Atomic Layer Deposition of Al-W-Fluoride on LiCoO2 Cathodes: Comparison of Particle- and Electrode-Level Coatings.

Atomic layer deposition (ALD) of the well-known Al2O3 on a LiCoO2 system is compared with that of a newly developed AlW x F y material. ALD coatings (∼1 nm thick) of both materials are shown to be effective in improving cycle life through mitigation of surface-induced capacity losses. However, the behaviors of Al2O3 and AlW x F y are shown to be significantly different when coated directly on cathode particles versus deposition on a composite electrode composed of active materials, carbons, and binders. Electrochemical impedance spectroscopy, galvanostatic intermittent titration techniques, and four-point measurements suggest that electron transport is more limited in LiCoO2 particles coated with Al2O3 compared with that in particles coated with AlW x F y . The results show that proper design/choice of coating materials (e.g., AlW x F y ) can improve capacity retention without sacrificing electron transport and suggest new avenues for engineering electrode-electrolyte interfaces to enable high-voltage operation of lithium-ion batteries.

Introduction

To meet the ever-increasing demands placed on lithium-ion batteries for both transportation applications and portable electronics, higher energy lithium-ion batteries with longer calendar life, lower cost, and greater safety are under development. An obvious strategy for obtaining higher energies is to operate batteries at higher voltages, thus extracting more energy per charge from the system. However, battery lifetimes under high-voltage operation (typically ≳4.2 V vs graphite) are short-lived owing to deleterious side reactions between charged electrodes and organic electrolytes.[1−3] Coating the surface of electrodes in an effort to mitigate unfavorable side reactions has proved to be a promising route for improving various performance metrics of lithium-ion batteries.[4−6] Several coating techniques and processes, including physical mixing, wet-chemical routes (sol–gel), and physical/chemical vapor deposition (PVD/CVD), are, among others, available to modify electrode surfaces. Atomic layer deposition (ALD), a modified CVD process, has been used extensively to enhance the stability and safety of positive and negative electrodes.[7−10] ALD is versatile compared with traditional coating methods, in that protection layers can be deposited not only on electrode material powders but also directly on electrode laminates that are composed of active materials, carbons, and polymeric binders. This unique capability, combined with the self-limiting nature of ALD surface chemistry, makes ALD attractive compared with conventional wet-chemical processes. Furthermore, ALD does not require postheating treatments as do wet-chemical methods.

Previous reports of ALD-coated laminates and powders have shown both routes to be effective in improving the durability and performance of lithium-ion cells.[7,8,11,12] Coating directly on active materials, in general, allows scale-up (gram to kilograms) and easy implementation into current battery manufacturing processes.[13] In addition, processing of active materials can be extended to various coating materials that require high deposition temperatures (>250–300 °C) such as some fluorides and sulfides because the crystallinity and morphologies of active materials are not affected by the typical temperatures used. On the contrary, ALD processing of electrodes inherently limits the choice to lower temperature coating materials (<150–200 °C) to prevent the decomposition of polymeric binders in the laminates during the ALD process. ALD processes that require substrate temperatures significantly higher than the melting temperature of binders (∼160–180 °C) cannot be used for laminate coating. In addition, new ALD strategies, for example, roll-to-roll or spatial ALD, need to be developed for compatibility with current battery manufacturing processes.[14,15]

Although metal oxides such as Al2O3, ZrO2, and TiO2 have received the most attention in ALD protective coating studies, these materials are electrical insulators so that directly coating the particles may hinder charge transport (ionic and electronic) between particles. On the other hand, ALD coatings directly on electrode laminates should not affect particle-to-particle transport because diffusion paths for charge transport are established in the laminate-making process, prior to the ALD coating. In addition, coating laminates may also provide some level of protection for the inactive components, such as polymeric binders and carbons, which are susceptible to reaction with electrolytes. Nevertheless, as mentioned above, direct coating of electrodes limits the choice of coating materials owing to the lower temperature requirements.

With these considerations in mind, it is clear that the electrochemical properties of ALD-coated electrodes can be influenced by the choice of fabrication process and coating material. However, few comparisons between the two processes and the details involved exist with notable exceptions being works by Jung et al.[7] Therein, the authors reported that the ALD of Al2O3 coating on LiCoO2 particles and electrodes did not show appreciable differences in terms of discharge capacities.[7] Al2O3 coatings on graphite anode laminates were reported to have superior electrochemical properties compared to Al2O3 coatings directly on the graphite powder.[10] However, detailed electrochemical properties including voltage profiles, rate capabilities, and transport properties of electrons/ions were not reported.

