Engineering the Transformation Strain in LiMnyFe1–yPO4 Olivines for

DOI: 10.1021/acs.nanolett.5b05146. Publication Date (Web): March 1, 2016. Copyright © 2016 American Chemical Society. *(D.B.R.): E-mail: [email protected]...
0 downloads 13 Views 3MB Size
Letter pubs.acs.org/NanoLett

Engineering the Transformation Strain in LiMnyFe1−yPO4 Olivines for Ultrahigh Rate Battery Cathodes Dorthe B. Ravnsbæk,*,†,‡ Kai Xiang,† Wenting Xing,† Olaf J. Borkiewicz,§ Kamila M. Wiaderek,§ Paul Gionet,∥ Karena W. Chapman,§ Peter J. Chupas,§ Ming Tang,⊥ and Yet-Ming Chiang*,† †

Department of Material Science and Engineering, Massachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge, Massachusetts 02139, United States ‡ Department of Physics, Chemistry, and Pharmacy, University of Southern Denmark, Campusvej 55, 5230, Odense M, Denmark § X-ray Science Division, Advanced Photon Source, Argonne National Laboratory, 9700 S. Cass Avenue, Argonne, Illinois 60439, United States ∥ A123-Systems, 200 West Street, Waltham, Massachusetts 02451, United States ⊥ Department of Materials Science and NanoEngineering (MSNE), Rice University, 6100 Main MS-325, Houston, Texas 77005-1827, United States S Supporting Information *

ABSTRACT: Alkali ion intercalation compounds used as battery electrodes often exhibit first-order phase transitions during electrochemical cycling, accompanied by significant transformation strains. Despite ∼30 years of research into the behavior of such compounds, the relationship between transformation strain and electrode performance, especially the rate at which working ions (e.g., Li) can be intercalated and deintercalated, is still absent. In this work, we use the LiMnyFe1−yPO4 system for a systematic study, and measure using operando synchrotron radiation powder X-ray diffraction (SRPXD) the dynamic strain behavior as a function of the Mn content (y) in powders of ∼50 nm average diameter. The dynamically produced strain deviates significantly from what is expected from the equilibrium phase diagrams and demonstrates metastability but nonetheless spans a wide range from 0 to 8 vol % with y. For the first time, we show that the discharge capacity at high C-rates (20−50C rate) varies in inverse proportion to the transformation strain, implying that engineering electrode materials for reduced strain can be used to maximize the power capability of batteries. KEYWORDS: Li-ion batteries, cathode, rate capability, misfit strain, lithium manganese iron phosphate, operando, X-ray diffraction, phase transformation

O

livines such as LiFePO4 (LFP)1 have become commercially important as cathode materials for lithium ion batteries due to their high safety, long cycle life, low cost, and above all for their outstanding rate capability when prepared in nanoscale form.2,3 Numerous studies have focused on explaining the charge−discharge mechanism in olivine cathodes, and several phase transformation models have been proposed.4−26 From these investigations, it becomes evident that at or close to thermodynamic equilibrium, the storage/release of lithium ions in LFP is accompanied by a first-order phase transition between lithiated and delithiated states.4,5 The large Li-miscibility gap results in a large lattice misfit of 6.5 vol % between equilibrium coexisting phases in the LFP/FP system6,7 and even larger misfit of 11.7% for its Mn counterpart, LMP/MP.27 The use of nanoscale particles has become ubiquitous in olivine cathodes and crystallite size reduction as well as cation doping has been shown to decrease the width of the miscibility gap.2,6−11 In addition, several theoretical studies12−14 focusing on phase transformation mechanisms in LFP under nonequilibrium © 2016 American Chemical Society

conditions have predicted that Li extraction and insertion can bypass the first-order phase transformation altogether, instead forming metastable solid solutions, LxFP (0 < x < 1). It has been suggested that high current rates and high overpotential can promote this behavior.13,14 Under static conditions, the metastable LxFP phase is expected to relax to LFP and FP.12 There has been a few reports of experimental evidence for formation of nonequilibrium LxFP phases,24,25 most recently, Liu et al.26 observed nonequilibrium solid solutions in 180 nm particle size LiFePO4 during cycling at high current rates (≥10C) by operando powder X-ray diffraction (PXD). It was observed that the compositional variations only arise during fast Li extraction and insertion but also that some two-phase LFP/ FP coexistence always persists; a single nonequilibrium LxFP phase was never observed. Received: December 16, 2015 Revised: February 18, 2016 Published: March 1, 2016 2375

