Letter pubs.acs.org/NanoLett
Development of Microstrain in Aged Lithium Transition Metal Oxides Eung-Ju Lee,† Zonghai Chen,‡ Hyung-Ju Noh,† Sang Cheol Nam,§ Sung Kang,§ Do Hyeong Kim,§ Khalil Amine,*,‡,⊥ and Yang-Kook Sun*,† †
Department of Energy Engineering, Hanyang University, Seoul 133-791, South Korea Chemical Sciences and Engineering Division, Argonne National Laboratory, 9700 South Cass Avenue, Argonne, Illinois 60439, United States § Institute of Industrial Science and Technology, San32 Hyoja-Dong, Nam-Gu, Pohang-City, Gyeongbuk 790-330, Korea ⊥ King Abdulaziz University, Faculty of Science, Chemistry Department, 80203 Jeddah, Saudi Arabia ‡
ABSTRACT: Cathode materials with high energy density for lithium-ion batteries are highly desired in emerging applications in automobiles and stationary energy storage for the grid. Lithium transition metal oxide with concentration gradient of metal elements inside single particles was investigated as a promising high-energydensity cathode material. Electrochemical characterization demonstrated that a full cell with this cathode can be continuously operated for 2500 cycles with a capacity retention of 83.3%. Electron microscopy and high-resolution X-ray diffraction were employed to investigate the structural change of the cathode material after this extensive electrochemical testing. It was found that microstrain developed during the continuous charge/discharge cycling, resulting in cracking of nanoplates. This finding suggests that the performance of the cathode material can be further improved by optimizing the concentration gradient to minimize the microstrain and to reduce the lattice mismatch during cycling. KEYWORDS: Coprecipitation, Micro-Strain, Ni Rich, Cathode, Lithium, Battery High specific capacity, up to 250 mAh·g−1, and good capacity retention during charge/discharge cycling of LMR-NMC have been consistently reported in the open literature.7,10−13 However, the high specific capacity can only achieved when the materials are electrochemically activated at a high potential, for example, 4.7 V vs Li+/Li. In addition, LMR-NMC needs to work in a fairly wide potential window, for example, between 2.5 and 4.6 V vs Li+/Li, to maintain the high specific capacity and high energy density, although it has a medium average working potential of about 3.6 V vs Li+/Li. The wide working potential window significantly challenges the electrochemical stability of carbonate-based solvents, and a high potential electrolyte is needed to take full advantage of LMR-NMC. In addition, the electrochemical activation of LMR-NMC during the initial charge results in the structural instability of LMN-NMC, which in turn leads to a continuous potential decay (also called “voltage fade”) during normal charge/discharge cycling.10 The voltage fade will contribute to the continuous decay of the energy density of the battery system, resulting in significantly lower end-of-life energy density, which is used to determine the minimum size of batteries to meet requirements on both cycle life and energy density. Compared with LMR-NMC, nickel-rich oxides can deliver a fairly high specific capacity, up to 220 mAh·g−1, with a much narrower working potential window, a higher average working potential, an outstanding capacity retention, and no
A
worldwide effort is underway to develop advanced energyrelated technologies that would reduce the global consumption of nonrenewable fossil fuels and alleviate the negative impact of greenhouse gas emission.1 Among available energy storage technologies, the high-energy-density lithium-ion battery is the most promising and emerging technology for electrification of automobiles to reduce the fossil fuel consumption in the transportation sector. However, the deployment of long-range plug-in hybrid electric vehicles and pure electric vehicles requires advanced energy storage technology with a substantially higher energy density and lower cost than state-ofthe-art lithium-ion technologies. In a conventional lithium-ion battery, LiCoO2 is widely used as the cathode material, whereas graphite, like mesocarbon microbeads, is widely adopted as the anode material. In practice, the graphitic anode can store a fairly large amount of charge, more than 300 mAh·g−1, whereas the LiCoO2 cathode can only store a relatively smaller amount of charge, up to 148 mAh·g−1. The degree of lithium utilization for LiCoO2 is thus limited to 50% to maintain reasonable electrochemical performance.2 Clearly, the energy density of the cathode material has been a bottleneck. Hence, a major effort has been in progress to develop alternative lithium transition metal oxides with a higher degree of lithium utilization and specific energy density.3−9 Recently, two classes of cathode materials have emerged as the most promising candidates:8,9 nickel-rich oxides and lithium−manganese-rich nickel−manganese-cobalt oxides [LMR-NMC or xLi2MnO3·(1 − x)LiTMO2, where TM = Ni, Mn, CO].7,10 © XXXX American Chemical Society
Received: June 18, 2014
A
dx.doi.org/10.1021/nl5022859 | Nano Lett. XXXX, XXX, XXX−XXX
Nano Letters
Letter
voltage fade during normal cycling. We had previously demonstrated that a well-protected nickel-rich oxide can be cycled for 800 cycles with 95% capacity retention.8,9 It has been previously shown that nickel-rich oxides not only can deliver a high specific capacity but also have a tendency to react with nonaqueous electrolyte. Hence, there is a need to create a more stable barrier between the nickel-rich core and the electrolyte. Sun et al.8,9 have reported that full concentration gradient (FCG) materials with a nickel-rich core and a reduced nickel concentration at the outer layer take advantage of both the high capacity of the nickel-rich core and the high stability of the nickel-poor shell. In the previous work, Sun et al. discovered that an aligned nanorod network was formed in the FCG materials; this morphological feature helps to improve the percolation property of materials and to reduce the number grain boundaries in the diffusion pathway of the lithium ions. As a result, a higher rate capability was observed in cell tests with FCG cathode materials having nanorod structures.8 In this work, we sought to maximize the aspect ratio of the nanostructure to reduce the number of grain boundaries in the radial direction. The precursor was designed to have uniform distribution of manganese (25%) in each spherical particle. The concentration of nickel gradually decreased from the center (75%) toward the outer layer (55%), whereas the concentration of cobalt increased from the center (0%) toward the outer layer (20%). Physical Characterization. Electron probe microanalysis (EPMA) was used to characterize the distribution of the three transition metal elements (Ni, Co, and Mn) in the radial direction of a particle. The concentration profiles of the transition metal in the precursor and the lithiated particle are shown in Figure 1a and b, respectively. Intermixing of the transition metals
occurred during the sintering at high temperature, which can lead to an increase of nickel concentration in the outer layer. Therefore, a constant-concentration layer of 0.5 μm was designed as the outer layer to suppress the increase of nickel concentration after sintering. The scanning electron microscopy (SEM) images in Figure 2a and c show that both the precursor and the lithiated powder
Figure 2. SEM images of (a,b) precursor particles and (c,d) lithiated particles. (e) TEM image of cross-section of lithiated particle showing the aligned nanostructure inside a single particle. (f) Electron diffraction pattern on a single nanoplate showing that the c axis (in R3m ̅ space) is perpendicular to the basal plane of the plate.
