Probing the Complexities of Structural Changes in Layered Oxide

Jan 19, 2018 - Published as part of the Accounts of Chemical Research special issue “Energy Storage: Complexities Among Materials and Interfaces at ...
2 downloads 3 Views 9MB Size
Download PDF
Article Cite This: Acc. Chem. Res. 2018, 51, 290−298

pubs.acs.org/accounts

Probing the Complexities of Structural Changes in Layered Oxide Cathode Materials for Li-Ion Batteries during Fast Charge−Discharge Cycling and Heating Published as part of the Accounts of Chemical Research special issue “Energy Storage: Complexities Among Materials and Interfaces at Multiple Length Scales”. Enyuan Hu,†,§ Xuelong Wang,†,‡,§ Xiqian Yu,*,‡ and Xiao-Qing Yang*,† †

Chemistry Division, Brookhaven National Laboratory, Upton, New York 11973, United States Beijing National Laboratory for Condensed Matter Physics, Institute of Physics, Chinese Academy of Sciences, Beijing 100190, China

‡

CONSPECTUS: The rechargeable lithium-ion battery (LIB) is the most promising energy storage system to power electric vehicles with high energy density and long cycling life. However, in order to meet customers’ demands for fast charging, the power performances of current LIBs need to be improved. From the cathode aspect, layer-structured cathode materials are widely used in today’s market and will continue to play important roles in the near future. The high rate capability of layered cathode materials during charging and discharging is critical to the power performance of the whole cell and the thermal stability is closely related to the safety issues. Therefore, the in-depth understanding of structural changes of layered cathode materials during high rate charging/discharging and the thermal stability during heating are essential in developing new materials and improving current materials. Since structural changes take place from the atomic level to the whole electrode level, combination of characterization techniques covering multilength scales is quite important. In many cases, this means using comprehensive tools involving diffraction, spectroscopy, and imaging to differentiate the surface from the bulk and to obtain structural/chemical information with different levels of spatial resolution. For example, hard X-ray spectroscopy can yield the bulk information and soft X-ray spectroscopy can give the surface information; X-ray based imaging techniques can obtain spatial resolution of tens of nanometers, and electron-based microcopy can go to angstroms. In addition to challenges associated with different spatial resolution, the dynamic nature of structural changes during high rate cycling and heating requires characterization tools to have the capability of collecting high quality data in a time-resolved fashion. Thanks to the advancement in synchrotron based techniques and high-resolution electron microscopy, high temporal and spatial resolutions can now be achieved. In this Account, we focus on the recent works studying kinetic and thermal properties of layer-structured cathode materials, especially the structural changes during high rate cycling and the thermal stability during heating. Advanced characterization techniques relating to the rate capability and thermal stability will be introduced. The different structure evolution behavior of cathode materials cycled at high rate will be compared with that cycled at low rate. Different response of individual transition metals and the inhomogeneity in chemical distribution will be discussed. For the thermal stability, the relationship between structural changes and oxygen release will be emphatically pointed out. In all these studies being reviewed, advanced characterization techniques are critically applied to reveal complexities at multiscale in layer-structured cathode materials. life.1−3 However, for large-scale commercialization of EVs, we still face many great challenges for future LIBs:4,5 in order to have longer driving range, the energy density needs to be significantly improved; the cycling and calendar life need to be extended, and the cost needs to be lowered; in order to meet consumer demands for fast charge, the batteries need to have high rate capability for charging and discharging, especially for

1. INTRODUCTION Global climate change and environment concerns have accelerated worldwide research on renewable energy sources to reduce the use of fossil energy. Rechargeable batteries are the key technologies for such renewable energies. Among the efforts for reducing fossil fuel usage in transportation, the development of various types of electric vehicles (EVs) is a very important part of it and the rechargeable battery is the key technology. The lithium-ion battery (LIB) is the most promising system with high energy density and long cycling © 2018 American Chemical Society

