Letter pubs.acs.org/macroletters
Electrochemical Effects in Thermoelectric Polymers William B. Chang,† Haiyu Fang,‡ Jun Liu,§ Christopher M. Evans,‡ Boris Russ,∥ Bhooshan C. Popere,‡ Shrayesh N. Patel,⊥ Michael L. Chabinyc,⊥ and Rachel A. Segalman*,‡ †
Department of Materials Science and Engineering, University of California, Berkeley, California 94720, United States Departments of Chemical Engineering and Materials and ⊥Department of Materials, University of California, Santa Barbara, California 93106, United States § Department of Mechanical and Aerospace Engineering, North Carolina State University, Raleigh, North Carolina 27695, United States ∥ Molecular Foundry, Lawrence Berkeley National Lab, Berkeley, California 94720, United States ‡
S Supporting Information *
ABSTRACT: Conductive polymers such as PEDOT:PSS hold great promise as flexible thermoelectric devices. The thermoelectric power factor of PEDOT:PSS is small relative to inorganic materials because the Seebeck coefficient is small. Ion conducting materials have previously been demonstrated to have very large Seebeck coefficients, and a major advantage of polymers over inorganics is the high room temperature ionic conductivity. Notably, PEDOT:PSS demonstrates a significant but short-term increase in Seebeck coefficient which is attributed to a large ionic Seebeck contribution. By controlling whether electrochemistry occurs at the PEDOT:PSS/electrode interface, the duration of the ionic Seebeck enhancement can be controlled, and a material can be designed with long-lived ionic Seebeck enhancements.
T
disorder.11−15 Polymers based on ethylenedioxythiophene, known as PEDOT, have been shown to have the highest thermoelectric performance.3,4,16−19 PEDOT:PSS is a phase separated blend of an electrically conductive conjugated polymer PEDOT and the ion conducting polymer polystyrenesulfonic acid (H-PSS). H-PSS is hygroscopic leading to efficient ion conduction at room temperature under high relative humidity conditions due to water uptake.20−23 The movement of ions under a temperature gradient in the mixed conductor PEDOT:PSS has been shown to result in a high ionic contribution to the overall Seebeck coefficient in addition to the electronic component at the initial application of the thermal gradient, as shown by Wang et al. However, the ionic Seebeck effect is transient and decays rapidly with time, enhancing the thermoelectric power factor for approximately 100 s.24 Fundamental understanding of the kinetics of the transient thermoelectric voltage in mixed conductors has been developed in inorganic systems, but not in polymer systems.25 The role of electrochemistry in determining the duration of the ionic Seebeck effect is investigated in mixed ion-electron conducting polymer blends. PEDOT:PSS is used as a model mixed conductor and is modified to include ions that can participate in electrochemical reactions at the electrodes. To
hermoelectrics are of interest as a solid-state technology to harvest useable electrical power from waste heat.1 Conventional thermoelectrics are typically inorganic materials such as Bi2Te3, and research aims for waste heat recovery at high temperature.2 However, many near room temperature applications, such as flexible thermoelectric fabrics for cooling, are enabled by the processability and flexibility intrinsic to polymers.3 Compared to inorganic materials, polymers typically have lower reported thermoelectric performance, as characterized by the figure of merit ZT defined in eq 1.