DOI: 10.1021/acs.nanolett.5b05146 Nano Lett. 2016, 16, 2375−2380

Letter

Nano Letters

Figure 1. Phase transition behavior. Lower angular region of the operando synchrotron radiation powder X-ray diffraction data measured during charge and discharge at C/10-rate and corresponding unit cell volumes as a function of overall Li-content in the cathodes of LiMnyFe1−yPO4 of 50 nm particles with (A) y = 0.0 (i.e., LFP), (B) y = 0.1, (C) y = 0.2, (D) y = 0.4, (E) y = 0.6 and (F) y = 0.8. The minor gaps in the PXD data in B, C, and D are due to a short interruption in the synchrotron beam.

The evidence for metastable solid solutions leaves open the question whether high rate requires low transformation strain, because coherency strain energy exists even between solid solution compositions of differing molar volume. Because atom mobility along lattice discontinuities comprising an incoherent interface (e.g., dislocations, stacking faults) can be more rapid

than within a perfect lattice, without detailed knowledge of all transport coefficients it cannot be known a priori which configuration has faster phase propagation. Meethong et al. previously suggested that reduction of transformation strain should be a selection criterion for high rate olivines, but their proposal has not to date been systematically tested.6 2376

DOI: 10.1021/acs.nanolett.5b05146 Nano Lett. 2016, 16, 2375−2380

Letter

Nano Letters

Figure 2. TEM and discharge voltage profiles. (A) Typical TEM image of the nanoscale olivine cathode powder. Here shown for composition y = 0.4. (B−F) Voltage versus specific capacity curves measured at different galvanostatic C-rates for the series of LiMnyFe1−yPO4 (LMFP, y = 0.1−0.8) samples. All samples have an average particle size of ∼50 nm.

Here, we first show that the dynamic transformation strain in the LiMnyFe1−yPO4 system is a strong function of y, and second we use this relationship to test the correlation between transformation strain and electrochemical rate-capability. A family of identically prepared compositions varying in y (0.0, 0.1, 0.2, 0.4, 0.6, and 0.8) and having nearly identical crystallite size (50 nm) is tested in identical electrode and cell configuration for rate capability at up to 50C rate (i.e., a full discharge half-cycle in 1/50th of an hour). Operando transformation strain is measured using synchrotron radiation powder X-ray diffraction (SR-PXD). We find a close correlation between measured transformation strain and the capacity realized at high C-rates from 20 to 50C, demonstrating for the first time the selection criterion of minimizing transformation strain for maximizing power. In addition, we find that the phase evolution sequence for all high rate compositions (0 < y < 0.8) deviates significantly from what is expected from the equilibrium phase diagram. Finally, we show that the transformation sequence differs significantly between charge and discharge, constituting further evidence of metastability. The tunable misfit in LiMnyFe1−yPO4 is a result of mixed Mn/Fe occupancy producing two voltage steps, corresponding to the Fe2+/Fe3+ and Mn2+/Mn3+ redox couples respectively, allowing the full misfit between the fully lithiated and fully delithiated endmembers, LiMny Fe 1−y PO 4 (LMFP) and MnyFe1−yPO4 (MFP), to be separated into two smaller strain steps27 due to the existence of an intermediate crystalline phase LixMnyFe1−yPO4 (x < 1), here denoted LxMFP. The two redox reactions are

Each step entails structural changes to the olivine crystal lattice mainly due to the change in ionic radius accompanying oxidation and reduction between r(Fe3+) = 69 pm and r(Fe2+) = 75 pm as well as r(Mn3+) = 72 pm and r(Mn2+) = 81 pm, respectively.29 According to published phase diagrams there is very limited solubility of the endmember compounds but a fairly wide solution regime for the intermediate phase.30,31 As we show, under dynamic cycling the phase sequence deviates far from these expectations. The operando SR-PXD data were measured using the AMPIX battery test cell32 at beamline 11-ID-B at the Advanced Photon Source, Argonne National Laboratory using an X-ray wavelength of 0.2114 Å. Data were collected during charge and discharge of six LiMnyFe1−yPO4 compositions, y = 0.0, 0.1, 0.2, 0.4, 0.6, and 0.8, at C/10 rate. The unit cell volumes were extracted by Rietveld refinement as a function of the net Licontent in the electrodes, determined galvanostatically, and are shown in Figure 1A−F. Full experimental details are provided in Supporting Information. The results for y = 0.4 have been previously reported28 but are included here for completeness. Galvanostatic discharge curves for each of the Mn-containing samples measured at C-rates from C/5 to 50C are shown in Figure 2. Consider first the LMFP → LxMFP transition occurring on the Fe-plateau during charging (left side of each panel in Figure 1). For the materials with Mn-content 0.0 < y ≤ 0.4, Figure 1A−D, we see that as Li-ions are being extracted the intermediate LxMFP phase forms via a first order phase transformation on the Fe2+/Fe3+ plateau. On the basis of the Vegard relation between LMFP and MFP endmembers, the Licontent in the intermediate phase upon formation can be determined and is x = 0.67, 0.72, and 0.78 for samples of y = 0.1, 0.2, and 0.4, respectively. With increasing Mn-content (Figure 1A−D), the misfit between LMFP and LxMFP systematically decreases, and for the two samples of highest Mn-content, y = 0.6 and 0.8, the LMFP → LxMFP transition occurs via a complete solid solution reaction with zero misfit.

