An Oxygen‐Scavenger Sulfide Coating Enabling Long‐Term Stable Nickel‐Rich Cathodes

TL;DR

A new sulfide coating helps stabilize nickel-rich battery cathodes by capturing oxygen and forming a protective layer, improving battery performance and safety.

Contribution

A novel ZrS2 coating is introduced that scavenges oxygen and converts into a protective Zr(SO4)2 layer on NMC811 cathodes.

Findings

ZrS2 nanocoating scavenges oxygen and converts into Zr(SO4)2, preventing electrolyte decomposition.

The coating stabilizes the cathode structure, suppresses microcracking, and reduces transition metal dissolution.

ZrS2-coated NMC811 cathodes show exceptional long-term stability and performance.

Abstract

Oxygen release is a major issue associated with layer‐structured metal oxide cathodes in lithium batteries, which can further cause a series of problems, such as irreversible phase transition, microcracking, and electrolyte decomposition. Eventually, these issues jointly result in cell performance degradation and safety hazards. Thus, it is very significant to tackle oxygen release for achieving long‐term stable cyclability, but very challenging. Although intensive efforts have been invested to date, there still lacks a feasible solution. In this study, nanoscale ZrS2 coatings are applied on prefabricated LiNi0.8Mn0.1Co0.1O2 (NMC811) cathodes directly via atomic layer deposition (ALD). Very encouragingly, we reveal that this ALD‐deposited conformal ZrS2 nanocoating can serve as an exceptional oxygen scavenger and then convert into a stable sulfate (Zr(SO4)2) coating. Such an in situ conversion is very beneficial and effective for protecting the electrolyte from decomposition. In addition, the resultant Zr(SO4)2 coating further inhibits undesirable reactions, stabilizes the interface between NMC811 and the electrolyte, suppresses microcracking, mitigates transition metal dissolution, and maintains the structural stability of the NMC811 cathode. Consequently, the ZrS2‐coated NMC811 cathode has demonstrated extraordinary performance. Thus, this study advances the understanding of interface engineering while paving a new technical pathway for commercializing NMC811 cathodes.

ZrS2 as a novel coating can scavenge oxygen released from NMC lattices while converting to Zr(SO4)2. In return, such a sulfide‐sulfate conversion results in many benefits leading to high‐performance NMC cathodes.

Figures

Schematic illustrations of a) bare NMC811 and b) ALD‐ZrS2‐coated NMC811 electrodes and their difference in performance.

Characterizations of ZrS2‐coated NMC811 electrodes. a) SEM and EDS elemental mapping images of 20‐ZrS2 electrodes. b) SEM and EDS line scans of 20‐ZrS2 electrodes. c) STEM and EDS elemental mapping of 20‐ZrS2 electrodes. d) High‐resolution XPS analyses of Zr 3d and S 2p of ALD‐deposited ZrS2 films on Si wafers.

Electrochemical performance of uncoated and ZrS2‐coated NMC811 electrodes at room temperature in air. a) Rate capability of 0‐, 5‐, 10‐, 20‐, and 40‐ZrS2 electrodes in the voltage range of 3.0–4.3 V. b) The 1st ch/disch profiles of 0‐, 5‐, 10‐, 20‐, and 40‐ZrS2 electrodes tested at 0.1 C ch/disch in the voltage range of 3.0–4.3 V. c) Long‐term cyclability of 0‐, 5‐, 10‐, 20‐, and 40‐ZrS2 electrodes tested at 1 C ch/disch in the voltage range of 3.0–4.3 V.

Long‐term electrochemical performance of 0‐ZrS2 and 20‐ZrS2 electrodes tested at 0.5 C ch/1 C disch in the voltage range of 3.0–4.3 V at room temperature using a temperature chamber: a) Long‐term cyclability and Coulombic efficiency, b) Ch/disch profiles, c) dQ/dV curves, and d) Average discharge voltages with cycling.

SEM imaging on morphological evolutions of a) 0‐ZrS2 and b) 20‐ZrS2 electrodes with cycling at room temperature, tested at 1 C ch/disch.

Structural analyses of 0‐ZrS2 and 20‐ZrS2 electrodes after 500 ch/disch cycles tested at 1 C ch/disch in the voltage range of 3.0–4.3 V at room temperature: HR‐TEM images of NMC811 particles of a) the 0‐ZrS2 and b) 20‐ZrS2 electrodes, and c) STEM‐EDS image and elemental mapping of ZrS2‐coated NMC after 500 charge‐discharge cycles.

Raman spectra of Zr(SO4)2.4H2O, ZrS2, 0‐ZrS2, 20‐ZrS2, and 40‐ZrS2 electrodes: a,b) commercial Zr(SO4)2.4H2O powder, commercial ZrS2 powder, and the 0‐ZrS2 electrode, c) the 20‐ZrS2 electrodes after 0, 5, 20, 50, and 100 ch/disch cycles at 1 C, and d) the 40‐ZrS2 electrodes after 0, 5, 20, 50, and 100 ch/disch cycles at 1 C in the voltage windows of 3–4.3 V.

XPS spectra of Zr 3d, Li 1s, F 1s, and O 1s of a) the 20‐ZrS2 and b) 0‐ZrS2 electrodes after different ch/disch cycles at 1 C in the voltage windows of 3–4.3 V, compared to the pristine 20‐ZrS2 electrode.

SEM and EDS analysis on Li anodes. SEM imaging and EDS mapping on Li anodes of a) Li||0‐ZrS2 and b) Li||20‐ZrS2 cells after 100, 300, and 500 ch/disch cycles. c) Li consumption and d) Ni concentration of Li anodes in Li||0‐ZrS2 and Li||20‐ZrS2 cells after 100, 300, and 500 ch/disch cycles. The cells were cycled at 1 C ch/disch in the voltage windows of 3–4.3 V.

