Precisely Engineering Interfaces for High-Energy Rechargeable Lithium Batteries

Abstract

While we are pursuing a fully electrified society, high-energy rechargeable batteries are undergoing intensive investigation. In this respect, atomic and molecular layer deposition (ALD and MLD) have been drawing increasing interest, due to their unmatched capabilities to precisely modify electrodes’ surfaces for better electrochemical performance. In this work, we reviewed the recent studies using ALD/MLD for interface engineering of several important electrode materials, including nickel (Ni)-rich metal oxide cathodes, silicon (Si), and lithium (Li) anodes in lithium-ion and lithium metal batteries. We particularly discussed the most promising coatings from these studies and explored the underlying mechanisms based on experiments and modeling. We anticipate that this work will inspire more studies using ALD/MLD as an important technique for securing new solutions for batteries.

1. Introduction

In an operating lithium-ion battery (LIB) cell, chemical and electrochemical interfaces play an important role in dictating the cell’s performance. Despite the tremendous progress and the state-of-the-art development made to date in the LIB industry, it is still necessary to advance our fundamental understanding of battery interfaces located between an electrode (anode or cathode) and an electrolyte, wherein the electrode/electrolyte interfaces are chemically and electrochemically active zones. On the cathode side, transition metal oxides (e.g., LiCoO₂) often undergo electrolyte-driven surface reactions, including transition metal dissolution, oxygen release, oxidization of the electrolyte, and the formation of a cathode electrolyte interphase (CEI). On the anode (e.g., graphite and silicon) side, electrolytes are prone to being reduced with the formation of solid electrolyte interphase (SEI) with the consumption of Li inventory and the rapid increase in cell impedance. Without proper control over these interfaces, the cycling performance of LIBs will be limited, and the battery safety will eventually be compromised due to the out-of-control thermal runaway.

In the last few decades, it has been recognized that engineering interfaces with protective coatings can form stable artificial SEI and CEI layers on anodes and cathodes, respectively. As a result, the coatings could stabilize these interfaces through separating active electrodes from electrolytes, improve thermal stability, suppress uncontrolled exothermic thermal decomposition pathways of electrolytes, and thereby make LIBs safer under abuse, degraded, or fault conditions. Therefore, the surface coating has become an important strategy to apply a protective layer over anodes or cathodes for taming the unwanted parasitic reactions at an electrified interface between an electrode (an anode or a cathode) and an electrolyte. Thus, multiple methods have been investigated, e.g., wet chemistry, physical vapor deposition (PVD), and chemical vapor deposition (CVD). Their practices did generate lots of compelling outcomes, but also exposed their limitations. They all commonly lacked the capability to control surface coatings accurately. This is often critical for pursuing the best cell performance. In this context, atomic layer deposition (ALD) became a new tool for the surface coating of batteries and greatly advanced our fundamental knowledge and technical innovations in the past decade. As a novel thin-film synthesis technique, ALD has its unrivaled capabilities in multiple aspects: (1) high precision in film growth at the atomic level [1], (2) extremely uniform film coverage over large-scale flat substrates [2], (3) unmatched conformal coating over high-aspect-ratio substrates [3], (4) moderate process temperature even down to room temperature [4,5]. So far, ALD is the only method that has been able to provide coatings over prefabricated electrodes while also can coat powder-based electrode materials [6]. All these unique features make ALD an irreplaceable tool for tackling the interfaces of LIBs and systems beyond LIBs. ALD is known for the growth of inorganic materials. As a supplement to ALD, furthermore, molecular layer deposition (MLD) is designed for organic materials and organic–inorganic hybrid materials growth [7]. ALD and MLD are based on the same operational principle but bear two different names. Recently, many more studies have used MLD methods to investigate battery interfaces. Inorganic coatings via ALD have been widely investigated as documented in the literature [6,8,9]. Compared to inorganic coatings, organic coatings via MLD provide some extra benefits, such as their excellent elasticity and flexibility. They have a better ability to accommodate volume changes. In this respect, organic coatings show great potential to tackle issues facing Si and Li anodes, as will be discussed in this article. Through the deposition of coatings on anodes, cathodes, solid electrolytes, or separators, ALD and MLD offer precise control over interface engineering in LIBs and beyond. In contrast to conventional coatings that often yield non-uniform thickness or incomplete surface coverage, ALD/MLD coatings are highly controllable in their thickness while highly tunable in their composition. These benefits could minimize the interfacial impedance growth while protecting electrolytes from decomposition. These unique features make ALD/MLD particularly attractive in seeking new solutions to a series of promising electrode materials, such as nickel-rich oxide cathodes, lithium metal anodes, and Si-based anodes, where interfacial stability is critical.

In addition to providing a passivating protective coating at the electrode/electrolyte interface, ALD/MLD are also known to enable the design of functional interfaces. Based on well-selected precursors, for instance, ALD has deposited Li-conductive films [10], such as LiNbO₃ [11], Li₃PO₄ [12], LiAlO₂ [13], and Li_xAl_yS [14]. Recently, we have developed several novel Li-conducting lithicones, including LiGL (GL = glycerol) [15,16], LiTEA (TEA = triethanolamine) [17], and LiHQ (HQ = hydroquinone) [18]. These functional interfaces form artificial SEIs on Li-metal anodes, enhance mechanical robustness, provide an electronic insulation region, and facilitate Li-ion conductivity across the electrode’s interfaces. According to the reported literature, the tunability of ALD/MLD is far more versatile than conventional coating techniques. Moreover, ALD/MLD are now being scaled up through spatial and roll-to-roll technologies, bringing atomic/molecular precision to industrial electrode sheets. This scalability bridges the gap between laboratory breakthroughs and factory production, making ALD/MLD not only superior in improving battery performance but also increasingly feasible for commercial adoption. Thus, in recent years, the stability of ALD/MLD coatings at electrode/electrolyte interfaces has become an important strategy in the tunable engineered interface that helps in delivering the longer-lived and safer rechargeable lithium batteries.

The previous literature has found that the benefits of ALD/MLD coatings lie in multiple aspects [19,20]: (1) protecting electrolytes from decomposition, (2) mitigating transition metal dissolution of cathodes, (3) inhibiting SEI and CEI formation, (4) suppressing Li dendrite growth of Li anodes. Despite many advancements made to date, a comprehensive understanding and a systematic analysis remain lacking. To highlight these issues, the purpose of this work is to give an overview of the importance of ALD/MLD’s roles in two main categories: (1) ALD/MLD coatings over anodes and (2) ALD/MLD coatings over cathodes. To address the important basic science related to these unique interfaces, the fundamental studies and insights will be addressed based on the ALD/MLD growth, experimental characterization, and computational simulation. Following this section, in this review, we will present a brief discussion on ALD/MLD growth. In a subsequent section, we focus on discussing interface engineering via ALD/MLD. In the last section, we conclude this review by providing perspectives on future research directions.

2. The Principle and Growth Mechanism of ALD and MLD

2.1. The Basics of ALD and MLD

Very distinctly, ALD and MLD are vapor-phase processes proceeding film growth in a layer-by-layer mode. They divide an overall reaction into two or more gas–solid surface reactions, while each surface reaction is self-limiting. To better explain the operation principle and growth mechanism of both ALD and MLD, we illustrate their processes in Figure 1. Figure 1a illustrates the widely used ALD process of Al₂O₃ (alumina) using trimethylaluminum (TMA, Al(CH₃)₃) and H₂O as precursors. The overall reaction between TMA and H₂O can be described as follows:

Al(CH₃)₃ + 3H₂O → Al₂O₃ + 3CH₄(g)

(1)

In the ALD process, however, the Al₂O₃ film is achieved via two half gas–solid surface reactions as follows:

|-OH + Al(CH₃)₃(g) → |-O-Al(CH₃)₂ + CH₄(g)

(2)

|-O-Al(CH₃)₂ + 2H₂O(g) → |-OAl(OH)₂ + 2CH₄(g)

(3)

where “|” indicates substrate surfaces while “(g)” signifies gas phases. The surface chemistry of the ALD Al₂O₃ is mainly based on ligand exchanges between –OH and –CH₃ to arrange atoms accurately in a layer-by-layer mechanism. As illustrated in Figure 1a, the ALD Al₂O₃ operates with four repeatable steps, i.e., Pulse A/Purge A/Pulse B/Purge B. The Pulse A (i.e., TMA) causes a self-limiting gas–solid reaction on the substrate surface, as shown in Equation (2). The following Purge A removes all the over-supplied precursor A (i.e., TMA) and byproducts (e.g., CH₄). Then, the Pulse B (i.e., H₂O) causes another self-limiting gas–solid surface reaction on the substrate surface, as shown in Equation (3). Thereafter, there is a Purge B to remove all the oversupplied precursor B (H₂O) and byproduct (CH₄). These four steps form one ALD cycle with the deposition of a layer of Al₂O_3, while their continuous repetition increases the Al₂O₃ film linearly to a desirable thickness. Through varying precursors, in principle, ALD can grow any inorganic materials [21,22], such as elements, oxides, sulfides, nitrides, fluorides, and so on.