Received: October 10, 2017 Published: January 19, 2018 290

DOI: 10.1021/acs.accounts.7b00506 Acc. Chem. Res. 2018, 51, 290−298

Article

Accounts of Chemical Research

lithium iron phosphate (LFP)18 and lower cost than LiCoO2 materials.19

charging. In addition, safety characteristic is a very important issue, since the size of the battery for EV is several orders of magnitude larger than those for consumer electronics and could cause disastrous consequences if it catches fire. Unfortunately, most of the cathode materials used in LIBs today release highly reactive oxygen at elevated temperatures, which could react with the carbon and electrolyte in the battery cells to cause dangerous thermal runaway.4,6 Therefore, improving the thermal stability of cathode materials against oxygen release is critically important for safety. In-depth understanding of the structural changes and their effects on rate capability during high rate cycling and on thermal stability during heating will provide valuable information for new material design, current material improvement, and preventive procedure development during operations for LIBs. The structural changes during heating and high rate cycling are dynamic, not static, and fast data collection capability for characterization tools is a must have. In addition, since the structural changes take place in a wide range of length scale,7,8 the combination of characterization techniques covering multilength scales from the electrode level,9 to the particle level10 and the grain and grain boundary level,11,12 and all the way to the atomic level is desirable.13 Advances in all kinds of synchrotron based techniques, such as time-resolved X-ray diffraction (TRXRD), time-resolved X-ray absorption (TR-XAS), and transmission X-ray microscopy (TXM), have provided powerful tools with fast data collection capability for doing kinetic studies using both spectroscopy and imaging at various levels of length scales. However, in order to reach atomic level resolution, transmission electron microscopy (TEM) and scanning transmission electron microscopy (STEM) techniques are needed. A schematic illustration is shown in Figure 1,

2. STRUCTURE CHANGES DURING HIGH RATE CHARGE−DISCHARGE CYCLING The fundamental understanding of the structure evolution of electrode materials during the fast charge−discharge process, especially during fast charge is very important for designing battery systems for high power applications, such as in hybrid electric vehicles and pure electric vehicles. However, most of the research results reported in the literature are focused on relatively slow processes near the thermodynamic equilibrium state only, not much on the fast cycling kinetics, mainly due to the lack of fast data collection techniques with high signal-tonoise ratio. Thanks to the high flux X-ray beam of the synchrotron light source, many new time-resolved techniques have been developed for such dynamic studies. Here are some examples. Because of the large penetration depth of X-ray probing beam, X-ray diffraction (XRD) is widely used to obtain reliable average structural information throughout the whole sample. Using time-resolved XRD (TR-XRD) technique, Zhou et al.20 investigated the phase transition behaviors of layered cathode material LiNi1/3Mn1/3Co1/3O2 (NMC)14 in cells cycled at various charging rates, especially at high charging rates. Figure 2 shows the in operando XRD patterns of NMC cathode collected during the first charge process at current rates of 0.1C, 1C, 10C, 30C, and 60C. During these measurements, data were collected while the cell was still running. At low current rate of 0.1C, the phase transition process contains two solid solution reaction regions (hexagonal phase of H1 and H2) at the early and late stages of charging, and one two-phase coexisting region (H1 + H2) in the middle stage of charging. Such phase transition route is different at high current rate as indicated by in operando XRD with each pattern taking about 4 s to be collected. When the C rate is increased to 10C, a new broad peak emerges between H1 and H2 phases and becomes more and more pronounced at 30C and 60C, indicating the formation of intermediate phase. In addition, the peak representing this intermediate phase at 60C charging rate is much broader than that at low rate, indicating the high inhomogeneity of Li content in this intermediate phase. At the end of charge, the intermediate phase is fully converted to the H2 phase even at the very high rate of 60C. To confirm this intermediate phase formation observed by TR-XRD, they used scanning transmission electron microscopy (STEM) to investigate the structural changes of NMC samples after half way charging (55 s) at a high rate of 30C, and the results are shown in Figure 3. STEM has the capability to probe a relevant location of a sample with very high spatial resolution offering structural information at atomic scale. A large number of atomic resolution high-angle annular dark-field (HAADF) and annularbright-field (ABF) STEM images were collected. For STEM, the contrast varies as Z1.7 for the HAADF images, but varies as Z1/3 for the ABF images where “Z” is the atomic number. Since HAADF is more sensitive to the heavy atoms, transition metal (TM) ions are clearly observed from the HAADF image in bright dots in Figure 3a,b,c, while the lighter atoms of Li and O are shown more clearly in the ABF image (Figure 3d,e,f). The uniformly distributed TM layers can be seen from Figure 3a,b,c), indicating that the structure of TM layers is well preserved after halfway charge at high rate. On the other hand, the ABF image taken at the same sample position shows two different types of domains in the green and red dashed squares