ZT =
S2σ T κ
(1)
Current conducting polymers based on thiophene derivatives have an electrical conductivity (σ) that exceeds 1000 S/cm with Seebeck coefficients (S) near 20 μV/K,3,4 and a reported inplane thermal conductivity (κ) of 1 W/(m K) and thru-plane thermal conductivity of 0.2 W/(m K).5 Methods to control the electrical conductivity in conjugated polymers are well-known, but strategies to maximize the Seebeck coefficient without compromising electrical conductivity are of great interest, because the Seebeck coefficient and electrical conductivity are normally inversely coupled.1 Improvements to the Seebeck coefficient of inorganic materials are typically achieved through band structure engineering and nanostructuring,6−10 but this strategy is more complex in conjugated polymers due to differences in charge transport mechanism and structural © XXXX American Chemical Society
Received: January 25, 2016 Accepted: March 10, 2016
455
DOI: 10.1021/acsmacrolett.6b00054 ACS Macro Lett. 2016, 5, 455−459
Letter
ACS Macro Letters
PH1000 PEDOT:PSS to form a composition of 33 wt % Ag:PSS to 66 wt % PEDOT:PSS (see SI for details). The PEDOT:Ag:PSS is composed of 2:1 PEDOT:PSS to Ag:PSS, resulting in PEDOT:Ag:PSS being 12 wt % Ag, and is held constant throughout this work. The incorporation of Ag:PSS within PEDOT:PSS results in a blend that exhibits electrical conductivity along the PEDOT-rich domains and either silver ion or proton transport in the percolated PSS-rich regions.31−35 The Ag+ ions also participate in electrochemistry with silver electrodes. The most commonly accepted interpretation of the morphology of PEDOT:PSS is that PEDOT-rich domains within an insulating PSS-rich matrix,31−35 as seen in the morphological schematic Figure 1B. Ionomers such as PSS typically exhibit improved ionic conductivities with increasing water content at higher relative humidity due to water uptake in the ion conducting domains.36 The role of electrochemistry in the ionic thermoelectric response in both PEDOT:PSS and PEDOT:Ag:PSS at high relative humidity can be seen in Figure 2. Figure 2 shows the voltage drop across a two kΩ resistor generated by a 3 μm thick film of PEDOT:PSS (green) under different temperature gradient intervals at a constant RH = 100%. The PEDOT:PSS film resistance is 1 kΩ. The actual vapor density ranges slightly between 18 and 23 g/m3 as the saturation vapor density is temperature dependent. The experiment is divided into five segments of 30 min each. Each elevated temperature segment is returned to ΔT = 0 K in order to probe transient ion effects. The trend in the decay of the thermoelectric voltage with (green) PEDOT:PSS is clear: the thermoelectric voltage and current from the ionic Seebeck effect decays exponentially over the 30 min segments, and the residual voltage and current baseline is solely due to the electronic contribution from PEDOT. The same phenomenon has been observed by Wang et al. in PEDOT:PSS.24 Upon relaxation of the temperature gradient in PEDOT:PSS (ΔT = 0 K), a symmetric voltage and current is produced due to coupled relaxation of both ion and hole concentration gradients. For systems where no proton electrochemistry occurs, such as PEDOT:PSS with gold electrodes, the ionic thermoelectric voltage decays rapidly in an exponential manner,24 and is due to ion motion resulting in ionic Seebeck coefficients much larger than the electronic Seebeck coefficient.37 However, the addition of Ag:PSS and silver electrodes result in a significantly prolonged ionic Seebeck effect when measured in the same geometry and applied temperature intervals (Figure 2, black). The decay in thermovoltage of PEDOT:Ag:PSS cannot be modeled by a simple exponential term as seen with PEDOT:PSS. Rather, a linear decay model is better suited for predicting the voltage decay in PEDOT:Ag:PSS, as seen in SI-6. It is observed that the addition of a species capable of electrochemistry changes the thermoelectric voltage decay from an exponential function to a linear one.The presence of a chemical reaction allows generation and consumption of silver ions, leading to a time-persistent ionic Seebeck enhancement that decays linearly in time. This is because the PEDOT:Ag:PSS system with silver electrodes is a mixed conductor where silver ions enter and exit the polymer conductor through silver electrodes, as shown in Figure 1. The lack of a symmetric peak upon relaxation is indicative of active electrochemistry, as electrochemistry at the polymer−electrode interface prevents buildup of Ag+ ions.
