Mitigating the Surface Degradation and Voltage Decay of Li1.2Ni0.13Mn0.54Co0.13O2 Cathode Material through Surface Modification Using Li2ZrO3.

In the quest to tackle the issue of surface degradation and voltage decay associated with Li-rich phases, Li-ion conductive Li2ZrO3 (LZO) is coated on Li1.2Ni0.13Mn0.54Co0.13O2 (LNMC) by a simple wet chemical process. The LZO phase coated on LNMC, with a thickness of about 10 nm, provides a structural integrity and facilitates the ion pathways throughout the charge-discharge process, which results in significant improvement of the electrochemical performances. The surface-modified cathode material exhibits a reversible capacity of 225 mA h g-1 (at C/5 rate) and retains 85% of the initial capacity after 100 cycles. Whereas, the uncoated pristine sample shows a capacity of 234 mA h g-1 and retains only 57% of the initial capacity under identical conditions. Electrochemical impedance spectroscopy reveals that the LZO coating plays a vital role in stabilizing the interface between the electrode and electrolyte during cycling; thus, it alleviates material degradation and voltage fading and ameliorates the electrochemical performance.

Introduction

Lithium-rich phases have been widely investigated as prospective high-capacity cathode materials for Li-ion batteries to be used in plug-in hybrid electric vehicles or electric vehicles. In particular, the lithium-rich transition metal oxide with the general formula of xLi2MnO3·(1 – x)LiMO2 (M = Mn, Co, Ni) is a significant milestone in materials design owing to its well-balanced capacity/power/cost, which is tailored for high operating voltage in comparison with the existing state of the art commercial cathode materials, such as LiCoO2, LiMn2O4, and LiFePO4.[1−5] Yet, there are vital snags which impede the commercial progress, such as (i) large irreversible capacity loss in the first cycle, (ii) inevitable phase transformation from layered to spinel phase leading to continuous decay in the discharge capacity,[6−9] (iii) oxygen loss from the crystal lattice which triggers serious safety concerns and low initial Coulombic efficiency, (iv) structural instability due to Mn dissolution, (v) voltage decay,[10−12] (vi) weak interfacial stability between the electrode and electrolyte, and (vii) thermal stability. Recently, several strategic approaches have been attempted to tackle the associated shortcomings of lithium-rich layered cathodes by adopting procedures such as surface coating, acid treatment, anion/cation doping,[13,14] innovative synthesis method,[15,16] substitution of heavy and bulky elements like Sn, Ti, Mo, and so forth.[17,18]

All of these techniques have beefed up the commercialization aspects of Li-rich layered cathode materials. Among all strategies, surface modification has proven to be an effective method to surmount the shortcomings of Li-rich phases. Surface coatings with metal oxides,[19−23] metal phosphates,[24,25] and metal fluorides[26] have thus been pursued to enhance the physical and chemical properties of lithium-rich cathode materials. Such surface modification improves the cycling stability, rate capability, and thermal stability of cathodes, as a result of suppression of oxygen-ion vacancies in the crystal lattice at the end of the first charge. Further, surface coating suppresses the undesired electrochemical reactivity of the electrolyte at the electrode/electrolyte interface by preventing direct contact between the electrode and electrolyte. However, most of these coating materials do not favor interfacial charge transfer of the electrode material; further, they act as a barrier for Li+ conduction and increase the diffusion path length of Li ion, resulting in inferior rate capability.

To overcome the limitation of the impeded lithium ion diffusion through the surface barrier, lithium-ion conducting material has been introduced as a protective layer on lithium-rich phases. This ion-conducting layer facilitates lithium-ion transportation between the cathode material and electrolyte; it also stabilizes their interface at high voltages.[27,28] In recent years, Li2ZrO3 (LZO) coating has gained immense interest for its role in the stability of cathodes.[29−31] In this context, Wu et al. have demonstrated a high rate and long cycle life performance of the LZO-coated LiNi0.5Co0.2Mn0.3O2 cathode.[32,33]

In the present article, we envisage the LZO coating to tackle specific issues associated with Li-rich layered phases, such as O2 evolution, reactivity of evolved O2 with the electrolyte, Mn dissolution, and phase transformation. The monoclinic LZO is structurally related to the layered Li1.2Ni0.13Mn0.54Co0.13O2 (LNMC). Hence, it is anticipated to enhance the structural integrity owing to an epitaxial relation between the host cathode and the Li-ion conducting surface layer. This epitaxial interface may play a beneficial role and effectively separate the bulk electrode from the electrolyte and stabilize the structure and the composition during high-voltage cycling. This article reports a systematic investigation of the role of the Li-ion conducting LZO surface layer in overcoming the issues with LNMC.