maintained a spherical morphology with a uniform particle size. Figure 2b shows a zoomed view of a precursor particle, indicating self-assembly of nanoplates. After the lithiation, the nanoplates fused together to form submicron structures (see Figure 2d). The transmission electron microscopy (TEM) image in Figure 2e shows that the plated structures in the final powder have a thickness of about 50 nm, and the aligned plates extended from the surface to the center of the particle (Figure 2e). A localized electron diffraction on a single plate indicates that the c axis (based on R3m ̅ notation) is perpendicular to the basal plane of the plate or the ab planes in the material run continuously from the outer layer toward the center (Figure 2f). This special orientation of plates can facilitate lithium diffusion during charge/discharge cycling. Figure 3 shows the voltage profile of a Li/FCG half cell that was charged to 4.3 V at 25 °C and was discharged to 2.7 V at different temperatures. A constant current of C/5 was used for both charge and discharge. The cathode material delivered a specific capacity of about 190 mAh·g−1 at 25 °C, and its discharge capacity was about 110 mAh·g−1 at a temperature as low as −20 °C. The good low temperature performance is believed to originate from the reduced number of grain boundaries, which reduce the lithium-ion transport properties at low temperatures.
Figure 1. EPMA scan of the atomic ratio of transition metals as a function of the distance from the particle center to the surface for (a) the precursor and (b) the lithiated powder. B
dx.doi.org/10.1021/nl5022859 | Nano Lett. XXXX, XXX, XXX−XXX
Nano Letters
Letter
Figure 3. Voltage profiles of a Li/FCG half cell that was discharged with a constant current of C/5 to 2.5 V vs Li+/Li at different temperatures.
Electrochemical Characterization. The electrochemical performance of the FCG material was carried out in tests using both Li/FCG button cells and graphite/FCG pouch cells. Figure 4 shows the discharge capacity of Li/FCG button cells that were continuously charged and discharged with a constant current of C/2 for 100 cycles. Three button cells were tested at 25 °C with an upper cutoff potential of 4.3, 4.4, and 4.5 V vs Li+/Li, respectively. The cell charged to 4.3 V delivered an initial capacity of 177.6 mAh·g−1, and 95.9% capacity retention was obtained after 100 cycles. Raising the upper cutoff potential to 4.5 V led to a moderate increase of the initial discharge capacity (from 177.6 mAh·g−1 to 192.6 mAh·g−1). Meanwhile, the capacity retention was reduced to 92.1%. When the cells were tested at 55 °C, a higher initial discharge capacity was obtained, and a faster capacity fade was also observed, as expected. At the harshest test condition, when cycled between 2.7 and 4.5 V vs Li+/Li at 55 °C, the cell still maintained a capacity retention of 93.7% after 100 cycles (see Figure 4b). These short-term performance data can be enough to meet the cycle-life requirements for portable electronics, which generally require 500 cycles with 80% capacity retention. The difference in capacity retention among the different cycling conditions is also marginal for shortterm applications. However, a small performance difference can accumulate over an extended period of operation time for such applications as electric vehicles and stationary applications for the grid and can have a substantial negative impact on the end-oflife energy density. For long-term testing, four pouch cells using graphite as the anode and FCG as the cathode were assembled and continuously charged and discharged with a constant current of 1C for 2500 cycles. Figure 5 shows the discharge capacity of the pouch cells as a function of the cycle number. When charged up to 4.2 V at 25 °C, the cell lost 16.7% of its initial discharge capacity, whereas the one cycled up to 4.4 V at 55 °C lost 55% of its initial discharge capacity. It is common sense that the side reaction between the lithiated graphite with the nonaqueous electrolyte substantially contributes to the capacity fade of full cells. However, it is difficult to quantify the individual contribution from both the anode side and the cathode side. Therefore, the tested cells were dissembled to examine the potential change of the cell components. Scanning electron microscopy (SEM) images showed
Figure 4. Discharge capacity of Li/FCG half cells tested at (a) 25 °C and (b) 55 °C and different cutoff potentials.
Figure 5. Discharge capacity of graphite/FCG full cells 25 and 55 °C and different cutoff potentials.
that the integrity of both the negative electrode and the positive electrode was maintained after extensive cycling; however, some microcrack was observed inside particles of FCG after extensive cycling. C
dx.doi.org/10.1021/nl5022859 | Nano Lett. XXXX, XXX, XXX−XXX
Nano Letters
Letter
Figure 6. Cross-sectional SEM images for (a) pristine FCG powder and FCG material recovered from a full cell cycled up to (b) 4.2 V at 25 °C, (c) 4.2 V at 55 °C, (d) 4.4 V at 25 °C, and (e) 4.4 V at 55 °C.