Figure 1. Schematic illustration of challenges associated with characterizing cathode materials of Li-ion batteries for vehicle applications, as well as techniques using (synchrotron based) X-ray photon and electron beam.

indicating challenges associated with characterizing layered cathode materials for vehicle application and how synchrotron X-ray based and electron beam based techniques can help to address these challenges. This Account will focus on the review of recent kinetic and thermal studies on layer-structured transition metal oxide cathode materials, especially the LiNixMnyCozO2 (x + y + z = 1, called NMC) materials,14,15 as well as their derivatives, the Lirich NMC materials,16,17 since their applications in LIBs have been expanded rapidly due to their higher energy density than 291

DOI: 10.1021/acs.accounts.7b00506 Acc. Chem. Res. 2018, 51, 290−298

Article

Accounts of Chemical Research

Figure 2. In operando XRD of NMC during the first charge. Contour plot of the (003) diffraction peak of Li1−xNi1/3Co1/3Mn1/3O2 with increasing x between x = 0 and 0.7 during the first charge process at different C rates (0.1C, 1C, 10C, 30C, 60C). Data were collected at X14A at NSLS with a wavelength of 0.7747 Å. Adapted with permission from ref 20. Copyright 2016 John Wiley and Sons.

middle position of two TM columns (actually, it should be in the middle of two oxygen columns, but the TM layers have much stronger contrast). The middle position corresponds to the typical octahedral sites as shown in the pristine sample. It looks like the lithium ions in Figure 3f are located at the pseudotetrahedral sites, as identified by the deviation away from the middle point of two adjacent TM layers. During the highrate charging, the higher overpotential can provide extra driving force for the motion of lithium ions, causing a largely inhomogeneous lithium ion distribution and structural defects within the sample particles. It is assumed that the tetrahedral site occupation might become energetically favorable for Li in some Li-poor regions and the intermediate phases are formed with such occupation. The occurrence of regions with different contents and occupations of Li distributions might be the origin of the intermediate phase observed with XRD (as explained in the previous section). It is likely that intermediate phase formed during high rate charging serves as a buffer to reduce the local stress and strain produced by inhomogeneity during charging. This investigation of intermediate phase formation during high rate charging for NMC samples was inspired by the previous reported studies in the literature. Similar intermediate phase formation during high rate charging was reported in olivine structured LiFePO4 (LFP) systems.21−23 For example, Liu et al.22 reported a metastable solid solution phase rather than a commonly two-phase separation during high-rate cycling by HR-XRD. Orikasa et al.23 observed a metastable crystal phase of LixFePO4 during high current charging/discharging, which contributes to the high-rate performance of LFP system. Furthermore, Gu et al.24 observed a new intermediate “Listaging” phase of Li0.5FePO4 using STEM technique with atomic spatial resolution. It is worth mentioning that while in operando XRD and ex situ STEM can sometimes provide complementary information about the same intermediate phase, as in layered NMC, it may not be always true in other situations. For example, for LiFePO4, the intermediate phase can only last for around 30 min, which makes it difficult to study using ex situ characterization afterward. Therefore, it is necessary to first identify the period in which the intermediate phase can be preserved (if possible, by in situ XRD) before carrying out ex situ measurements.

Figure 3. Structure of half charged NMC at 30C rate. (a) Typical HAADF image taken along the [110] zone axis of the NMC electrode after 55 s charging at the current rate of 30C. (b, c) The zoom-in image of the areas marked with orange and pink squares, respectively. (d) Corresponding ABF images of NMC electrode after 55 s charging. (e, f) The zoom-in image of the areas marked with orange and pink squares, respectively. The blue and green dots indicate the TM ions and Li ions, respectively. Adapted with permission from ref 20. Copyright 2016 John Wiley and Sons.

in Figure 3d and in more details in Figure 3e,f. The location of the Li columns in the green dashed square region in Figure 3e is different from that in the red dashed square region in Figure 3f. In Figure 3f, the Li columns are no longer located in the 292