Full text

Introduction

1

While electric vehicles (EVs) are penetrating the market, the limits of state‐of‐the‐art lithium‐ion batteries (LIBs) have been concerning consumers, such as range anxiety and unsatisfactory lifetime. To this end, high‐energy rechargeable batteries are urgently needed, and new electrode chemistries are undergoing intensive investigation. Among promising candidates are layer‐structured metal oxide cathodes, such as LiNi* x Mn y Co z O_2_ (NMCs, x + y + z = 1). Compared to the dominant LiCoO_2_ cathodes, NMCs feature their lower Co contents and higher capacities. Particularly, any increase in the Ni content of NMCs can further boost capacity while reducing cost. Consequently, a higher Ni content is always preferred. To date, NMCs with x ≤ 0.6 have been commercialized, while LiNi_0.8_Mn_0.1_Co_0.1_O_2_ (NMC811) is underway.^[^ 1, 2, 3 ^]^ In comparison to NMCs of x ≤ 0.6, it is much more challenging to commercialize NMC811, for the latter is more vulnerable structurally.^[^ 4 ^]^ The biggest issue of NMC811 is oxygen release from lattices at a much lower potential (≈4.0 V vs 4.4 V for NMCs of x ≤ 0.6),^[^ 4, 5, 6 ^]^ which can further cause Ni/Li cationic mixing, irreversible layered‐spinel‐rocksalt phase transition, transition metal (TM) ion dissolution, microcracking, and electrolyte decomposition. These issues eventually lead to cell capacity fading and safety hazards. To address these issues, various strategies have been practiced, such as elemental doping,^[^ 7, 8 ^]^ addition of electrolyte additives,^[^ 9, 10 ^]^ and surface coating.^[^ 11, 12, 13, 14 ^]^ Of these methods, surface coating is very facile and effective.^[^ 15 ^]^ Through applying a layer of a coating material on NMCs, surface coating can isolate NMCs from contacting with electrolytes, reduce undesirable reactions at interfaces, and mitigate decomposition of electrolytes. Traditionally, surface coating has been dominantly performed on battery powders via drying coating approaches and wet chemistry processes prior to electrode fabrication.^[^ 15 ^]^ Dry coating methods (e.g., ball milling) were not able to form a complete layer over cathodes, while wet chemistry methods (e.g., sol‐gel) may cause more residual lithium compounds (e.g., LiOH and Li_2_CO_3_) and are prone to lead to non‐uniform and thick coatings, ranging from several tens to several hundreds of nanometers.^[^ 11, 15 ^]^ In the past decade, atomic layer deposition (ALD) as a novel vapor‐phase technique has emerged as a new thrust for surface coating of LIBs and emerging battery systems, featuring its unparalleled capabilities enabling extremely uniform and conformal nanoscale coatings at a moderate temperature (≤ 200 °C).^[^ 16, 17, 18, 19, 20 ^]^ It can be performed on either battery powders before electrode fabrication or prefabricated electrodes. Particularly, it is to date the only technique that can coat prefabricated electrodes directly. In the case of NMCs, a variety of coatings via ALD has been investigated, including oxides (Al_2_O_3_, TiO_2_, ZrO_2_, ZnO, MgO, and Li x Ti y O, and Li x Zr y O),^[^ 21, 22, 23, 24, 25, 26 ^]^ fluorides (AlF_3_, AlF_3_/AlW x F y *, and LiAlF_4_),^[^ 27, 28 ^]^ nitrides (TiN),^[^ 29 ^]^ and phosphates (Li_3_PO_4_, AlPO_4_, TiPO, and TiPON).^[^ 30, 31, 32, 33 ^]^ In general, these ALD coatings could improve the mechanical properties of NMC811, mitigate microcracking, inhibit HF attacks, and thereby alleviate TM ion dissolution. However, all these could not completely prevent oxygen release from lattices and therefore could not protect electrolytes from decomposition. Oxygen release is regarded as the dominant driver for electrolyte decomposition and very detrimental to the performance of NMC cathodes as well as cells.^[^ 34 ^]^ This is because that electrolyte decomposition generates acids (e.g., HF), water, and gases.^[^ 34 ^]^ These products could further cause decomposition of lithium salts (e.g., LiPF_6_) and a series of unfavorable reactions on NMCs, including NMC corrosion, dissolution of TM ions, formation of cathode electrolyte interphase (CEI), microcracking, and irreversible phase transitions, as illustrated in Scheme 1a. As a result, cell failures eventually could occur through a gradually aggravated degradation process. Thus, it is very significant to solve the issues of oxygen release.

Schematic illustrations of a) bare NMC811 and b) ALD‐ZrS2‐coated NMC811 electrodes and their difference in performance.

To this end, recently we studied sulfides for the first time as a new class of coatings.^[^ 35 ^]^ We hypothesized that many sulfides are reactive to oxygen with the production of stable sulfates, and such sulfide‐sulfate conversions would be able to scavenge the released oxygen of NMCs to prevent electrolytes from oxidation. In other words, sulfide coatings constitute an antioxidative layer for electrolytes. Additionally, the resultant sulfates could still play multifunctional roles as coatings. Specifically, the resultant sulfates would be a mechanical reinforcement layer of NMCs to minimize the evolution of microcracking. Through coating on prefabricated NMC cathodes, they could also maintain the electrodes’ integrity. Furthermore, the resultant sulfate coatings would act as an artificial CEI to stabilize the interface between NMCs and electrolytes, which could minimize parasitic reaction and dissolution of TM ions. All these benefits jointly could maximize the performance of NMC cathodes. Very importantly, we first demonstrated such a creative thought with nanoscale Li_2_S coatings via ALD^[^ 36 ^]^ and convincingly verified the feasibility of our innovative pathway for tackling oxygen release.^[^ 35 ^]^ This is very encouraging and deserves more exploration. In this respect, we further hypothesized that sulfides may differ in their capacity as oxygen scavengers, and a sulfide with a higher capacity of scavenging oxygen may better improve the performance of layered metal oxide cathodes. To this end, in this study, we investigated ZrS_2_ as a new coating of NMC811, for ZrS_2_ per mole has a capacity twice that of Li_2_S to scavenging oxygen as described as follows:

$$[eqn]$$ $$[eqn]$$

Considering the density difference of Li_2_S (1.66 g cm^−3^) and ZrS_2_ (3.82 g cm^−3^) and supposing the same coating thickness, ZrS_2_ still outperforms Li_2_S in scavenging oxygen, ≈1.5 times higher. Thus, ZrS_2_ should be a more desirable coating to scavenge oxygen released from NMCs. Additionally, the resultant Zr(SO_4_)2 can also remove trace water in the cell via a hydration process, leading to Zr(SO_4_)2.xH_2_O (x ≤ 4). This is another potential benefit of the proposed ZrS_2_ coating. Based on all these hypotheses, in this study we deposited nanoscale ZrS_2_ coatings of different thicknesses on prefabricated NMC811 electrodes via an ALD process developed in one of our studies.^[^ 37 ^]^ Very excitingly, our study revealed that there exists an optimal thicknesses for the nanoscale ZrS_2_ coatings, ≈2 nm (comparable to the one for the Li_2_S coating^[^ 35 ^]^), and the desirable ZrS_2_ nanocoating could dramatically improve the performance of NMC811 cathodes to achieve an extremely long cycling life of over 1300 charge/discharge cycles while still sustain a capacity of ≈60% at a charge rate of 0.5 C and a discharge rate of 1 C in the voltage windows of 3–4.3 V, where 1 C is defined as 200 mA g^−1^. At the same conditions, in comparison, the capacity of bare NMC811 cathodes dropped to 60% of its initial value after ≈300 charge/discharge cycles. Moreover, the desired ZrS_2_ coating also remarkably improved the rate capability and Coulombic efficiency (CE) of NMC811 cathodes. Particularly, our study revealed that the ZrS_2_ coating has a unique capability other than non‐sulfide coatings. It can scavenge oxygen released from NMC lattices and protect the electrolyte from oxidation. Such an oxidization turns ZrS_2_ into zirconium sulfate, Zr(SO_4_)2. Moreover, the resultant Zr(SO_4_)2 continues to play multiple roles as a coating: 1) a stable artificial interface inhibiting undesirable reactions, 2) a mechanical strengthener suppressing microcracking and continuous oxygen release, and 3) a protection layer mitigating TM ion dissolution. Additionally, it can remove trace water in the cell via a hydration process. Furthermore, we also found that these benefits generated on the cathode bring forth assets to Li anodes, for they tremendously reduce crosstalk and thereby protect Li from consumption. These advantages resulted from the ZrS_2_ coating are illustrated in Scheme 1b. Compared to the former Li_2_S coating, particularly, this study also demonstrated that this ZrS_2_ coating exhibits some more benefits: 1) Enabling much better electrode performance, in terms of sustainable capacity and cyclability and 2) being air‐stable to facilitate commercial application while the Li_2_S coating is highly reactive to air.^[^ 36 ^]^

All these compelling improvements in the performance of NMC811 cathodes distinctly indicate that sulfides are an important class of surface coatings, unexplored but deserving further exploration. Thus, this study has provided a groundbreaking solution for commercializing Ni‐rich cathodes and related layer‐structured cathodes that suffer from oxygen release and related issues. It has important implications for commercializing these promising cathodes for high‐energy lithium batteries.