study both conduction processes it is critical to choose appropriate electrodes and ionic species; therefore, silver ions with silver electrodes were selected as an electrochemically active addition to PEDOT:PSS where oxidation and injection at the electrodes can occur, while protons with silver electrodes were selected as the counter system where protons may be reduced to H2, but there is no external source of protons. Importantly, the PSS domains respond to an increase in relative humidity via uptake of water, leading to an increase in ionic conductivity and Seebeck coefficient while allowing for control of the ionic Seebeck effect. Through the application of a transport model, insight into thermoelectric transport in mixed conductors is achieved. Notably, the transient decay in the ionic thermoelectric voltage for ions that do not participate in electrochemistry is better understood. The proposed model incorporates the observed phenomenon of the thermoelectric transport of the holes and ions in systems where there is no chemical reaction at the electrodes and in the situation where ions are injected and extracted from the system. The electrochemically active material with an ion-enhanced Seebeck is no longer a pure thermoelectric; because net transport of silver ions from one electrode to the other is required, when all the silver at one end of the system is consumed, the ionic Seebeck enhancement will ultimately vanish. Any proposed device would require cycling operations for indefinite use. Extensive details on experimental techniques can be found in the Supporting Information. The ionic Seebeck effect in PEDOT:PSS is dependent on both the chemical structure and the morphology of the biphasic material. Figure 1A depicts
Figure 1. Chemical structures of PEDOT:PSS and Ag:PSS are presented in the top right. These materials have the morphology of a mixed conductor, sketched to the left, with Ag+ ions and protons (H+) transported by the PSS rich domains and hole (h+) transport in the PEDOT rich regions. L, the distance between PEDOT rich domains, would increase with humidity due to water swelling the PSS-rich regions. Three separate electrode geometries were used to separate ionic and electronic conductivity as shown in the bottom right: Au, Ag:Nafion/Ag, and H:Nafion/Ag:Nafion/Ag electrodes. Ag:Nafion/ Ag foil contacts were used to measure Ag+ ionic conductivity in PEDOT:Ag:PSS, because the Ag:Nafion membrane blocks electron transport and only the Ag+ ions will react electrochemically with Ag foil.
the chemical structures of the two materials systems investigated, one being the conventional PEDOT:PSS, which has been significantly studied as a thermoelectric previously,4,17,26−30 and the other being PEDOT:PSS with added silver polystyrenesulfonate (Ag:PSS). Ag:PSS was formed through ion exchange and blended with commercial Clevios 456
DOI: 10.1021/acsmacrolett.6b00054 ACS Macro Lett. 2016, 5, 455−459
Letter
ACS Macro Letters
Figure 2. Thermoelectric response to stepwise temperature gradients applied over 30 min intervals to PEDOT:PSS (green) and PEDOT:Ag:PSS (black) 3 μm drop-coated films on glass. Thermoelectric voltage measured across a 2 kΩ resistor. Upon relaxation of the temperature gradient (ΔT = 0 K), a symmetric, opposite voltage is observed as the ion flux is transiently reversed. PEDOT:Ag:PSS demonstrates a lower maximum thermoelectric voltage, but is much more time stable compared to PEDOT:PSS, which quickly decays exponentially. L, the distance between PEDOT rich domains, would increase with humidity due to water swelling the PSS-rich regions.
PEDOT:PSS or PEDOT:Ag:PSS the ratio of ionic conductivity to electrical conductivity is 0.01 and 0.1, respectively, at the highest relative humidity conditions.