DOI: 10.1021/acs.accounts.7b00506 Acc. Chem. Res. 2018, 51, 290−298

Article

Accounts of Chemical Research

(Ni−O) experiences dramatic changes in both position and intensity within the first 100 s, indicating that most of the charge compensation occurred at the Ni sites. The peak intensities decreased first (from 0 s to ∼60 s) due to the Jahn− Teller distortion caused by the oxidation of Ni2+ to Ni3+, then turned around to increase with further oxidation of Ni3+ to Ni4+. The EXAFS features remain unchanged after 160 s, indicating that the oxidation of Ni2+ to Ni4+ was almost completed within the first 3 min. Compared to Ni, the evolution of the local structure around Co and Mn sites caused by Li+ extraction stretched over a longer period of time as displayed in Figure 4: the first coordination shell peak (Co−O and Mn−O) intensities show a continuous decrease. Specifically, Co−O peak experiences most of the intensity change within 200 s. After that, no significant Co−O peak intensity changes can be observed. The evolution of the Mn−O peak intensity generally follows a similar trend to that of the Co−O peak, but the changes are continuous over the whole observation time scale (900 s), indicating much slower delithiation kinetics around Mn sites. Recently, a new synchrotron based technique, transmission X-ray microscopy (TXM), has been developed and applied to battery material studies, especially for in situ studies. Compared with XAS, which provides average information over a relatively larger volume of the sample (illuminated by the X-ray beam), TXM can focus on a nanometer area with higher spatial resolution. Compared with TEM, which has atomic resolution, TXM is able to do in situ experiments with spatial resolution between XAS and TEM and provide a powerful tool to do in situ studies at the particle level. In addition, TXM is able to do chemical and oxidation state distribution mapping over an area of interest, giving us opportunity to visualize the morphological, chemical, and oxidation state changes of individual particles. Using TXM, Xu et al.28 carried out an in situ investigation on a single LiCoO2 particle in a cell during cycling at different rates. In these experiments, the cell was stopped after cycling to a certain state to collect the images. As shown in Figure 5, the cell was sequentially charged and discharged at different rates (0.2C, 1C, and 10C) between 3 and 4.6 V. The chemical mapping based on the oxidation state of Co for a selected

While XRD provides information on structural change during electrochemical cycling, X-ray spectroscopy technique, especially X-ray absorption near edge spectroscopy (XANES) can identify the contribution of individual transition metals in an elementally selective way and therefore has been widely used to study battery materials, which are typically multitransition metal systems. Despite such wide application, a limited number of studies on the kinetic characteristics of these transition metal elements have been reported in the literature. It is very important to understand the unique contribution of each transition metal to the material’s kinetic properties for designing cathode materials with better rate capability. In operando time-resolved X-ray absorption spectroscopy (TRXAS) has the capability to monitor structural changes around certain transition metal ions in the electrode material during high rate charge−discharge cycling. Yu et al.25 reported their studies on the contribution from different transition metals to the bulk kinetic properties of Li1.2Ni0.15Co0.1Mn0.55O2. This material belongs to the family of Li-rich Mn-rich layerstructured materials.16 Many researchers believe that this material is an intergrowth composite of LiMO2 (M represents transition metal ions) and Li2MnO3 components.17,26,27 In order to start the redox reactions of all three transition metals simultaneously during charge, a 5 V constant-voltage charging condition, rather than the conventional constant current charging was applied during the collection of TR-XAS data. This 5 V charging voltage is higher than all of the Ni2+/Ni3+, Ni3+/Ni4+, and Co3+/Co4+ redox potentials and the activation voltage of Li2MnO3. The data collection continued for 900 s while the 5 V voltage was continuously applied. The data collection time for each spectrum was optimized to be 15 s to balance the signal-to-noise ratio and the number of spectra collected. The extended X-ray absorption fine structure (EXAFS) part of collected spectra was then Fourier transformed to yield data in R space. The results are shown in a 2D view in Figure 4.

Figure 4. Five volt constant voltage charging was applied on the Li1.2Ni0.15Co0.1Mn0.55O2 electrode. Ni, Co, and Mn reacted simultaneously, which was recorded using a time-resolved XAS technique. Projection view of the corresponding Ni−O, Co−O, and Mn−O peak magnitudes of the Fourier transformed K-edge spectra as functions of charging time. Adapted with permission from ref 25. Copyright 2014 John Wiley and Sons.

