Bohang Song‡
a,
Cuifeng Zhoub,
Yu Chenc,
Zongwen Liub,
Man On Laia,
Junmin Xuec and
Li Lu*a
aMaterials Science Laboratory, Department of Mechanical Engineering, National University of Singapore, Singapore 117576. E-mail: luli@nus.edu.sg
bSchool of Chemical and Biomolecular Engineering, The University of Sydney, Sydney, Australia
cDepartment of Materials Science and Engineering, National University of Singapore, Singapore 117576
First published on 1st September 2014
Li-rich Li(Li0.2Mn0.54Ni0.13Co0.13)O2 cathode coated with carbon layer has been prepared by a hydrothermal approach followed by a post annealing process. The cathode after surface modification exhibits both enhanced cyclability and improved rate capability compared with the pristine one without coating. The carbon coating process causes a phase transformation from Li2MnO3-like domain to cubic-spinel domain with Fdm symmetry at surface regions of particles. As a consequence, the valence state of Mn on the surface accordingly varies. Such transformed surface spinels, as well as the wrapped carbon layers as a result of this coating strategy are believed to be responsible for the enhanced electrochemical performance.
Although the high capacity is very attractive, this family of cathode materials still suffers from several drawbacks: poor cyclability and rate capability and a reducing output voltage during cycling. These drawbacks are believed to be partially associated with the sensitivity of particle surfaces according to recent discoveries on these materials. Gu et al.10 found that in the Li(Li0.2Ni0.2Mn0.6)O2 system, Ni prefers to segregate at the surface facets of synthesized particles, and the particle facets tend to terminate at transition metal (TM) layers, while the Li fast diffusion channels in these layers are almost perpendicular to the channels inside. In fact, both of the phenomena nearby surface regions could induce a high diffusion barrier against Li transportation, leading to poor rate capability. Concerning cyclability, the dissolution of TM ions in high potential ranges from layered structure is always recognized as a main reason for the capacity fade, as suggested by Amine11 and Song.12 Another surface initiated effect related to the capacity fade is a phase transformation caused by the migration of TM ions, which is driven by thermodynamic instability of Li and O vacancies.5,8,13 Because of the above reasons, such particle surfaces easily experience internal stress leading to local non-crystallization and further micro-cracks on surfaces during cycling.14
To address the issues associated to the sensitivity of the particles' surfaces, various methodologies have been used for surface modifications through various coatings of Li–Ni–PO4,15 AlPO4,16 Al17 and RuO218 as the conductive surface layers to enhance the rate capability. On the other hand, Al2O3,19 ZnO coating layers20 and surface nitridation21 were also applied to improve the cycle performance. Besides these traditional protection or conductivity effects that resulted from coating layers, AlF322 and MnO223 were also found to contribute to surface spinel formation after corresponding surface modifications. Such surface spinels initiating from original Li2MnO3-like domains were reasonably believed to be able to improve both cyclability and rate performance. As a matter of fact, similar improvements because of surface spinels were also observed in (NH4)2SO4,24 graphene25 and carbon black26 treated Li-rich layered systems. Nevertheless, due to the reduction concerns of carbon coating, only a few works are reported on the effects of carbon coating on such Li-rich layered systems,27 although carbon coating is always regarded as the most effective method for the enhancement in olivine systems. Therefore, in this work we intend to develop a facile way to achieve carbon coating layers on Li(Li0.2Mn0.54Ni0.13Co0.13)O2 particles to improve electrochemical performance without largely changing the chemical composition of the layered compounds.