The transient thermoelectric voltage due to ion motion in PEDOT:PSS can be modeled using charged particle transport equations. The flux (J) of a charged ion with charge z1q within the temperature gradient is dependent on the local concentration gradient, temperature gradient, and electric field as shown in eq 2. dc dT dV J = z1qD − cqDT −σ dx dx dx
z qD dc z qD dc dV = 1 1 1 + 2 2 2 dx σ1 + σ2 dx σ1 + σ2 dx
(4)
First, ions and holes are both expected to migrate in response to a temperature gradient, generating a voltage gradient. If the ions do not participate in an electrochemical reaction, the ion migration will ultimately result in a large concentration gradient of ions generating an internal electric field. Second, charge carriers will then move under the thermoelectric voltage, with the extent of motion governed by the transference number of each carrier, setting up a concentration gradient to oppose the internal electric field. This will manifest as a peak in the thermoelectric voltage over time, followed by a relaxation process and decay. Third, once the temperature gradient is released, both concentration gradients of ions and holes will return to equilibrium, which will result in a symmetric peak in the thermoelectric voltage. Finally, we expect that in a system with electrodes that allow an electrochemical reaction, there will be a reduced ion concentration gradient due to electrochemistry at the electrodes, resulting in a constant flux of ions from hot to cold. In the second scenario when there is electrochemistry at the electrodes, the ionic Seebeck effect is from two sources: the entropy of the chemical reaction and the thermodiffusion of ions seen in eq 5, which results in ions having much larger Seebeck coefficients than holes and electrons.38 The first term is related to the entropy of the chemical reaction, and only applies to ions undergoing an electrochemical reaction and is on the order of hundreds of microvolts per kelvin. The second term is concentration dependent, and applies to both electrons and ions. Typically, ions have larger thermodiffusion coefficients (DT) than electrons, resulting in ionic Seebeck
(2)
where D is the mass diffusion coefficient; DT is the thermal diffusion coefficient, σ is the electrical conductivity, c is the concentration, q the fundamental unit of charge, and z the charge on the ion. The thermal diffusion coefficient is due to the Soret effect, by which a temperature gradient induces a concentration gradient of molecules. An extension of eq 2 can be directly applied to model a system of two different, equivalently charged particles, as shown in eq 3, where the two carriers experience the same temperature gradient and electric field. dc1 dc dT + z 2qD2 2 − (c1z1qD1, T + c 2z 2qDT ,2) dx dx dx dV − (σ1 + σ2) (3) dx
J = z1qD1
By analyzing this equation in the case where both the temperature gradient (dT/dx) and flux are zero, a relationship can be written between the concentration gradients of the two species and the electric field in the system. The ionic species will be denoted as species 1 and 2 will be the holes in PEDOT:PSS, which are free to enter and leave the system at the electrodes. From kinetic theory, the diffusion coefficient is directly proportional to the conductivity by the relation kT D = B 2 σ . By substituting into eq 4, the species transference nq number is critical in determining which carrier contributes most to flux under a thermoelectric voltage. It is noted that in 457
DOI: 10.1021/acsmacrolett.6b00054 ACS Macro Lett. 2016, 5, 455−459
Letter
ACS Macro Letters coefficients of ∼1 mV/K and electronic Seebeck coefficient ∼30 μV/K.
S=
F ΔS zcDT + nF σT
(5)