Figure 5. In situ monitoring of the chemical inhomogeneity in a single particle of LiCoO2 up to 20 cycles. Panel a is the chemical map of the particle at its pristine state. Panel b is at the charged state (at 4.6 V). Panels c−e are the chemical maps at discharged state (at 3 V) after the particle went through cycles of different rates at 1C, 10C, and 0.2C, respectively. Panel f is the map at the discharged state after 20 cycles at 0.2C. All chemical maps are color coded to the corresponding pie charts. The red area represents the domains at charged state, and the green area represents the domains at discharged state. Reproduced with permission from ref 28. Copyright 2017 American Chemical Society.

The peaks on the Fourier transformed EXAFS plot correspond to characteristic bonds within the nearest shells. The peak position in the plot is usually 0.3−0.4 Å shorter than the actual bond distance because of phase shift. In the case of layered material, the first peak corresponds to transition metal− oxygen bond, which lies within the first shell around transition metal. Figure 4 shows that the first coordination peak of Ni 293

DOI: 10.1021/acs.accounts.7b00506 Acc. Chem. Res. 2018, 51, 290−298

Article

Accounts of Chemical Research

material, which is usually accompanied by oxygen release. The newly released oxygen can react with the carbon and binder in the cathode, and even with the flammable organic solvent in the electrolyte, leading to the thermal runaway.31 Therefore, it is very important to study the correlation between structural changes and oxygen release of cathode materials during heating by recording the information about both of them simultaneously. Nam et al. reported their results of combined mass spectroscopy (MS) and X-ray diffraction (XRD) during heating using a specially designed heating cell at synchrotron facilities.32 The strong X-ray source at synchrotron beamlines and the fast 2D area detector provided high quality time-resolved XRD data, and the MS provided gas evolution correlating to the structural changes during heating. Taking advantage of this technique, studies were carried out on Li0.33Ni0.8Co0.15Al0.05O2 (NCA). It was reported that during heating, two distinct oxygen-release peaks with their centers around 230 and 300 °C appeared in the MS results (Figure 7). These peaks are clearly

particle at the end of each discharge process and the end of 1C charge process is plotted in Figure 5. Using the XANES spectra of LiCoO2 powder at fully charged and discharged state as references, they rated the state of charge (SOC) of each spot in the particle. It can be seen from these mappings, the distribution of SOC in this particle is inhomogeneous, especially when high current rate was used. In an ideal equilibrium situation, the LiCoO2 particle is supposed to return to the fully discharged state (marked by green) when cell is discharged to 3 V. However, there were still some parts of the particle staying at charged state (marked by red) for all cycling rates. The portion of particle recovered back to discharged state gets smaller when the rate gets higher. The recovery percentage of the particle bounced back to a higher level in following low rate cycling. The inhomogeneity at particle level is induced by high rate cycling. It is worth noting that the electrochemical behavior of this particular particle may not be able to represent the whole cell, especially in capacity retention. However, the inhomogeneity at particle level is representative. Furthermore, the 3D chemical map of a LiCoO2 particle recovered from cycling (Figure 6) shows a crack in particle associated with the

Figure 6. Ex situ 3D chemical map of a selected particle recovered from a cell that was cycled at 0.2C and disassembled at 3 V. Panels a, b and c are the same particle in different viewing angles indicated by the arrows on the bottom. The chemical components are color coded in the 3D rendering with the discharged stat shown in green and the charged state shown in red (the portion of the particle that failed to return to the discharged state at 3 V). Reproduced with permission from ref 28. Copyright 2017 American Chemical Society.

Figure 7. (a) Time-resolved (TR) XRD patterns and (b) results from simultaneously measured mass spectroscopy (MS) that trace oxygen gas released from the overcharged Li0.33Ni0.8Co0.15Al0.05O2 during heating up to 500 °C. The cathode sample in a glass tube was heated from 25 to 500 °C in 4 h during the TR-XRD and MS measurements (heating rate 2.0 °C min−1). The left panel shows the ideal crystalstructure models for rhombohedral, spinel, and rock-salt structures. Adapted with permission from ref 32. Copyright 2013 John Wiley and Sons.

inactive domain. The appearance of the crack suggests a close relationship between morphological defects and chemical inhomogeneity. Since the large chemical inhomogeneity is induced by high rate cycling as shown in Figure 6, morphological degradation of the electrode particle could be a consequence of high rate charge−discharge cycling. Similar chemical inhomogeneity and cracks in LiCoO2 particles during high rate cycling was also observed by other researchers using different techniques.29 A spatial correlation between Li content distribution and fracture direction was found. Furthermore, atomic scale observation of structural evolution in LiCoO2 particle during ultrafast charging was also reported by Gong et al.30 using TEM.