Results and Discussion

Figure a,b compare the X-ray diffraction patterns of bare LNMC and LZO surface-modified LNMC (LZO @ LNMC). Both the patterns are indexed as hexagonal α-NaFeO2 structures with R3̅m space group. The well-defined splitting of the reflections, (006)/(102) and (108)/(110), manifests in the crystallization of the layered structure, without formation of the spinel phase. The weak reflections appearing between 20 and 30° (2θCu) can be indexed considering a √3ahex × √3bhex superstructure in the ab 3d metal planes as a result of an ordering between the Li, Ni, Co, and Mn ions similar to the monoclinic Li2MnO3 structure (represented in the C2/m space group).[35,36] The integrated intensity ratio of the (003) to (104) reflections used for quantification of cation mixing in the lattice being 1.40, which is greater than 1.2, provides evidence of the lesser degree of cation mixing in the Li layer, whereas in the case of the surface-modified pattern, the I(003)/I(104) ratio increases to 1.47.[37] This indicates that the degree of cation mixing depreciates by surface modification. Further, it suggests that there is no severe change in the host layered structure of LZO @ LNMC, except for some peaks that are enlarged in the inset of Figure b. The peaks marked in “*” are the characteristics of LZO corresponding to the monoclinic system with 2/c space group, which is consistent with the JCPDS card: 01-070-8744. Figure shows the Rietveld refinement of XRD patterns for surface-modified and pristine material using Fullprof software. The lattice parameters (a, b, and c) of the surface-modified sample are same as those of the pristine pattern, suggesting that the surface coating with LZO does not cause any undue bulk structural changes. The structural parameters determined by Rietveld refinement for both pristine and coated samples using R3̅m and C2/m phases are listed in Tables S1 and S2, respectively.

Figure 1

Powder XRD patterns recorded with Cu Kα radiation for (a) pristine LNMC, (b) 5% LZO @ LNMC. The diffraction peaks in the inset figure marked with “*” correspond to LZO.

Figure 2

Rietveld refinement pattern of powder XRD data for (a) LNMC, and (b) 5% LZO @ LNMC, with the experimental data (black dots), calculated pattern (red line), difference curve (blue line), and Bragg diffraction positions (yellow ticks for R3̅m space group and green ticks for C2/m space group).

FESEM images of bare and 5 wt % LZO-coated LNMC particles are displayed in Figure a,b. No significant difference is observed in the morphology and grain size between pristine and LZO @ LNMC, whereas a closer observation of the surface reveals the relatively smooth grains for the sample of surface-modified LZO. EDS has been employed to elucidate the uniform distribution and analyze the elemental composition of the surface-modified sample. The atomic ratio of Ni/Mn/Co/Zr is 15.11:63.4:15.35:6.03, which is in concordance with that of the targeted composition of the phase, that is, 0.12:0.51:0.12:0.05, respectively. Elemental mapping in Figure d–h of the surface-modified sample clearly demonstrates the uniform distribution of zirconium element on the surface.

Figure 3

FESEM images of (a) bare LNMC, (b) 5% LZO-coated LNMC and (c) EDAX profile, (d–h) elemental mapping images of 5% LZO-coated sample.

HRTEM studies have been carried out to acquire further information pertaining to the interface between the LZO coating and bulk LNMC material (Figure a,b). Both the figures correspond to high-resolution transmission electron microscopy (HRTEM) images of the rectangle tagged interfacial area of their inset figures. The images show uniform surface layers of thickness of around 10 nm. In Figure a, the lattice spacings of 4.7 nm for bulk and 4.3 nm for the surface coating are in good concurrence with (003) and (110) interplanar spacing of isostructural hexagonal LNMC and tetragonal LZO, which is complemented by the XRD results. The (110) plane of LZO, oriented along the direction of (003) plane of LNMC shows perfect epitaxial alignment.