To further understand the ALD process with respect to electrode versus particle coating and choice of materials, we report herein the effects of ALD coatings on powders versus laminates. We choose as a baseline the well-known Al2O3 on a LiCoO2 system and compare with a newly developed AlWF material. Despite being ultrathin (∼1 nm), and regardless of the ALD process and a coating material, we find that the electrochemical performance of LiCoO2 is dramatically improved. However, owing to the more insulating nature of Al2O3, coated LiCoO2 particles showed a large overpotential during lithium insertion, whereas AlWF-coated particles did not. This result reveals the importance of charge-transport considerations through coating layers when designing effective coating materials.

Experimental Section

Atomic Layer Deposition

Al2O3 layers were deposited using trimethylaluminum (TMA, Sigma-Aldrich, USA) and H2O, and AlWF layers were deposited using TMA and tungsten hexafluoride (WF6, Sigma-Aldrich, USA). Substrate (electrode laminates) temperatures were kept at 150 and 200 °C for Al2O3 and AlWF, respectively. For ALD coating on particles, LiCoO2 powder (Sigma-Aldrich, USA) was loaded into a customized stainless steel sample tray covered with a mesh to allow facile precursor diffusion and to keep the powders in the holder. Eight cycles of TMA–N2–H2O–N2 or four cycles of TMA–N2–WF6–N2 were used to deposit 1 nm coatings of Al2O3 or AlWF, respectively, on the powders or the laminates. Detailed procedures can be found in the previous reports.[11,12]

Electrochemical Measurements

The LiCoO2 composite electrodes were prepared by mixing a slurry of the oxide (Sigma-Aldrich, USA), Super P, and polyvinylidene difluoride (PVDF), followed by casting on aluminum foil and drying under vacuum at 110 °C overnight. The oxide/Super P/PVDF mass ratios were 84:8:8 in the Li metal half-cell studies. The 2032-type coin cells were assembled in an Ar-filled glovebox (water and oxygen ≤0.5 ppm) with Gen2 electrolyte (1.2 M LiPF6 in a mixture of ethylene carbonate and ethyl–methyl carbonate with a mass ratio of 3:7). Charge–discharge measurements were recorded on a MACCOR cycler at room temperature under different rates with various upper cutoff voltages ranging from 4.4 to 4.6 V (vs Li metal). The galvanostatic intermittent titration technique (GITT) measurements for powder-coated LiCoO2 electrodes and uncoated LiCoO2 electrodes were conducted between voltage windows of 2.5–4.4 V. Prior to the GITT measurements, a formation cycle at 20 mA/g between 2.5 and 4.4 V was conducted on each cell. The GITT protocol consisted of 10 mA/g current pulses for 10 min, followed by 300 min relaxation periods for voltage windows in the charge and discharge direction.

Transmission Electron Microscopy

Transmission electron microscopy (TEM) analysis was performed by the Evans Analytical Group (Sunnyvale, CA) on the thick films of Al2O3 and AlWF, deposited on silicon wafers. Cross-sectional TEM samples were prepared using the in situ focused ion beam (FIB) liftout technique on an FEI Strata DualBeam FIB/scanning electron microscope. The samples were capped with a protective layer of carbon prior to FIB milling and were imaged with an FEI Tecnai TF-20 FEG/TEM operated at 200 kV.

Results and Discussion

LiCoO2 powders and electrode laminates were coated with films of Al2O3 and AlWF. Film thicknesses were limited to ∼1 nm to mitigate thickness effects of coating materials.[7] All coated LiCoO2 particles and electrodes showed excellent capacity retention compared to uncoated LiCoO2 when cycled above 4.4 V, as expected, whereas significant impedance growth was observed for uncoated samples as depicted in Figure S1. Table summarizes the discharge capacities measured during the 1st, 2nd, and 50th cycles at the different upper cutoff voltages tested for LiCoO2 powders and laminates, bare and coated with Al2O3 and AlWF. The discharge capacities of all coated samples were similar. However, the voltage profiles were different between the powder and laminate coatings. Figure a,b shows charge–discharge voltage profiles for the coated LiCoO2 powders (straight lines) and laminates (dotted lines) on the 2nd cycle. For all of the samples, the main LiCoO2 electrochemical features are observable; for example, a long plateau at ∼3.9 V and transition peaks at ∼4.1 V (∼120–140 mA h/g). The typical sharp knee in the discharge at ∼3.8 V remains strong for all but the Al2O3-coated particles (Figure a), and we attribute this to the development of a strong overpotential on discharge for this sample. As such, reinsertion of lithium ions into the host structures is impeded and thus requires more energy. Such behavior may be resulted from the insulating nature of the coating that imposes kinetic barriers for electron and lithium-ion transport. It is noteworthy that the overpotential during lithium insertion (discharge) is higher than that during lithium extraction (charge). It is not clear why lithium insertion is more difficult than extraction with coating layers, but this could relate to the formation of Li2MO2-like phases on the oxide particle surface owing to sluggish lithium-ion diffusion through the coating material.[16] The laminate coatings, however, showed only small overpotentials. We attribute these smaller overpotentials to the fact that the electronic and ionic pathways between the active materials, carbon, and current collector within the composite electrode are maintained after the ALD laminate coating, in agreement with earlier reports.[10] It is interesting to note that AlWF-coated LiCoO2 showed little difference between the particle and laminate coatings compared to Al2O3 coatings. Although small overpotentials were observed for the AlWF coatings, they were significantly less pronounced than, for example, Al2O3-coated LiCoO2 particles. We speculate that the much higher electronic conductivity of AlWF reduces the overpotential compared with the electronically insulating Al2O3 coatings.[12]