MLD follows the same working principle as that of ALD but takes different precursors for growing pure organic materials or organic–inorganic hybrid materials. Through adopting two homobifunctional precursors, as shown in Figure 1b, MLD grows thin films of pure polymers. Coupling one ALD precursor and one MLD precursor, additionally, MLD is able to produce hybrid organic–inorganic films (as illustrated in Figure 1c). To date, MLD has deposited a variety of polymers (e.g., nylon [23,24], polyazomethines [25,26,27,28], polyureas [29,30,31,32,33], polyamides [23,24], poly(3,4-ethylenedioxythiophene) [34,35], polyimide-polyamides [36], polythioureas [37], and polyethylene terephthalate [38]) and many hybrid materials [15,16,17,18,39]. Furthermore, ALD and MLD processes can be integrated into a super-process for growing laminated materials with highly tunable layer thickness and layer number of both inorganic and organic components. Such a combination makes ALD and MLD unparalleled in searching for novel materials. In short, ALD and MLD feature their unrivaled capabilities for creating new materials in a controllable way as well as their incredible tunability for desired properties.

Owing to these unmatched merits of ALD and MLD as thin film techniques, they were investigated for modifying interfaces of LIBs and other emerging battery systems in the past decades. The first study was reported in 2007, in which an ALD TiN film was applied on Li₄Ti₅O₁₂ (LTO) powders, while the electrode made from the resultant TiN-coated LTO powders exhibited improved conductivity [40]. Thereafter, there were no other studies using the ALD technique to modify battery components until 2010, when there were two studies [41,42] that unveiled a new area using ALD as well as MLD for engineering battery interfaces accurately. These two studies demonstrated that sub-nano Al₂O₃ coatings (<1 nm) could be grown on either electrode powders or prefabricated electrodes using ALD. Specifically, Jung et al. [41] demonstrated that the 2-ALD-cycle (~0.2 nm) Al₂O₃ coating on LiCoO₂ powders or prefabricated electrodes could dramatically improve the resultant electrodes’ performance, showing a comparable capacity retention of 89% after 120 charge/discharge cycles in the voltage range of 3.3–4.5 V. In contrast, the bare LiCoO₂ electrodes could only sustain a capacity retention of 45% after 120 charge/discharge cycles [41]. At the same time, Jung et al. [42] also demonstrated that a 5-ALD-cycle (~0.5 nm) Al₂O₃ coating over prefabricated graphite electrodes can remarkably boost the resultant graphite electrodes’ performance, which was tested at 50 °C, accounting for a capacity retention of 98% vs. 26% of bare graphite electrode after 200 charge/discharge cycles. They believed that the excellent performance of ALD-coated electrodes was due to the coatings’ protection effects, including inhibiting uncontrolled SEI/CEI generation, improving mechanical integrity, and minimizing interfacial reactions. To date, a long list of surface coatings via ALD has been formed, including nitrides [40] and oxides [41,42,43,44,45] to fluorides [46,47], sulfides [14], phosphates [48], and oxynitrides [49]. In addition to LIBs, ALD has also been used for addressing interfacial issues in emerging battery systems, including Li-S [50,51], Li-O₂ [52], Na-based [53], K-based [54], Zn-based [55], and Al-based batteries [56]. Along with the efforts of ALD, there have also been efforts using MLD for modifying the interfaces of various types of batteries since 2013. Leobl et al. [57] made the first attempt to modify LIBs with an alucone coating via MLD. So far, a variety of MLD polymeric coatings have been reported for battery interface engineering [58], such as Li-metal anodes [15,16,17,18,59], Na metal anodes [60], and cathodes of Na-ion batteries [61].

2.2. Theoretical Studies

To date, ALD is a key technique in modern materials synthesis, enabling the fabrication of high-quality, conformal inorganic thin films. Whereas for MLD, it is a relatively unexplored counterpart for organic thin films synthesis techniques, despite various studies reported in recent years. To further advance these two synthesis methods, i.e., ALD/MLD as a new advanced gas-phase route for novel metal–organic or inorganic–organic thin films growth [62,63] for next-generation battery applications, a systematic study in multiscale simulation [64,65] of both ALD and MLD could be useful (Figure 2). With the support of high-throughput data mining, accurate atomistic simulation, mesoscale simulation, and advanced continuum modeling at the macroscale level, the exploration of various engineered surfaces/interfaces (e.g., crystalline coordination-polymer, crystalline-amorphous boundaries, metal–organic-framework-like structures) that are useful in battery applications could potentially be achieved using ALD/MLD techniques. These would enable the synthesis of fundamentally new materials, which are unattainable through conventional synthesis methods. As highlighted in Figure 2, the proposed multiscale simulation workflow could possibly accelerate the rational design of electrochemically stable, low-resistance, and mechanically robust ALD/MLD-coated interfaces, which is crucial for next-generation high-performance batteries. For example, by using a high-throughput data mining method, it is possible to help in identifying a set of oxide coatings that are predicted to be stable against Li metal (Section 3). Using atomistic simulation (e.g., Density Functional Theory (DFT) calculations), we can quantify the useful interfacial adhesion energies, electronic properties, and Li-ion diffusion for these coatings, which are selected by data mining. With the support of DFT simulation and experimental data, mesoscale phase-field modeling can predict the morphological evolution of coated interfaces and determine whether those coatings can help to suppress interfacial void formation or lithium dendrite growth. Then, by using macroscale continuum level simulation, it is possible for us to evaluate the overall battery performance improvements (e.g., cell resistance, overpotential reduction) based on these new coatings.

For ALD/MLD applications in batteries, ideally, a bottom-up cell fabrication and nanopatterning technique (e.g., area-selective ALD/MLD) [64] could be used, as a well-tuned functional technique, as inspired by the semiconductor industry. To achieve this new bottom-up paradigm in electrode patterning and cell fabrication, an exhaustive basic understanding of materials properties and synthesis processes from an atomistic to macroscopic scale is needed [66,67]. To aid the experimental studies, computational studies using multiscale simulation (Figure 2) address the issues while simultaneously examining the materials properties, surface kinetics, and precursor gas flow in reactors during the ALD/MLD synthesis is timely and necessary [64,65,66,67,68,69]. Therefore, alongside advancements in experimental methods and characterization techniques, developing multiscale simulations (Figure 2) is essential for providing high-fidelity, robust design guidelines to optimize the process operation and control during ALD/MLD synthesis in real-world applications.

As stated in Section 2.1, the synthesis processes involved in ALD and MLD can be very complicated, yet versatile. In principle, ALD/MLD can be utilized to deposit various types of inorganic or organic materials on virtually any substrates with a differing nature of surface and geometries (e.g., nanoparticles, layered structures, nanotubes, and porous media) [62,63,67] for surface/interface engineering in various battery applications. Depending on the conditions during synthesis, the design rationale for ALD/MLD processes are generally dictated by fine-tuning in operating temperature, pressure, precursor molecular structures, precursor surface chemistry, exposure time, etc. to control the reaction steps and the self-limiting deposition process that are governed by thermodynamics and kinetics during the synthesis processes [62,63,67] A common misapprehension about ALD/MLD is that it delivers a complete monolayer (ML) in each cycle. However, there are studies that suggest that these synthesis processes deposit less than a single ML of materials in each cycle [70], and this could be problematic in actual applications. To achieve the optimum growth per cycle (GPC) to improve the performance of any given engineered interface, a systematic study, supported by multiscale simulation (Figure 2), could be useful. For example, in a recent work by Tezsevin et al. [70] based on DFT simulation and mesoscale stochastic lattice modeling, it was found that the influence of the random sequential adsorption (RSA) mechanism, where precursor molecules arrive one-by-one randomly and adsorb at surface sites on a substrate, is an essential and important factor in determining the surface packing and ML coverage. According to Tezsevin et al. [70], the RSA mechanism yields a ~22% to 40% lower ML coverage relative to the ordered packing, with estimated GPC values of ~0.05Å relative to experimental data, considering a variation in the precursor shape and size. Therefore, this suggests that a systematic study based on mesoscale simulation can offer a precise prediction of the GPC of the ALD/MLD synthesis process and offer valuable predictions regarding potential variations in ALD/MLD films deposited experimentally when different precursors are used [62,63,70,71].

