In Situ TEM Analysis of Organic–Inorganic Metal-Halide Perovskite

Letter pubs.acs.org/NanoLett

In Situ TEM Analysis of Organic−Inorganic Metal-Halide Perovskite Solar Cells under Electrical Bias Quentin Jeangros,*,†,§ Martial Duchamp,‡,⊥ Jérémie Werner,† Maximilian Kruth,‡ Rafal E. Dunin-Borkowski,‡ Bjoern Niesen,† Christophe Ballif,† and Aïcha Hessler-Wyser† †

Photovoltaics and Thin-Film Electronics Laboratory, Ecole Polytechnique Fédérale de Lausanne (EPFL), Institute of Microengineering (IMT), Rue de la Maladière 71B, Neuchâtel CH-2000, Switzerland § Department of Physics, University of Basel, Klingelbergstrasse 82, Basel CH-4056, Switzerland ‡ Ernst RuskaCentre for Microscopy and Spectroscopy with Electrons and Peter Grünberg Institute, Forschungszentrum Jülich D-52425, Germany S Supporting Information *

ABSTRACT: Changes in the nanostructure of methylammonium lead iodide (MAPbI3) perovskite solar cells are assessed as a function of current−voltage stimulus by biasing thin samples in situ in a transmission electron microscope. Various degradation pathways are identified both in situ and ex situ, predominantly at the positively biased MAPbI3 interface. Iodide migrates into the positively biased charge transport layer and also volatilizes along with organic species, which triggers the nucleation of PbI2 nanoparticles and voids and hence decreases the cell performance. KEYWORDS: Characterization, in situ transmission electron microscopy, microstructure, degradation, photovoltaics, perovskite solar cell

O

in cell morphology during J−V characterization were assessed (i) in situ by contacting samples to a microelectromechanical systems chip using a focused ion beam (FIB) scanning electron microscope (SEM) workstation5,6 (Figure 1a) and also (ii) ex situ by comparing FIB-prepared lamellae extracted from cells, either as deposited or after they had been electrically characterized, as detailed in ref 3. FIB samples for ex situ characterization were prepared using a Zeiss NVision 40, while samples assessed in situ were prepared in an FEI Helios Nanolab and contacted to either single-tilt or double-tilt DENSSolutions TEM holders and eventually to a Keithley 2602A sourcemeter. The single-tilt holder was preferred for energy-dispersive X-ray spectroscopy (EDX) analysis (as the needles used to electrically contact the chip to the holder do not shadow the EDX signal emerging from the area of interest), whereas the double-tilt design enabled a more precise sample tilt alignment and hence assessment of interfaces. In all cases, the FIB-prepared samples were thinned using a final voltage of 5 kV to reduce Ga+-induced damage and possible Ga+-rich surface short-circuits, which are particularly detrimental to biasing experiments. A final thickness of about 200 nm was chosen, as thinner samples were seen in preliminary experi-

rganic−inorganic metal-halide perovskite solar cells are emerging as a promising photovoltaic technology to harvest solar energy, with initial efficiencies now surpassing 22%:1 an impressive increase from the first reported value of 3% in 2009.2 In addition to potentially low manufacturing costs, the optical properties of such cells can be tailored to form efficient tandems when combined with high-efficiency silicon solar cells.3 However, the commercialization of this technology remains hindered by stability issues, as various mechanisms (some still under debate) degrade perovskite cells upon operation.4 A detailed analysis of changes in the active layers caused by the application of an electrical bias is therefore essential to understand some of the mechanisms that degrade cell efficiency. Here, applied voltage (V) and current density (J) are correlated directly to changes in the active layer structure on the nanometer scale by electrically biasing perovskite solar cells in situ in a transmission electron microscope (TEM). The cell architecture investigated is described elsewhere3 and consists of a Au contact, a hole transport layer (HTL) of 2,2′,7,7′-tetrakis[N,N-di(4-methoxyphenyl)amino]-9,9′-spirobifluorene (Spiro-OMeTAD), a planar MAPbI3 perovskite absorber, an electron transport layer (ETL) made of a phenyl-C61-butyric acid-methyl-ester/polyethylenimine stack and an indium tin oxide (ITO) front electrode. These layers were deposited onto either an n-type Si wafer for in situ characterization, where a conductive substrate is required, or aluminoborosilicate glass for ex situ characterization. Changes © 2016 American Chemical Society

Received: July 28, 2016 Revised: September 30, 2016 Published: October 24, 2016 7013

DOI: 10.1021/acs.nanolett.6b03158 Nano Lett. 2016, 16, 7013−7018

Letter

Nano Letters

Figure 1. (a) SEM image of a FIB-prepared TEM lamella connected to a biasing chip for in situ TEM electrical measurements; (b) STEM HAADF image and corresponding EDX map of the initial microstructure (the magenta region results from the superposition of the red Pb and blue I signals), with the arrows showing the presence of I in the HTL; (c) J−V measurements for the sample shown in (b) obtained in situ in the TEM (the measurement sequence is indicated by the arrow).