Results and Discussion

2

The Exceptional Effects of ZrS2 Coatings on the Cathode Performance

2.1

ALD features its accuracy in depositing conformal films over various substrates. In this study, the ZrS_2_ nanofilms of different thicknesses were coated over NMC811 electrodes and formed a high‐quality conformal layer to protect NMC811 electrodes. To exemplify the resultant conformal coverage of ZrS_2_, the 20‐ZrS_2_ electrode (NMC811 electrode coated by 20‐ALD‐cycle ZrS_2_) was characterized using scanning electron microscopy (SEM), scanning transmission electron microscopy (STEM), energy‐dispersive X‐ray spectroscopy (EDS), and X‐ray photoelectron spectroscopy (XPS), as illustrated in Figure 1. Corresponding to the SEM image of 20‐ZrS_2_ electrode in Figure 1a, EDS mapping images clearly revealed that Zr and S were deposited over the NMC powders uniformly. This was further corroborated by the EDS line scanning in Figure 1b. Furthermore, STEM and EDS mapping results shown in Figure 1c disclosed that Zr and S were deposited over NMC powders uniformly while they were diffused into the interparticle spaces conformally. XPS peaks of Zr 3d and S 2p in Figure 1d verified that the deposited films were ZrS_2_.^[^ 37 ^]^ Furthermore, SEM images of all bare and coated electrodes in Figure S1 (Supporting Information) show no difference in morphology and also support that the ZrS_2_ coatings on NMC811 electrodes are uniform and conformal. This is consistent with our previous studies.^[^ 24, 25, 35 ^]^ In addition, high‐resolution synchrotron‐based X‐ray diffraction (XRD) was employed to investigate the crystal structures of NMC811 before and after the ALD ZrS_2_ coating, i.e., bare NMC811 (0‐ZrS_2_) and 20‐ZrS_2_, respectively. As illustrated in Figure S2 (Supporting Information), the diffraction patterns for both bare (Figure S2a, Supporting Information) and ZrS_2_‐coated NMC811 (Figure S2b, Supporting Information) were indexed to a rhombohedral layered structure with space group R 3¯ m, characteristic of a well‐ordered α‐NaFeO_2_‐type lattice. Rietveld refinement yielded precise lattice parameters (as included in Table S1, Supporting Information), showing that the intensity ratio between the (003) reflection to the (104) reflection (i.e., I_(003)/I(104)) is 1.38 and 1.47 for 0‐ZrS_2 and 20‐ZrS_2_, respectively. The I_(003)/I(104)_ ratio is a crucial parameter to determine cation mixing in NMC, while a value below 1.2 indicates significant cationic position exchange between Li^+^ and Ni^2+^, which causes structural deformation.^[^ 38, 39 ^]^ Thus, our 0‐ZrS_2_ and 20‐ZrS_2_ electrodes have no significant cation mixing between Li^+^ and Ni^2+^. Particularly, the latter has a slightly higher c/a and I_(300)/I(104)_ ratio, implying the potential to have better electrochemical performance. The presence and sharpness of the (101), (006), and (012) reflections confirm the rhombohedral symmetry and layered ordering of NMC811. These peaks were instrumental in refining the lattice parameters and assessing structural integrity, with no evidence of secondary phases or stacking disorder. Moreover, the (018) and (110) reflections were well‐resolved, confirming both the long‐range stacking order along the c‐axis and the in‐plane TM distribution. The distinguished doublet features of (006)/(012) and (018)/(110) lattice planes signify no deficiency of Li^+^ in the crystal structure, for these doublets tend to merge/overlap if the crystal matrix is Li^+^ deficient. Overall, all these results affirm the phase purity and structural conformity of the reported material. Particularly, there are no XRD peaks related to ZrS_2,_ and this is consistent with the amorphous phase of the ZrS_2_ coating.^[^ 37 ^]^

Characterizations of ZrS2‐coated NMC811 electrodes. a) SEM and EDS elemental mapping images of 20‐ZrS2 electrodes. b) SEM and EDS line scans of 20‐ZrS2 electrodes. c) STEM and EDS elemental mapping of 20‐ZrS2 electrodes. d) High‐resolution XPS analyses of Zr 3d and S 2p of ALD‐deposited ZrS2 films on Si wafers.

To explore the film‐thickness effects of the ZrS_2_ coatings, we first investigated the rate capacity of these coated and uncoated electrodes in the voltage windows of 3–4.3 V, as shown in Figure 2a. Very evidently, all these ZrS_2_‐coated electrodes exhibited much better performance than the bare electrode, 0‐ZrS_2_, while 20‐ZrS_2_ showed the best rate capacity. Furthermore, we examined the first charge/discharge (ch/disch for short) profiles of these electrodes (Figure 2b) and there had no distinct differences observed. That means that the ZrS_2_ coatings did not contribute to the ch/disch processes remarkably. We further conducted electrochemical evaluations on the long‐term cyclability of these electrodes in the voltage windows of 3–4.3 V at 1 C ch/disch. As shown in Figure 2c, all the ZrS_2_‐coated electrodes performed better than the bare electrode, i.e., 0‐ZrS_2_. Again, among them, 20‐ZrS_2_ was the best one, accounting for a capacity retention of ≈75% vs 15% for 0‐ZrS_2_ after 250 ch/disch cycles. In other words, 20‐ZrS_2_ boosted a capacity improvement by ≈60%. These results indicated that the ZrS_2_ coating thickness of 20 ALD cycles is optimal, i.e., ≈2 nm. To take all these results together, one is easy to conclude that the ZrS_2_ coatings are very beneficial for helping NMC811 electrodes harvest much better performance, while 20‐ZrS_2_ is the best. The underlying mechanisms may be that a thinner ZrS_2_ coating (i.e., 5‐ and 10‐ZrS_2_) was not sufficient to protect NMC811 electrodes from existing issues, while a thicker coating (i.e., 40‐ZrS_2_) has inhibited Li‐ions from transporting through the coating layer. Similar outcomes also have been revealed with other coatings in our previous studies.^[^ 22, 24, 25, 26, 35 ^]^ We further analyzed the improvements of the 20‐ZrS_2_ electrode compared to the 0‐ZrS_2_ electrode in cyclability, ch/disch profiles, dQ/dV profiles, and the evolution of average discharge voltage in Figure S3 (Supporting Information). Associating with the improved cyclability (Figure S3a, Supporting Information) and ch/disch behaviors (Figure S3b, Supporting Information), the 20‐ZrS_2_ electrode exhibited much better reversibility of phase transitions (Figure S3c, Supporting Information) and much less drop in discharge voltage (0.1 V vs 0.4 V for 0‐ZrS_2_ after 250 ch/disch cycles, Figure S3d, Supporting Information). All these together corroborated the beneficial effects of the 20‐ZrS_2_ coating.

Electrochemical performance of uncoated and ZrS2‐coated NMC811 electrodes at room temperature in air. a) Rate capability of 0‐, 5‐, 10‐, 20‐, and 40‐ZrS2 electrodes in the voltage range of 3.0–4.3 V. b) The 1st ch/disch profiles of 0‐, 5‐, 10‐, 20‐, and 40‐ZrS2 electrodes tested at 0.1 C ch/disch in the voltage range of 3.0–4.3 V. c) Long‐term cyclability of 0‐, 5‐, 10‐, 20‐, and 40‐ZrS2 electrodes tested at 1 C ch/disch in the voltage range of 3.0–4.3 V.