Morphological Change of Microstructure. To investigate the change of microstructure inside particles, we employed focus ion beam (FIB) to remove part of a material particle for electron microscopy study. Figure 6a shows the SEM image obtained from the cross section of a pristine particle (before electrochemical characterization). It shows that the pristine particle is very dense and has few voids inside the particle. However, numerous voids developed after electrochemical cycling for 2500 cycles (Figure 6b−e), particularly when the material was characterized at a harsh condition between 2.7 and 4.4 V at 55 °C, as is evident by a large crack throughout the whole particle (Figure 6e). Similarly, TEM images showed fewer voids in the pristine particle compared with the cycled one under various conditions (see Figure 7). Miller et al. previously reported a similar finding for LiNi0.8Co0.15Al0.05O2 that was characterized after cycling at a milder condition (4.0 V vs Li+/Li).14 The crack in the LiNi0.8Co0.15Al0.05O2 particles was believed to be caused by anisotropic expansion of different grains inside a single particle that led to substantial microstrain over the grain boundary.14 In the FCG material, the nanoplate structures were well aligned inside the particle, minimizing the mechanical strain caused by the anisotropic expansion. However, the concentration gradient of the transition metal elements along the radial direction can lead to a slight mismatch of c parameters during the insertion/ removal of lithium. It is believed that the small microstrain along the nanoplates led to the crack of the plates and development of voids inside the particle. Compared to LiNi0.8Co0.15Al0.05O2 without the aligned structure, FCG had
much smaller microstrain during charge/discharge cycling and, hence, could survive a much harsher condition before the development of large cracks. High-Resolution X-ray Diffraction Study. To validate the development of microstrain after electrochemical cycling, the recovered electrode materials before and after extensive cycling were further characterized by high resolution X-ray diffraction (HRXRD). Figure 8 shows the HRXRD patterns for FCG material before and after the extensive cycling. The lack of any peak around 5.5−6° (2θ) implies that the FCG material has a pure layer structure based on a hexagonal a-NaFeO2 structure with space group R3̅m without superstructure. Figure 9 shows zoomed views of the HRXRD patterns for FCG material before and after the extensive cycling. Peak broadening was observed for all the detectable peaks, but only peaks (003), (108̅), and (110) are shown in Figure 9 for better visualization. After electrochemical testing, peak (003) shifted toward a lower 2θ value, indicating an expansion on the c axis after cycling. In addition, the right shift of (110) shift suggests a contraction of the a/b axis after cycling. The HRXRD data were further fitted using a layer structure (R3̅m model) to quantify the origin of the peak broadening. Figure 10 clearly shows that the hexagonal structure (R3̅m) was well maintained even after extensive charge/discharge testing. Figure 10 shows a typical fit of the pristine powder using an R3̅m structure, and the key fitting parameters obtained are summarized in Table 1. It can be seen that the a/b value slightly D
dx.doi.org/10.1021/nl5022859 | Nano Lett. XXXX, XXX, XXX−XXX
Nano Letters
Letter
Figure 7. Cross-sectional TEM images for (a) pristine FCG powder and FCG material recovered from a full cell cycled up to (b) 4.2 V at 25 °C, (c) 4.2 V at 55 °C, (d) 4.4 V at 25 °C, and (e) 4.4 V at 55 °C.
decreased by about 0.3% after extensive cycling, and the c value slightly increased by about 0.3%. In addition, Williamson−Hall analysis of on (00l) peaks of HRXRD data was also carried out for a better quantification on the impact of microstrain development. The instrumental
broadening (βinst) was corrected corresponding to each (00l) peak of FCG using eq 1 βhkl = E
2 ′ 2 − βinst βhkl
(1)
dx.doi.org/10.1021/nl5022859 | Nano Lett. XXXX, XXX, XXX−XXX
Nano Letters
Letter
Figure 8. High-resolution X-ray diffraction patterns of FCG powders showing no superstructure after long-term electrochemical testing. Figure 10. Typical fitting results to FCG powder using R3̅m model. The experimental data shown here are for the pristine FCG powder.
Table 1. Cell Parameters and Peak Profile Parameters for FCG Powder before and after Long-Term Electrochemical Testing sample condition
a/b, Å
c, Å
pristine 4.2 V at 25 °C 4.2 V at 55 °C 4.4 V at 25 °C 4.4 V at 55 °C
2.870(1) 2.863(6) 2.865(5) 2.865(5) 2.861(5)
14.215(6) 14.247(4) 14.230(1) 14.232(7) 14.256(7)
ε=
βhkl 4tan θ
(3)
In eq 2, K is the crystallite shape factor and was assumed to be 1 to simplify the calculation, and λ is the wavelength of the detecting X-ray (0.0413 nm). Figure 11 shows the results of the
Figure 9. High-resolution X-ray diffraction patterns of FCG powders showing the significant peak broadening after long-term electrochemical testing.
Figure 11. Comparison of (a) microstrain, and (b) crystallite size before and after extensive cycling for 2500 cycles.
In above equation, βhkl ′ is the measured halfway-full-width (HWFW) of each peak. Then, the crystallite size (D) and the microstrain (ε) can be calculated using following equations
D=
Kλ βhkl cos θ
Williamson−Hall analysis on (00l) peaks. It is shown that the pristine material has a crystallite size of about 120 nm, which is close to the thickness of nanoplates (see Figure 2e). In addition, a fairly large microstrain (∼0.12%) was also observed for the pristine material; this is believed originated from the concentration gradient of transition metals in the particles that caused
(2) F
dx.doi.org/10.1021/nl5022859 | Nano Lett. XXXX, XXX, XXX−XXX
Nano Letters
Letter
were collected at sector 11-BM of the Advanced Photon Source at Argonne National Laboratory. The wavelength of the X-ray source was 0.413362 Å. The morphology of the prepared powders was characterized using scanning electron microscopy (SEM, Hitachi Model S-4800). The local composition of the FCG materials and cross sections of particles were prepared by embedding the particles in an epoxy and mechanical polishing was applied to obtain the cross section. Line scans of the polished surfaces for prepared FCG precursor and lithiated FCG powders were analyzed via an electron probe microanalysis (EPMA, JEOL model JXA-8100) with an electron beam size of 1 by 1 μm. Transmission electron microscopy (TEM) samples were prepared by focused ion beam and examined using a JEOL model JEM-2100F instrument.