Figure 2. (a) STEM HAADF micrographs showing changes in morphology of the MAPbI3 layer when a current density of 20 mA cm−2 is passed through the cell with +6 V applied to the HTL (different sample than in Figure 1), with i and ii showing magnified views of the regions highlighted in (a) (the dashed line represents the initial interface); (b) HAADF images recorded when biasing for 10 min at +6 V, with the current increasing from 60 to 200 mA cm−2, along with corresponding HAADF intensity profiles integrated over the width of the shaded area; (c) HAADF images acquired when biasing for 10 min at −6 V, which results in a current density of −35 to −40 mA cm−2, and corresponding HAADF intensity profiles extracted from the shaded area. Arrows and circles highlight the formation of nanoparticles and the resulting shrinkage of the MAPbI3 phase during biasing, while arrowheads show a polarity-dependent increase in HAADF intensity at the position of the charge transport layers.

7014

DOI: 10.1021/acs.nanolett.6b03158 Nano Lett. 2016, 16, 7013−7018

Letter

Nano Letters

Nanoparticles continue to form at the interface with the HTL when the sample is kept at +6 V for 10 min (arrows in Figure 2b). The changes in HAADF intensity with biasing confirm that the MAPbI3−HTL interface continues to retract, while nanoparticles grow at this interface. Moreover, the HAADF intensity in the HTL region in the vicinity of the MAPbI3 phase increases slightly under these biasing conditions, which is indicative of ionic migration of heavy elements (arrowhead in the profiles in Figure 2b). The current that passes through the cell increases from 60 to 200 mA cm−2 during this sequence of images. As was also observed in Figure 1c, the setup becomes more conductive during electrical characterization, presumably due to structural changes occurring at the FIB-prepared contacts but also cell degradation. However, a constant current on the order of −35 to −40 mA cm−2 is measured when the voltage is reversed to −6 V for 10 min. These J−V breakdown conditions result in the growth of some nanoparticles in the perovskite phase close to the HTL layer (circles in the images in Figure 2c) but also in void formation within the MAPbI3 located at the interface with the ETL (arrows in Figure 2c). The shrinkage of the active layer upon biasing at the ETL side is further demonstrated by the intensity profiles shown in Figure 2c, with this shrinkage coinciding with a slight increase in intensity at the position of the ETL (arrowhead in the profiles in Figure 2c). On the other hand, the I-rich layer present in the HTL does not appear to evolve during this sequence, as highlighted in the intensity profiles shown in Figure 2c. Additional images capturing the microstructural evolution of the cell as a function of polarity and current density are shown in Supplementary Figure 3. Overall, these microstructural changes are also observed in regions that were not previously irradiated by the electron beam during biasing, hence confirming that they arise from bias and not from electron irradiation (see Supplementary Figure 4). In fact, the PbI2 nanoparticles that form when a current is passing through the sample become less visible upon electron irradiation, presumably due to an amorphization and redispersion process (Supplementary Figure 5). EDX and HRTEM imaging reveal that the nanoparticles, formed during electrical biasing, are made of PbI2 (Figure 3), a degradation product of MAPbI3.11,14−16 Moreover, a lower atomic I-to-Pb ratio of 2.4 is measured by EDX (including the regions where I diffused in the HTL) in comparison to results before biasing (I-to-Pb ratio of 2.7), indicating that some I, probably along with organic species, has volatilized from the sample (assuming a constant Pb concentration). As already observed in Figure 2, a thin void is observed at the position of the interfaces, notably between MAPbI3 and the HTL (arrowhead in Figure 3a). In addition, a small Pb EDX signal is found on the Spiro-OMeTAD side of the interface (arrows in Figure 3a,b) after shrinkage of the MAPbI 3 layer. PbI2 nanoparticles, such as those observed in Figure 3c, appear along defined lines (dashed in Figure 3a), along which current may flow and trigger degradation to PbI2. However, the activephase MAPbI3 still remains in certain regions after biasing (Figure 3d). Under these degradation conditions, Au does not appear to diffuse from the contact to the other layers of the cell,17 at least not within the detection limit of EDX (on the order of 0.1 at%). Some of these bias-induced processes are also observed in samples that underwent ex situ J−V characterization under onesun illumination (Figure 4; characterization performed from 1.1 to −0.2 V and then from −0.2 to 1.1 V, hence ending with the