Furthermore, we investigated the effects of different testing protocols on the performance of NMC811 electrodes. As illustrated in Figure 3, 0‐ZrS_2_ and 20‐ZrS_2_ electrodes were tested under the testing protocol: charging at 0.5 C and discharging at 1 C (i.e., Ch 0.5 C/Disch 1 C). Remarkably, the 20‐ZrS_2_ electrode enabled an extremely long cyclability up to 1300 cycles while sustaining 75% of its initial capacity after 500 cycles and 60% after 1300 cycles. Such an impressive cyclability accounted for a capacity fading rate of 0.03% per cycle. In contrast, the 0‐ZrS_2_ electrode faded in capacity continuously while only sustained 25% of its initial capacity after 500 cycles and 5% after 1000 cycles, accounting for a capacity fading rate of 0.095% per cycle. In other words, the 20‐ZrS_2_ electrode improved the capacity sustainability by 55% after 1000 ch/disch cycles and reduced the capacity fading rate by 3 times. Compared to the optimal 20‐Li_2_S electrode reported in one of our previous studies,^[^ 35 ^]^ this 20‐ZrS_2_ also showed significant improvement at the same testing conditions, accounting for a sustainable capacity of 145 mAh g^−1^ for the 20‐ZrS_2_ electrode vs 122 mAh g^−1^ for the 20‐Li_2_S electrode and a capacity retention of 75% for the 20‐ZrS_2_ electrode vs 71% for the 20‐Li_2_S electrode after 500 ch/disch cycles (see Figure S4a, Supporting Information). In addition to its exceptional cyclability, the 20‐ZrS_2_ electrode also exhibited excellent Coulombic efficiency (CE, ≈100%, Figure 3a) during the whole 1300 cycles. Very differently, the CE of the 0‐ZrS_2_ electrode started to fluctuate with increased intensity after ≈250 charge/discharge cycles. Figure 3b illustrates the ch/disch profiles of both 0‐ZrS_2_ and 20‐ZrS_2_ in 1000 cycles. Clearly, one can easily conclude that the 20‐ZrS_2_ electrode is much more stable in sustaining capacity than that of the 0‐ZrS_2_ electrode. This has also been witnessed with the dQ/dV profiles in Figure 3c. The dQ/dV profiles of 0‐ZrS_2_ exhibit an aggressive polarization progress with increased cycles, while the ones of 20‐ZrS_2_ remain stable in the whole process with some minor retrogression. These electrochemical behaviors of both 0‐ZrS_2_ and 20‐ZrS_2_ were further reflected in their cell voltage stability, as illustrated in Figure 3d. The cell voltage of 0‐ZrS_2_ faded quickly from 3.8 to 3.1 V, while the cell voltage of 20‐ZrS_2_ dropped from 3.8 to 3.7 V in 1000 cycles. That is to say that 0‐ZrS_2_ had a much larger voltage drop of ≈0.7 V vs a drop of ≈0.1 V for 20‐ZrS_2_. All these implied that 0‐ZrS_2_ had experienced serious phase transitions irreversibly, whereas 20‐ZrS_2_ stayed stably in structure. In addition to the 0‐ZrS_2_ and 20‐ZrS_2_ electrodes, we also investigated 5‐ZrS_2_, 10‐ZrS_2_, and 40‐ZrS_2_ electrodes under 0.5 C ch/1 C disch, as illustrated in Figure S4b (Supporting Information). Again, these results verified that all the coated (5‐, 10‐, 20‐, and 40‐ZrS_2_) electrodes perform better than the uncoated (0‐ZrS_2_) electrode while the 20‐ZrS_2_ one is the best one. According to previous studies,^[^ 22, 24, 25, 26, 35 ^]^ we understood that the underlying mechanism responsible for the optimization of these ZrS_2_ coatings is a compromise between Li^+^ diffusion and protection effects. For the 5‐ and 10‐ZrS_2_ electrodes, their ZrS_2_ coatings were not thick enough to protect NMC811 from reacting with the electrolyte, while, for the 40‐ZrS_2_ electrode, the ZrS_2_ coating could well separate NMC811 from the electrolyte but made the Li^+^ diffusion through the coating layer sluggish. In comparison, the 20‐ZrS_2_ electrode has the optimal ZrS_2_ coating, which enabled an excellent protection while did not evidently affect the Li^+^ diffusion. Furthermore, we also demonstrated the improvements of the 20‐ZrS_2_ electrode under the testing protocol at 0.5 C ch/disch in Figure S5 (Supporting Information). Very evidently, the 20‐ZrS_2_ electrode outperformed the 0‐ZrS_2_ electrode in capacity retention (Figure S5a, Supporting Information), ch/disch behaviors (Figure S5b, Supporting Information), reversibility of phase transitions (Figure S5c, Supporting Information), and discharge voltage stability (Figure S5d, Supporting Information).

Long‐term electrochemical performance of 0‐ZrS2 and 20‐ZrS2 electrodes tested at 0.5 C ch/1 C disch in the voltage range of 3.0–4.3 V at room temperature using a temperature chamber: a) Long‐term cyclability and Coulombic efficiency, b) Ch/disch profiles, c) dQ/dV curves, and d) Average discharge voltages with cycling.

In addition to these afore‐discussed results tested at room temperature, we further investigated the performance of both 0‐ZrS_2_ and 20‐ZrS_2_ electrodes at an elevated temperature of 45 °C. As shown in Figure S6 (Supporting Information), the 20‐ZrS_2_ electrode nearly did not show any capacity drop (Figure S6a,c, Supporting Information) while the 0‐ZrS_2_ electrode has exhibited an evident capacity drop in 150 ch/disch cycles (Figure S6a,b, Supporting Information). Compared to the performance of the 0‐ZrS_2_ and 20‐ZrS_2_ electrodes at room temperature (inset of Figure S6a, Supporting Information, which is part of Figure 3a), this elevated temperature has helped maintain higher capacities. This may be due to the improved conductivity of cell components (electrode materials and electrolytes). Moreover, we also examined the effects of air exposure on the performance of both 0‐ZrS_2_ and 20‐ZrS_2_ electrodes, as shown in Figure S7 (Supporting Information). It is very evident that air exposure produced very negative effects on the 0‐ZrS_2_ electrode and led to significantly reduced cyclability and a quickly dropped capacity. Specifically, the 0‐ZrS_2_ electrode after a 3‐day exposure started to drop in capacity evidently after ≈60 ch/disch cycles (Figure S7a,b, Supporting Information) while the 0‐ZrS_2_ electrode after a 6‐day exposure began to fade in capacity after ≈30 ch/disch cycles (Figure S7d,e, Supporting Information). In sharp contrast, the 20‐ZrS_2_ electrode was not affected evidently by air exposure, as witnessed by Figure S7a,c (Supporting Information) for the 3‐day exposure and by Figure S7d,f (Supporting Information) for the 6‐day exposure. This is particularly beneficial, for NMC cathodes are very vulnerable to air exposure. This has posed many challenges for the transportation and storage of NMC cathodes. The ZrS_2_ coating apparently paves a technically feasible solution to these challenges.

Take all these results together, one is easy to conclude that the 20‐ZrS_2_ coating has dramatically improved the performance of NMC811 electrodes. Thus, it is critical to gain a better understanding of the roles played by the 20‐ZrS_2_ coating. To this end, we explored the underlying mechanisms responsible for the different electrochemical behaviors and performance of 0‐ZrS_2_ and 20‐ZrS_2_ electrodes using SEM, TEM, Raman, and XPS.