mismatch on the lattice parameters. After the extensive cycling, the microstrain was almost doubled (to about 0.24%), and a consistent reduction in the crystallite size was also observed (see Figure 11). In summary, an FCG lithium transition metal oxide with wellaligned nanoplate structures was synthesized by a modified coprecipitation process. It was demonstrated that FCG material can be continuously cycled up to 2500 cycles with good capacity retention. The morphology characterization and structure analysis revealed that microstrain developed during the electrochemical testing, resulting in cracking inside the particle. The aligned nanoplate structure helps to reduce the number of grain boundaries and the structure strain across the boundary caused by the anisotropic expansion of different grains. Further improvement in electrochemical performance of FCG material can be achieved by optimizing the concentration gradient of transition metals to reduce the microstrain developed during cycling. Experimental Details. Synthesis of FCG Material. Spherical FCG [NixCoyMnz](OH)2 precursor was synthesized via the coprecipitation method.8,15 A Ni-poor aqueous solution (molar ratio of Ni:Co:Mn = 0.55:0.20:0.25) consisting of NiSO4·6H2O, CoSO4·7H2O, and MnSO4·5H2O was slowly pumped from tank 2 into a Ni-rich (molar ratio of Ni:Mn = 0.75:0.25) stock solution in tank 1. The homogeneously mixed solution was then fed into a continuously stirred tank reactor (CSTR, 4 L) in a replenished N2 atmosphere. After the composition of tank 1 reached the desired surface composition, the feeding of tank 2 was stopped. Concurrently, a 4.0 mol L−1 NaOH solution (aq) and the desired amount of a NH4OH (2.0 mol L−1) chelating agent solution (aq) were pumped separately into the reactor. The concentration of the solution (2.0 mol L−1 total for transition metal elements), pH (11.6), temperature (30 °C), and stirring speed (1000 rounds per minute) of the mixture in the reactor were carefully controlled. During the early stage of the coprecipitation process, [Ni0.75Mn0.25](OH)2 (center composition) was first precipitated. Nickel−cobalt−manganese hydroxide with a different composition was then gradually piled onto the formed [Ni0.75Mn0.25](OH)2 particles to result in a roughly linear composition change of Ni and Co toward the outer surface of the particles. The precursor powders were obtained through filtering, washing, and drying at 100 °C overnight. The obtained FCG [Ni0.60Co0.15Mn 0.25 ](OH) 2 , FCG [Ni 0.59 Co 0.16 Mn 0.25 ](OH) 2 , and [Ni0.56Co0.19Mn0.25](OH)2 were mixed with LiOH·H2O, and the mixture was calcined at 845 °C for 10 h. Electrochemical Characterization. For fabrication of the cathodes, the above synthesized powders were mixed with carbon black and poly(vinylidene fluoride) (85:7.5:7.5) in N-methylpyrrolidinone. The obtained slurry was coated onto Al foil and roll-pressed. Preliminary cell tests were performed with a 2032 button cell using lithium metal as the anode. Longterm cycle-life tests were performed in a laminated-type full cell (35 mAh) wrapped with an Al pouch. Mesocarbon microbead graphite (Osaka Gas) was used as the anode. The electrolyte solution was 1.2 M LiPF6 in ethylene carbonate and ethyl methyl carbonate (3:7 in volume). The cells were charged and discharged at 25 and 55 °C and between 3.0 and 4.2 to 4.4 V by applying a constant 1C current (35 mA corresponds to 180 mA g−1). Morphology Characterization. To recover the powder from the cycled cells, we opened the cycled pouch-type full cells in an Ar-filled dry room. After the remaining electrolyte was carefully removed from the surface of the electrode, the cathode materials were recovered from the current collector. The HRXRD patterns
■
AUTHOR INFORMATION
Corresponding Authors
*E-mail: [email protected] (Y. S.). *E-mail: [email protected] (K. A.). Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS Research at Hangyang University was supported by the Human Resources Development Program (No. 20124010203310) of the Korea Institute of Energy Technology Evaluation and Planning (KETEP) grant funded by the Korea government Ministry of Trade, Industry, and Energy and also by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MEST; No. 2009-0092780). Support from David Howell (Team Lead), Tien Duong, and Peter Faguy of the Vehicle Technologies Program, Hybrid and Electric Systems, at the U. S. Department of Energy, Office of Energy Efficiency and Renewable Energy, is gratefully acknowledged. The authors acknowledge the use of the Advanced Photon Source (APS) of Argonne National Laboratory, which is supported by the U. S. Department of Energy, Office of Science, Office of Basic Energy Science.
■
REFERENCES
(1) Hawkins, T. R.; Gausen, O. M.; Stromman, A. H. Int. J. Life Cycle Assess. 2012, 17, 997−1014. (2) Chen, Z. H.; Lu, Z. H.; Dahn, J. R. J. Electrochem. Soc. 2002, 149, A1604−1609. (3) Lu, Z. H.; Dahn, J. R. J. Electrochem. Soc. 2002, 149, A815−A822. (4) Lu, Z. H.; Beaulieu, L. Y.; Donaberger, R. A.; Thomas, C. L.; Dahn, J. R. J. Electrochem. Soc. 2002, 149, A778−A791. (5) Kang, S. H.; Pol, V. G.; Belharouak, I.; Thackeray, M. M. J. Electrochem. Soc. 2010, 157, A267−A271. (6) Zonghai, C.; Yan, Q.; Amine, K.; Sun, Y. K. J. Mater. Chem. 2010, 20, 7606−7612. (7) Johnson, C. S.; Li, N.; Lefief, C.; Vaughey, J. T.; Thackeray, M. M. Chem. Mater. 2008, 20, 6095−6106. (8) Sun, Y.-K.; Chen, Z.; Noh, H.-J.; Lee, D.-J.; Jung, H.-G.; Ren, Y.; Wang, S.; Yoon, C. S.; Myung, S.-T.; Amine, K. Nat. Mater. 2012, 11, 942−947. (9) Sun, Y.-K.; Myung, S.-T.; Park, B.-C.; Prakash, J.; Belharouak, I.; Amine, K. Nat. Mater. 2009, 8, 320−324. (10) Croy, J. R.; Gallagher, K. G.; Balasubramanian, M.; Chen, Z.; Ren, Y.; Kim, D.; Kang, S.-H.; Dees, D. W.; Thackeray, M. M. J. Phys. Chem. C 2013, 117, 6525−6536. (11) Bettge, M.; Li, Y.; Sankaran, B.; Rago, N. D.; Spila, T.; Haasch, R. T.; Petrov, I.; Abraham, D. P. J. Power Sources 2013, 233, 346−357. (12) Li, Y.; Bettge, M.; Polzin, B.; Zhu, Y.; Balasubramanian, M.; Abraham, D. P. J. Electrochem. Soc. 2013, 160, A3006−A3019. G
dx.doi.org/10.1021/nl5022859 | Nano Lett. XXXX, XXX, XXX−XXX
Nano Letters
Letter
(13) Mohanty, D.; Sefat, A. S.; Kalnaus, S.; Li, J.; Meisner, R. A.; Payzant, E. A.; Abraham, D. P.; Wood, D. L.; Daniel, C. J. Mater. Chem. A 2013, 1, 6249−6261. (14) Miller, D. J.; Proff, C.; Wen, J. G.; Abraham, D. P.; Bareno, J. Adv. Ener. Mater. 2013, 3, 1098. (15) Noh, H.-J.; Chen, Z.; Yoon, C.-S.; Lu, J.; Amine, K.; Sun, Y.-K. Chem. Mater. 2013, 25, 2109−2115.