ments to be more prone to degradation to PbI2 during FIB sample preparation. Air exposure during transfer from the FIB to the TEM was kept below 5 min to minimize interaction with air/moisture, as prolonged transfers were observed to result in the formation of large PbI2 crystals impinging on neighboring contacts (Supplementary Figure 1).7 In situ characterization was carried out in a probe-Cs-corrected FEI Titan Chemistem microscope operated at 200 kV with a low beam current of approximately 50 pA to reduce irradiation damage and surface C deposition due to the electron beam, which may also lead to surface short-circuits. Analysis involved scanning TEM (STEM) imaging using a high-angle annular dark-field (HAADF) detector, with the electron beam remaining blanked in between image acquisitions to reduce irradiation damage. EDX was performed using the single-tilt holder and mainly restricted to after biasing as preliminary experiments showed that the dose required for EDX mapping may result in amorphization of the perovskite phase, PbI2 formation and C deposition.8,9 In addition, high-resolution TEM (HRTEM) was performed after the in situ experiments in an image-Cscorrected FEI Titan operated at 300 kV (with transfer in air of 8 V is required to pass a current through the TEM lamella, with this value then decreasing upon further electrical testing (Figure 1c; measurement sequence marked with an arrow). This high resistance is likely related to series resistance at the FIB-deposited Pt contacts. The voltage required to pass a current of a few tens of mA cm−2 is hence much larger in the TEM sample than in a full cell (the initial voltage is ∼8 V for this in situ experiment compared to 1 V ex situ).3 This voltage then decreases upon electrical characterization, presumably due to a decrease in both series resistance at the FIB-prepared contacts and shunt resistance at the cell level (due to its degradation). These two effects are difficult to deconvolve. Nevertheless, the curves measured in situ in the TEM are diode-shaped, as expected for a photovoltaic cell, until breakdown J−V conditions are reached. The current density is computed assuming a constant TEM lamella cross-section (10−8 cm2). The application of +6 V on the HTL side of the cell results in a current density of 20 mA cm−2 (a value close to the shortcircuit current of the cell, which is on the order of 15−20 mA cm−2)3 and induces microstructural changes after a few minutes. Nanoparticles appear at the positions of structural defects and at the interface with the HTL, which shrinks slightly as a result of this process (Figure 2a, insets i and ii). 7015