Figure 4

(a, b) High-resolution transmission electron microscopy images showing epitaxial orientation of LZO on LNMC particles.

Similarly, in Figure b, the (112) plane of LZO are oriented around 60° with respect to the (003) plane of the core LNMC lattice. This cross-wire epitaxial arrangement results in negligible interfacial lattice mismatch {(4.7 – 1.99/cos 60)/4.7 = 15%}. Such a slant epitaxial growth in the nanoheterostructure is reported for other systems.[38,39] The lattice mismatch calculation used here is adopted from the previous works.[38,39] The epitaxial growth with the preferred orientation and alignment of LZO on LNMC provides structural integrity and possesses a Li-ion diffusion pathway.

X-ray photoelectron spectroscopy (XPS) of Li1s, Ni2p, Co2p, and Mn2p core levels in the pristine and coated samples have been carried out to understand the oxidation states of these elements in these materials. Ni2p, Co2p, Mn3p, Zr3d, O1s, and C1s core-level spectra are shown in Figure . The Li1s peak observed at 55.5 eV (given in Figure S1) in both samples corresponds to Li+ species.[40] The Ni2p3/2 peak at 854.6 with the satellite peak at 860.7 eV is attributed to Ni2+ species.[41] Co2p3/2,1/2 peaks at 780 and 794.9 eV, respectively, can be assigned to either Co2+ or Co3+ species. However, the presence of a weak satellite peak at 789.6 eV related to the 2p3/2 main peak, with a distance of around 9 eV, explicitly confirms the presence of Co3+ in both of the samples.[42,43] Mn2p3/2,1/2 core level peaks at 642.4 and 654.0 eV in the coated and uncoated samples are assigned to Mn4+ species, respectively.[41] Zr3d5/2,3/2 peaks at 182.3 and 184.4 eV in the surface-modified sample shown in Figure are associated with Zr4+ species, respectively.[44] The envelopes of O1s core-level spectra in both the samples are broad, which are deconvoluted into three peaks corresponding to the oxide-adsorbed hydroxyl and water molecules. The main peak at 530.1 eV stands for oxide species, whereas the higher binding energy peaks at 531.8 and 533.9 eV are related to OH– and H2O species, respectively.[45] The LZO-coated sample shows a relatively large amount of adsorbed OH– and H2O species compared to that in the uncoated sample. Similarly, C1s spectra shown in the figure are broad in nature and can be deconvoluted into several component peaks. The sharp peak appearing at 284.5 eV corresponds to adventitious carbon species. The broad peak at 286.5 eV is ascribed to oxidized carbon species containing C–O bonds. The third peak at 289.4 eV is assigned to the surface-adsorbed carbonate group.

Figure 5

XPS spectra of pristine and LZO @ LNMC samples: Ni2p, Co2p, Mn2p, and Zr3d core levels and deconvoluted O1s and C1s core levels.

The capacity versus voltage profile for the Li half-cells of the pristine and 5% LZO @ LNMC cathodes cycling at C/5 current rate in the potential range of 1.75–4.8 V are shown in Figure a. The charge profile is composed of a sloping region below 4.4 V, followed by a long plateau region around 4.5 V. Both of the samples show almost the same specific capacity in the sloping region, which corresponds to the oxidation of the transition metals, such as Ni2+ to Ni4+ and Co3+ to Co4+. The specific capacity of the plateau region begins at around 4.5 V, which is accompanied by the extraction of Li from the 3d metal layer and also the removal of lattice oxygen. It is to be noted here that the LZO-coated cathode exhibits a lower amplitude in the plateau region. This difference in capacity is speculated to be attributed to the slump in the oxygen loss from the lattice. The pristine Li-rich NMC exhibits an initial charge capacity of 329 mA h g–1, whereas the 5% LZO-coated LNMC shows initial charge capacity of 306 mA h g–1, part of this decrease in capacity is also attributed to the presence of 5% (in weight) electrochemically inactive LZO in the electrode material. The bare LNMC exhibits a discharge capacity of 234, with 95 mA h g–1 of irreversibility, whereas the 5% surface-modified sample exhibits an initial capacity of 225 mA h g–1, with 80 mA h g–1 irreversibility. This decrease in irreversible capacity is mainly attributed to the suppression of solid-electrolyte interface (SEI) formation by the LZO surface barrier.