![]() | ||
Fig. 1 Powder XRD patterns of the pristine carbon coated (C-3, C-15) and post annealed (C-3-H, C-15-H) LLNCM samples. |
Fig. 2 compares the morphologies of the pristine carbon coated (C-3, C-15) and post annealed (C-3-H, C-15-H) particle. The distribution of particle size of about 100 to 200 nm with facets could be observed in the pristine sample, while the particles tend to aggregate and grow in size without a clear facet after coating process, as seen in both C-3 and C-15 samples. The increase in particle size is caused by a hydrothermal reaction and might also be associated with carbon coatings which interconnect small crystals. A thin carbon layer about 3 nm was evenly coated on the particles from the TEM observation (Fig. 3), whereas parts of carbon were accumulated to form spheres. Elemental mappings of the C-3 particle reveal uniform distribution of C, Mn, Ni and Co (Fig. 4) after carbon coating. Similar but not equal, with the increase in duration time of the reaction from 3 to 15 h, more carbon layers in the C-15 sample tend to grow into carbon spheres, (Fig. 2) which was confirmed in the bright field TEM image (Fig. 3). The more transparent spheres around a LLNCM particle represent the carbon spheres. According to TGA results (Fig. S1†), the carbon contents in C-3 and C-15 are approximately 6.3 wt% and 6.6 wt%, respectively. A sudden loss in weight from 150 to 300 °C was caused by loss of functional groups on these synthesized carbons. After a post annealing process, the particle morphologies of C-3-H and C-15-H samples barely change, but the average particle sizes are apparently reduced compared to those before annealing, as observed in Fig. 2. Moreover, the corresponding TEM images of both C-3-H and C-15-H samples (Fig. S2†) exhibit non-uniform transparency in the form of nano-domains compared to dissimilar characters in transparency of the pristine sample (Fig. S2†). It may indicate a local disorder or rearrangement of the original layered structure as a result of surface treatment, which is highly related to local spinel-like transformation.
![]() | ||
Fig. 2 SEM images of the pristine, carbon coated (C-3, C-15) and post annealed (C-3-H, C-15-H) LLNCM particles. Arrows inside indicate the carbon spheres in the C-15 sample. |
HRTEM characterization of the C-15-H sample is shown in Fig. 5. Fig. 5(a) reveals the general appearance of the particles with electron diffraction pattern (EDP) and diffraction index in Fig. 5(b) and (c). Two sets of diffraction can be indexed, layered triclinic structure and a spinel cubic structure as confirmed. Fig. 5(d) shows HRTEM on the surface of a particle. Further FFT images for sub areas as shown in Fig. 5(e) and (f) reveal the spatial relationship between these two phases. Green dash lines in Fig. 5(d) highlighting the layered domains tend to be enclosed by spinel domains, which are terminated on the surface of the particle. The zone axis with respect to two phases are [1−11]layered and [110]spinel.
The transformation on the surface regions is believed to be induced during the hydrothermal processing where the presence of carbon coating increases the acidity of the hydrothermal solution resulting in ions exchanged between H+ and Li+ at high pressure. The final annealing may assist the phase transformation and aggregation of carbon. On the other hand, the further reduced carbon layers as confirmed by both TGA results (Fig. S1†) and the TEM image (Fig. S3†) could be partially sustained on particle surfaces after heat treatment. Based on the above observations and the synthesis procedure, the mechanisms of carbon coating and transformation on the particles' surfaces in C-3 (C-15) and C-3-H (C-15-H) are schematically illustrated in Fig. 6.
XPS was applied to investigate the solid-state chemistry of particle surfaces of all the pristine and modified C-3, C-3-H samples. Spectra of C 1s, O 1s, Li 1s, Mn 2p, Ni, 2p and Co 2p orbital are shown in Fig. 7. Curve fittings were performed based on the binding energy of C 1s (C–C sp2 bond) at 284.5 eV. As shown in C 1s spectra, several peaks corresponding to 284.5, 285.5 and 287.6 eV could be recognized as C–C, C–OH and CO, respectively.31,32 The higher intensities of both C–OH and C
O peaks in the C-3 sample indicate the presence of functional oxygen groups on the surface carbon layers rather than highly-carbonized layers after the coating process. However, these functional groups partially decompose after the post annealing process judged from the drastic decrease in the intensities of the C-3-H sample. The similar trend is also observed in the O 1s spectra. A larger amount of functional oxygen groups on the particle surface for the C-3 sample compared to the amount of the pristine sample was able to function as the protection layer on particles to mitigate the corrosion effects of the electrolyte, and thus enhances the cyclability. Moreover, these functional oxygen groups could also decrease the surface electrical conductivity of particles. Consequently, these characteristics of the functional groups significantly affect the electrochemical performance and will be shown later. On the other hand, the surface modification has a fewer effects on the variations of the valence states of both Ni and Co, according to the comparable peaks fitted in Ni 2p and Co 2p spectra. Ni2+ (854.6 eV) and Co3+ (780.1 eV) dominate in all the samples with trace amounts of Ni3+ (855.9 eV) and Co4+ (781.1 eV) at the synthesized state. It is interesting to note that the valence state of Mn substantially varies as a result of carbon coating and post annealing, as seen from both the Mn 2p and Mn 3p (close to Li 1s) spectra. Mn4+ with respect to 642.9 eV peak is the majority constituent in the pristine sample, while the majority constituent switches to Mn3+ in both the C-3 and C-3-H samples. The decrease in average valence state of Mn may imply formation of spinel, or at least a change in electronic configuration related to the Mn at surface regions. As also suggested in HRTEM results in Fig. 5, it is likely to deduce that the transformed spinel is the LiMn2O4-like cubic spinel, despite that the accurate chemical composition in the local structure is hardly detected. However, the statistical feature of the chemical compositions on surface regions of the particle could be confirmed by XPS.