For both ALD and MLD, the film properties and growth characteristics are uniquely derived from different precursor gas pulses (Figure 1). To assist the gas-surface reactions during the growth process, the metal and organic precursor in ALD and MLD synthesis with proper reactive counterparts (e.g., halogenide ions, organic ligands, functional groups) needs to be well-chosen [62,63,70,71]. To optimize the design interface for any specific battery application, fine-tuning and a well-chosen precursor molecule from a vast selection of molecular candidates in inorganic and organic precursors is not trivial. To aid the experimental design and preliminary selection of the appropriate precursor [62,63,67,70,71], a high-throughput screening and data mining study based on accurate DFT calculations and prediction of structure-property relationship on a vast precursor molecules dataset (Figure 2) could be a useful strategy to guide precursor selection in experimental synthesis. Complementary to the modeling of the deposition processes using the aforementioned mesoscale simulation, the unique structures and material properties (e.g., electronic, chemical, mechanical properties, and ion diffusion in bulk or interfaces) of thin films obtained using ALD/MLD have been widely studied using atomistic simulation [64,65,66,68,69], and this has been proven to be extremely useful for improving the basic understanding and material design that utilizes ALD/MLD. Thus, with the improved understanding of the material properties obtained from atomistic simulation and mesoscale modeling (Figure 2), a reliable macroscopic modeling of area-selective ALD/MLD (ASALD/ASMLD) [64,65,66,68,69] is obtained. As an accurate bottom-up nanopatterning technique in device manufacturing, the model could possibly be applied in future battery cell design.

3. Accurately Engineering Interfaces of LIBs and Beyond

3.1. The Cathode/Electrolyte Interfaces

Cathodes are taking a critical role in rechargeable batteries, for they are closely related to cell voltage, capacity, energy and power density, lifetime, and safety. At present, multiple LIB cathodes are commercialized, including spinel LiMn₂O₄ (LMO), olivine LiFePO₄ (LFP), layered LiCoO₂ (LCO), layered LiNi_0.8Co_0.15Al_0.05O₂ (NCA), layered LiNi_xMn_yCo_zO₂ (NMCs, x + y + z = 1), and lithium- and manganese-rich (LMR) layer-structured cathode materials (commonly described as xLi₂MnO₃∙(1 − x)LiTMO₂, TM = Ni, Co, Mn, etc., 0 < x < 1). Among these cathodes, NMCs are an important type of cathode with less Co but higher capacity. The Ni-Mn-Co composition can be varied and therefore result in different NMCs, such as NMC111 (LiNi_1/3Mn_1/3Co_1/3O₂), NMC442 (LiNi_0.4Mn_0.4Co_0.2O₂), NMC532 (LiNi_0.5Mn_0.3Co_0.2O₂), NMC622 (LiNi_0.6Mn_0.2Co_0.2O₂), NMC811 (LiNi_0.8Mn_0.1Co_0.1O₂), and others with even higher Ni contents. In these NMC cathode materials, each transition metal ion plays its particular role. Specifically, Ni ions are mainly responsible for capacity through the Ni²⁺/Ni³⁺ and Ni³⁺/Ni⁴⁺ redox couples, Co ions can inhibit Ni/Li cationic mixing during the synthesis and cycling processes while enhancing rate capability, and Mn ions can improve the structural and thermal stability. Thus, an increased Ni content enables higher capacity and lower weight in NMCs. To date, NMCs of x ≤ 0.6 have been commercialized, but it is particularly challenging to commercialize NMC811 or the ones with an even higher Ni content. To this end, it is imperative to address a series of interfacial and structural issues, including the undesirable reactions at the interface, dissolution of TM ions, irreversible phase transition, microcracking, and lattice oxygen release. In this article, we discuss the latest development of ALD and MLD coating on Ni-rich NMCs (x ≥ 0.6).

Metal oxides are the most studied ALD coatings for NMCs [9], such as Al₂O₃ [72], ZrO₂ [73,74], and TiO₂ [45]. In addition, a few studies have also investigated fluorides, nitrides, and phosphates as the surface coatings of NMC cathodes, e.g., AlF₃ [75], AlW_xF_y [76], TiN [77], and AlPO₄ [78,79]. These materials are inert to electrolytes and can constitute a protective layer between NMC cathodes and electrolytes. They could reduce undesirable interfacial reactions, suppressing dissolutions of TM ions, inhibiting oxygen release from NMC lattices, and thereby sustaining the interfacial and structural stability. In this respect, our recent work on TiO₂ ALD coatings on NMC622 is one of the representative case studies [45]. In the study, we studied the effects of different thicknesses of ALD TiO₂ on prefabricated NMC622 electrodes directly. The ALD TiO₂ film thickness was varied by adjusting ALD cycles (20, 40, 60, and 100). We noticed an island growth of the ALD TiO₂ on NMC powders, which implied that the ALD TiO₂ grew preferentially on certain locations, such as defective sites. The ALD TiO₂-coated NMC622 electrodes were tested in NMC/Li half-cells. We found that the cells with an ALD-coated electrode have a lower static leakage current, compared to the cells with a bare electrode. The static leakage current is an indicator of interfacial parasitic reactions. We measured the static leakage current using a home-built high-precision leakage current measurement system to quantify the rate of solvent oxidation at different potentials. Figure 3a shows examples of the obtained current relaxation curves of three exemplary NMC/Li half cells. The cells were held at a constant potential of 4.4 V vs. Li/Li⁺ for 40 h so that the state of charge (SoC) or lithium concentration in the working electrode remained unchanged. The initial high current is due to the charging of the double-layer capacitor and/or the working electrode, while the relatively flat region was related to any gain of electrons from the oxidation of the electrolyte solvent. An exponential decay function (inset of Figure 3a) is used to extract the static leakage current (y₀) from the current relaxation curves. The resultant static leakage current value is used as a quantitative indicator of the rate of the solvent oxidation reaction. We found that, by performing a series of potentiostatic hold experiments from 4.1 to 4.6 V vs. Li/Li⁺, we found that the static leakage current values (y₀) increase with increased potentials, indicating the severe electrochemical oxidation of electrolyte components at the surface of the NMC (Figure 3b). Compared to bare NMC, all the ALD-modified NMC samples have smaller static leakage currents. Particularly, we found that a coating of 20 ALD TiO₂ cycles could effectively alleviate the electrolyte oxidation, while a thicker coating was not helpful for further reducing static leakage currents. We also investigated NMC/Li half cells and demonstrated that the 20-cycle ALD TiO₂ coating (i.e., ALD-20) evidently improved the cell capacity retention at both 4.4 and 4.5 V versus Li/Li⁺ (Figure 3c). Furthermore, we also confirmed the beneficial effects of the ALD-20 electrodes in the NMC/graphite full cells (Figure 3d). To address the benefits of the ALD TiO₂ coating, we measured the amounts of TMs (i.e., Ni, Mn, and Co) on the graphite anodes after cycling and noticed that there were much fewer TMs on the graphite side in the cells using an ALD-coated NMC electrode (Figure 3e). This should be attributed to the reduced dissolution of TMs by the ALD coating. Furthermore, we also witnessed that the ALD coating inhibits microcracking by improving the mechanical properties of the coated electrodes (Figure 3f). Furthermore, we found that the ALD coating could suppress the decay of the average discharge voltage (Figure 3g). These findings reflect that the ALD coating boosts the structural stability of the NMC622 electrodes. In the previous work, it was clearly demonstrated that the ALD technique provides a useful pathway to achieve accurate interface engineering, while the ALD TiO₂ coating exhibits multiple benefits through forming a protective oxide layer between the NMC cathode and the electrolyte, including inhibiting interfacial parasitic reaction, boosting the NMC cathode’s mechanical integrity, suppressing dissolution of TM ions, and reducing microcracking of cathode.