Figure 6

(a) First cycle charge–discharge profiles of the bare and 5% LZO @ LNMC, (b) cycling performance of bare and 5% LZO @ LNMC, (c) rate capability plot of bare and surface-modified LNMC, (d) discharge profiles at C/5 rate, and (e) corresponding differential capacity curves of the pristine and 5% LZO @ LNMC samples in the voltage range of 1.75–4.8 V.

To evaluate the effect of the LZO coating on LNMC, charge–discharge cycles have been conducted continuously at a current rate of C/5 between 1.75 and 4.8 V. Figure b shows the cycle number versus specific discharge capacity for pristine LNMC and LZO-coated LNMC samples. The surface-modified cathode shows a superior retention of 85% after 100 cycles. In contrast, the bare sample shows a rapid decline in the specific discharge capacity, with retention of 57% after 100 cycles under identical conditions. The rate capability test has been carried out for both the samples at different current rates. It can be seen from Figure c that the LZO surface-modified cathode delivers excellent discharge capacities of 225, 196, 160, 100, and 55 mA h g–1 at C/5, C/2, 1C, 2C, and 3C rates, respectively. On the other hand, the pristine material delivers 234, 170, 125, 60, and 30 mA h g–1 at C/5, C/2, 1C, 2C, and 3C rates, respectively. This large difference in rate performance is mainly attributed to the existence of fast lithium-ion diffusion pathway in the LZO surface barrier.

The comparison of the discharge voltage profile as a function of the cycle numbers for pristine and LZO surface-modified LNMC is displayed in Figure d. There is an abrupt decline in the discharge capacity as well as voltage fading for uncoated LNMC due to the surface deterioration owing to the side reactions at the electrode and electrolyte interface.[46,47] On the other hand, the surface-modified cathode exhibits a fairly stable discharge capacity and comparatively lower voltage decay. The differential capacity (dQ/dV) plots of both materials are shown in Figure e. The comparison of the dQ/dV plots gives a better understanding of voltage fading. The peak seen in the dQ/dV plot above 4 V corresponds to reduction of Ni4+/Ni2+, the peak around 3.7 V corresponds to Co4+/Co3+, and the peak below 3.5 V is ascribed to the reduction of Mn4+ and of oxygen.[48,49] The uncoated LNMC shows that the peaks below 3.5 V correspond to Mn reduction shifting toward lower potential, also a new peak starts to evolve at lower potential. This voltage hysteresis is attributed to the gradual accumulation of spinel environment in the crystal structure. The layered to spinel phase transformation is attributed to decrease in the Mn content in the structure due to dissolution.[6,8] In contrast, the structural integrity between LZO coating and the surface of cathode impede the diffusion of Mn and hence limit the layered to spinel phase transformation.