Mn 3 + y Fe3 +1 − y PO4 + y Li+ + ye− ↔ Li yMn 2 + y Fe3 +1 − y PO4 (1)

Li yMn 2 + y Fe3 +1 − y PO4 + (1 − y)Li+ + (1 − y)e− ↔ LiMn 2 + y Fe2 +1 − y PO4

(2) 2377

DOI: 10.1021/acs.nanolett.5b05146 Nano Lett. 2016, 16, 2375−2380

Letter

Nano Letters

Figure 3. (A) Correlation between lattice misfit and discharge power performance for nanoparticulate olivine LiMnyFe1‑yPO4. Left axis: Volume misfit during discharge as a function of Mn-content, y in LiMnyFe1−yPO4 for the phase transformations at the Fe-plateau, that is from LMFP to LxMFP (black lines and squares) and at the Mn-plateau (blue lines and circles). Note that a volume misfit of 0% corresponds to solid solution reactions. Right axis: Discharge capacity (blue shaded curves) measured at current rates of C/5, 20C, 35C, and 50C as a function of Mn-content, y in LiMnyFe1−yPO4. (B) Volume misfit during charge and discharge as a function of Mn-content. Greater hysteresis in the misfit between charge and discharge occurs for the Fe-plateau and increases with Mn content.

(equivalent to slightly different Li-content, x) to the data. Hence, the broadening cannot be subscribed to nanoscale twophase coexistence. In contrast, the approach of fitting two phases was successful for the material of y = 0.1, showing that LMFP nucleates with a unit cell volume close to that of LxMFP. In this region, the Li-content within the intermediate phase reaches a plateau and covers a relatively narrow compositional range of xLi = 0.50−0.58. Hence, for this sample it seems that the structural rearrangement occurring during Fe-reduction is too large to accommodate a solid solution transformation. Galvanostatic discharge curves at C/5 rate to 50C rate are shown in Figure 2B−F for each of the Mn-bearing compositions along with a TEM image of a typical powder microstructure (Figure 2A). (The current at a given C-rate is defined with respect to the theoretical charge capacity of the compound.) Three cells of each composition were tested under identical conditions and found to exhibit closely agreeing data. The relative capacities on the Mn2+/Mn3+ and Fe2+/Fe3+ plateaus correspond closely to the Mn/Fe ratio of the sample. In Figure 2A, the average capacity of three cells are plotted against C-rate for each composition. At low C-rates, for example, 1C and below, we observed no great sensitivity of discharge capacity to C-rate for Mn contents 0.0 ≤ y ≤ 0.4, whereas for y = 0.6 and 0.8, the total capacity is lower even at C/5. However, at 20C rate and above the samples are clearly differentiated. In Figure 3A, contours showing the capacity realized at C/5, 20C, 35C and 50C rate are plotted for each of the compositions, along with the misfit strain observed upon discharge. Note that the strain on the Fe and Mn plateaus are plotted separately and that as one diminishes the other increases. The correspondence between misfit strain and capacity at high C-rate is clearly seen. There is a broad maximum in capacity for y = 0.1, 0.2, and 0.4, coinciding with the composition regime where the smallest misfit strains are observed upon discharge. The y = 0.2 composition is unique in having near-zero strain for both the Fe and Mn transitions; we hypothesize that in composite electrodes formulated to allow still higher C-rates (e.g., with highly dilute loadings and excess carbon), still finer differentiation of the rate capability may be possible. For example, the zero-transformation-strain composi-