As shown in Table 1, two important observations are summarized: (1) compositional ratio of Li relative to a sum of TM ions reduces from 1.04 of the pristine sample to 0.21 of C-3 and to 0.73 of C-3-H and (2) the relative contents of Ni, Co and Mn also varies. It is important to note that although the above data indicate the spinel formation in the C-3 sample, a subsequent heat treatment is still necessary to complete such transformations according to distinct electrochemical behaviors in the following section.
Li | Transition metals | O | ||||
---|---|---|---|---|---|---|
Mn | Ni | Co | Total | |||
Pristine | 1.02 | 0.54 | 0.22 | 0.22 | 0.98 | 2.8 |
C-3 | 0.35 | 0.91 | 0.45 | 0.30 | 1.65 | 5.9 |
C-3-H | 0.84 | 0.84 | 0.18 | 0.13 | 1.15 | 2.6 |
Fig. 8 compares the first cycle charge/discharge curves at a current density of 50 mA g−1 and the respective dQ/dV plots of all the samples. The charge/discharge capacities, as well as the coulombic efficiencies are tabulated in Table 2. The pristine LLNCM delivers a charge capacity of 300 mA h g−1 and discharge capacity of 253 mA h g−1 with 84% efficiency, while these values drastically reduce for C-3 and C-15 samples. It is noted that the coated layers of carbon or spheres are of poor electronic conductivity because of their poor carbonization feature, as supported by the presence of functional oxygen groups in XPS C 1s and O 1s results. However, the charge capacities decrease, whereas both discharge capacities and coulombic efficiencies increase for C-3-H and C-15-H samples, i.e. 263 mA h g−1 with 96.7% and 264 mA h g−1 with 96.4%, respectively. These improvements in discharge capacity and initial coulombic efficiency after post annealing could be ascribed to two reasons: the formation of spinel by consuming the amount of the Li2MnO3-like component, which is confirmed by the reduced 4.5 V charge plateau, but elongated the 2.7 V discharge plateau and the better carbonized layers or spheres as a result of post annealing as also supported by XPS results in Fig. 7. Although the pristine LLNCM exhibits a typical deintercalation/intercalation profile of Li rich layered structures,2 the typical electrochemical profile apparently changed after modification (samples C-3, C-15, C-3-H and C-15-H). The dQ/dV plot of the pristine sample first shows an oxidation of Ni2+ and Co3+ to higher states related to the 4.1 V peak (O1), followed by a redox-like peak at about 4.6 V, which is commonly recognized because of the oxygen release. Both of the two peaks clearly shift to higher potentials after coating (C-3, C-15) because of the polarization effects from coating layers, but they shift back to the comparable positions of the pristine sample after the further heating process (C-3-H, C-15-H). In a typical discharge process of the pristine sample, three reduction peaks noted as R1, R2 and R3 were observed. The origin of R1 is still unclear at the present,33 while R2 and R3 are believed to be associated with Ni4+/Ni2+ in addition to Co4+/Co3+ redox couples in a layered structure and Mn4+/Mn3+ redox in a layered structure, respectively. The only peaks appearing in C-3-H and C-15-H samples are R4 and R5 and are attributed to the Mn4+/Mn3+ redox in a layered-spinel-mixed complex structure and in a surface-formed spinel structure.34 It is worth pointing out that (1) R3 contributions in C-3 and C-15 samples significantly reduce compared to the pristine sample, suggesting a less activation of the Li2MnO3-like component, which is responsible for reduced discharge capacities, and (2) the existence of only R4 and R5 peaks confirms phase transformation after the heating process, which not only took place at surface regions, but also possibly initiated in bulk regions because of the significant contributions to the capacity. Although XRD and XPS results imply the possible spinel formation in the coated samples C-3 and C-15, the disappearance of R4 and R5 peaks still indicates that the leaching of Li incorporated with the rearrangement of electronic configuration related to Mn occurred, but the local structure was sustained in the layered form before the subsequent heating process.
