiOH chemistry is running in pure O₂ merely. At the Li anode side, H₂O also brings unexpected anode erosion via a fast N₂-attacking reaction to form Li₃N and LiOH. Thus, to obtain stable and durable battery performance in an air atmosphere, the penetration of CO₂ and H₂O should be strictly restrained (other varied contaminants like NO_x and organic gas have not been considered in present studies). Utilizing SSEs with high air compatibility is a prerequisite, wherein not only the (electro)chemical stability to CO₂ and H₂O is required but also the hydrophobic property is indispensable when a protective layer is absent. The dense physical structure and/or the hydrophobic groups of SSEs can be effective against H₂O, however, their blocking behavior is the lack of in-depth understanding. To get into that, the H₂O diffusion pattern in the free volumes or specific sites of SSEs should be well explored to provide guidance for SSE design. Besides, assembling a water-proof but oxygen-permeable layer at the air-facing side of the cathode is an ideal solution for LABs. A thin thickness of the layer is highly preferred, for reducing the influences on mass transfer and total energy density. Whereas, the knottiest problem is the undesirable size discrepancy between O₂ (0.346 nm) and H₂O (0.265 nm) molecules,²⁵³ nullifying the pore sieving strategy and bringing a great challenge for building up O₂-selective but H₂O-screening gas channels. Undoubtedly, the surface engineering of these water-proof layers with nonpolar molecular structure is required, to exhibit high selectivity for nonpolar O₂ molecules while repelling polar H₂O molecules. The exquisite physicochemical design of building up an exclusive highway for O₂ at the cathode is a radical solution for realizing practical LABs, while lacking intensive study in both fundamental and experimental research yet.

Although great challenges remain, we believe the booming development of in situ characterization techniques and computational methods will provide further insights into the mechanisms for solid-state LABs. Underpinned by the theoretical guidance and the progress of material science, in the future, well-designed strategies targeting the specific bottleneck in different systems will endow practical LAB devices. When that happens, this energy-rich technology would revolutionize power supply from wearable devices to varied vehicles, and even large-scale energy storage systems.

ACKNOWLEDGMENTS

This study was financially supported by the National Key R&D Program of China (Grant No. 2021YFA1202300), the National Natural Science Foundation of China (Grant Nos. 22075132 and 22209069), the Natural Science Foundation of Jiangsu Province, China (Grant Nos. BK20211556 and BK20220783), Jiangsu Province Carbon Peak and Neutrality Innovation Program (Grant No. BE2022002-2), the Shenzhen Science and Technology Innovation Committee (Grant Nos. RCYX20200714114524165, JCYJ20210324123002008, and 2021Szvup055), and Guangdong Basic and Applied Basic Research Foundation (Grant Nos. 2023A1515011437, 2022A1515110736, and 2022A1515010026). The authors also thank Fundamental Research Funds from the Central Universities and Frontiers Science Center for Critical Earth Material Cycling Fund.

CONFLICTS OF INTEREST STATEMENT

The authors declare no conflicts of interest.