The corresponding voltage−capacity curves (Figure 2) have initially a clear voltage plateau at 3.45 V that is characteristic of the Fe2+/Fe3+ couple (Figure 2B−D), but at the highest Mn concentrations (Figure 2E,F), the Fe “plateau” is no longer flat but is sloped, consistent with a solid solution. As Li is further extracted during charging, we notice that the compositions with the largest strain on the Fe2+/Fe3+ plateau have the smallest misfit on the Mn2+/Mn3+ plateau and vice versa. For samples of y = 0.1 and 0.2, Figure 1B,C, the transition from LxMFP to MFP follows essentially solid solution behavior; no distinct two-phase coexistence is detected. In contrast, in Figure 1D−F, this misfit at high charge state (e.g., on the Mn2+/Mn3+ plateau) increases as y increases but at the same time the misfit for LMFP to LxMFP decreases and even vanishes in Figure 1E,F. Assuming that a high misfit strain on either of the Fe or Mn plateaus can inhibit lithiation/delithiation, these data suggest that an intermediate Mn content is desirable for the fastest kinetics. Upon discharge (right side of each panel in Figure 1), we see that no composition simply reverses its behavior during charge; there is always some hysteresis. From this alone, one can conclude that metastability is present, because both sequences cannot represent a single equilibrium. The behavior of the pure LFP and y = 0.8 LMFP (Figure 1A,F) are the closest to symmetric, each exhibiting two-phase coexistence with a large misfit strain between the phases. At intermediate Mn content, 0.1 ≤ y ≤ 0.4, the misfit strain on discharge is reduced, and for y = 0.2 a continuous solid solution is reached. These compositions make for the most interesting comparisons with electrochemical performance. In Figure 1, the horizontal error bars visible in panels C−E are significant because they indicate that there is a distribution of Li-compositions present in LxMFP. The magnitude of the compositional variation is extracted from peak broadening observed during the phase transitions or solid solution reactions. There may be contributions to peak broadening from disorder/defects/crystallites; however, we assume that it completely originates from compositional variation so as not to underestimate this contribution. At no point could satisfactory description of the data via Rietveld refinement be achieved by fitting two phases of slightly different cell parameters 2378

DOI: 10.1021/acs.nanolett.5b05146 Nano Lett. 2016, 16, 2375−2380

Letter

Nano Letters tions may be able to deliver still greater capacity at C-rates beyond 50C. We conclude that a strong correlation between transformation strain and the discharge power performance of the material exists as shown in Figure 3A. We assume that a metastable solid solution is possible under high driving force in most of these compositions, given recent results showing such behavior at 10C rate for pure LFP of significantly larger particle size (180 nm).26 The sensitivity shown here suggests that even metastable transformation pathways benefit from reduced transformation strain. The rate controlling mechanism(s) at atomic scale are not yet clear, however. One may consider whether the minimum in transformation strain also coincides with a maximum in electronic or ionic transport. Obtaining such data in a dynamically formed metastable solid solution is difficult although measurements for quenched disordered LiFePO4 have been reported.33 For LiMnyFe1−yPO4, the limited data that exists34,35 suggests that electronic conductivity is higher than ionic conductivity for intermediate y values, presumably measured in samples in a phase-equilibrated state. The misfit strains upon charge, Figure 3B, show a similar trend with Mn-composition, y, in that the smallest misfits are observed for y = 0.2 and 0.4. However, note that the charge misfit strains for the Fe-plateau are larger than those upon discharge, especially in the regime where solid solution behavior occurs for this plateau (y ≥ 0.2). On the other hand, there is little hysteresis for the Mn plateau and for pure LFP. The origin of the hysteresis on the Fe plateau is currently unclear and warrants further study, as does the correlation of this behavior with charge rate capability in these same compositions. In order to observe the behavior after the initial cycle and at higher rates, operando SR-PXD data were also collected over two consecutive charge−discharge cycles at 1C-rate. This was conducted only for samples of y ≤ 0.6 because the y = 0.8 sample exhibits too low a capacity at 1C rate to permit the phase transformation at the Mn-plateau to occur to an extent observable by SR-PXD. Results for the material with y = 0.2, shown in Figure 4, are representative of the other material compositions, all shown in Supporting Information. We found that at 1C rate all materials show similar phase evolution behavior as under C/10 conditions, that is, extended solid solution formation and highly hysteretic behavior is retained at the higher rate. This can be seen upon comparing Figure 4A (1C) with Figure 1C (C/10) for the y = 0.2 sample. Between the first and second cycle, the main difference observed is that the two-phase LMFP → LxMFP transformation becomes somewhat more continuous during the second charge, suggesting that the hysteresis will diminish, and solid solution behavior will be more pronounced with repeated cycling. However, the transformation sequence remains qualitatively the same as at the lower rate. Thus, within a series of nanoscale LiMnyFe1−yPO4 olivines carefully prepared and tested to exclude extrinsic variables we observe a direct correlation between measured transformation strain and the cathode discharge capacity at high C-rates. Note that the measured differences in capacity between different compositions are significant and meaningful for real-world applications. For example at y = 0.2 where the misfit strain is zero, 50% of the theoretical capacity (170 mAh/g) is obtained at 20C, compared to the pure LFP, which with a misfit of 5.4 vol % only delivers 21% of the theoretical capacity at 20C. For high power applications such as hybrid electric vehicles and