H
dx.doi.org/10.1021/nl5022859 | Nano Lett. XXXX, XXX, XXX−XXX
Development of Microstrain in Aged Lithium Transition Metal Oxides Eung-Ju Lee,† Zonghai Chen,‡ Hyung-Ju Noh,† Sang Cheol Nam,§ Sung Kang,§ Do Hyeong Kim,§ Khalil Amine,*,‡,⊥ and Yang-Kook Sun*,† †
Department of Energy Engineering, Hanyang University, Seoul 133-791, South Korea Chemical Sciences and Engineering Division, Argonne National Laboratory, 9700 South Cass Avenue, Argonne, Illinois 60439, United States § Institute of Industrial Science and Technology, San32 Hyoja-Dong, Nam-Gu, Pohang-City, Gyeongbuk 790-330, Korea ⊥ King Abdulaziz University, Faculty of Science, Chemistry Department, 80203 Jeddah, Saudi Arabia ‡
ABSTRACT: Cathode materials with high energy density for lithium-ion batteries are highly desired in emerging applications in automobiles and stationary energy storage for the grid. Lithium transition metal oxide with concentration gradient of metal elements inside single particles was investigated as a promising high-energydensity cathode material. Electrochemical characterization demonstrated that a full cell with this cathode can be continuously operated for 2500 cycles with a capacity retention of 83.3%. Electron microscopy and high-resolution X-ray diffraction were employed to investigate the structural change of the cathode material after this extensive electrochemical testing. It was found that microstrain developed during the continuous charge/discharge cycling, resulting in cracking of nanoplates. This finding suggests that the performance of the cathode material can be further improved by optimizing the concentration gradient to minimize the microstrain and to reduce the lattice mismatch during cycling. KEYWORDS: Coprecipitation, Micro-Strain, Ni Rich, Cathode, Lithium, Battery High specific capacity, up to 250 mAh·g−1, and good capacity retention during charge/discharge cycling of LMR-NMC have been consistently reported in the open literature.7,10−13 However, the high specific capacity can only achieved when the materials are electrochemically activated at a high potential, for example, 4.7 V vs Li+/Li. In addition, LMR-NMC needs to work in a fairly wide potential window, for example, between 2.5 and 4.6 V vs Li+/Li, to maintain the high specific capacity and high energy density, although it has a medium average working potential of about 3.6 V vs Li+/Li. The wide working potential window significantly challenges the electrochemical stability of carbonate-based solvents, and a high potential electrolyte is needed to take full advantage of LMR-NMC. In addition, the electrochemical activation of LMR-NMC during the initial charge results in the structural instability of LMN-NMC, which in turn leads to a continuous potential decay (also called “voltage fade”) during normal charge/discharge cycling.10 The voltage fade will contribute to the continuous decay of the energy density of the battery system, resulting in significantly lower end-of-life energy density, which is used to determine the minimum size of batteries to meet requirements on both cycle life and energy density. Compared with LMR-NMC, nickel-rich oxides can deliver a fairly high specific capacity, up to 220 mAh·g−1, with a much narrower working potential window, a higher average working potential, an outstanding capacity retention, and no
A
worldwide effort is underway to develop advanced energyrelated technologies that would reduce the global consumption of nonrenewable fossil fuels and alleviate the negative impact of greenhouse gas emission.1 Among available energy storage technologies, the high-energy-density lithium-ion battery is the most promising and emerging technology for electrification of automobiles to reduce the fossil fuel consumption in the transportation sector. However, the deployment of long-range plug-in hybrid electric vehicles and pure electric vehicles requires advanced energy storage technology with a substantially higher energy density and lower cost than state-ofthe-art lithium-ion technologies. In a conventional lithium-ion battery, LiCoO2 is widely used as the cathode material, whereas graphite, like mesocarbon microbeads, is widely adopted as the anode material. In practice, the graphitic anode can store a fairly large amount of charge, more than 300 mAh·g−1, whereas the LiCoO2 cathode can only store a relatively smaller amount of charge, up to 148 mAh·g−1. The degree of lithium utilization for LiCoO2 is thus limited to 50% to maintain reasonable electrochemical performance.2 Clearly, the energy density of the cathode material has been a bottleneck. Hence, a major effort has been in progress to develop alternative lithium transition metal oxides with a higher degree of lithium utilization and specific energy density.3−9 Recently, two classes of cathode materials have emerged as the most promising candidates:8,9 nickel-rich oxides and lithium−manganese-rich nickel−manganese-cobalt oxides [LMR-NMC or xLi2MnO3·(1 − x)LiTMO2, where TM = Ni, Mn, CO].7,10 © XXXX American Chemical Society
Received: June 18, 2014
A
dx.doi.org/10.1021/nl5022859 | Nano Lett. XXXX, XXX, XXX−XXX
Nano Letters
Letter
voltage fade during normal cycling. We had previously demonstrated that a well-protected nickel-rich oxide can be cycled for 800 cycles with 95% capacity retention.8,9 It has been previously shown that nickel-rich oxides not only can deliver a high specific capacity but also have a tendency to react with nonaqueous electrolyte. Hence, there is a need to create a more stable barrier between the nickel-rich core and the electrolyte. Sun et al.8,9 have reported that full concentration gradient (FCG) materials with a nickel-rich core and a reduced nickel concentration at the outer layer take advantage of both the high capacity of the nickel-rich core and the high stability of the nickel-poor shell. In the previous work, Sun et al. discovered that an aligned nanorod network was formed in the FCG materials; this morphological feature helps to improve the percolation property of materials and to reduce the number grain boundaries in the diffusion pathway of the lithium ions. As a result, a higher rate capability was observed in cell tests with FCG cathode materials having nanorod structures.8 In this work, we sought to maximize the aspect ratio of the nanostructure to reduce the number of grain boundaries in the radial direction. The precursor was designed to have uniform distribution of manganese (25%) in each spherical particle. The concentration of nickel gradually decreased from the center (75%) toward the outer layer (55%), whereas the concentration of cobalt increased from the center (0%) toward the outer layer (20%). Physical Characterization. Electron probe microanalysis (EPMA) was used to characterize the distribution of the three transition metal elements (Ni, Co, and Mn) in the radial direction of a particle. The concentration profiles of the transition metal in the precursor and the lithiated particle are shown in Figure 1a and b, respectively. Intermixing of the transition metals
occurred during the sintering at high temperature, which can lead to an increase of nickel concentration in the outer layer. Therefore, a constant-concentration layer of 0.5 μm was designed as the outer layer to suppress the increase of nickel concentration after sintering. The scanning electron microscopy (SEM) images in Figure 2a and c show that both the precursor and the lithiated powder
Figure 2. SEM images of (a,b) precursor particles and (c,d) lithiated particles. (e) TEM image of cross-section of lithiated particle showing the aligned nanostructure inside a single particle. (f) Electron diffraction pattern on a single nanoplate showing that the c axis (in R3m ̅ space) is perpendicular to the basal plane of the plate.