DOI: 10.1021/acs.nanolett.6b03158 Nano Lett. 2016, 16, 7013−7018

Letter

Nano Letters

sample that was degraded ex situ is constant after the FIB preparation, the thickness of the thin TEM lamella that was degraded in situ becomes inhomogeneous during these experiments, giving rise to additional HAADF contrast at the position of these PbI2 nanoparticles. The microstructural changes observed in situ and ex situ occur preferentially at the positively biased MAPbI3 interface, in agreement with optical micrographs of individual MAPbI3 layers under bias reported elsewhere.20,21 While these observations were explained in literature in terms of ionic migration within the perovskite layer, additional degradation pathways are triggered here by the presence of the neighboring hole and electron selective contacts. Indeed, iodide and vacancies have been reported to migrate under bias, not only within the MAPbI3 phase but also into the HTL (under forward conditions).22−25 In this regard, the HAADF intensity profiles and EDX data shown in Figures 2 and 4 suggest that I− ions migrate mostly from the positive MAPbI3 interface into the charge transport layer and less within the MAPbI3 phase itself (as the HAADF intensity at the center of the perovskite phase does not evolve noticeably with bias). Additionally, the presence of Pb on the positive side of the interfacial void might be induced by electromigration. Furthermore, bias accelerates the volatilization of I and organic species, as suggested by the decrease in I-to-Pb ratio measured by EDX, especially in the case of the samples tested in situ. This volatilization mechanism is known to occur in the form of hydrogen iodide (HI) and methylamine (CH3NH2) or methylammonium iodide (MAI) but usually happens over a longer time scale in the absence of bias, e.g., after a few days of storage in air or vacuum14,15 or during aging for 20 days at 85 °C under one-sun illumination when encapsulated.16 Overall, HRTEM imaging and spectroscopies demonstrate that these transport mechanisms of I-related species result in the formation of PbI2 at the positive MAPbI3 interface (Figure 3), in agreement with Raman measurements reported in the literature.21 These polarity-dependent migration, volatilization and resulting PbI2 and void formation processes are indicative of an energy dissipation mechanism occurring preferentially at the positive MAPbI3 interface, presumably due to the presence of recombination centers. Overall, while PbI2 may reverse back to MAPbI3 under certain conditions,26 the bias-induced changes observed here are irreversible, similarly to the findings of ref 21, both in the case of the thin TEM sample degraded in situ and also in the case of a full perovskite solar cell held for 20 min at a similar current density to that employed in situ (60 mA cm−2; see Supplementary Figure 8). While reports have demonstrated perovskite cells with different designs exhibiting stable performance over a few hundreds of hours,4,27,28 degradation is accelerated here due to (i) the enhanced volatilization of species through the large free surfaces of the TEM sample, as confirmed by the larger decrease in I-to-Pb ratio measured by EDX in thin samples tested situ when compared to bulk samples tested ex situ (2.7 to 2.4 in situ, see Figures 1 and 3; 2.6 to 2.5 ex situ, see Figure 4) and (ii) the large bias and current densities employed in situ, which are a few times higher than cell-operation conditions. In summary, electrical biasing of thin samples in situ in a TEM provides insight into bias-induced degradation mechanisms of MAPbI3 perovskite cells at the nanometre scale. Several processes that may contribute to a decrease in cell efficiency during operation have been identified, including the

Figure 3. (a) STEM HAADF image of the Spiro-OMeTAD−MAPbI3 interface after electrical experiments in the TEM and corresponding EDX maps of Pb, I, and C; (b) EDX spectra acquired at positions i to iv shown in (a), highlighting the presence of Pb (arrows) on the HTL side of the interfacial void (arrowhead in a); (c) HRTEM image and corresponding Fourier transform of a PbI2 nanoparticle; (d) HRTEM image and Fourier transform of two epitaxial MAPbI3 grains. Both Fourier transforms were indexed using the software JEMS18 according to the hexagonal phase of PbI2 and the tetragonal phase of MAPbI3,19 respectively.

HTL being positively biased3). As for the in situ experiments, an interfacial void forms at the MAPbI3−HTL interface (marked by dots in Figure 4b,d), with some I and Pb remaining at the Spiro-OMeTAD side of the void (arrowhead in Figure 4d). Figure 4d also suggests that negative I− ions, i.e., iodide, migrate under bias into the positively biased hole transport layer, which would explain the changes in HAADF intensity observed in Figure 2. Indeed, the I EDX signal in the HTL increases with respect to the I signal in MAPbI3 during electrical characterization, indicating a migration of iodide toward the HTL. In addition, Supplementary Figure 3 suggests that the iodide that migrated into the HTL under forward conditions may migrate back toward the MAPbI3 phase when the polarity is reversed. The MAPbI3−ETL interface did not degrade significantly ex situ as the sample was tested mainly under forward conditions (Supplementary Figure 6). PbI2 domains are also detected after ex situ characterization (Supplementary Figure 7). However, the effect is less clear in ex situ than in in situ conditions, presumably due to local differences in sample thickness. While the thickness of the 7016

DOI: 10.1021/acs.nanolett.6b03158 Nano Lett. 2016, 16, 7013−7018

Letter

Nano Letters

Figure 4. (a) STEM HAADF micrograph and corresponding EDX map of an as deposited sample (I-to-Pb atomic ratio of 2.6 including the diffused layer of I in the HTL); (b) STEM HAADF micrograph and corresponding EDX map of a sample from the same batch that was characterized ex situ (I-to-Pb atomic ratio of 2.5 including the diffused layer of I in the HTL); (c) magnified STEM HAADF image recorded from the MAPbI3−HTL interface after deposition shown alongside the corresponding net EDX intensity counts profile for the I L, Pb L, and C K edges (background removed); (d) corresponding STEM HAADF image and EDX net counts profile after electrical characterization. Dots mark the formation of a void at the MAPbI3−HTL interface and arrowheads the accumulation of species across this interfacial void. Arrows highlight the bias-induced change in ratio between the I EDX intensities in the MAPbI3 phase (80 nm from the interface with the HTL) and in the HTL (25 nm into the SpiroOMeTAD).