correlated with the two structural phase-transitions shown in the time-resolved XRD (TR-XRD) patterns in the left panel of Figure 7. The beginning of first noticeable oxygen release observed at about 180 °C is close to the starting temperature of the structural transition from the initial layered structure to the spinel structure, as shown in the TR-XRD results. After the intensity of the first peak reached the maximum at about 230 °C and started decreasing, a second oxygen release peak arose around 250 °C and continued to 500 °C, with its center at about 300 °C. This second peak relates to the phase transition from the disordered spinel structure to the NiO-like rock-salt structure between 250 and 500 °C as shown in the TR-XRD. More O2 gas was released during the second phase transition than during the first one. Direct correlation between structural changes and oxygen release during the thermal decomposition of LixNi0.8Co0.15Al0.05O2 cathode materials overcharged to x = 0.5, 0.33, and 0.1 levels was reported by Bak et al.33 This study

3. CORRELATIONS BETWEEN STRUCTURAL CHANGES AND OXYGEN RELEASE DURING HEATING As mentioned before, studying the structural changes of layered oxide cathode materials during heating is very important in providing guidance for the improvement of thermal stability, which is critically related to the safety characteristics of these cathode materials. Cathode heating, caused by electric short circuit, mechanical impact, or other accidental events could trigger dangerous thermal runaway. During such processes, the fast temperature increase induces structural changes of cathode 294

DOI: 10.1021/acs.accounts.7b00506 Acc. Chem. Res. 2018, 51, 290−298

Article

Accounts of Chemical Research

stability. For the unheated overcharged NCA, TEM imaging (shown in Figure 9) revealed three structures, the layered

showed that the state of charge affects both the structural changes that occur in these materials and the evolution of O2 and CO2 gases during thermal decomposition. The evolution of both O2 and CO2 gases are well-correlated with the phase transitions that occur during thermal decomposition. The higher the level of overcharging (i.e., overdelithiation) leads to the poorer the thermal stability and the larger amounts of oxygen release at lower temperatures. In particular, highly overcharged cathode samples (i.e., Li0.1Ni0.8Co0.15Al0.05O2) show a severe oxygen release at a temperature as low as 175 °C, accompanied by a sudden structural change from the layered to the disordered spinel phase. Bak et al. also reported their systematic studies on the correlations between the structural changes and oxygen release for LiNixMnyCozO2 (NMC, x + y + z = 1) materials with different Co/Ni ratios (NMC433, NMC532, NMC622, and NMC811).34 As shown in Figure 8, the effects of the Co/Ni

Figure 9. HRTEM images viewed along (a) [1̅2̅1]R and (b) [001]R directions for an overcharged Li0.15Ni0.8Co0.15Al0.05O2 particle. The insets in panel a show the magnified images from the small rectangle areas, indicating the rock-salt (RS), spinel (S), and rhombohedral (R) structures. In panel b, although both the spinel and rock-salt structures have hexagonal pattern, the periodicity of the former is twice as that of the latter. Reproduced with permission from ref 32. Copyright 2013 John Wiley and Sons.

(located in the core, far from the surface), the spinel (the shell near the surface), and the rock-salt (at the surface layer or edge of the particle). Figure 10 shows the in situ TEM results of overcharged NCA, indicating that upon heating, spinel phase (shell) propagates toward the core of the particle and the rocksalt phase (surface) grows toward the spinel phase region (shell). For NMC333, the overcharged particles also have layered phase in the core, the spinel phase in the shell. However, the great difference is, rather than the small amount

Figure 8. Mass spectroscopy profiles for the oxygen (O2, m/z = 32) collected simultaneously during measurement of TR-XRD and the corresponding temperature region of the phase transitions for NMC samples (lower panel). Reproduced with permission from ref 34. Copyright 2014 American Chemical Society.