In the ideal scenario two, a symmetric spike in the thermoelectric voltage is not observed upon relaxation of the temperature gradient, as there are no ion concentration gradients being formed due to consumption and injection of silver at the electrodes. In order to observe a constant enhanced thermoelectric voltage from the ionic Seebeck effect, it is necessary to design a material system that minimizes the accumulation and depletion of ions due to the thermal gradient, and instead allows a constant flux of ions from hot to cold. This constant flux behavior has also been observed in certain thermogalvanics.39 This is achieved here through the addition of Ag:PSS and silver electrodes. This is seen in Figure 2, where the ionic Seebeck voltage and current no longer displays the exponential behavior or symmetric thermoelectric voltage upon removal of the temperature gradient as observed in PEDOT:PSS. Although PEDOT:Ag:PSS has a lower conductivity than PEDOT:PSS, the Seebeck coefficient over time is much higher, resulting in an stable ion enhanced power factor, higher thermoelectric voltages and larger currents. A comparison of the output power of each respective material is shown in SI-12, where PEDOT:Ag:PSS outperforms PEDOT:PSS. Over the entire experimental time scale of 30 min with Δ = 1 K and a 2 kΩ resistor, PEDOT:PSS outputs 8.6 nJ of electrical energy, while PEDOT:Ag:PSS outputs 20.8 nJ of energy. The transient decay of ion-enhanced polymer thermoelectric can be modeled by the diffusion of the holes caused the thermoelectric voltage built by the ionic Seebeck effect (Vion). This can be modeled to first-approximation by Fick’s second law, and from eq 3, it follows that a concentration gradient of charge carriers is directly related to the thermoelectric voltage. Equation 6 models the experimental transient thermoelectric voltages well, which is derived from Fick’s second law and eq 4. V = Vione−λ ionDiont + Vione λh+Dh+t
Figure 3. (A) Humidity dependent Seebeck coefficient measurements of PEDOT:PSS (green circles) and PEDOT:Ag:PSS (black squares) show an exponential increase in the Seebeck coefficient due to the ion motion when above a threshold relative humidity of 60%. The Seebeck coefficient is calculated via the maximum observed thermoelectric voltage prior to the onset of decay. (B, C) Electrical conductivity (circle) and ionic conductivity (square) of PEDOT:PSS (B, green) and PEDOT:Ag:PSS (C, black) is measured as a function of humidity.
humidity in both PEDOT:PSS and PEDOT:Ag:PSS. The maximum Seebeck coefficient of PEDOT:Ag:PSS is almost an order of magnitude smaller than PEDOT:PSS at the highest relative humidity, but much more stable over time, which is not reflected in Figure 3A. The ionic Seebeck effect can only be observed above a threshold relative humidity of 60%; below this threshold both materials show purely electronic Seebeck coefficients of 18 μV/K for PEDOT:PSS and 8 μV/K for PEDOT:Ag:PSS. At a relative humidity of 100%, the maximum observed Seebeck coefficient at short time for PEDOT:PSS is 2.4 mV/K, and 0.10 mV/K for PEDOT:Ag:PSS. Figure 3B,C is the humidity controlled electronic (circles) and ionic (square) conductivity of PEDOT:PSS (green) and PEDOT:Ag:PSS (black). Increasing the humidity that the mixed conductor is exposed to, the electronic conductivity decreases in both systems, which is attributed to the swelling of the insulating PSS domains in the polymer morphology. This effect is observed to saturate at a relative humidity of 75% for PEDOT:PSS and 82% for PEDOT:Ag:PSS. Overall, increasing the relative humidity causes the electrical conductivity to decrease by a factor of 3 for PEDOT:PSS and PEDOT:Ag:PSS; the final electrical conductivity at RH = 100% is seen to be 0.15 and 0.01 S/cm, respectively. We used time-domain thermoreflectance (TDTR) method to measure the thru-plane thermal conductivity of PEDOT:Ag:PSS thin film, as shown in SI-10. The measured thru-plane thermal conductivity of 0.24 W/(m K) demonstrates that even with the addition of silver ions to the polymer, the thermal conductivity is unchanged from PEDOT:PSS (0.3 W/(m K)).5 We find that the maximum thermoelectric power factor in PEDOT:PSS at RH = 100% to be 80 μW/mK2 using the initial value before the transient decay, and with the assumption that the in-plane and thru-plane thermal conductivity is similar, results in a zT = 0.13. This is very close to the zT observed for optimized PEDOT:PSS with DMSO, which results from much higher electrical conductivity but lower Seebeck coefficients.4 The theoretical limits of mixed conductor thermoelectrics are intrinsically tied to the ratio of the electrical and ionic conductivity, and the lifetime of said device is calculated in the SI.