While the aforementioned coatings did help improve the performance of NMC cathodes, they were also exposed to their limited conductivity. The low ionic conductivity of ALD surface coatings could slow the diffusion of Li ions out/in the cathode during charge/discharge processes [80,81]. To this end, there has been an increasing interest in developing Li-ion conductors as surface coatings of NMCs, such as Li₃PO₄ [82,83], LiTaO₃ [84], Li_xZr_yO [79], and LiAlF₄ [46]. Particularly, computational calculations have revealed that, compared to binary oxides, Li-containing counterparts could enable better ionic conductivity [85]. To illustrate the benefits of these Li-containing coatings, we have conducted two studies on the ALD ZrO₂ [73] and Li_xZr_yO [79] coatings recently. To produce Li_xZr_yO coatings, we developed a super-ALD process through combining two sub-ALD processes of LiOH and ZrO₂, as illustrated in Figure 4a. The LiOH sub-ALD [86] adopted lithium tert-butoxide (LTB) and H₂O, while the ZrO₂ sub-ALD [74] used tetrakis(dimethylamido)zircominum (TDMA-Zr) and H₂O as the precursors. In experiments, the LiOH and ZrO₂ sub-ALD processes could be tuned in their sub-ALD cycles, m and n, respectively. The different m/n ratios could result in different Li_xZr_yO in composition. To study the effects of different Li_xZr_yO in composition, we investigated 1:1 and 1:2 Li_xZr_yO (i.e., 1:1 and 1:2 LZO) as the coatings of prefabricated NMC622 electrodes. We found that the 1:1 ALD-LZO coating could remarkably improve the insertion and extraction rates of Li ions across the cathodes. We further found that the 20-cycle 1:1 LZO could lead to the best performance, as shown in Figure 4b. Compared to the results of the ALD ZrO₂ coatings of different thicknesses [73], the 20-cycle 1:1 LZO coating exhibited a much better performance, in terms of rate capability and long-term cyclability. As illustrated by Figure 4b, specifically, the NMC622 electrode coated by 20-cycle 1:1 LZO could sustain a capacity of >100 mAh g⁻¹ after 100 charge/discharge cycles (Figure 4b), while the NMC622 electrodes coated by ZrO₂ could only sustain a capacity of ~80 mAh g⁻¹ after 50 charge/discharge cycles (Inset of Figure 4b). Thus, this confirmed that Li-conductive ALD coatings are more favorable than the ALD coatings with low ionic conductivity, such as ALD ZrO₂.

In addition to all the previously discussed oxide coatings, we recently discovered that sulfides are an important class of ALD coatings that have been overlooked [87]. In contrast to oxides, we discovered that sulfides could play very unique roles and can serve as oxygen scavengers. This new finding is very significant, because the oxygen released from NMC lattices can lead to electrolyte oxidation and produce many detrimental species (e.g., H₂O, CO, and CH₄). Furthermore, in addition to scavenging oxygen released from NMC lattices, we found that sulfides can convert into stable sulfates. The resultant sulfates can act as chemically and electrochemically stable surface coatings to maintain the interfacial and structural stability of NMC cathodes. Specifically, our recent study clearly demonstrated the sulfide–sulfate conversion using the Li₂S ALD coating with the distinct benefits of the sulfide conversion and the resultant sulfate coating [87]. As illustrated in Figure 5a, we applied a Li₂S coating over a prefabricated NMC cathode via an ALD process [51]. In the following charging process, the oxygen released from the NMC cathode changed the Li₂S coating to a Li_xS_yO coating. To verify the benefits of the Li₂S coating, a comparative study with Li₂S ALD coating on a NMC811 cathode and a bare NMC811 cathode was conducted. The study revealed that, compared to a bare NMC811 cathode (ALD-0), the Li₂S coating could help dramatically improve the performance of the Li₂S-coated NMC811 cathodes (ALD-20, i.e., 20-cycle ALD Li₂S-coated NMC811, Figure 5b) in achieving long-term stable cyclability, accounting for a sustainable capacity of 122 vs. 20 mAh g⁻¹ for Li₂S-coated and bare NMC811 electrode after 500 charge/discharge cycles, respectively. Importantly, we first determined that the Li₂S-Li_xS_yO conversion with the Li₂S coating by analyzing the cycled NMC811 cathodes with X-ray photoelectron spectroscopy (XPS) (Figure 5c). The XPS spectra revealed that, after cycling, the Li₂S coating changed from Li₂S to Li₂SO₃ and Li₂SO₄. We believe that this study unveils a new area for NMC cathode interface engineering, which needs to be further explored. In addition to protecting the electrolyte from oxidation, the resultant Li_xS_yO coating protected the NMC811 from microcracking and mitigated irreversible phase transition. Thus, the Li₂S coating helped sustain both interfacial and structural stability of NMC811.

In addition to the previously discussed ALD coatings, recently, there has been an attempt to combine ALD and MLD coatings to create a hybrid coating to address the issues in NMC811 electrodes [88]. In that study, researchers deposited TiO₂ via ALD and titanium terephthalate (Ti-TPA) via the MLD technique. The resultant hybrid ALD/MLD coating was named TPA-Ti, which was deposited with five supercycles and yields ~8 nm thick. In that experiment, each supercycle consisted of 12 sub-ALD cycles of TiO₂ and 1 sub-MLD cycle of TPA-T, and the NMC811 electrode coated by the ALD-MLD process was named NMC-TPA-Ti, while the bare NMC811 electrode was named NMC-ref. The researchers studied their performance and revealed that, using graphite as the anodes, the NMC-TPA-Ti cathodes could achieve a capacity retention of ~86% versus 56% of the NMC-ref after 500 charge/discharge cycles in the voltage range of 2.9–4.3 V at 1 C (200 mAh g⁻¹). Their postmortem analysis further clarified that there was thicker CEI formation and irreversible transition in the NMC-ref, while the NMC-TPA-Ti electrode exhibited reduced cracking formation and enhanced structural stability. This study inspired us to adopt both ALD and MLD for designing new surface coatings with exceptional properties to address NMC and other cathodes’ issues.

ALD Cathode Modeling

As a complement to experiments, theoretical studies based on simulation are useful for scientists in order to understand the complex science of interfaces in batteries. To aid in the basic understanding of the electrochemical stability of spinel lithium-rich oxide cathodes, early studies focused on the stability of the cathode/electrolyte interface. From an early study based on first-principles simulation [89], it was found that the electrolyte decomposition was initiated by the electron transfer to the adsorbed solvent molecules (e.g., ethylene carbonate, EC) at the electrode/electrolyte interface, depending on whether or not the electrode was coated by alumina ALD. For the bare electrode, the adsorbed EC molecule decomposes within picoseconds during the reduction process, according to the ab initio molecular dynamics (AIMD) simulation [89]. In contrast, the oxide ALD coating provided a passivated layer and contributed significantly to reducing the rate of electron transfer (Figure 6), and this was confirmed by microgravimetric measurements, which demonstrated that the ALD coating reduced electrolyte decomposition [89]. To verify the theoretical prediction made by Leung et al. [89], carbon films deposited onto Cu were used as electrodes to study the passivating role of the ALD alumina coatings with respect to the electrolyte (1 M LiPF₆ in a 1:1 volume mixture of EC and diethyl carbonate) reductive decomposition in the experiment. Based on the voltametric and gravimetric responses of the uncoated and alumina-coated Pulsed Laser Deposition (PLD) carbon-film electrodes, they found that a higher overpotential was required to drive solvent decomposition, and a lower quantity of mass addition took place with the 0.55 nm and 1.1 nm alumina coatings present at the electrode [89]. A greater overpotential and reduced mass uptake were found in the 1.1 nm alumina coating compared to the thinner 0.55 nm coating. This was attributed to the thicker alumina film, which provided a more effective kinetic barrier for reducing the reductive solvent decomposition and byproduct deposition on the electrode. Thus, a thicker alumina layer was anticipated to reduce electron tunneling, thereby slowing solvent decomposition and delaying SEI formation. In addition, despite the complex solid–liquid state at the cathode/electrolyte interface, the advanced atomistic simulation method, such as AIMD modeling, was useful for understanding the mechanism behind the dissolution of transition metal (TM) ions (e.g., Mn) [89,90]. Based on AIMD simulation, it was possible to demonstrate that the migration of TM ions from the spinel cathode surface caused the subsequent negative effects on the graphite anode SEI passivating layers through the dissolution of TM ions into the organic liquid electrolyte (Figure 6). Thus, this indicated that applying a thin ALD coating could enhance cathode stability by forming a protective interface to stabilize the cathode surface, reducing TM dissolution, and unwanted electrolyte decomposition.