First, we examined the morphological evolutions of the 0‐ZrS_2_ and 20‐ZrS_2_ electrodes after different ch/disch cycles, as illustrated in Figure 4. We did not observe any evident changes to the NMC powders of both the 0‐ZrS_2_ (Figure 4a) and 20‐ZrS_2_ (Figure 4b) electrodes after 100 ch/disch cycles. Very strikingly, however, we observed cracks on some NMC powders of the 0‐ZrS_2_ electrode after 200 ch/disch cycles, while the cracks became larger and larger after 300 and 400 cycles (Figure 4a), as indicated by the arrows. After 500 ch/disch cycles, particularly, we noticed a lot of completely broken powders from the 0‐ZrS_2_ electrode. This ever‐aggravating cracking phenomenon along with cell cycling should underlie the continuous capacity fading, the gradually exacerbated irreversibility of phase transitions, and the constant average discharge voltage dropping. This is because cracking of NMC powders exposes new surface areas, releases more oxygen, brings about more undesired interfacial reactions, causes electrolyte degradation, generates more CEI, increases cell impedance, and loses more active materials. It has been regarded as one of the main reasons for NMC cathodes.^[^ 40 ^]^ In sharp contrast, we did not observe any evident morphological changes with the 20‐ZrS_2_ electrode after different cycles up to 500 ch/disch cycles (Figure 4b). This clearly corroborates the compelling performance of the 20‐ZrS_2_ electrode. This also evidences that the 20‐ZrS_2_ coating has dramatically strengthened the mechanical integrity of the 20‐ZrS_2_ electrode and the NMC powders.

SEM imaging on morphological evolutions of a) 0‐ZrS2 and b) 20‐ZrS2 electrodes with cycling at room temperature, tested at 1 C ch/disch.

Besides these observations on the morphological evolutions with cycling, we further investigated the beneficial effects of the 20‐ZrS_2_ coating on the structure of the NMC811 powders. To this end, we applied focused ion beam (FIB) to cut the cycled 0‐ZrS_2_ and 20‐ZrS_2_ powders and then used high‐resolution transmission electron microscopy (HR‐TEM) and STEM to observe their near‐surface areas. HR‐TEM revealed that there is a very thick rocksalt layer of ≈20 nm with the cycled 0‐ZrS_2_ electrode (Figure 5a), while the cycled 20‐ZrS_2_ electrode was not observed for such a rocksalt layer but covered by a thin 5‐nm amorphous layer (Figure 5b) after 500 ch/disch cycles at 1 C. It indicates that the near‐surface area of the 0‐ZrS_2_ electrode had experienced a serious irreversible phase transition from its initial layered structure to the NiO‐like rocksalt structure. This was witnessed by the (003) planes of the layered structure (having a lattice spacing of 4.7 Å) and the (111) planes of NiO‐like structure (having a lattice spacing of 2.4 Å) (Figure 5a).^[^ 41 ^]^ The 5‐nm amorphous layer over the NMC powders of the 20‐ZrS_2_ electrode (Figure 5b) was suspected to be ZrS_2_ or/and the converted Zr(SO_4_)2. In this respect, we applied STEM‐EDS to analyze the cycled 20‐ZrS_2_ electrode, as illustrated in Figure 5c. The STEM image clearly showed that there is a coating layer over the NMC powder, and the EDS mapping further exhibited that the coating layer consists of Zr, S, Ni, Mn, Co, and O. Zr and S are due to the ALD coating of ZrS_2_ while Ni, Mn, and Co may be due to some extraction from the bulk NMC. Particularly, O is likely from oxygen release, which might have turned ZrS_2_ into Zr(SO_4_)2, as described in Equation (2). If so, the nominal 2‐nm thick ZrS_2_ should have become a 4.3‐nm Zr(SO_4_)2, which is calculated based on Equation (2) using the crystalline ZrS_2_ density (3.82 g cm^−3^) and the crystalline Zr(SO_4_)2 (3.22 g cm^−3^). This calculation is consistent with our observation in Figure 5b, showing a coating layer of ∼5 nm after 500 ch/disch cycles. To verify this hypothesis, we applied Raman and XPS to study the 20‐ and 40‐ZrS_2_ electrodes before and after cycling.

Structural analyses of 0‐ZrS2 and 20‐ZrS2 electrodes after 500 ch/disch cycles tested at 1 C ch/disch in the voltage range of 3.0–4.3 V at room temperature: HR‐TEM images of NMC811 particles of a) the 0‐ZrS2 and b) 20‐ZrS2 electrodes, and c) STEM‐EDS image and elemental mapping of ZrS2‐coated NMC after 500 charge‐discharge cycles.

The Unique ZrS2‐Zr(SO4)2 Conversion Mechanism for Scavenging Oxygen

2.2

Using Raman spectroscopy, we measured the spectra of commercial Zr(SO_4_)2.4H_2_O and ZrS_2_ as well as the 0‐ZrS_2_ (i.e., bare NMC811) electrode as our references, as duplicated in Figure 6a,b. The Raman spectra of the commercial Zr(SO_4_)2.4H_2_O powder exhibits two strong peak at 1040 and 1027 cm^−1^ due to Zr‐O‐S stretching vibrations,^[^ 42, 43, 44 ^]^ while the Raman spectra of the commercial ZrS_2_ powder show two strong characteristic peaks: A_1g_ peak at 332 cm^−1^ and the E_g_ peak at 250 cm^−1^, consistent to those reported in literature.^[^ 45 ^]^ Additionally, the Raman spectra of bare NMC811 electrode (0‐ZrS_2_) revealed only one main peak at 528 cm^−1^, indicating Ni‐O vibrations and that the NMC811 powders are fresh.^[^ 46 ^]^ Furthermore, we also measured the Raman spectra of a commercial ZrO_2_ powder, as shown in Figure S8 (Supporting Information). Particularly, we compared the spectra of the commercial ZrO_2_, ZrS_2_, and Zr(SO_4_)2.4H_2_O powders before and after 1‐h air exposure (Figure S8, Supporting Information). We had two conclusions: 1) the characteristic peaks of ZrO_2_ are very different from those of ZrS_2_ and Zr(SO_4_)2.4H_2_O and 2) these powders are not air‐sensitive. These conclusions are significant for us to conduct further measurements on cycled 20‐ZrS_2_ and 40‐ZrS_2_ electrodes, as shown in Figure 6c,d, respectively.

Raman spectra of Zr(SO4)2.4H2O, ZrS2, 0‐ZrS2, 20‐ZrS2, and 40‐ZrS2 electrodes: a,b) commercial Zr(SO4)2.4H2O powder, commercial ZrS2 powder, and the 0‐ZrS2 electrode, c) the 20‐ZrS2 electrodes after 0, 5, 20, 50, and 100 ch/disch cycles at 1 C, and d) the 40‐ZrS2 electrodes after 0, 5, 20, 50, and 100 ch/disch cycles at 1 C in the voltage windows of 3–4.3 V.