(6)
λh+ is determined from the boundary conditions and geometry of the film, while Dh+ is the diffusion coefficient of holes in PEDOT. The transient thermoelectric voltage and the corresponding exponential decay coefficient for PEDOT:PSS in Figure 2 will correspond to the diffusion coefficient of holes kT in PEDOT:PSS calculated using the relation D = nqB 2 σ . The diffusion coefficient for holes extracted from Figure 3 is 0.116 cm2/s, and can be converted to an electrical conductivity of 0.296 S/cm, using an effective carrier concentration of 1020 cm−3. This closely matches the experimental conductivity measured in Figure 3, compared to an Ag+ concentration of approximately 0.5e19 cm−3. The calculated ion diffusion coefficient is 0.011 cm2/s. Using a carrier concentration of 1019 cm−3 calculated from the molar density of HPSS, we calculate a proton conductivity of 0.005 S/cm. Measurements of the Seebeck coefficient of PEDOT:PSS and PEDOT:Ag:PSS were then performed at controlled relative humidity conditions to further investigate the mechanism leading to ionic Seebeck coefficients. The results are depicted in Figure 3A, where the Seebeck coefficients of PEDOT:PSS and PEDOT:Ag:PSS are displayed. As the Seebeck coefficient in Figure 3A is plotted on a logarithmic scale, the increase in Seebeck coefficient is exponential with respect to relative
■
ASSOCIATED CONTENT
S Supporting Information *
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acsmacrolett.6b00054. Experimental, measurement, and synthetic details (PDF). 458
DOI: 10.1021/acsmacrolett.6b00054 ACS Macro Lett. 2016, 5, 455−459
Letter
ACS Macro Letters
■
(30) See, K. C.; Feser, J. P.; Chen, C. E.; Majumdar, A.; Urban, J. J.; Segalman, R. A. Nano Lett. 2010, 10, 4664. (31) Nardes, A. M.; Kemerink, M.; Janssen, R. A. J.; Bastiaansen, J. A. M.; Kiggen, N. M. M.; Langeveld, B. M. W.; van Breemen, A. J. J. M.; de Kok, M. M. Adv. Mater. 2007, 19, 1196. (32) Timpanaro, S.; Kemerink, M.; Touwslager, F. J.; De Kok, M. M.; Schrader, S. Chem. Phys. Lett. 2004, 394, 339. (33) Nardes, A. M.; Janssen, R. A. J.; Kemerink, M. Adv. Funct. Mater. 2008, 18, 865. (34) Lang, U.; Müller, E.; Naujoks, N.; Dual, J. Adv. Funct. Mater. 2009, 19, 1215. (35) Wei, Q.; Mukaida, M.; Kirihara, K.; Ishida, T. ACS Macro Lett. 2014, 3, 948. (36) Shetzline, J. A.; Creager, S. E. J. Electrochem. Soc. 2014, 161, H917. (37) Wei, Q.; Mukaida, M.; Kirihara, K.; Naitoh, Y.; Ishida, T. Appl. Phys. Express 2014, 7, 031601. (38) Newman, J. Ind. Eng. Chem. Res. 1995, 34, 3208. (39) Weininger, J. L. J. Electrochem. Soc. 1964, 111, 769.
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected]. Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS
■
REFERENCES
We would like to thank Prof. David Cahill at University of Illinois Urbana−Champaign for helpful discussion and TDTR resources. This work was supported by AFOSR MURI FA955012-1-0002. CME and BCP were supported by the MRSEC Program of the National Science Foundation under Award No. DMR 1121053.