ratio on the thermal structural stability can be clearly seen. For NMC422 and NMC532 samples with higher Co/Ni ratios, the phase transition to both LiMn2O4 type and Co3O4 type spinels occurred at higher temperatures and the Co3O4 type spinel remained as the dominating phase all the way up to 600 °C. In contrast, for the NMC622 and NMC811 samples with lower Co/Ni ratios, the phase transitions to both LiMn2O4 type and Co3O4 type spinels started and finished at much lower temperatures, and the phase transition to rock-salt phase was completed at 550 and 365 °C, respectively. The structural changes and oxygen release differences between NCA and NMC agreed very well with an earlier study about the phase distribution of them at the particle level using in situ transmission electron microscopy (TEM) during heating. Wu et al. used high resolution TEM (HRTEM) to study highly delithiated NCA and NMC333 and found that their particles both feature the core−shell−surface structure after overcharging (over delithiation).35 However, the detailed phase distributions differ significantly between NCA and NMC, which has direct consequence on their respective thermal

Figure 10. HRTEM images taken from an overcharged Li0.15Ni0.8Co0.15Al0.05O2 particle (a) before heating, and after heating at (b)100 °C, (c) 200 °C, and (d) 300 °C. The insets are the corresponding SAEDPs taken from a large area (approximately 240 nm in diameter) of the same particle. RS, rock-salt; S, spinel; R, rhombohedral. Reproduced with permission from ref 32. Copyright 2013 John Wiley and Sons. 295

DOI: 10.1021/acs.accounts.7b00506 Acc. Chem. Res. 2018, 51, 290−298

Article

Accounts of Chemical Research

kinetically sluggish. This not only pushes the phase transition to occur at a higher temperature but also kinetically stabilizes the LiMn2O4 type phase over a large temperature range. The role of Co and Mn are further confirmed by thermal stability study of a series of NMC materials with varying element ratios. It shows that the thermal stability of NMC materials improves as the concentration of Co and Mn increase. The benefit of having Co and Mn (especially Co) in the material also extends to higher temperature when the second step reaction is initiated. At high temperature, the cations may become mobile enough that kinetics is no longer an issue. The reaction route is mainly determined by thermodynamics or by the phase diagram. For the Li−Ni−O2 phase diagram, NiO rock salt is a very stable phase, and for the Li−Co−O2 phase diagram, Co3O4 spinel is the stable one.40 Consequently, nickel-rich materials tend to form rock-salt phase at high temperature and cobalt-rich materials tend to form Co3O4 type spinel. Because Co3O4 formation leads to less oxygen release than rock-salt formation, adding Co to layered material helps to improve the thermal stability.

of rock-salt phase at the surface in NCA, small amount of O1 phase (CdI2 type structure featuring hexagonal close-packing of oxygen, rather than cubic close-packing of oxygen as in layered, spinel, and rock salt) was observed on the surface. Upon heating, the surface O1 phase transforms into a Co3O4 type spinel and the shell spinel phase propagates toward the core. This O1 to Co3O4 type spinel phase transition pushed the spinel to rock salt transition to a very high temperature, contributing to the good thermal stability of NMC333. It is interesting to note that O1 phase is usually formed when layered material is highly delithiated (e.g., with 0.9Li extracted).36,37 The fact that layered phase being the core, spinel phase being the shell and O1 phase being the surface suggests the considerable structural inhomogeneity within the particle. As a result, such inhomogeneity induced during the electrochemical process can greatly influence the thermal stability of the material. Aside from reasons mentioned above, different kinetic and thermodynamic properties of transition metal elements also contribute to the difference on thermal stability between NMC333 and NCA. For the first step of phase transition (from layered to LiMn2O4-type spinel), theoretical calculation shows that while the layered to spinel phase transition is always thermodynamically driven (for delithiated cathode), at what specific temperature such transition takes place is really a kinetic issue.38 This is because it requires the migration of transition metal cations from the original octahedral sites in the transition metal layer to the octahedral sites in the adjacent lithium layer. It turned out such migration is in fact via the intermediate tetrahedral site as this path is the most energetically favorable one. Based on this consideration, octahedral site stabilization energy (OSSE), which calculates the energy difference between octahedral and tetrahedral occupation is a good indicator of the mobility of transition metal cations.39 Figure 11 shows detailed OSSE of various