With the information in Figure 6a,b, we investigated 20‐ZrS_2_ (Figure 6c) and 40‐ZrS_2_ (Figure 6d) electrodes before (0 cycle) and after 5–100 ch/dish cycles. It is not surprising that all these electrodes commonly exhibit a peak at 528 cm^−1^, ascribed to Ni‐O vibrations of NMC811 powders and indicated by the black dash lines in in Figure 6c,d. However, we did not observe any evident peaks related to ZrS_2_ from pristine 20‐ZrS_2_ (0 cycle for ch/disch, Figure 6c) and 40‐ZrS_2_ (0 cycle for ch/disch, Figure 6d) electrodes. To this end, we examined the regions of 300–400 cm^−1^ more closely, which are shown in insets of Figure 6c,d, as guided by the blue arrows. Then, we did observe one weak peak at 332 cm^−1^ for both pristine 20‐ZrS_2_ (0 cycle for ch/disch, Figure 6c) and 40‐ZrS_2_ (0 cycle for ch/disch, Figure 6d) electrodes, as indicated by the yellow arrows. This evidence that both pristine 20‐ZrS_2_ (0 cycle for ch/disch, Figure 6c) and 40‐ZrS_2_ (0 cycle for ch/disch, Figure 6d) electrodes were coated with a layer of ZrS_2_ film via ALD. For cycled 20‐ZrS_2_ (5‐100 cycles for ch/disch, Figure 6c) and 40‐ZrS_2_ (5–50 cycles for ch/disch, Figure 6d) electrodes, we noticed that the ZrS_2_ peak at 332 cm^−1^ became unobservable but the Zr(SO_4_)2 peak at 1040 cm^−1^ emerged after 5 ch/disch cycles, as indicated by the red dash line. Particularly, the Zr(SO_4_)2 peak at 1040 cm^−1^ become stronger after 20 ch/disch cycles for 20‐ZrS_2_ and 10 ch/disch cycles for 40‐ZrS_2_. Then, this peak intensity was relatively stable and did not change evidently with cycles. These results strongly suggested that ZrS_2_ converted into Zr(SO_4_)2 or/and its hydrates. To better illustrate the evolution of the Zr(SO_4_)2 peak at 1040 cm^−1^ with cycles, furthermore, we use a peak intensity ratio between the NMC811 peak at 528 cm^−1^ and the Zr(SO_4_)2 peak at 1040 cm^−1^as an index, as illustrated in Figure S9 (Supporting Information). It shows that the Zr(SO_4_)2/NMC811 peak ratio increases continuously with cycles but levels off after 50 ch/disch cycles. This implies two possibilities: 1) all ZrS_2_ has been converted into Zr(SO_4_)2 or/and its hydrates in 50 ch/disch cycles or 2) no more ZrS_2_ converted into Zr(SO_4_)2 or/and its hydrates after 50 ch/disch cycles. Considering that 40‐ZrS_2_ doubled the amount of the ZrS_2_ coating of 20‐ZrS_2_, the results in Figure S9 (Supporting Information) apparently corroborate the latter, that is, no more ZrS_2_ converted into Zr(SO_4_)2 or/and its hydrates after 50 ch/disch cycles or the ZrS_2_‐Zr(SO_4_)2 conversion was significantly slowed.

To further verify the ZrS_2_‐Zr(SO_4_)2 conversion, we also conducted XPS analysis on 20‐ZrS_2_ electrodes before and after cycling, as illustrated in Figure 7. As references, we also analyzed commercial ZrS_2_, Zr(SO_4_)2.4H_2_O, and ZrO_2_ powders, as shown in Figure S10 (Supporting Information). It can be seen that the commercial ZrS_2_ powder was a bit oxidized with some Zr(SO_4_)2. To track the evolution of the ZrS_2_ coating with ch/disch cycling, Figure 7a shows the XPS spectra of the pristine 20‐ZrS_2_ electrode and the 20‐ZrS_2_ electrodes after 5, 20, 50, and 100 ch/disch cycles, as indicated in the figure. The Zr 3d spectra revealed that the ZrS_2_ coating converted into Zr(SO_4_)2 completely after 50 ch/disch cycles. Along with this conversion, we noticed a gradually increased P 2s peak at ≈191–192 eV due to insoluble Li_x_PF_y_ or/and Li_x_PO_y_F_z_ compounds (which are typical decomposition products of LiPF_6_).^[^ 47, 48, 49, 50 ^]^ In comparison, the P 2s peak of the 0‐ZrS_2_ electrode has become very strong after 5 ch/disch cycles (Figure 7b), indicating that the ZrS_2_ coating has evidently mitigated the electrolyte from decomposition. We did not include the S 2p spectra, for they were very noisy and could not be well analyzed.

XPS spectra of Zr 3d, Li 1s, F 1s, and O 1s of a) the 20‐ZrS2 and b) 0‐ZrS2 electrodes after different ch/disch cycles at 1 C in the voltage windows of 3–4.3 V, compared to the pristine 20‐ZrS2 electrode.

Additionally, we mainly analyzed the spectra of Li 1s, F 1s, and O 1s, as shown in Figure 7. The Li 1s peak of the pristine 20‐ZrS_2_ electrode (Figure 7a) at ≈54.2 eV is assigned to the Li in the NMC811 lattices^[^ 51 ^]^ while the one at ≈56.2 eV is due to LiF^[^ 52 ^]^ probably produced during electrode fabrication. With increased ch/disch cycles, the LiF peak intensity increased but seemed stabilized after 50 ch/disch cycles. The production of LiF means electrolyte decomposition, but stabilization indicates no more decomposition. From the evolution of the Li 1s spectra, one can easily conclude that the electrolyte decomposition was limited but did not continue after 50 ch/disch cycles. In comparison, the LiF peak of the 0‐ZrS_2_ electrode became dominant while the lattice Li peak from NMC11 was not observable after 5 ch/disch cycles (Figure 7b). These results again indicated that the 0‐ZrS_2_ electrode had significant electrolyte decomposition while the 20‐ZrS_2_ electrode could effectively inhibit the electrolyte from decomposition. Corresponding to the Li 1s spectra, we observed a LiF peak at ≈685.7 eV^[^ 53, 54 ^]^ and a peak assigned to C‐F of PVDF at ≈687.8 eV ^[^ 53, 54 ^]^ from the F 1s spectra of the pristine 20‐ZrS_2_ electrode (Figure 7a). After 20 ch/disch cycles, the LiF increased while the C‐F peak decreased, which might also be contributed from P‐F of Li_x_PF_y_. Particularly, the intensity ratio between the two peaks remained unchanged after 50 ch/disch cycles. Furthermore, the O 1s peak of the pristine 20‐ZrS_2_ electrode at ≈529.8 is due to the lattice oxygen of NMC811 (Figure 7a).^[^ 55 ^]^ The binding energy of this lattice oxygen peak increased to ≈530 eV after 5 ch/disch cycles, dropped back to ≈529.8 eV after 20 ch/disch cycles, continuously dropped to ≈529 eV after 50 ch/disch cycles, and then remained unchanged in the subsequent 50 ch/disch cycles. The binding energy evolution of the O 1s peak of the lattice oxygen with cycling implies that the lattice oxygen became more stable during the first 50 ch/disch cycles and stabilized thereafter. In comparison, the binding energy of the O 1s peak of the lattice oxygen for the 0‐ZrS_2_ electrode is much higher, ≈533 eV in 50 ch/disch cycles (Figure 7b); that is, the lattice oxygen of the uncoated NMC811 electrode is more active. Compared to the 20‐ZrS_2_ electrode, in other words, the uncoated NMC811 electrode (i.e., 0‐ZrS_2_) is more prone to lose lattice oxygen. Additionally, the O 1s peak of the pristine 20‐ZrS_2_ electrode at ≈532.4 eV is attributed to O─C═O (Figure 7a).^[^ 51, 56 ^]^ Within 50 ch/disch cycles, the O 1s peak at ≈532.4 eV increased with the decreased lattice oxygen peak. It may imply some decomposition of the electrolyte and formation of CEI. Upon the completion of the first 50 ch/disch cycles, the O 1s peak at ≈532.4 eV shifted to ≈532 eV and further moved to ≈531.6 eV after 100 ch/disch cycles. However, the peak ratio between the two O 1s peaks was relatively stable from 50 to 100 ch/disch cycles.

Take all these Raman and XPS results together, one is easy to conclude in two aspects: 1) the ZrS_2_ has converted into Zr(SO_4_)2 or/and its hydrates if there is any water existing in the electrolyte and 2) some limited electrolyte decomposition occurred in the first 50 cycles but then became negligible in the following cycles.