In recent years, despite the extensive use of ALD coating in various cathodes (e.g., NMC) [89], an in-depth understanding of the role of coatings in fine-tuning battery performance in terms of extended lifetime remains lacking. For batteries in operation, the task of obtaining the detailed local structural change and the electrochemistry of ALD coatings embedded at the electrode interfaces from both the simulation and experiment is not trivial and often very challenging [91]. For example, in the cathode study, a protective coating, such as Al₂O₃, can be used to improve the LiCoO₂ (LCO), LiNiO₂ (LNO), and LiNi_xMn_yCo_1-x-yO₂ (NMC) cathodes [91]. As such, these alumina coatings with different film thicknesses have been exposed to different material interfaces and cell operating conditions. To optimize coating performance, it is critical to understand the role of their local structural (e.g., Al and O bonds) changes in various environments. For the Al₂O₃ ALD coating, the disordered four- and six-coordinated Al-O environments are generally found [91]; however, an accurate quantitative and qualitative study on the distribution of the different Al-O environments in different cathodes due to different synthesis methods and at various cell operating conditions remains lacking. A greater understanding of the coating’s local structure (e.g., Al-O bonding environments), in response to different ALD coating-cathode interface and cell operating conditions, is extremely useful when optimizing these ALD coatings. From the theoretical point of view, a greater basic understanding of ALD growth (Figure 2), materials properties, and their interfaces (Figure 6) is critical to optimize the roles of ALD coating. Eventually, this will help to fine-tune for area deposition, minimize excessive coating, maximize the ions’ conductive pathways across the cathode interface, thus mitigating cathode surface degradation and maintaining high-capacity retention and cyclability of batteries.

NMC811 is a well-known Ni-rich cathode material for LIBs, yet the degradation and parasitic side reactions at the cathode/electrolyte interface remain a challenge. In recent studies, there were mixed results reported for alumina ALD coatings of the NMC811 cathode. According to Huang et al. [92], there were no huge differences between the alumina ALD-coated and the bare commercial-grade NMC811 cathode with regard to long-term cycling stability at different upper cutoff voltages, rate capability, stability against ambient storage, and the charge transfer kinetics. In contrast, according to Chen et al. [93], an improved NMC811 surface stability coated by alumina ALD was confirmed by monitoring the gaseous degradation species via electrochemical mass spectrometry and X-ray spectroscopic (XPS) study of the electrochemically aged cathodes through examining the variations in the oxidation states of Ni and O, and local cathode structures. According to Chen et al. [93], the study suggested that alumina ALD coatings exhibited an important dual role, i.e., a protective layer against electrolyte degradation, and a synthetic passivating layer that mitigated oxygen loss and inhibited phase changes in NMC811 surfaces. Thus, these two conflicting experimental studies suggest that a further understanding of the basic roles of ALD coating is necessary.

To aid in the experimental efforts, a systematic theoretical study that focuses on the ALD-cathode coating is critical. To improve the coating design, a theoretical study that focuses on the surface chemistry of ALD with solvent or salt in electrolytes, and the basic understanding of the Li-ion conductive pathways and diffusivities within ALD coatings and across cathode interfaces with different thicknesses in coating is needed (Figure 6). As reported in a recent study [94], DFT calculations indicate that there are several factors that influence the Li-ion diffusivity of the CeO₂ ALD coating, for example, local structures (e.g., crystal and amorphous configuration), film thickness, surface and bulk diffusion, as well as neighboring atomic species. Overall, these factors possibly contribute to the experimentally observed trade-off in Li-ion diffusivity. To optimize the Li-ion conducting in the ALD coating design from a bottom-up approach, we suggest that a high-throughput computational screening and data mining, which utilize the advanced machine learning methods (Figure 6) in data analysis and the interatomic potentials for molecular dynamics simulation that were employed in searching for robust Li-ion solid state electrolytes (SSEs), could be useful [95,96,97,98]. The identification of new Li-ion SSE candidates through an exhaustive search using data mining will provide useful insights towards the further design and fine-selection of metal precursors [62] in ALD, as highlighted in Figure 2. Meanwhile, in order to conduct a high-throughput screening for new candidates for the robust Li-ion conducting coating via ALD, an exhaustive theoretical study of Li-ion diffusivity in various amorphous solids (e.g., oxides, sulfides, halides, etc.) [97], instead of crystalline structures, is necessary. As inspired by He et al. and their CeO₂-coated LMO study [94], a good Li-ion diffusivity in a CeO₂ amorphous layer with an optimum thickness, in bulk LMO, and at the CeO₂-LMO interface is critically important in ALD-cathode design. Therefore, in addition to the basic understanding of Li-ion diffusivity in bulk amorphous structures, a comprehensive theoretical modeling of Li-ion diffusion pathways of any given Li-ion-conducting amorphous film with different thicknesses at the interface of the cathode (e.g., NMC811) is necessary. With this indispensable knowledge, a more robust computational design of Li-ion-conducting ALD coating at the cathode could be possible.

3.2. The ALD/Anode Interfaces

LIBs’ capacities are dictated by the capacities of cathodes and anodes. To improve batteries’ capacities, ideally, one needs to optimize the cathode and anode capacities. To couple with promising cathodes, such as NMCs, two anode materials are among the most promising choices, i.e., Si and Li, due to their extremely high capacities, 3579 and 3860 mAh g⁻¹ at room temperature, respectively. The former is for the next-generation LIBs, while the latter is for the emerging Li-metal batteries (LMBs). However, these anode materials are still not commercialized, due to their interface instability. To address this issue, ALD and MLD have become two new surface coating techniques in the past decade to improve the stability of the interface between anodes and electrolytes. Through these techniques, the protected Si and Li anodes can achieve much better performance than traditional LIBs.

3.2.1. Si Anodes

Si as anode promises a very high gravimetric energy density, ~10 times higher than that of commercial graphite anodes (372 mAh g⁻¹). Additionally, it has a low discharge voltage of ~0.2 V versus Li/Li⁺. On the other hand, Si ranks second in abundance in the Earth’s crust, and this implies its cost-effectiveness. All these benefits jointly make Si an excellent anode candidate for LIBs. Unfortunately, Si experiences large volume changes up to ~300% during the charge/discharge processes in LIBs at room temperature [99,100]. Such large volume expansion generates many more issues, including electrochemical pulverization, continuous SEI formation, high consumption of electrolytes, and continuous increases in cell impedance. These issues are reflected by rapid cell performance fading and eventually cell failure. In order to tackle these issues, surface coating is effective and serves multifunctional roles, including maintaining the mechanical integrity of Si anodes, suppressing SEI formation, and boosting electrode performance. To this end, many inorganic coatings via ALD have been developed [8], but all these attempts have been vulnerable to the weakness of the ALD coatings in terms of elasticity, which makes it impossible to tackle the large volume changes in Si. Previously, we conducted a comprehensive survey on various ALD coatings over Si electrodes, including Al₂O₃, TiO₂, ZnO, HfO₂, and TiN [8]. For more details, readers can refer to the original reference [8]. Due to their inorganic nature, these ALD coatings generally lack flexibility while having limited elasticity. As a consequence, they find it difficult to accommodate the huge volume changes in Si electrodes. Alternatively, polymeric coatings via MLD were investigated. As an alternative to ALD, Piper et al. [58] reported the first attempt to apply a layer of an alucone (denoted as AlGL with a unit structure of [(CH₂CHCH₂)O₃Al) on the prefabricated Si electrodes (Figure 7a). They reported that a 5 nm thick MLD-AlGL coating (Figure 7b) could remarkably boost the coated Si anodes’ performance (Figure 7c). They showed that the AlGL coating could maintain the Si electrode network. Specifically, the MLD-AlGL-coated Si anode only had a 17.5% volume increase after 20 electrochemical cycles, while a 50% volume increase was observed in the bare Si electrode. Thus, this suggests that the MLD-AlGL coating is exceptional in dealing with the extreme volumetric changes in the Si electrodes.