by J.W. The work was supervised by B.N., R.D.B., C.B., and A.H. The manuscript was written by Q.J. with contributions from all of the authors. All authors have given approval to the final version of the manuscript.

formation of PbI2 nanoparticles and voids. These microstructural changes occur preferentially at the MAPbI3 interface with the positively biased charge transport layer, presumably due to an interface energy dissipation process that depends on polarity. Under forward conditions, the degradation of MAPbI3 occurs at the interface with the hole transport layer and is triggered by both the migration of iodide into the positive hole transport layer and the volatilization of iodine and organic species. These results highlight the need for a detailed analysis of the MAPbI3−hole transport layer interactions during cell operation to improve the stability of the device.

■

Funding

The European Union 7th Framework Programme (grant agreement 312483-ESTEEM2-I3), the projects Swiss National Science Foundation PV2050 and Disco, Nano-Tera and the Swiss Federal Office of Energy, under grant SI/501072−01. Notes

The authors declare no competing financial interest.

■

ASSOCIATED CONTENT

S Supporting Information *

ACKNOWLEDGMENTS The authors thank the Interdisciplinary Centre for Electron Microscopy of EPFL for FIB and TEM access, D. Meertens and V. Migunov for help with sample preparation and biasing experiments, and N. Wyrsch and F.-J. Haug for fruitful discussions.

The following files are available free of charge. Supporting Information (PDF) The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.nanolett.6b03158. Figures showing the microstructural effects of air exposure, sample thickness, in situ and ex situ J−V characterization and electron beam irradiation in addition to J−V curves obtained in situ and ex situ. (PDF)

■

■

REFERENCES

(1) NREL. Research Cell Efficiency Records. http://www.nrel.gov/ pv/. (2) Kojima, A.; Teshima, K.; Shirai, Y.; Miyasaka, T. J. Am. Chem. Soc. 2009, 131 (17), 6050−6051. (3) Werner, J.; Weng, C.-H.; Walter, A.; Fesquet, L.; Seif, J. P.; De Wolf, S.; Niesen, B.; Ballif, C. J. Phys. Chem. Lett. 2016, 7 (1), 161− 166. (4) Leijtens, T.; Eperon, G. E.; Noel, N. K.; Habisreutinger, S. N.; Petrozza, A.; Snaith, H. J. Adv. Energy Mater. 2015, 5 (20), 1−23. (5) Duchamp, M.; Xu, Q.; Dunin-Borkowski, R. E. Microsc. Microanal. 2014, 20 (1), 1638−1645. (6) Duchamp, M.; Migunov, V.; Tavabi, A.; Mehomic, A.; Munde, M.; Kenyon, A. J.; Dunin-Borkowski, R. E. Resolut. Discovery 2016, DOI: 10.1556/2051.2016.00036.

AUTHOR INFORMATION

Corresponding Author

*E-mail: quentin.jeangros@epfl.ch. Present Address ⊥

School of Materials Science and Engineering, Nanyang Technological University, 50 Nanyang Avenue, Singapore 639798

Author Contributions

The experiments were designed by Q.J. with the help of M.D. and performed by M.K., M.D., and Q.J. using samples prepared 7017