Table 1

Summarized Electrochemical Performances of Uncoated and Coated LiCoO2 When Cycled between 2.5–4.4 and 2.5–4.5 V (vs Li/Li+)

sample	1st cycle (mA h/g)	2nd cycle (mA h/g)	50th cycle (mA h/g)	capacity retention (%)
uncoated LiCoO2	170.33 (4.4 V)	169.85	146.30	86.14
	194.08 (4.5 V)	192.69	107.40	55.74
Al2O3 particle	158.98 (4.4 V)	162.27	157.38	96.99
	175.72 (4.5 V)	176.96	167.93	94.90
Al2O3 laminate	153.46 (4.4 V)	165.81	162.45	97.97
	192.57 (4.5 V)	191.68	170.42	88.91
AlWxFy particle	166.10 (4.4 V)	167.61	165.48	98.73
	180.62 (4.5 V)	183.79	174.58	94.99
AlWxFy laminate	164.07 (4.4 V)	164.78	161.95	98.28
	188.12 (4.5 V)	189.17	183.04	96.76

Figure 1

Voltage profiles of (a) Al2O3- and (b) AlWF-coated LiCoO2 powders (straight lines) and laminates (dotted lines) when cycled between 4.4 and 2.5 V and (c,d) cycling stabilities of uncoated and coated LiCoO2 at 4.4–2.5 and 4.5–2.5 V.

Figure c,d shows cycling stabilities of particle- and laminate-coated LiCoO2 (Al2O3 and AlWF) cycled to upper cutoff voltages of 4.4 and 4.5 V along with uncoated LiCoO2. At voltages ≥4.4 V (vs Li/Li+), LiCoO2 is known to destabilize.[17] In particular, surface damage (e.g., oxygen loss and transition metal migration) may lead to slow phase transformation and progressive capacity fade. Therefore, 4.4 and 4.5 V were chosen for long-term cycling studies to judge the relative efficacy of coated materials for protecting LiCoO2 surfaces. All of the coated samples exhibited stable electrochemical performance compared to the uncoated samples (see Figure S1). Capacity differences between the powder and laminate coatings were small at both 4.4 and 4.5 V, except for the case of Al2O3 at 4.5 V. At 4.5 V, the Al2O3-coated laminates achieved ∼15 mA h/g more than the Al2O3-coated particles, likely because of the high impedance associated with coated particles as discussed above (see also Supporting Figures). However, the higher capacity of the coated laminate sample could not be maintained with cycling and possibly signals a more rapid breakdown of the Al2O3 film at this high voltage (4.5 V). By contrast, the AlWF coatings, on both powders and laminates, exhibited consistent capacities at both 4.4 and 4.5 V. It is also expected that the coatings provide superior rate capabilities than uncoated LiCoO2 owing to the absence of electrolyte oxidative products during high voltage of operation as reported earlier.[12] In addition, the capacity achieved at 4.5 V for the AlWF-coated laminates was similar to the capacity of the Al2O3-coated laminate at 4.5 V; however, this capacity could be maintained with cycling for the AlWF material with little change in the voltage profile. These data reveal the superior performance of the AlWF coatings on particles and laminates compared with that of Al2O3 coatings.