In a recent study, another metalcone, niobicone, was reported for the coating of Si nanowire (NW) electrodes to improve the cell performance [101]. As reported, the niobicone layer of NbHQ (having a unit structure of [OC₆H₄O]₂Nb) was deposited on Si NW electrodes for 50 MLD cycles (~17.7 nm) (Figure 8a), and this resulted in the anode Si NW@NbHQ. In this previous study, some of the Si NW@NbHQ electrodes were further annealed at 500 °C for 2 h in N₂ atmosphere, leading to the annealed electrodes, i.e., Si NW@NbHQ-500 (Figure 8b). After the thermal annealing, the NbHQ layer was verified to be ~15.7 nm (Figure 8c). The thermal annealing yielded the cross-linked NiHQ layer, and this significantly enhanced the mechanical properties of the Si NW@NiHQ-500 electrode (as witnessed in Figure 8d), accounting for the enhanced Young’s modulus of 44.2 GPa, as compared to 12.8 GPa for the non-thermally annealed Si NW@NbHQ electrode. Electrochemical tests revealed that the Si NW@NbHQ-500 performed the best, compared to Si NW@NbHQ and bare Si electrodes in rate capability (Figure 8e) and long-term cyclability (Figure 8f). Fang et al. summarized the merits of the thermally annealed NbHQ layer in several aspects: (1) enhanced tolerance of stress during volume expansion and resistance to the side reactions from the electrolyte, (2) generation of Nb⁴⁺ species from the reduction in NbHQ-500 to participate in SEI formation, (3) fast Li-ion transport kinetics and lower in stress distribution within the thin SEI [101].

In addition to the previously discussed hybrid metalcone coatings, a polyamide (PA) coating was also used to modify a prefabricated Si electrode [102]. In the study, it was observed that the Si nanoparticles changed morphologically from spherical particles to irregularly shaped particles during charge/discharge processes. The scanning transmission electron microscopy (STEM) results revealed that the PA coating remained in intimate contact with the Si nanoparticles during electrochemical tests. The tests showed that the 0.5 nm PA coating had dramatically increased the coated Si electrode’s capacity sustainability, cyclability, and CE. Conversely, the 3 nm and 15 nm thick PA coatings did not help in improving the Si electrodes’ performance. Wallas et al. [102] emphasized that, compared to the MLD-AlGL coating discussed above, this PA coating was an all-organic polymer and much thinner.

3.2.2. Li-Metal Anodes

In addition to Si, Li metal as anode can also achieve an extremely high capacity, 3860 mAh g⁻¹, and has a low potential (i.e., −3.04 V versus SHE0). Thus, Li metal is another ideal anode material. Unfortunately, Li-metal anodes face two serious issues: Li dendritic growth during plating and uncontrollable growth of SEI. The former endangers cell safety by growing into the cathode side, while the latter continuously consumes electrolytes and Li inventory, thus increasing cell impedance. These two issues are reflected in the cell performance, i.e., low CE and continuous capacity fading. To address these challenges, several strategies have been investigated, such as the development of new robust liquid electrolytes, solid-state electrolytes, surface coating, separator modification, and Li-metal framework design [103]. Among these strategies, surface coating is facile and effective, but its effectiveness is highly related to the properties of surface coating materials. In this respect, ALD and MLD recently emerged as two new useful techniques and have been attracting more research interest.

In an earlier study, Kozen et al. [104] coated Li anodes with nanoscale Al₂O₃ films via ALD, followed by other ALD coatings, including ZrO₂ [105], TiO₂ [106], LiF [107], Li₃PO₄ [108], and Li_xAl_yS [14]. To protect the Li anodes, a coating is ideal to be electronically insulating but ionic conductive. To overcome the ionically insulating nature of Al₂O₃, ZrO₂, and TiO₂, we developed an ionically conductive coating material, Li_xAl_yS (LAS, enabling an ionic conductivity of 2.5 × 10⁻⁷ S/cm at room temperature) by combining two individual sub-ALD processes of Li-S [51] and Al-S [109] in a sub-cycle ratio of 1:1. We coated the 1:1 LAS film over Li metal as a 50 nm thick protective film [14]. Testing in symmetric Li||Li cells, the pristine Li symmetric cell had a resistivity (R_SEI) value of ~2500 Ω, around five times higher than that of the ALD LAS-coated Li symmetric cell after 68 h storage. This implied that the ALD LAS layer could protect the Li anodes from direct contact with the liquid electrolyte. We further tested 50 nm thick LAS-coated Cu foils in asymmetric Li||Cu cells. We found that the Li||LAS-coated Cu cell exhibited a very stable CE for 170 cycles, while the Li||pristine Cu cell gradually dropped in CE in the first 90 cycles and then decreased sharply after ~135 cycles. In addition, we observed that numerous long filaments formed on the pristine Cu foil, while no obvious Li dendritic structures were observed on the LAS-coated Cu foil. These results demonstrated that the ALD LAS layer can stabilize the interface between the Li-metal anode and electrolyte and can suppress Li dendritic formation, due to its decent ionic conductivity. Similar favorable effects were later verified in subsequent work reported by Harrison et al. [110], in which an in situ electrochemical STEM (EC-STEM) was used to directly visualize Li deposition and stripping with and without the ALD LAS protective coating. In addition, LiF [107] and Li₃PO₄ [108] were also verified for their effectiveness as ALD coatings of Li anodes, but their layer thicknesses were limited to less than 10 nm due to their limited ionic conductivity.

Along with the ALD studies discussed above, there have been several attempts using the MLD technique in the development of novel polymeric coatings to protect Li anodes. In this respect, there are several earlier studies focused on alucone [52], zircone [84], and polyurea [111] as the coatings for Li anodes. Compared to ALD coating, these polymeric coatings showed promising protective results and could help improve the electrochemical stability and lifetime of Li-metal anodes. Furthermore, Zhao et al. [112] conducted a study investigating bilayer hybrid protective films consisting of an MLD organic layer (AlEG) and an ALD inorganic layer (Al₂O₃). This unique ALD-MLD combination features a high tunability in bilayered film thickness and the bilayer property through accurately adjusting the inorganic layer and the organic layer. In the study, Zhao et al. found that the combination of 50-ALD-cycle Al₂O₃ (50Al₂O₃) and 50-MLD-cycle AlEG (50AlEG) was optimal. Particularly, they demonstrated that, using Li||Li symmetric cells with a carbonate electrolyte, the 50AlEG/50Al₂O₃/Li performed better than 50Al₂O₃/50AlEG/Li and the bare Li in different testing conditions. Zhao et al. explained that the underlying mechanism lay in the better mechanical strength of the 50AlEG/50Al₂O₃ bilayer, which helped suppress crack formation. Using an ether electrolyte instead, the 50AlEG/50Al₂O₃/Li still exhibited the best performance. Furthermore, the 50AlEG/50Al₂O₃/Li anodes were used to couple with a sulfur cathode and an LFP cathode in the full cells. Compared to the pristine Li foil, the 50AlEG/50Al₂O₃/Li showed a much better performance for the full cells of Li-S and Li-LFP, in terms of sustainable capacity and cyclability, which suggests the usefulness of the ALD-MLD hybrid coating.

Different from all the Li-free polymeric coatings discussed above, we recently developed three different lithicones as surface coatings of Li anodes, i.e., LiGL [8,9], LiTEA [10], and LiHQ [11], featuring their Li-containing nature and good ionic conductivity of up to 10⁻⁴ S cm⁻¹ at room temperature. Their unit structures are (CH₂CHCH₂)(OLi)₃, N(CH₂CH₂OLi)₃, and LiOCH₆H₄OLi, respectively. These lithicones showed very promising protection effects on Li anodes in Li||Li symmetric cells and Li||NMC811 full cells. As reported in [8,9,10,11], these lithicone surface coatings could well protect Li anodes from corrosion and achieve long-term cyclability of Li||Li cells. The lithicone-coated Li anodes could dramatically boost the Li||NMC811 cells’ performance, and it was found that coupling the lithicone-coated Li anodes via MLD with Li₂S-coated NMC811 via ALD could achieve the best cell performance [17,18].

Among these lithicones, the LiGL MLD process exhibited the highest GPC of ~2.7 nm cycle⁻¹ at 150 °C and could be coated on graphene nanosheets (GNS) conformally (Figure 9a–c) [8]. Similar to ALD, the LiGL coating thickness on Li-metal anodes was tunable through adjusting the LiGL MLD cycles, and the resultant Li-metal anodes were named LiGLX, where X is the MLD cycle number. For example, LiGL10 means the Li-metal anodes coated by a 10-MLD-cycle LiGL film, and so on. We particularly noted that it is desirable to coat Li anodes with a LiGL film over 200 nm, which can achieve an ultra-long cyclability of over 20,000 Li-stripping/plating cycles (over 10,000 h) without failures (Figure 9d). In addition, the LiGL coatings thicker than 200 nm could also tolerate a high areal current density, 7.5 mA cm⁻² or higher. Furthermore, we verified that a sufficiently thick LiGL coating (e.g., LiGL60) could well protect Li anodes from corrosion (i.e., SEI and dendrites) (Figure 10a,b), as confirmed by SEM. Additionally, XPS analysis further confirmed that, when compared to the thick SEI layer (indicated by the thickness of fluorine (F) due to the decomposition of LiTFSI (the lithium salt used in the electrolyte)), which formed on bare Li metal after 10 Li-stripping/plating cycles (Figure 10c), the LiGL90-coated Li metal was SEI-free after 10 (Figure 10d) and 50 Li-stripping/plating cycles (Figure 10e).