maintained a spherical morphology with a uniform particle size. Figure 2b shows a zoomed view of a precursor particle, indicating self-assembly of nanoplates. After the lithiation, the nanoplates fused together to form submicron structures (see Figure 2d). The transmission electron microscopy (TEM) image in Figure 2e shows that the plated structures in the final powder have a thickness of about 50 nm, and the aligned plates extended from the surface to the center of the particle (Figure 2e). A localized electron diffraction on a single plate indicates that the c axis (based on R3m ̅ notation) is perpendicular to the basal plane of the plate or the ab planes in the material run continuously from the outer layer toward the center (Figure 2f). This special orientation of plates can facilitate lithium diffusion during charge/discharge cycling. Figure 3 shows the voltage profile of a Li/FCG half cell that was charged to 4.3 V at 25 °C and was discharged to 2.7 V at different temperatures. A constant current of C/5 was used for both charge and discharge. The cathode material delivered a specific capacity of about 190 mAh·g−1 at 25 °C, and its discharge capacity was about 110 mAh·g−1 at a temperature as low as −20 °C. The good low temperature performance is believed to originate from the reduced number of grain boundaries, which reduce the lithium-ion transport properties at low temperatures.
Figure 1. EPMA scan of the atomic ratio of transition metals as a function of the distance from the particle center to the surface for (a) the precursor and (b) the lithiated powder. B
dx.doi.org/10.1021/nl5022859 | Nano Lett. XXXX, XXX, XXX−XXX
Nano Letters
Letter
Figure 3. Voltage profiles of a Li/FCG half cell that was discharged with a constant current of C/5 to 2.5 V vs Li+/Li at different temperatures.
Electrochemical Characterization. The electrochemical performance of the FCG material was carried out in tests using both Li/FCG button cells and graphite/FCG pouch cells. Figure 4 shows the discharge capacity of Li/FCG button cells that were continuously charged and discharged with a constant current of C/2 for 100 cycles. Three button cells were tested at 25 °C with an upper cutoff potential of 4.3, 4.4, and 4.5 V vs Li+/Li, respectively. The cell charged to 4.3 V delivered an initial capacity of 177.6 mAh·g−1, and 95.9% capacity retention was obtained after 100 cycles. Raising the upper cutoff potential to 4.5 V led to a moderate increase of the initial discharge capacity (from 177.6 mAh·g−1 to 192.6 mAh·g−1). Meanwhile, the capacity retention was reduced to 92.1%. When the cells were tested at 55 °C, a higher initial discharge capacity was obtained, and a faster capacity fade was also observed, as expected. At the harshest test condition, when cycled between 2.7 and 4.5 V vs Li+/Li at 55 °C, the cell still maintained a capacity retention of 93.7% after 100 cycles (see Figure 4b). These short-term performance data can be enough to meet the cycle-life requirements for portable electronics, which generally require 500 cycles with 80% capacity retention. The difference in capacity retention among the different cycling conditions is also marginal for shortterm applications. However, a small performance difference can accumulate over an extended period of operation time for such applications as electric vehicles and stationary applications for the grid and can have a substantial negative impact on the end-oflife energy density. For long-term testing, four pouch cells using graphite as the anode and FCG as the cathode were assembled and continuously charged and discharged with a constant current of 1C for 2500 cycles. Figure 5 shows the discharge capacity of the pouch cells as a function of the cycle number. When charged up to 4.2 V at 25 °C, the cell lost 16.7% of its initial discharge capacity, whereas the one cycled up to 4.4 V at 55 °C lost 55% of its initial discharge capacity. It is common sense that the side reaction between the lithiated graphite with the nonaqueous electrolyte substantially contributes to the capacity fade of full cells. However, it is difficult to quantify the individual contribution from both the anode side and the cathode side. Therefore, the tested cells were dissembled to examine the potential change of the cell components. Scanning electron microscopy (SEM) images showed
Figure 4. Discharge capacity of Li/FCG half cells tested at (a) 25 °C and (b) 55 °C and different cutoff potentials.
Figure 5. Discharge capacity of graphite/FCG full cells 25 and 55 °C and different cutoff potentials.
that the integrity of both the negative electrode and the positive electrode was maintained after extensive cycling; however, some microcrack was observed inside particles of FCG after extensive cycling. C
dx.doi.org/10.1021/nl5022859 | Nano Lett. XXXX, XXX, XXX−XXX
Nano Letters
Letter
Figure 6. Cross-sectional SEM images for (a) pristine FCG powder and FCG material recovered from a full cell cycled up to (b) 4.2 V at 25 °C, (c) 4.2 V at 55 °C, (d) 4.4 V at 25 °C, and (e) 4.4 V at 55 °C.