![]() | ||
Fig. 8 First charge/discharge curves with corresponding dQ/dV plots of the pristine, C-3, C-15, C-3-H and C-15-H cathodes tested at a current density of 50 mA g−1, 2.0–4.8 V at room temperature. |
Charge capacity (mA h g−1) | Discharge capacity (mA h g−1) | First coulombic efficiency (%) | |
---|---|---|---|
Pristine | 300 | 253 | 84.3 |
C-3 | 188 | 116 | 61.7 |
C-15 | 199 | 142 | 71.4 |
C-3-H | 272 | 263 | 96.7 |
C-15-H | 273 | 264 | 96.4 |
Fig. 9 shows the cycle performance and rate capability of all the samples. At 0.2 C (Fig. 9(a)), the pristine sample is able to provide 112 mA h g−1 discharge capacity after 100 cycles with 44% capacity retention, whereas a continuous increase in the capacities for both C-3 and C-15 samples were observed in the first 8 charge/discharge cycles, and 92% for C-3 and 91% for C-15 can be obtained. The increase in capacity during charge/discharge upon the first 8 cycles is because of the continuous activation of the Li2MnO3 component. As discussed in the previous XPS section, the apparent functional oxygen groups in surface carbon layers may slow down the kinetics of the related activation of Li2MnO3. However, it could be further electrochemically reduced to more conductive carbon layers by the gradual formation of Li2O during Li intercalation and deintercalation.35,36 The further reduced carbon layers not only facilitate the activation of Li2MnO3, but also act as protection layers to prevent thick formation of SEI layers and dissolution of transition metals because of HF attacking.37–39 In fact, after the initial activation, C-3 and C-15 materials are capable of delivering capacities as high as 42 and 54 mA h g−1, respectively, at a high rate of 20 C compared with only 13 mA h g−1 for the pristine sample as shown in Fig. 9(b). On the other hand, the post annealed samples C-3-H and C-15-H possess not only similar initial capacities compared to the pristine sample, but also the enhanced ability of capacity retention, i.e. 70% and 68%, respectively. It suggests that the sustained carbon layers at particle surfaces after post annealing are still able to function as the protection layers. Furthermore, more than 100 mA h g−1 capacity could be achieved at 10 C from both the samples, which might be attributed to increased extrinsic conductivity of carbon coating, as well as the transformed surface spinels providing fast diffusion channels of Li.40,41 Moreover, it is very important to note that the surface spinels may not be associated with the improved cycleability. On the contrary, they may even negatively contribute to the cyclability since spinel is an unstable phase when exposed to electrolyte (Mn2+ dissolution). To better understand the structural evolution process upon cycling, Fig. S4† compares the charge/discharge and dQ/dV curves at various cycling stages of all the samples. As widely accepted, the family of Li-rich layered cathodes is inevitably involved with a phase transformation to the spinel ordered structure during deintercaltion/intercalation of Li.12,13,42 The pristine material underwent phase transformation in the first 100 cycles. All the other cathodes after surface modification exhibit similar behaviors, indicating that neither surface coating of carbon nor subsequent heat treatment could contribute to the suppression of this structural evolution process, as also supported by arrows in dQ/dV plots.
Footnotes |
† Electronic supplementary information (ESI) available: Additional information includes characterizations by TGA and TEM of carbon-coated samples. See DOI: 10.1039/c4ra04976d |
‡ Present address: Department of Engineering Science, University of Oxford, Parks Road, Oxford OX1 3 PJ, UK. |
This journal is © The Royal Society of Chemistry 2014 |