The Benefits of the ZrS2 Coating on the Anode and the Cell

2.3

Besides the benefits of the ZrS_2_ coatings to the NMC811 cathode, we also investigated the possible benefits generated to the Li anode. To this end, we applied SEM to observe the cross‐sections of Li anodes coupled with the 0‐ZrS_2_ (Figure 8a) and 20‐ZrS2 cathode (Figure 8b). For the Li anodes coupled with the 0‐ZrS_2_ cathodes (Figure 8a), we found that the initial 500‐µm thick Li metal was reduced to 250, 113, and 50 µm after 100, 300, and 500 ch/disch cycles, respectively. The elemental mapping of these cross‐sections clearly showed the SEI layers containing P, C, O, and F elements, while the unused Li metal layers were black. For the Li anodes coupled with the 20‐ZrS_2_ cathodes (Figure 8b), in comparison, we found that the initial 500‐µm thick Li metal were reduced to 450, 312, and 300 µm after 100, 300, and 500 ch/disch cycles, respectively. Apparently, the 20‐ZrS_2_ cathode significantly reduced the consumption of Li metal anodes, as summarized in Figure 8c. This should be ascribed to less harmful species generated from the 20‐ZrS_2_ cathode and fewer negative effects produced by the 20‐ZrS_2_ cathode. In other words, the 20‐ZrS_2_ cathode significantly mitigated the crosstalk between the NMC811 cathode and the Li anode. We further verified this by detecting the Ni content on the Li anodes after different ch/disch cycles, as shown in Figure 8d. Very remarkably, there were significant Ni contents detected on the Li anodes coupled with the 0‐ZrS_2_ cathodes, which nearly increased linearly from ≈0.07 at.% after 100 ch/disch cycles to ≈0.26 at.% after 300 cycles and ≈0.55 at.% after 500 cycles. For the Li anodes coupled with the 20‐ZrS_2_ cathodes, In sharp contrast, only ≈0.05 at.% of Ni were detected on the Li anode after 500 cycles, 10 times lower than the one detected on the Li anode coupled with the 0‐ZrS_2_ cathode. All these are very encouraging for developing Li||NMC811 LMBs.

SEM and EDS analysis on Li anodes. SEM imaging and EDS mapping on Li anodes of a) Li||0‐ZrS2 and b) Li||20‐ZrS2 cells after 100, 300, and 500 ch/disch cycles. c) Li consumption and d) Ni concentration of Li anodes in Li||0‐ZrS2 and Li||20‐ZrS2 cells after 100, 300, and 500 ch/disch cycles. The cells were cycled at 1 C ch/disch in the voltage windows of 3–4.3 V.

To summarize, the ZrS_2_ coating exhibited multiple benefits on the NMC811 cathode and the Li anode: 1) evidently mitigate irreversible structural transition, 2) remarkably protect the electrolyte from oxidation, 3) significantly reduce interfacial reactions, 4) greatly enhance the mechanical integrity of NMC811 cathodes, 5) tremendously alleviate microcracking of cathodes, 6) considerably decrease TM ion dissolution and shuttling, 7) notably improve the air stability of NMC811, and 8) visibly protect Li anodes from consumption. All these benefits were further reflected on the Li||NMC811 cell's impedance, as shown in Figure S11 (Supporting Information). The Li||0‐ZrS_2_ cell's impedance was comparable to that of the Li||20‐ZrS_2_ cell after the 1st ch/disch cycle (Figure S11a, Supporting Information). After 50 ch/disch cycles, however, the Li||0‐ZrS_2_ cell's impedance was evidently larger than that of the Li||20‐ZrS_2_ cell (Figure S11b, Supporting Information). This was even more distinct after 100 cycles (Figure S11c, Supporting Information). To better compare these cells’ impedances after different cycles, we summarized their values in Figure S11d (Supporting Information), in which R_f_ is the Li^+^ diffusion resistance in the cathode surface layer, and R_ct_ is charge transfer resistance at the electrode/electrolyte interface. The equivalent circuit is shown in Figure S11e (Supporting Information), in which R_s_ is the solution resistance and the small intercept, while W is Warburg impedance corresponding to the straight line at low frequency and reflects the Li^+^ diffusion through the solid electrode. Clearly, the 0‐ZrS_2_ cell's impedance increased dramatically with cycles, while the 20‐ZrS_2_ cell's is much smaller and more stable.

In short, the ZrS_2_ coatings demonstrated tremendous benefits for protecting NMC811 cathodes and for achieving high performance. Thus, this study is inspiring and has great potential for commercializing NMC811 cathodes.

Conclusion

3

In this study, we investigated nanoscale ZrS_2_ films as a novel coating material for addressing the structural and interfacial issues of NMC811 cathodes. We deposited these nanofilms on prefabricated NMC811 electrodes using ALD at 150 °C. To achieve the best electrochemical performance of NMC811 electrodes, we found that the optimal coating thickness of ZrS_2_ coatings is around 2 nm (20 ALD cycles of ZrS_2_). Very compellingly, the resultant 20‐ZrS_2_ electrode (i.e., the NMC811 electrode coated by 20‐cycle ALD ZrS_2_) still could sustain 60% of its initial capacity after 1300 ch/disch cycles while the 0‐ZrS_2_ electrode (i.e., bare NMC811 electrode) dropped to 60% of its initial capacity after 300 ch/disch cycles. Such a dramatic improvement in the cathode performance made it very interesting to unveil the underlying mechanisms. To this end, we convincingly disclosed that ZrS_2_ could serve as a scavenger for oxygen released from NMC lattices. As a consequence, the ZrS_2_ coating protected the electrolyte from oxidation and helped avoid many harmful reactions. By reacting with oxygen, In addition, the ZrS_2_ coating converted into Zr(SO_4_)2 or/and its hydrates, while the resulting Zr(SO_4_)2 continued to protect NMC811 powders from microcracking, undesirable interfacial reactions, any further oxygen release, and TM ion dissolution. These beneficial effects on the NMC811 cathode were further reflected on the Li anode. As a return, Li anode was consumed much slower and the cell exhibited a much lower impedance, compared to the cell without the ZrS_2_ coating. Thus, this study not only reported a new coating but actually pave a feasible pathway for commercializing NMC811 cathodes.

Experimental Section

4

Electrode Preparation

In this study, the NMC811 electrode was prepared with a recipe of 86 wt.% NMC811 powder (MSE Supplies, USA), 7 wt.% polyvinylidene fluoride (PVDF, HSV900, MTI Corporation, USA), and 7 wt.% carbon black (Timical Super C65, MTI, USA). The mixture was added with a certain amount of N‐methyl‐2‐pyrrolidinone (NMP, 99.5% in purity, Sigma–Aldrich, USA) and sufficiently mixed using a mixer (THINKY AR‐100, USA), resulting in a homogenous slurry. The slurry was then cast onto an aluminum (Al) foil. The resulting NMC laminates were dried under vacuum at 120 °C for 12 h. The mass loading of the NMC811 active material was ≈8.0 mg cm.^−2^