Morphological Change of Microstructure. To investigate the change of microstructure inside particles, we employed focus ion beam (FIB) to remove part of a material particle for electron microscopy study. Figure 6a shows the SEM image obtained from the cross section of a pristine particle (before electrochemical characterization). It shows that the pristine particle is very dense and has few voids inside the particle. However, numerous voids developed after electrochemical cycling for 2500 cycles (Figure 6b−e), particularly when the material was characterized at a harsh condition between 2.7 and 4.4 V at 55 °C, as is evident by a large crack throughout the whole particle (Figure 6e). Similarly, TEM images showed fewer voids in the pristine particle compared with the cycled one under various conditions (see Figure 7). Miller et al. previously reported a similar finding for LiNi0.8Co0.15Al0.05O2 that was characterized after cycling at a milder condition (4.0 V vs Li+/Li).14 The crack in the LiNi0.8Co0.15Al0.05O2 particles was believed to be caused by anisotropic expansion of different grains inside a single particle that led to substantial microstrain over the grain boundary.14 In the FCG material, the nanoplate structures were well aligned inside the particle, minimizing the mechanical strain caused by the anisotropic expansion. However, the concentration gradient of the transition metal elements along the radial direction can lead to a slight mismatch of c parameters during the insertion/ removal of lithium. It is believed that the small microstrain along the nanoplates led to the crack of the plates and development of voids inside the particle. Compared to LiNi0.8Co0.15Al0.05O2 without the aligned structure, FCG had
much smaller microstrain during charge/discharge cycling and, hence, could survive a much harsher condition before the development of large cracks. High-Resolution X-ray Diffraction Study. To validate the development of microstrain after electrochemical cycling, the recovered electrode materials before and after extensive cycling were further characterized by high resolution X-ray diffraction (HRXRD). Figure 8 shows the HRXRD patterns for FCG material before and after the extensive cycling. The lack of any peak around 5.5−6° (2θ) implies that the FCG material has a pure layer structure based on a hexagonal a-NaFeO2 structure with space group R3̅m without superstructure. Figure 9 shows zoomed views of the HRXRD patterns for FCG material before and after the extensive cycling. Peak broadening was observed for all the detectable peaks, but only peaks (003), (108̅), and (110) are shown in Figure 9 for better visualization. After electrochemical testing, peak (003) shifted toward a lower 2θ value, indicating an expansion on the c axis after cycling. In addition, the right shift of (110) shift suggests a contraction of the a/b axis after cycling. The HRXRD data were further fitted using a layer structure (R3̅m model) to quantify the origin of the peak broadening. Figure 10 clearly shows that the hexagonal structure (R3̅m) was well maintained even after extensive charge/discharge testing. Figure 10 shows a typical fit of the pristine powder using an R3̅m structure, and the key fitting parameters obtained are summarized in Table 1. It can be seen that the a/b value slightly D
dx.doi.org/10.1021/nl5022859 | Nano Lett. XXXX, XXX, XXX−XXX
Nano Letters
Letter
Figure 7. Cross-sectional TEM images for (a) pristine FCG powder and FCG material recovered from a full cell cycled up to (b) 4.2 V at 25 °C, (c) 4.2 V at 55 °C, (d) 4.4 V at 25 °C, and (e) 4.4 V at 55 °C.
decreased by about 0.3% after extensive cycling, and the c value slightly increased by about 0.3%. In addition, Williamson−Hall analysis of on (00l) peaks of HRXRD data was also carried out for a better quantification on the impact of microstrain development. The instrumental
broadening (βinst) was corrected corresponding to each (00l) peak of FCG using eq 1 βhkl = E
2 ′ 2 − βinst βhkl
(1)
dx.doi.org/10.1021/nl5022859 | Nano Lett. XXXX, XXX, XXX−XXX
Nano Letters
Letter
Figure 8. High-resolution X-ray diffraction patterns of FCG powders showing no superstructure after long-term electrochemical testing. Figure 10. Typical fitting results to FCG powder using R3̅m model. The experimental data shown here are for the pristine FCG powder.
Table 1. Cell Parameters and Peak Profile Parameters for FCG Powder before and after Long-Term Electrochemical Testing sample condition
a/b, Å
c, Å
pristine 4.2 V at 25 °C 4.2 V at 55 °C 4.4 V at 25 °C 4.4 V at 55 °C
2.870(1) 2.863(6) 2.865(5) 2.865(5) 2.861(5)
14.215(6) 14.247(4) 14.230(1) 14.232(7) 14.256(7)
ε=
βhkl 4tan θ
(3)
In eq 2, K is the crystallite shape factor and was assumed to be 1 to simplify the calculation, and λ is the wavelength of the detecting X-ray (0.0413 nm). Figure 11 shows the results of the
Figure 9. High-resolution X-ray diffraction patterns of FCG powders showing the significant peak broadening after long-term electrochemical testing.
Figure 11. Comparison of (a) microstrain, and (b) crystallite size before and after extensive cycling for 2500 cycles.
In above equation, βhkl ′ is the measured halfway-full-width (HWFW) of each peak. Then, the crystallite size (D) and the microstrain (ε) can be calculated using following equations
D=
Kλ βhkl cos θ
Williamson−Hall analysis on (00l) peaks. It is shown that the pristine material has a crystallite size of about 120 nm, which is close to the thickness of nanoplates (see Figure 2e). In addition, a fairly large microstrain (∼0.12%) was also observed for the pristine material; this is believed originated from the concentration gradient of transition metals in the particles that caused
(2) F
dx.doi.org/10.1021/nl5022859 | Nano Lett. XXXX, XXX, XXX−XXX
Nano Letters
Letter
were collected at sector 11-BM of the Advanced Photon Source at Argonne National Laboratory. The wavelength of the X-ray source was 0.413362 Å. The morphology of the prepared powders was characterized using scanning electron microscopy (SEM, Hitachi Model S-4800). The local composition of the FCG materials and cross sections of particles were prepared by embedding the particles in an epoxy and mechanical polishing was applied to obtain the cross section. Line scans of the polished surfaces for prepared FCG precursor and lithiated FCG powders were analyzed via an electron probe microanalysis (EPMA, JEOL model JXA-8100) with an electron beam size of 1 by 1 μm. Transmission electron microscopy (TEM) samples were prepared by focused ion beam and examined using a JEOL model JEM-2100F instrument.