Figure 4. Repeated charge−discharge cycling at high C-rate. (A) Operando SR-PXD for LiMn0.2Fe0.8PO4 during two consecutive charge−discharge cycles at 1C rate. Enlargement of the (211)(020)peak is shown for the 1st charge (B) and discharge (C) as well as for the 2nd charge (D) and discharge (E).

start−stop (microhybrid) applications, these differences are meaningful and are why LMFP is now considered a second generation olivine cathode to succeed LFP.



ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.nanolett.5b05146. Experimental details, discharge rate capability, Charge, misfit strains, and operando PXD data of 1C-rate charge and discharge. (PDF)



AUTHOR INFORMATION

Corresponding Authors

*(D.B.R.): E-mail: [email protected]. Phone: 0045 6550 7635. Fax: 0045 6615 8780. *(Y.-M.C.): E-mail: [email protected]. Phone: (617)253-6471. Fax: (617)253-6201. Author Contributions

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This work was supported by DOE Project Number DESC0002626. Use of the Advanced Photon Source, an Office of 2379

DOI: 10.1021/acs.nanolett.5b05146 Nano Lett. 2016, 16, 2375−2380

Letter

Nano Letters

(26) Liu, H.; Strobridge, F. C.; Borkiewicz, O. J.; Wiaderek, K. M.; Chapman, K. W.; Chupas, P. J.; Grey, C. P. Science 2014, 344, 1252817. (27) Li, G.; Azuma, H.; Tohda, M. Electrochem. Solid-State Lett. 2002, 5, A135−A137. (28) Ravnsbæk, D. B.; Xiang, K.; Xing, W.; Borkiewicz, O. J.; Wiaderek; Gionet, P.; Chapman, K. W.; Chupas, P. J.; Chiang, Y.-M. Nano Lett. 2014, 14, 1484−1491. (29) Shannon, R. Acta Crystallogr., Sect. A: Cryst. Phys., Diffr., Theor. Gen. Crystallogr. 1976, 32, 751−767. (30) Yamada, A.; Kudo, Y.; Liu, Y. Y. J. Electrochem. Soc. 2001, 148, A1153−A1158. (31) Malik, R.; Zhou, F.; Ceder, G. Phys. Rev. B: Condens. Matter Mater. Phys. 2009, 79, 214201. (32) Borkiewicz, O. J.; Shyam, B.; Wiaderek, K. M.; Kurtz, C.; Chupas, P. J.; Chapman, K. W. J. Appl. Crystallogr. 2012, 45, 1261− 1269. (33) Tan, H. J.; Dodd, J. L.; Fultz, B. J. J. Phys. Chem. C 2009, 113, 2526−2531. ́ (34) Molenda, J.; Ojczyk, W.; Swierczek, K.; Zając, W.; Krok, F.; Dygas, F.; Liu, R.-S. Solid State Ionics 2006, 177, 2617−2624. (35) Molenda, J.; Ojczyk, W.; Marzec, J. J. J. Power Sources 2007, 174, 689−694.

Science User Facility operated for the U.S. Department of Energy (DOE) Office of Science by Argonne National Laboratory, was supported by the U.S. DOE under Contract No. DE-AC02-06CH11357. D.B.R. acknowledges the Carlsberg Foundation and the Villum Foundation for funding. M.T. acknowledges support from DOE project number BESC0014435.