ZrS2 ALD Coating

The ZrS_2_ ALD process was developed in one of our recent studies.^[^ 37 ^]^ In this study, we deposited ZrS_2_ onto prefabricated NMC811 electrodes at 150 °C using a Savannah 200 ALD system (Veeco, USA) integrated with an argon (Ar)‐filled glove box. Such an integrated ALD‐glovebox system ensured that, after the ZrS_2_ coating, the coated NMC811 electrodes would not be exposed to air and could be made into coin cells directly. The ZrS_2_ ALD process was performed using tetrakis(dimethylamido)zirconium (TDMA‐Zr, Strem Chemicals, USA) and 4 at.% H_2_S (balanced by Ar) as the precursors. It needs to point out that H_2_S should be employed with caution, due to its toxicity. To date, H_2_S was the most widely used precursor as the sulfur source for growing sulfide via ALD.^[^ 57 ^]^ To ensure a sufficient vapor pressure, TDMA‐Zr was heated to 75 °C. Before the ALD deposition, NMC811 electrodes were loaded and stabilized inside the ALD chamber for at least 1 h. Each ALD cycle consisted of four sequential steps, i.e., dosing TDMA‐Zr‐purging‐dosing H_2_S‐purging. Each of these steps consumed a certain time, i.e., t_1_ − t_2_ − t_3_ − t_4_ in sequence, where t_1_, t_2_, t_3_, and t_4_ were set as 0.1, 10, 1, and 10 s, respectively. The NMC electrodes were coated with conformal ZrS_2_ nanofilms of different thicknesses by varying ALD cycles from 5 to 10, 20, and 40. The growth per cycle (GPC) of the ALD ZrS_2_ at 150 °C was ≈ 1 Å cycle^−1^.^[^ 37 ^]^ Nominally, thus, the corresponding film thicknesses were 0.5, 1, 2, and 4 nm, respectively. To identify these uncoated (bare) and ALD‐coated NMC811 electrodes, we denoted them as **0‐ZrS_2_ ** (bare), **5‐ZrS_2_ ** (5‐cycle ALD ZrS_2_), **10‐ZrS_2_ ** (10‐cycle ALD ZrS_2_), **20‐ZrS_2_ ** (20‐cycle ALD ZrS_2_), and **40‐ZrS_2_ ** (40‐cycle ALD ZrS_2_), respectively. Additionally, ZrS_2_ was also deposited on a silicon (Si) wafer for 100 ALD cycles for analysis using X‐ray photoelectron spectroscopy (XPS).

Molecular Layer Deposition (MLD)

MLD, a sister technique of ALD, was specifically for growing polymeric and hybrid films.^[^ 58 ^]^ Using either trimethylaluminum (TMA, Strem Chemicals, USA) or diethylzinc (DEZ, Strem Chemicals, USA) to pair with glycerol (GL, Sigma–Aldrich, USA) as precursors, we deposited a layer of aluminum‐based GL (i.e., AlGL) films (25‐100 MLD cycles, ≈6–25 nm thick) or zinc‐based GL (i.e., ZnGL) films (25‐100 MLD cycles, ≈6–25 nm thick) over the ALD ZrS_2_ nanofilms (which were deposited on NMC811 cathodes and Cu and Al foils) as a capping layer. Such an AlGL or ZnGL MLD coating has been used in one of our previous studies and performed at 150 °C.^[^ 35 ^]^ This MLD AlGL or ZnGL capping layer could well help protect ZrS_2_‐coated samples from air exposure during the process of installing the samples for XPS analysis. The AlGL or ZnGL deposition process was conducted in four steps for each MLD cycle: 1) dosing TMA/DEZ for 0.05 s, 2) purging for 60 s using Ar to remove oversupplied TMA/DEZ and the byproduct, methane (CH_4_)/ethane (C_2_H_6_), 3) dosing GL for 2 s, and 4) purging for 60 s using Ar to clean oversupplied GL and the byproduct, CH_4_/C_2_H_6_.

Electrochemical Evaluations

Both bare and ZrS_2_‐coated NMC811 cathodes and Li chips were punched into 7/16′’ discs and thereby assembled into CR2032 coin cells for electrochemical evaluations. Coin cells were assembled in the Ar‐filled glove box integrated with the ALD/MLD system. The O_2_ and H_2_O levels of the glovebox were strictly controlled below 0.01 ppm. The resultant Li||NMC811 cells used Celgard 2325 membrane as their separators. The electrolyte was 1.2 m LiPF_6_ in ethylene carbonate (EC)/ethylmethyl carbonate (EMC) (3:7 by weight) (Sigma‐Aldrich, USA). The assembled cells were rested for 10 hours at room temperature before electrochemical testing. Electrochemical testing was conducted using a constant current (CC) mode. Different testing protocols were adopted: 1) 0.5 C for both charge and discharge; 2) 0.5 C for charge and 1 C for discharge, and 3) 1 C for both charge and discharge. The electrochemical tests were conducted using a Neware battery test system. Rate capability was also conducted under various C‐rates (i.e., 0.1, 0.2, 0.5, 1, 2, and 5 C). These coin cells were tested in the voltage window of 3.0–4.3 V vs Li/Li⁺. Coin cells were mainly tested at room temperature without using a temperature chamber and exhibited some capacity fluctuations due to the fluctuating room temperature. Some coin cells were tested using a temperature chamber (Maccor, MTC‐020, USA) at room temperature and 45 °C, while they showed very smooth capacity changes with charge/discharge cycles. Electrochemical impedance spectroscopy (EIS, BioLogic SP 200, USA) was used for measuring the evolution of cell impedance in the frequency range from 100 kHz to 10 mHz at an amplitude voltage of 5 mV.

Materials Characterization

The morphological characteristics and elemental distribution of NMC811 electrodes were analyzed using advanced imaging and spectroscopy techniques. SEM (XL30, Philips FEI) equipped with EDS was performed at the Arkansas Nano & Bio Materials Characterization Facility, University of Arkansas (Fayetteville, AR, USA). We used a Renishaw inVia confocal micro‐Raman microscope for conducting Raman measurements at the University of Arkansas AIMRC facility, which has a 532 nm excitation laser with a 1200 lines/mm grating. Raman spectra were collected using a 50× objective lens. The system was calibrated using the silicon reference peak at 520.5 cm^−1^ prior to measurement to ensure spectral accuracy. Raman measurements were conducted on commercial Zr(SO_4_)2.4H_2_O, ZrS_2_, and ZrO_2_ powders (Sigma–Aldrich), and the received spectra were used as references to analyze cycled ZrS_2_‐coated electrodes. Raman measurements were also performed on bare and ZrS_2_‐coated electrodes before and after cycling.

XPS analysis was conducted on various samples using a Thermo Scientific Model K‐Alpha XPS system (Thermo Scientific LLC, Madison, Wisconsin) with a monochromatic Al Kα radiation (1486.7 eV) at Center for Integrative Nanotechnology Sciences (CINS) at the University of Arkansas at Little Rock (UALR).

TEM studies were conducted at the Center for Nanoscale Materials (CNM), Argonne National Laboratory (ANL, IL, USA). Specimens were prepared from pristine and cycled NMC811 electrodes using a FIB system (Thermo Fisher Scientific Helios 5 CX) with a standard lift‐out procedure. Elemental mapping was performed using STEM (FEI Talos F200X) equipped with a SuperX EDS detector. HR‐TEM experiments were carried out on a JEOL 2100F operated at 200 kV. Synchrotron XRD studies were carried out to analyze bare and ZrS_2_‐coated NMC811 samples at the 17‐BM‐B rapid acquisition powder diffraction beamline of Advanced Photon Source (APS, ANL, IL, USA). The X‐ray wavelength was 0.25251 Å and the diffraction data were collected with a Perkin‐Elmer Area detector in transmission mode with Q‐range up to 7 Å^−1^. The powder samples were sandwiched between Kapton tapes inside the Ar‐filled glovebox for the experiment. The instrument was calibrated using LaB_6_ as a standard and the collected 2D images were integrated using GSAS‐II software package to obtain the 1D XRD patterns.

Conflict of Interest

The authors declare no conflict of interest.

Author Contributions

X.M. conceived the concept, proposed the project, designed the experiments, conducted data analysis, and drafted and revised the manuscript. K.V.C. conducted electrochemical evaluations of cells, EIS measurements, SEM characterization, Raman measurements, and ALD and MLD growth. F.W. conducted XPS characterization, X.M. and K.V.C. conducted XPS and Raman analysis, Y.L. conducted TEM characterization. K.V.C. and X.M. conducted TEM analysis. J.C. and H.Z. performed XRD measurements. X.M. and W.L. discussed the project and data analysis.

Supporting information

Supporting Information