4. CONCLUDING REMARKS The high rate charge/discharge capability and safety characteristics are two of the main challenges to be addressed for the large-scale LIB application for electric vehicles. In this Account, we have reviewed major characterization techniques in multilength scales for monitoring the structural evolution and kinetic characteristics of various transition metal elements for layer-structured transition metal oxides such as LiNixMnyCozO2 (x + y + z = 1, called NMC), Li-rich materials, and LiCoO2 during the fast charge/discharge processes. Formation of the intermediate phase during high rate cycling was detected by combination of TR-XRD and STEM in LiNi1/3Mn1/3Co1/3O2, which can serve as a buffer to reduce the local stress and strain produced by inhomogeneity during charging. In Li-rich material, much slower kinetics of Mn than Co and Ni during quick charge/discharge process was monitored by TR-XAS and TEM. In addition, the chemical inhomogeneity and cracks on the particle level of LiCoO2 were observed by in situ TXM. We have also summarized the advanced characterization methods for studying the correlation between structural changes and oxygen release of layered cathode materials during heating. The crucial insights obtained from TR-XRD and in situ TEM show that the thermal stability of the cathode materials depends on the transition metal chemistry and phase inhomogeneity at the particle level. The advanced characterization techniques covering multilength scales play important roles in guiding the optimization of current layered electrode materials, as well as in designing novel cathode materials with high power capability and good thermal stability.

Figure 11. Tendency of migration (based on octahedral site stabilization energy) as a function of the electronic structure of 3d transition metal cations. Exchange energy is not considered in calculating the OSSE. Reproduced with permission from ref 39. Copyright 2016 John Wiley and Sons.

■

cations. It indicates that OSSE is highly dependent on specific electronic structure, which is determined by the oxidization state and spin state. From this aspect, the better thermal stability of NMC in comparison to NCA can be attributed to the different transition metal composition. For delithiated NMC333, there are large amounts of Co3+ (low spin), Co4+ (low spin), and Mn4+, all of which are highly stable in octahedral sites (indicated by the fairly negative OSSE). Therefore, the energy barrier of migrating to the intermediate tetrahedral site is high and the layered to spinel transition is

AUTHOR INFORMATION

Corresponding Authors

*X.Y. E-mail: [email protected]. *X.-Q.Y. E-mail: [email protected]. ORCID

Xiqian Yu: 0000-0001-8513-518X Xiao-Qing Yang: 0000-0002-3625-3478 Author Contributions §

E. Hu and X. Wang contributed equally.

296

DOI: 10.1021/acs.accounts.7b00506 Acc. Chem. Res. 2018, 51, 290−298

Article

Accounts of Chemical Research Notes

Microstructural Defects on Voltage Fade of Li- and Mn-Rich Cathodes. Nano Lett. 2016, 16, 5999−6007. (12) Yan, P.; Zheng, J.; Gu, M.; Xiao, J.; Zhang, J.-G.; Wang, C.-M. Intragranular cracking as a critical barrier for high-voltage usage of layer-structured cathode for lithium-ion batteries. Nat. Commun. 2017, 8, 14101. (13) Xu, B.; Fell, C. R.; Chi, M.; Meng, Y. S. Identifying surface structural changes in layered Li-excess nickel manganese oxides in high voltage lithium ion batteries: A joint experimental and theoretical study. Energy Environ. Sci. 2011, 4, 2223−2233. (14) Yabuuchi, N.; Ohzuku, T. Novel lithium insertion material of LiCo1/3Ni1/3Mn1/3O2 for advanced lithium-ion batteries. J. Power Sources 2003, 119, 171−174. (15) Delmas, C.; Menetrier, M.; Croguennec, L.; Saadoune, I.; Rougier, A.; Pouillerie, C.; Prado, G.; Grüne, M.; Fournes, L. An overview of the Li(Ni, M)O2 systems: syntheses, structures and properties. Electrochim. Acta 1999, 45, 243−253. (16) Lu, Z.; MacNeil, D.; Dahn, J. Layered cathode materials Li[NixLi(1/3−2x/3)Mn(2/3−x/3)]O2 for lithium-ion batteries. Electrochem. Solid-State Lett. 2001, 4, A191−A194. (17) Thackeray, M. M.; Johnson, C. S.; Vaughey, J. T.; Li, N.; Hackney, S. A. Advances in manganese-oxide ’composite’ electrodes for lithium-ion batteries. J. Mater. Chem. 2005, 15, 2257−2267. (18) Padhi, A. K.; Nanjundaswamy, K.; Goodenough, J. B. Phosphoolivines as positive-electrode materials for rechargeable lithium batteries. J. Electrochem. Soc. 1997, 144, 1188−1194. (19) Mizushima, K.; Jones, P.; Wiseman, P.; Goodenough, J. B. LixCoO2 (0