To further demonstrate the protective capability of the LiGL coating, very interestingly, we designed a novel experiment to demonstrate the compelling protective effects of the LiGL coating, in which ultra-long stripping/plating processes were performed for up to 24 h at 2 mA cm⁻² in Li||Li cells. After a 24 h stripping (Figure 11a), many erupted spots on the bare Li were observed, and these erupted spots were found to be not uniformly distributed. Remarkably, on the other side, a top layer, consisting of a large number of dendritic structures, was formed after a 24 h plating (Figure 11b). Very differently, the LiGL60-coated Li metal showed a smooth surface covered by many small cracks after a 24 h stripping (Figure 11c) and plating (Figure 11d). We further verified that these cracks were due to the mechanical press during cell assembly, and not due to the stripping and plating process. These results strongly suggested that the LiGL coating can effectively protect Li metal from corrosion. To further test this hypothesis, subsequently, a reversed 24 h plating/stripping process was carried out right after the initial 24 h stripping/plating process. As highlighted in Figure 12a,b, the bare Li-metal anodes had a significant formation of SEI and uncontrolled growth of Li dendrites, while the LiGL60-coated Li metal was found nearly intact and did not show any signs of corrosion (Figure 12c,d). In a subsequent study, the excellent protection capability of LiGL has also been further verified to improve the Li||NMC622 LMB cells [15]. As reported, the LiGL400 (~1000 nm thick) doubled the cyclability of the Li||NMC622 cells without any capacity drop, compared to bare Li||NMC622 cells. Based on these observations, we can conclude that the LiGL MLD coating can provide an exceptional protection to the Li-metal anodes due to the following excellent properties, i.e., high ionic conductivity, outstanding electrical insulation, good mechanical flexibility, and excellent chemical compatibility and stability. However, it is important to note that the cracks present in LiGL (Figure 12c,d) could potentially be a problem, indicating that further investigation on the mechanical resilience and properties of LiGL is needed.

In two subsequent studies [17,18], LiTEA and LiHQ MLD coatings were developed and coated on Li anodes. These studies revealed that both the LiTEA and LiHQ could evidently improve Li||Li cells’ performance, accounting for ultra-long cyclability of up to 10,000 Li-stripping/plating cycles (>4000 h) without failures at 5 mA cm⁻² and 1 mAh cm⁻². Similarly to LiGL, it was confirmed that there was little formation of SEI and no dendritic growth under the protection of sufficiently thick LiTEA and LiHQ coatings. According to the authors [10,11], it was demonstrated that the Li anodes coated with 200 MLD cycles of LiTEA (i.e., LiTEA200, ~72 nm) or 75 MLD cycles of LiHQ (LiHQ75, ~30 nm) could remarkably improve the cyclability of Li||NMC811 (Figure 13). For the NMC811 cathode, which was coated by 20 cycles of ALD Li₂S (i.e., Li₂S20) [87], the resultant LiTEA200||Li₂S20 (Figure 13a) or LiHQ75||Li₂S (Figure 13e) anode in LMB cells showed a significant improvement in cell cyclability. From the reported results [10,11], Ni content was detected on the Li anodes after 300 charge/discharge cycles based on energy-dispersive X-ray spectroscopy (EDX), and this is similar to the observation that was found on the Li anodes of bare Li||NMC811 and LiTEA200||NMC811 (Figure 13b,c). Whereas for the Li anode of LiTEA200||Li₂S20-NMC811 (Figure 13d), a much lower Ni content from EDX is found. Based on these results (Figure 13), we concluded that it would be most desirable to handle the issues of both Li anodes and cathodes synergically with optimal ALD and MLD coatings.

3.3. Anode Interface Modeling

From the discussion in Section 3.2.1 and Section 3.2.2, it is noteworthy to point out that the failure in both the Si and Li anode is due to the lack of fundamental understanding and lack of control of materials properties at the interfaces [113,114]. During the charging process, the non-uniform lithiation process on the anode is problematic, likewise during the discharge process. At the anode, the differences in the Si or Li diffusivity and deposition rate on the surface relative to the bulk lithiation yield immense internal structural stress, and cause the pulverization. As one of the approaches to tailor the interfaces of Si or Li anode (Section 3.2), surface coating via ALD/MLD thin film turns out to be a prevailing strategy [10,11,12,13,53,84] in providing a uniform and homogeneous protective layer with precisely controlled thickness, which is beneficial to electrochemical performance.

3.3.1. MLD-Si Anode Modeling

Compared to ALD, a systematic theoretical study of MLD coating on the anode is significantly lacking. For the basic computational study in the synthesis process (Figure 2) and the modulation in materials properties, the reported studies have been mostly focused on ALD, less on MLD [115,116]. For the pioneering theoretical study on MLD-coated films on a Si anode, the efforts were led by Balbuena et al., with the focus on aluminum alkoxide (alucone) film [117]. Based on DFT and AIMD simulations, the atomistic models of the alucone MLD layer in contact with the Si (001) surface and the lithiated Si surface (i.e., Li_xSi_y) were proposed and studied (Figure 14a). From DFT calculation, the optimized alucone layer (approximately 3nm thick) exhibits multi-crosslinks and consists of Al-O complexes featuring 3-fold or 4-fold oxygen coordination, as illustrated in Figure 14a. During the lithiation process, it was found that Li atoms bind favorably to the O atoms in alucone, and less into the Si bulk. From DFT calculation [117], the strong Li-O interaction in alucone suggests that the Li ions released from the alucone film were less probable. Thus, the alucone film lithiation would proceed to saturation at all favorable sites, and the volume expansion of the lithiated alucone layer would further accommodate the significant volumetric expansion of the silicon anode. Whereas for the alucone-film-coated lithiated silicone (i.e., Li_xSi_y) surface, Li atoms were observed migrating from the lithiated Li_xSi_y surface into the alucone layer, confirming that the alucone film is more favorable for lithiation than either Si or lithiated Si anode surface.

From DFT simulation [115], the strong Li binding in alucone film implies that the reduction may have been present on the surface of lithiated alucone film, and this was confirmed by the strong adsorption of ethyl carbonate (EC) molecules in AIMD simulation (Figure 14b) when both the alucone/Si and alucone/Li_xSi_y were exposed to electrolytes. According to the authors [117], different degrees of lithiation on the alucone film changed both the structural and electronic properties significantly, and a substantial improvement in electron conductivity was observed in the Li-rich alucone film (Figure 14c). In this case, the electronic conductivity of lithiated Li-rich alucone film was studied using the DFT-Green’s function method based on the Landauer formalism [115]. Experimental results suggested that the lithiated alucone film exhibited both electronic and ionic conductivity. According to the authors [115], the increased electronic conductivity indicated that the lithiated alucone film’s enhanced electronic conductivity capability may have promoted electrolyte molecule reduction, subsequently leading to the formation of a passivation layer on its surface.

3.3.2. MLD-Li Anode Modeling

In addition to the cathode, the anode surface plays an important role in the decomposition thermodynamics of the solid–liquid organic electrolyte interface in the Li-ion batteries. Due to the lower potential of lithium metal relative to graphite, the organic solvent (e.g., EC) decomposition on lithium metal was spontaneous and was much faster than that on the LiC₆ interface [118]. Thus, apart from the experiments, the thermodynamic and electrochemical instability of Li-metal anode exposed to liquid electrolytes was well-studied previously based on DFT and AIMD simulation [116,118,119,120,121,122,123]. According to the reported theoretical studies, electrolyte decomposition at the Li-metal anode depended not only on the electrode surface potential but also on the surface chemical composition and the kinetics of the chemical reactions [116]. These findings are useful for the basic understanding of SEI formation at the anode, and extensive efforts have been devoted to designing surface coating layers on Li-metal anodes to enable functionalized SEI formation [124,125,126]. Since then, the theoretical exploration for a thermodynamically stable, elastic-compliant, and high-ionic-conductivity-passivation coating for the Li-metal anode that is able to mitigate lithium dendrite growth and crack formation has become evidently important.