DOI: 10.1021/acs.nanolett.6b03158 Nano Lett. 2016, 16, 7013−7018

Letter

Nano Letters (7) Conings, B.; Drijkoningen, J.; Gauquelin, N.; Babayigit, A.; D’Haen, J.; D’Olieslaeger, L.; Ethirajan, A.; Verbeeck, J.; Manca, J.; Mosconi, E.; Angelis, F. D.; Boyen, H.-G. Adv. Energy Mater. 2015, 5 (1500477), 1−8. (8) Xiao, C.; Li, Z.; Guthrey, H.; Moseley, J.; Yang, Y.; Wozny, S.; Moutinho, H.; To, B.; Berry, J. J.; Gorman, B. P.; Yan, Y.; Zhu, K.; AlJassim, M. J. Phys. Chem. C 2015, 119, 26904−26911. (9) Klein-Kedem, N.; Cahen, D.; Hodes, G. Acc. Chem. Res. 2016, 49 (2), 347. (10) Matteocci, F.; Busby, Y.; Pireaux, J.-J.; Divitini, G.; Cacovich, S.; Ducati, C.; Di Carlo, A. ACS Appl. Mater. Interfaces 2015, 7 (47), 26176−26183. (11) Divitini, G.; Cacovich, S.; Matteocci, F.; Cinà, L.; Di Carlo, A.; Ducati, C. Nat. Energy 2016, 1 (2), 15012. (12) Xiao, M.; Huang, F.; Huang, W.; Dkhissi, Y.; Zhu, Y.; Etheridge, J.; Gray-Weale, A.; Bach, U.; Cheng, Y. B.; Spiccia, L. Angew. Chem., Int. Ed. 2014, 53 (37), 9898. (13) Cliff, G.; Lorimer, G. W. J. Microsc. 1975, 103 (2), 203−207. (14) Alberti, A.; Deretzis, I.; Pellegrino, G.; Bongiorno, C.; Smecca, E.; Mannino, G.; Giannazzo, F.; Condorelli, G. G.; Sakai, N.; Miyasaka, T.; Spinella, C.; La Magna, A. ChemPhysChem 2015, 16 (14), 3064− 3071. (15) Deretzis, I.; Alberti, A.; Pellegrino, G.; Smecca, E.; Giannazzo, F.; Sakai, N.; Miyasaka, T.; La Magna, A. Appl. Phys. Lett. 2015, 106 (13), 131904. (16) Han, Y.; Meyer, S.; Dkhissi, Y.; Weber, K.; Pringle, J. M.; Bach, U.; Spiccia, L.; Cheng, Y.-B. J. Mater. Chem. A 2015, 3 (15), 8139− 8147. (17) Domanski, K.; Correa-Baena, J.-P.; Mine, N.; Nazeeruddin, M. K.; Abate, A.; Saliba, M.; Tress, W.; Hagfeldt, A.; Grätzel, M. ACS Nano 2016, 10, 6306−6314. (18) Stadelmann, P. A. Ultramicroscopy 1987, 21 (2), 131−145. (19) Oku, T. Sol. Cells - New Approaches Rev. 2015, 77−101. (20) Xiao, Z.; Yuan, Y.; Shao, Y.; Wang, Q.; Dong, Q.; Bi, C.; Sharma, P.; Gruverman, A.; Huang, J. Nat. Mater. 2014, 14 (V), 193− 198. (21) Leijtens, T.; Hoke, E. T.; Grancini, G.; Slotcavage, D. J.; Eperon, G. E.; Ball, J. M.; De Bastiani, M.; Bowring, A. R.; Martino, N.; Wojciechowski, K.; McGehee, M. D.; Snaith, H. J.; Petrozza, A. Adv. Energy Mater. 2015, 5 (20), 1500962. (22) Azpiroz, J. M.; Mosconi, E.; Bisquert, J.; De Angelis, F. Energy Environ. Sci. 2015, 8 (7), 2118−2127. (23) Eames, C.; Frost, J. M.; Barnes, P. R. F.; O’Regan, B. C.; Walsh, A.; Islam, M. S. Nat. Commun. 2015, 6, 7497. (24) Li, C.; Tscheuschner, S.; Paulus, F.; Hopkinson, P. E.; Kießling, J.; Köhler, A.; Vaynzof, Y.; Huettner, S. Adv. Mater. 2016, 28 (12), 2446−2454. (25) Back, H.; Kim, G.; Kim, J.; Kong, J.; Kim, T. K.; Kang, H.; Kim, H.; Lee, J.; Lee, S.; Lee, K. Energy Environ. Sci. 2016, 9, 1258−1263. (26) Yuan, Y.; Wang, Q.; Shao, Y.; Lu, H.; Li, T.; Gruverman, A.; Huang, J. Adv. Energy Mater. 2016, 6 (2), 1501803. (27) Bush, K. A.; Bailie, C. D.; Chen, Y.; Bowring, A. R.; Wang, W.; Ma, W.; Leijtens, T.; Moghadam, F.; McGehee, M. D. Adv. Mater. 2016, 28, 3937. (28) Saliba, M.; Matsui, T.; Seo, J.-Y.; Domanski, K.; Correa-Baena, J.-P.; Nazeeruddin, M. K.; Zakeeruddin, S. M.; Tress, W.; Abate, A.; Hagfeldt, A.; Grätzel, M. Energy Environ. Sci. 2016, 9 (6), 1989−1997.

7018

DOI: 10.1021/acs.nanolett.6b03158 Nano Lett. 2016, 16, 7013−7018