Electrochemical impedance spectroscopy (EIS) has been employed to elucidate the improved performance of the surface-modified cathode. Figure a,b displays the Nyquist plots for fresh electrode and cycled electrode at various cycles of pristine and LZO-coated LNMC half-cells. The inflated high frequency region of Figure a,b are shown in Figure c,d respectively. As seen from the plots, the semicircle in the mid frequency region is merged with that of the high frequency region. This indicates the overlapping of the SEI layer resistance and charge-transfer resistance. The sloping region at low frequencies represents the Warburg resistance, which is attributed to Li+ diffusion through the bulk electrode. The EIS data have been simulated by Zsimpwin software using the equivalent circuit shown in the inset of Figure a. In the equivalent circuit, Rs represents the solution resistance, Rsl is the resistance of the SEI layer, Rct is the charge-transfer resistance, W represents the Warburg resistance, and CPE1 and CPE2 represent the double layer capacitance and capacitance of the passivation film, respectively. The values of Rs, Rsl, and Rct obtained from the simulation of EIS data are depicted in Table S3. The solution resistance (Rs) is negligible and almost same for all of the samples. Rct value of the pristine electrode before cycling is 196.5 Ω, and it drastically increases to 1396 Ω after 80 cycles. The reason for such a significant hike in the charge-transfer resistance is attributed to the undesirable reaction between electrode/electrolyte ensuing a slump in surface charge transfer, which is confirmed by increasing the SEI layer resistance (Rsl) from 324.2 to 809.7 Ω after 80 cycles. The charge-transfer resistance of the LZO-coated sample before cycling is 379.7 Ω, which is higher compared to that of the uncoated sample probably because of the presence of surface-adsorbed species, which are revealed from the XPS and also IR studies (Figure S2). The adsorbed species are removed during the first charge, resulting in drop of the Rct value to 121.3 Ω. The Rct and Rsl values together increase as a function of cycles but the increasing trend is much lower than that of the pristine sample. The increase in Rsl and Rct in the uncoated sample correlates well with the capacity fading observed during cycling. The loss of lithium content is attributed to the formation of undesirable SEI on the surface. On the contrary, the electrochemically passive LZO surface barrier provides greater benefits by protecting the active sites on the surface of LNMC from acidic electrolyte, induced with oxygen at high voltage. Also, the Li+ conducting surface layer facilitates diffusion of lithium and hence a better rate performance.

Figure 7

Nyquist plots of (a) bare LNMC, (b) 5% LZO-coated sample and (c, d) their corresponding enlarged plots, respectively; (e, f) plots of the real part of impedance [Z] as a function of the inverse square root of the angular frequency [ω–1/2].

In addition, lithium-ion diffusion coefficients for both LNMC and LZO @ LNMC have also been explored from the same set of EIS data using the following equations[50]where R is the gas constant, T is the absolute temperature, A is the area of the electrode, n is the number of electrons per molecule during charge–discharge process, F is the Faraday constant, C is the concentration of Li+ per unit cell and σ is the Warburg factor, which is obtained by linear fitting of the inclined lines at low frequencies, depicted in Figure e,f, and it has a relationship with Z (ω = 2πf), as shown in eq .The calculated diffusion coefficient values are plotted in Figure . Before cycling, the pristine sample shows a diffusion coefficient of 23.77 × 10–16, whereas that for the surface-modified electrode is 22.18 × 10–16. After the first cycle, DLi+ values of both the pristine and surface-coated material are reduced to 12.90 × 10–16 and 15.45 × 10–16, respectively. This decrease in the diffusion coefficient is attributed to the structural rearrangement after charge. In contrast, the decreasing diffusivity coefficient in the surface-coated sample is comparatively low due to suppression of oxygen evolution during the first charge. Upon further cycling, the pristine sample shows that the rate of Li+ diffusion drastically decreases with the increasing cycle number (10.02 × 10–16, 58.83 × 10–17, and 42.97 × 10–17 for the 20th, 40th, and 80th cycle, respectively), whereas in the case of the surface-modified sample it is almost constant (12.95 × 10–16, 12.47 × 10–16, and 13.60 × 10–16 for the 20th, 40th, and 80th cycle, respectively). Incidentally, the decreasing DLi+ in LNMC coincides with the capacity degradation (Figure b) as well as spinel phase transformation (Figure e), which is observed as a function of the cycle number. In contrast, the almost constant value of DLi+ for the surface-modified sample is in accordance with the capacity retention and mitigation of phase transformation.

Figure 8

Calculated Lithium-ion diffusion coefficient (DLi+) for LNMC and LZO @ LNMC at various cycles, from electrochemical impedance spectroscopy.

To demonstrate the structural integrity of the surface-modified cathode, HRTEM investigation of the electrode material has been conducted after the completion of 100 cycles (Figure ). The microscopic results confirm that the LZO coating is neither pulverized nor detached from the surface of the core active material (LNMC). The intactness after the prolonged cycle is attributed to the structural integrity that persists between LZO and LNMC, which is responsible for the improved performance. In summary, the strategy of LZO coating on the surface of LNMC can play a significant positive role and pave the way for commercial realization of Li-rich cathode materials as high-capacity electrodes.

Figure 9

High-resolution transmission electron microscopy images of LZO-coated LNMC electrode after completing 100 charge–discharge cycles.