Over the years, ALD and MLD techniques have been utilized to provide a protection layer to bare Li-metal anodes (Section 3.2.2). For ALD and MLD coatings, these coating materials, whether inorganic or organic, can be classified as fully lithiated or partially/non-lithiated. Many ALD oxide (e.g., CeO₂, Al₂O₃) coatings can be lithiated during battery operation, as the mobile Li⁺ ions can migrate into these oxides and be lithiated continually [116,117]. Once lithiated, the mechanical and electronic properties of ALD coatings become dependent on the degree of lithiation, similar to certain electrode materials [127]. To mitigate the mechanical failure and limited ionic conductivity of ALD-coated Li-metal anodes (Section 3.2.2), a comprehensive study (in simulation) is needed to understand the basic property changes dependent on the degree of lithiation (or Li-concentration) and film thickness of these amorphous solids. Compared to the organic polymeric films of MLD, the ALD inorganic film was found to be more vulnerable due to a lack of resilience in accommodating significant changes in stress and local structures (Section 3.1 and Section 3.2). For the MLD protective coating on the Li-metal anode, the pioneering work was first reported by Meng et al. [16], based on both theoretical and experimental studies. In this work, the coating materials were based on lithicone [16], i.e., LiGL in Figure 15a, which can be synthesized through a precisely controlled MLD process using lithium tert-butoxide (LTB) and glycerol (GL) as precursors. From the experiment, it is suggested that the LiGL lithicone could serve as an excellent polymeric protection film for lithium metal anodes (Section 3.2.2).

Like the MLD-coated Si anode (Section 3.2.1), the LiGL coating undergoes lithiation during battery operation. From Meng et al. [16], the mobile Li⁺ ions will migrate into these polymeric organic films and be lithiated continually according to depth profiles of XPS. According to the AIMD simulation (Figure 15b), it was found that Li atoms tend to migrate from Li metal into the bulk LiGL polymeric network and bind very favorably to the O atoms in the lithicone film. Consequently, this facile Li⁺ migration leads to anticipated Li⁺-ion diffusion and facilitates an efficient ionic conduction within the LiGL film and across the Li-metal anode interface. Due to the unique interwoven polymeric organic network in amorphous LiGL, the three-dimensional (3D) Li⁺-ion diffusion paths are observed from AIMD simulation (Figure 15d). Thus, according to the authors, the high ionic conductivity of the LiGL film could therefore be particularly important to stabilize the Li-metal anode interface and thereby decrease cell impedance during the Li-stripping/plating processes (Section 3.2.2). Apart from being an ionic conductor [16], the amorphous LiGL is predicted to be an electronic insulator with ~3.0 eV in band gap (Figure 15c) according to DFT calculation. According to Meng et al. [16], this electronic insulating feature persists even at room temperature (T = 300 K) and elevated temperature (T = 400 K) (Figure 15c,d), as indicated by the trend of the highest occupied molecular orbital/band (HOMO), the lowest unoccupied orbital/band (LUMO) and Fermi level of the amorphous LiGL bulk based on AIMD simulation. Thus, it is believed that this electronically insulating feature of the LiGL films could be extremely useful to help in suppressing the Li dendrite growth at room temperature and even at elevated temperatures.

4. Conclusions and Outlook

In this review, we reviewed and compared ALD and MLD studies for precise surface modification of several key electrode materials, including NMC622, NMC811, Si, and Li. These materials represent the most promising electrode material candidates for enabling high-energy rechargeable batteries, but they still face some serious challenges for practical application. This work presented that ALD and MLD are two powerful methods to conduct interface engineering precisely through applying surface coatings conformally and uniformly over electrodes. It is clearly demonstrated that the performance of the coated electrodes is closely related to the coatings’ properties.

In the case of NMC cathodes, we learned that Li-conductive oxide coatings performed better than binary metal oxides, ascribed to the improved ionic conductivity. Particularly, we reported that sulfides can serve as oxygen scavengers to tackle oxygen released from layer-structured metal oxide cathodes, for example, NMCs, LCO, and others. This is very significant for oxygen released from cathode lattices, as this could bring forth more issues, such as electrolyte oxidation and the production of undesirable reaction products. All these new coatings provide new solutions to cathode materials. In addition, one should notice that, while addressing interfacial issues, these surface coatings via ALD have helped maintain the cathodes’ structure. Thus, these ALD coatings provide a robust and inert interface layer and serve as a mechanical strengthener. As a consequence, this helps to mitigate interfacial issues, protect cathodes from microcracking and dissolution of transition metals, and inhibits cathodes from oxygen release and irreversible phase transition.

In the case of Si and Li anodes, we particularly summarized the recent work of MLD coatings, which provide flexible coatings to the anodes. In the case of Si, several alucones and one polymer have been investigated and have improved the performance of silicon anodes. The main merit of these MLD coatings is their flexibility to mitigate the significant volume change in Si anodes. In the case of lithium anodes, we particularly discussed the promising applications of lithicones, which exhibited better ionic conductivity than other metalcones and polymers. Although very promising, it bears noting that improvements to the mechanical properties of lithicones are needed.

Based on these encouraging results from ALD/MLD coatings, we expect that future studies should be able to bring us more exciting findings. First, ALD/MLD coatings with superionic conductivity (>10⁻⁴ S cm⁻¹ at room temperature) will be desirable for efficient Li-ion transport in both cathodes and anodes. Such coatings will also possibly be coated thick enough to secure sufficient mechanical strength to protect electrodes. Second, hybrid coatings, through combining ALD and MLD, may offer us an unlimited resource in the search for desirable coating materials. Such a combination will be able to deliver coatings in a tunable manner. Through adjusting ALD and MLD processes, we would be able to search for novel materials in a more controllable manner. In short, there is still plenty of room left for us to utilize ALD and MLD for seeking new solutions to high-energy rechargeable batteries. We expect to see more efforts invested in this area in future studies.

Based on the theoretical studies over the years, we found that for an ideal ALD or MLD coating on the cathode or anode, the ideal ALD/MLD coating should exhibit two key properties: high electronic insulation and efficient lithium-ion conductivity. Thus, a systematic investigation on the ionic/electronic conductivity of a wide range of ALD/MLD coatings is deemed necessary. However, for ALD and MLD films, the basic properties of these inorganic and organic coating materials are not static and can be classified into non-lithiated, partially lithiated, and fully lithiated, defined by the thermally and electrochemically driven lithium-ion migration across the electrode’s interfaces during the battery’s operation. Thus, once these ALD/MLD coatings were lithiated, the materials properties of these coatings become dependent on the degree of lithiation, like the electrodes. Meanwhile, the involvement of ALD/MLD film during the lithiation process might also affect the redox processes at the electrode and coating surface and may contribute to new types of electrode SEI when exposed to the electrolytes. To advance the design and functionalization of these ALD/MLD coatings, a basic understanding of electrolyte interaction, electron transfer, ion diffusion, materials properties changes, and interface responses are all important. Modeling these mechanisms is essentially non-trivial, as the processes involved generally occur at different spatial and temporal scales. Often, new models and understanding will be needed if the reactant materials (electrolytes and electrodes) are changing when a wide variety of material spaces are considered. To overcome the limitations and challenges of simulation models, it reminds us to envision these ALD/MLD-electrode interfaces as multifaceted, mosaic, and evolving problems. Thus, simulation techniques, ranging from electronic structures to atomistic simulation, continuum simulation to phenomenological models, high-throughput screening, data mining to machine learning studies, are all equally important and should be properly utilized to make useful predictions and to shed light on various interrelated problems.

In short, describing, understanding, and improving Li-ion batteries in operation is extremely challenging yet rewarding. Establishing a set of working structure–property relationships of ALD/MLD for various electrode designs is very challenging to both theories and experiments due to the complexities that involve wide varieties in materials, chemistries, physical processes, and device designs. Additionally, a lot of common computational studies are described by the ideal models, instead of a real complex system, which can be both a limitation and an advantage for a simulation. To fully utilize the state-of-the-art developments in big data science, a close synergy in the utilization of both experimental data and machine learning techniques could possibly be useful to help us further advance the development of ALD/MLD-electrode design to optimize Li-ion battery performance.