(1) Snyder, G. J.; Toberer, E. S. Nat. Mater. 2008, 7, 105. (2) Riffat, S.; Ma, X. Appl. Therm. Eng. 2003, 23, 913. (3) Zhang, Q.; Sun, Y.; Xu, W.; Zhu, D. Adv. Mater. 2014, 26, 6829. (4) Kim, G.-H.; Shao, L.; Zhang, K.; Pipe, K. P. Nat. Mater. 2013, 12, 719. (5) Liu, J.; Wang, X.; Li, D.; Coates, N. E.; Segalman, R. A.; Cahill, D. G. Macromolecules 2015, 48, 585. (6) Heremans, J. P.; Dresselhaus, M. S.; Bell, L. E.; Morelli, D. T. Nat. Nanotechnol. 2013, 8, 471. (7) Lan, Y.; Poudel, B.; Ma, Y.; Wang, D.; Dresselhaus, M. S.; Chen, G.; Ren, Z. Nano Lett. 2009, 9, 1419. (8) Poudel, B.; Hao, Q.; Ma, Y.; Lan, Y.; Minnich, A.; Yu, B.; Yan, X.; Wang, D.; Muto, A.; Vashaee, D.; Chen, X.; Liu, J.; Dresselhaus, M. S.; Chen, G.; Ren, Z. Science 2008, 320, 634. (9) Wang, H.; Pei, Y.; Lalonde, A. D.; Snyder, G. J. Thermoelectric Nanomaterials; Springer: Berlin, Heidelberg, 2013. (10) Zhang, Y.; Bahk, J.-H.; Lee, J.; Birkel, C. S.; Snedaker, M. L.; Liu, D.; Zeng, H.; Moskovits, M.; Shakouri, A.; Stucky, G. D. Adv. Mater. 2014, 26, 2755. (11) Okamoto, H.; Toyama, T.; Hattori, K. J. Non-Cryst. Solids 2004, 338−340, 341. (12) Novikov, S. V. arXiv 2013, DOI: 10.1134/S1023193512030081. (13) Mott, N. F. Philos. Mag. 1969, 19, 835. (14) Li, L.; Meller, G.; Kosina, H. Microelectron. J. 2007, 38, 47. (15) Mehraeen, S.; Coropceanu, V.; Brédas, J.-L. Phys. Rev. B: Condens. Matter Mater. Phys. 2013, 87, 195209. (16) Culebras, M.; Gómez, C. M.; Cantarero, a. J. Mater. Chem. A 2014, 2, 10109. (17) Bubnova, O.; Khan, Z. U.; Malti, A.; Braun, S.; Fahlman, M.; Berggren, M.; Crispin, X. Nat. Mater. 2011, 10, 429. (18) Kim, D.; Kim, Y.; Choi, K.; Grunlan, J. C.; Yu, C. ACS Nano 2010, 4, 513. (19) Chang, K.-C.; Jeng, M.-S.; Yang, C.-C.; Chou, Y.-W.; Wu, S.-K.; Thomas, M. A.; Peng, Y.-C. J. Electron. Mater. 2009, 38, 1182. (20) Rieke, P. C.; Vanderborgh, N. E. J. Membr. Sci. 1987, 32, 313. (21) Sone, Y.; Ekdunge, P.; Simonsson, D. J. Electrochem. Soc. 1996, 143, 1254. (22) Merle, G.; Wessling, M.; Nijmeijer, K. J. Membr. Sci. 2011, 377, 1. (23) Stavrinidou, E.; Leleux, P.; Rajaona, H.; Khodagholy, D.; Rivnay, J.; Lindau, M.; Sanaur, S.; Malliaras, G. G. Adv. Mater. 2013, 25, 4488. (24) Wang, H.; Ail, U.; Gabrielsson, R.; Berggren, M.; Crispin, X. Adv. Energy Mater. 2015, 5, n/a. (25) Janek, J.; Korte, C. Solid State Ionics 1996, 92, 193. (26) Bubnova, O.; Berggren, M.; Crispin, X. J. Am. Chem. Soc. 2012, 134, 16456. (27) Bubnova, O.; Crispin, X. Energy Environ. Sci. 2012, 5, 9345. (28) Park, T.; Park, C.; Kim, B.; Shin, H.; Kim, E. Energy Environ. Sci. 2013, 6, 788. (29) Yu, C.; Choi, K.; Yin, L.; Grunlan, J. C. ACS Nano 2011, 5, 7885. 459
DOI: 10.1021/acsmacrolett.6b00054 ACS Macro Lett. 2016, 5, 455−459