Coated (Al2O3 and AlWF) and uncoated LiCoO2 electrodes were characterized using electrochemical impedance spectroscopy (EIS) and GITT with current pulses (10 mA/g for 10 min) as reported earlier.[18]Figure a shows charge–discharge voltage profiles of all samples prior to EIS and GITT measurements. As expected, the Al2O3-coated LiCoO2 powder sample has a large overpotential, whereas the uncoated LiCoO2 has only a small overpotential when cycled at a slow current rate of 20 mA/g. GITT results (Figure b) for LiCoO2, with 0.35 < x < 0.55, corresponding to capacities of 100–150 mA h/g, show that lithium diffusion is slower in the coated materials, and little difference exists between the equilibrium potentials for the coated samples. Therefore, lithium transport across the ionically insulating Al2O3 and AlWF coatings is similar. Figure a shows representative Nyquist spectra of an AlWF-coated LiCoO2/Li half-cell between 10 kHz and 0.1 Hz with a 5 mV voltage perturbation. As expected, the spectrum consists of a high-frequency intersect (electrolyte impedance), mid-frequency (1k–10 Hz) semicircle corresponding to charge-transfer impedance at the electrode–electrolyte interface, and a low-frequency (<1 Hz) Warburg tail.[19,20] The Al2O3- and AlWF-coated sample curves began to deviate at ∼3.9 V when cycled at 20 mA/g, where EIS spectra at four different voltages starting at 3.8 V are shown in Figure b. The semicircle at high to mid frequencies at different discharge voltages, or depth of discharge (DOD), did not change significantly within and between samples, implying that the surface films at the electrode–electrolyte interface are stable at the given condition. The low-frequency tail below 1 Hz became straighter at low voltages because of low lithium diffusivity at low states of charge. Interestingly, magnitudes of impedance (normalized by mass) at the same potential, or the same DOD, are smaller for AlWF than for the Al2O3 coatings. For example, magnitudes of the impedance at 3.5 V and 0.1 Hz were found to be 3.08 and 3.40 Ω·g, whereas those at 90% DOD were found to be 2.87 and 3.40 Ω·g for AlWF and Al2O3, respectively. Because lithium-ion diffusion characteristics of the two coated materials were similar from GITT, the differences in EIS and overpotentials may be attributed to the differences in electronic transport.

Figure 2

(a) Voltage profiles of uncoated LiCoO2 and Al2O3- and AlWF-coated LiCoO2 before GITT and (b) GITT results at LiCoO2 (0.35 < x < 0.55).

Figure 3

(a) Representative Nyquist plot of coated LiCoO2 (AlWF) with frequencies and (b) Nyquist plot of Al2O3- and AlWF-coated LiCoO2 at different voltages or DOD.

Figure a shows a cross-sectional bright-field TEM image of an Al2O3–AlWF bilayer deposited on Si. The homogeneous, bright color of the Al2O3 layer suggests an amorphous, uniform material composed of light elements with low electron densities. The dark and light contrast of the AlWF layer, along with previous studies, suggests that AlWF coatings consist of small, W-containing particles embedded in a matrix of lower density material such as AlF3 when deposited under these conditions. A similar microstructure was measured for ALD AlMoF films deposited under similar conditions.[21] In addition, previous X-ray photoelectron spectroscopy studies on thick ALD AlWF films confirmed the metallic nature of tungsten or tungsten carbide, and resistivity values from four-point measurement on the film were found to be 10–2 Ω·cm, approximately 18 orders of magnitude smaller than that of Al2O3.[22] However, preliminary X-ray absorption spectroscopy data on AlWF-coated particles (not shown) suggest the presence of partially oxidized Wδ+ species. Therefore, it is likely that strong interactions between thin (∼1 nm) AlWF films and the underlying oxide particle substrate (e.g., LiCoO2) play an important and unique role in AlWF-coated, lithium–metal oxide systems. Similar phenomena are known for tungsten-based catalytic systems.[23] Further spectroscopic studies are ongoing and will be reported elsewhere.

Figure 4

(a) High-resolution transmission electron microscopy image of Al2O3 and AlWF layers deposited in silicon wafer and (b) schematic of charge transport through coating layers to active materials.

This set of results implies that the small overpotentials demonstrated for AlWF-coated LiCoO2 may be ascribed to enhanced electron transport with respect to Al2O3-coated LiCoO2. Furthermore, previous results suggest that transport properties in these coated materials may be tuned according to the nature of the integrated W-containing species, and further studies are ongoing to better understand the factors governing the performance.

Conclusions

Protective coatings of Al2O3 and AlWF, deposited via ALD on LiCoO2 particles and laminates, are shown to be promising options for effectively mitigating the capacity fade of LiCoO2 cathodes up to ∼4.5 V (vs Li/Li+). ALD coatings directly on laminates have small overpotentials relative to coatings deposited on cathode powders owing to the preservation of electronic pathways in coated, composite laminates. In the case of particle coating, Al2O3 layers were found to impose significantly higher impedance, especially during lithium insertion, relative to AlWF layers, which showed little energy penalty. Near equilibrium, open-circuit potentials from GITT measurements showed that lithium diffusion within Al2O3 and AlWF is similar. However, the incorporation of unique, metal-containing species in the AlWF layers, by way of the novel ALD process introduced,[12,22] may provide a way to enhance, or tune, the electronic properties of such films while simultaneously enhancing surface stability at high voltages.