Conclusions

A facile approach to prepare a uniform coating of a Li-ion conducting LZO surface layer over a high-capacity LNMC cathode is reported in the present study. The growth of LZO with the preferred orientation and alignment on LNMC provides a new strategy to design a robust surface coating, which can possess an ion pathway and thereby enhance the initial Coulombic efficiency and rate performances. This is mainly attributed to (a) the protective role of Li-ion conductive surface barrier between the bulk electrode and electrolyte and (b) improved Li+ migration facilitating lithium diffusion pathways on the surface of the cathode. The intactness of the surface coating significantly suppresses the layered to spinel structural transformation and consequently inhibits the voltage decay. Most importantly, the dissolution of Mn from the bulk cathode into the electrolyte is also drastically mitigated. The surface modification strategy proposed here broadens the scope of further investigation on lithium-rich phases toward commercialization aspects and triggers new insights into the inhabitation of O2 evolution at high voltage and safety concerns of high capacity electrodes.

Experimental Section

The LZO-coated LNMC sample was prepared by a two step procedure as follows. In the first step, nanocrystalline LNMC was synthesized using a solution combustion process, as reported in our previous publication.[34] The surface modification of solution combustion-synthesized LNMC with LZO was carried out as follows. Stoichiometric amounts of LiOH·H2O dissolved in methanol and Zr[O(CH2)3CH3]4 dissolved in carbon tetrachloride were mixed together, in which the nanocrystalline LNMC was added. The solution was kept for constant stirring at room temperature for 6 h. Then, a few drops of deionized water were added into the suspension to undergo hydrolysis reaction. The stirring was maintained for another 6 h, followed by gently heating at 60 °C to evaporate the complete solvent. The powder was then heated at 600 °C for 5 h. Hereafter, we refer to the sample, Li1.2Ni0.13Mn0.54Co0.13O2 as Li-rich NMC (LNMC) and Li2ZrO3-coated Li1.2Ni0.13Mn0.54Co0.13O2 as LZO @ LNMC. The phase characterization of these materials was primarily carried out by X-ray diffraction with a Bruker D8 Advance diffractometer, using a Cu Kα radiation (α1 = 1.54056 Å, α2 = 1.54439 Å). The morphology and size of the powder samples were obtained using field emission scanning electron microscope (FESEM MIRA3 LMU). HRTEM studies were carried out with a Tecnai-G2 F20 microscope.

XPS of these samples were recorded with a Thermo Fisher Scientific Multilab 2000 spectrometer fitted with nonmonochromatic Al Kα radiation (1486.6 eV) as the X-ray source operated at a power of 150 W (12.5 mA, 12 kV). The reported binding energies were corrected with reference to the C1s peak at 284.6 eV. For XPS analysis, powder samples were pelletized to 8 mm diameter, which were mounted on the sample holders and placed in the load-lock chamber in ultrahigh vacuum (UHV) at 8 × 10–8 mbar for 5 h to desorb any volatile species absorbed on the surface. After 5 h, the samples were introduced one by one inside the analyzer chamber, with UHV at 5 × 10–10 mbar. All of the spectra were recorded with pass energy of 40 eV and step increment of 0.05 eV. Infrared spectra were obtained using the FTIR spectrometer, Bruker Optick GmbH, using KBr pressed disk.

Electrochemical experiments were conducted on swagelok-type Li half-cells constructed in an argon-filled glovebox. The positive electrode was formulated with the ball-milled mixture at an 80:20 weight ratio of the active material and Super-P Li Carbon (Timcal Belgium). The lithium metal disc was used as a negative as well as reference electrode. The Whatman glass microfiber filter paper acted as a separator. The 1 M LiPF6 dissolved in a mixture of diethyl carbonate, dimethyl carbonate, and ethylene carbonate (2:1:2 ratio by volume) was used as the electrolyte. In a typical experiment, the cathode loading was 4–6 mg per cell. The galvanostatic cycling experiments were carried out at room temperature at C/5 rate, where C represents the theoretical exchange of 1 mol of lithium per formula unit in 1 h. The cells were cycled between the potential window of 1.75 and 4.8 V using a VMP3Z biologic multichannel potentiostat/galvanostat.