ABBREVIATIONS SR-PXD, synchrotron radiation powder X-ray diffraction; LFP, LiFePO4; FP, FePO4; LMP, LiMnPO4; MP, MnPO4; LMFP, LiMnyFe1−yPO4; MFP, MnyFe1−yPO4



REFERENCES

(1) Padhi, A. K.; Nanjundaswamy, K. S.; Goodenough, J. B. J. J. Electrochem. Soc. 1997, 144, 1188−1194. (2) Chung, S.-Y.; Bloking, J. T.; Chiang, Y.-M. Nat. Mater. 2002, 1, 123−128. (3) Delacourt, C.; Poizot, P.; Levasseur, S.; Masquelier, C. Electrochem. Solid-State Lett. 2006, 9, A352−A355. (4) Dodd, J. L.; Yazami, R.; Fultz, B. Electrochem. Solid-State Lett. 2006, 9, A151−A155. (5) Delmas, C.; Maccario, M.; Croguennec, L.; Le Cras, F.; Weill, F. Nat. Mater. 2008, 7, 665−671. (6) Meethong, N.; Huang, H.-Y. S.; Speakman, S. A.; Carter, W. C.; Chiang, Y.-M. Adv. Funct. Mater. 2007, 17, 1115−1123. (7) Meethong, N.; Kao, Y.-H.; Tang, M.; Huang, H.-Y.; Carter, W. C.; Chiang, Y.-M. Chem. Mater. 2008, 20, 6189−6198. (8) Meethong, N.; Huang, H.-Y. S.; Carter, W. C.; Chiang, Y.-M. Electrochem. Solid-State Lett. 2007, 10, A134−A138. (9) Meethong, N.; Kao, Y.-H.; Speakman, S. A.; Chiang, Y.-M. Adv. Funct. Mater. 2009, 19, 1060−1070. (10) Omenya, F.; Chernova, N. A.; Wang, Q.; Zhang, R.; Whittingham, M. S. Chem. Mater. 2013, 25, 2691−2699. (11) Omenya, F.; Chernova, N. A.; Zhang, R.; Fang, J.; Huang, Y.; Cohen, F.; Dobrzynski, N.; Senanayake, S.; Xu, W.; Whittingham, M. S. Chem. Mater. 2013, 25, 85−89. (12) Malik, R.; Ceder, G. Nat. Mater. 2011, 10, 587−590. (13) Bai, P.; Cogswell, D. A.; Bazant, M. Z. Nano Lett. 2011, 11, 4890−4896. (14) Orvanos, B. Electrochim. Acta 2014. (15) Singh, G. K.; Ceder, G.; Bazant, M. Z. Electrochim. Acta 2008, 53, 7599−7613. (16) Burch, D.; Bazant, M. Z. Nano Lett. 2009, 9, 3795−3800. (17) Tang, M.; Carter, W. C.; Chiang, Y.-M. Annu. Rev. Mater. Res. 2010, 40, 501−529. (18) Tang, M.; Carter, W. C.; Belak, J. F.; Chiang, Y.-M. Electrochim. Acta 2010, 56, 969−976. (19) Kao, Y.-H.; Tang, M.; Meethong, N.; Bai, J.; Carter, W. C.; Chiang, Y.-M. Chem. Mater. 2010, 22, 5845−5855. (20) Sharma, N.; Guo, X.; Du, G.; Guo, Z.; Wang, J.; Wang, Z.; Peterson, V. K. J. Am. Chem. Soc. 2012, 134, 7867−7873. (21) Bazant, M. Z. Acc. Chem. Res. 2013, 46, 1144−1160. (22) Orikasa, Y.; Maeda, T.; Koyama, Y.; Minato, T.; Murayama, H.; Fukuda, K.; Tanida, H.; Arai, H.; Matsubara, E.; Uchimoto, Y.; Ogumib, Z. J. J. Electrochem. Soc. 2013, 160, A3061−A3065. (23) Chueh, W. C.; Gabaly, F. E.; Sugar, J. D.; Bartelt, N. C.; McDaniel, A. H.; Fenton, K. R.; Zavadil, K. R.; Tyliszczak, T.; Lai, W.; McCarty, K. F. Nano Lett. 2013, 13, 866−872. (24) Orikasa, Y.; Maeda, T.; Koyama, Y.; Murayama, H.; Fukuda, K.; Tanida, H.; Arai, H.; Matsubara, E.; Uchimoto, Y.; Ogumi, Z. Chem. Mater. 2013, 25, 1032−1039. (25) Orikasa, Y.; Maeda, T.; Koyama, Y.; Murayama, H.; Fukuda, K.; Tanida, H.; Arai, H.; Matsubara, E.; Uchimoto, Y.; Ogumi, Z. J. Am. Chem. Soc. 2013, 135, 5497−5500. 2380

DOI: 10.1021/acs.nanolett.5b05146 Nano Lett. 2016, 16, 2375−2380