mismatch on the lattice parameters. After the extensive cycling, the microstrain was almost doubled (to about 0.24%), and a consistent reduction in the crystallite size was also observed (see Figure 11). In summary, an FCG lithium transition metal oxide with wellaligned nanoplate structures was synthesized by a modified coprecipitation process. It was demonstrated that FCG material can be continuously cycled up to 2500 cycles with good capacity retention. The morphology characterization and structure analysis revealed that microstrain developed during the electrochemical testing, resulting in cracking inside the particle. The aligned nanoplate structure helps to reduce the number of grain boundaries and the structure strain across the boundary caused by the anisotropic expansion of different grains. Further improvement in electrochemical performance of FCG material can be achieved by optimizing the concentration gradient of transition metals to reduce the microstrain developed during cycling. Experimental Details. Synthesis of FCG Material. Spherical FCG [NixCoyMnz](OH)2 precursor was synthesized via the coprecipitation method.8,15 A Ni-poor aqueous solution (molar ratio of Ni:Co:Mn = 0.55:0.20:0.25) consisting of NiSO4·6H2O, CoSO4·7H2O, and MnSO4·5H2O was slowly pumped from tank 2 into a Ni-rich (molar ratio of Ni:Mn = 0.75:0.25) stock solution in tank 1. The homogeneously mixed solution was then fed into a continuously stirred tank reactor (CSTR, 4 L) in a replenished N2 atmosphere. After the composition of tank 1 reached the desired surface composition, the feeding of tank 2 was stopped. Concurrently, a 4.0 mol L−1 NaOH solution (aq) and the desired amount of a NH4OH (2.0 mol L−1) chelating agent solution (aq) were pumped separately into the reactor. The concentration of the solution (2.0 mol L−1 total for transition metal elements), pH (11.6), temperature (30 °C), and stirring speed (1000 rounds per minute) of the mixture in the reactor were carefully controlled. During the early stage of the coprecipitation process, [Ni0.75Mn0.25](OH)2 (center composition) was first precipitated. Nickel−cobalt−manganese hydroxide with a different composition was then gradually piled onto the formed [Ni0.75Mn0.25](OH)2 particles to result in a roughly linear composition change of Ni and Co toward the outer surface of the particles. The precursor powders were obtained through filtering, washing, and drying at 100 °C overnight. The obtained FCG [Ni0.60Co0.15Mn 0.25 ](OH) 2 , FCG [Ni 0.59 Co 0.16 Mn 0.25 ](OH) 2 , and [Ni0.56Co0.19Mn0.25](OH)2 were mixed with LiOH·H2O, and the mixture was calcined at 845 °C for 10 h. Electrochemical Characterization. For fabrication of the cathodes, the above synthesized powders were mixed with carbon black and poly(vinylidene fluoride) (85:7.5:7.5) in N-methylpyrrolidinone. The obtained slurry was coated onto Al foil and roll-pressed. Preliminary cell tests were performed with a 2032 button cell using lithium metal as the anode. Longterm cycle-life tests were performed in a laminated-type full cell (35 mAh) wrapped with an Al pouch. Mesocarbon microbead graphite (Osaka Gas) was used as the anode. The electrolyte solution was 1.2 M LiPF6 in ethylene carbonate and ethyl methyl carbonate (3:7 in volume). The cells were charged and discharged at 25 and 55 °C and between 3.0 and 4.2 to 4.4 V by applying a constant 1C current (35 mA corresponds to 180 mA g−1). Morphology Characterization. To recover the powder from the cycled cells, we opened the cycled pouch-type full cells in an Ar-filled dry room. After the remaining electrolyte was carefully removed from the surface of the electrode, the cathode materials were recovered from the current collector. The HRXRD patterns
■
AUTHOR INFORMATION
Corresponding Authors
*E-mail: [email protected] (Y. S.). *E-mail: [email protected] (K. A.). Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS Research at Hangyang University was supported by the Human Resources Development Program (No. 20124010203310) of the Korea Institute of Energy Technology Evaluation and Planning (KETEP) grant funded by the Korea government Ministry of Trade, Industry, and Energy and also by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MEST; No. 2009-0092780). Support from David Howell (Team Lead), Tien Duong, and Peter Faguy of the Vehicle Technologies Program, Hybrid and Electric Systems, at the U. S. Department of Energy, Office of Energy Efficiency and Renewable Energy, is gratefully acknowledged. The authors acknowledge the use of the Advanced Photon Source (APS) of Argonne National Laboratory, which is supported by the U. S. Department of Energy, Office of Science, Office of Basic Energy Science.
■
REFERENCES
(1) Hawkins, T. R.; Gausen, O. M.; Stromman, A. H. Int. J. Life Cycle Assess. 2012, 17, 997−1014. (2) Chen, Z. H.; Lu, Z. H.; Dahn, J. R. J. Electrochem. Soc. 2002, 149, A1604−1609. (3) Lu, Z. H.; Dahn, J. R. J. Electrochem. Soc. 2002, 149, A815−A822. (4) Lu, Z. H.; Beaulieu, L. Y.; Donaberger, R. A.; Thomas, C. L.; Dahn, J. R. J. Electrochem. Soc. 2002, 149, A778−A791. (5) Kang, S. H.; Pol, V. G.; Belharouak, I.; Thackeray, M. M. J. Electrochem. Soc. 2010, 157, A267−A271. (6) Zonghai, C.; Yan, Q.; Amine, K.; Sun, Y. K. J. Mater. Chem. 2010, 20, 7606−7612. (7) Johnson, C. S.; Li, N.; Lefief, C.; Vaughey, J. T.; Thackeray, M. M. Chem. Mater. 2008, 20, 6095−6106. (8) Sun, Y.-K.; Chen, Z.; Noh, H.-J.; Lee, D.-J.; Jung, H.-G.; Ren, Y.; Wang, S.; Yoon, C. S.; Myung, S.-T.; Amine, K. Nat. Mater. 2012, 11, 942−947. (9) Sun, Y.-K.; Myung, S.-T.; Park, B.-C.; Prakash, J.; Belharouak, I.; Amine, K. Nat. Mater. 2009, 8, 320−324. (10) Croy, J. R.; Gallagher, K. G.; Balasubramanian, M.; Chen, Z.; Ren, Y.; Kim, D.; Kang, S.-H.; Dees, D. W.; Thackeray, M. M. J. Phys. Chem. C 2013, 117, 6525−6536. (11) Bettge, M.; Li, Y.; Sankaran, B.; Rago, N. D.; Spila, T.; Haasch, R. T.; Petrov, I.; Abraham, D. P. J. Power Sources 2013, 233, 346−357. (12) Li, Y.; Bettge, M.; Polzin, B.; Zhu, Y.; Balasubramanian, M.; Abraham, D. P. J. Electrochem. Soc. 2013, 160, A3006−A3019. G
dx.doi.org/10.1021/nl5022859 | Nano Lett. XXXX, XXX, XXX−XXX
Nano Letters
Letter
(13) Mohanty, D.; Sefat, A. S.; Kalnaus, S.; Li, J.; Meisner, R. A.; Payzant, E. A.; Abraham, D. P.; Wood, D. L.; Daniel, C. J. Mater. Chem. A 2013, 1, 6249−6261. (14) Miller, D. J.; Proff, C.; Wen, J. G.; Abraham, D. P.; Bareno, J. Adv. Ener. Mater. 2013, 3, 1098. (15) Noh, H.-J.; Chen, Z.; Yoon, C.-S.; Lu, J.; Amine, K.; Sun, Y.-K. Chem. Mater. 2013, 25, 2109−2115.
H
dx.doi.org/10.1021/nl5022859 | Nano Lett. XXXX, XXX, XXX−XXX
