Light-Induced Increase of Electron Diffusion Length in a p-n Junction Type CH 3 NH 3 PbBr 3 Perovskite Solar Cell

Transcription

1 Light-Induced Increase of Electron Diffusion Length in a p-n Junction Type CH 3 NH 3 PbBr 3 Perovskite Solar Cell Nir Kedem 1, *, Thomas M. Brenner 1, *, Michael Kulbak 1, Norbert Schaefer 2, Sergiu Levcenco 2, Igal Levine 1, Daniel Abou-Ras 2, Gary Hodes 1, David Cahen 1 *equal contribution 1. Dept. of Materials & Interfaces, Weizmann Inst. of Science, Rehovot, Israel Helmholtz-Zentrum Berlin für Materialien und Energie GmbH, Hahn-Meitner-Platz 1, Berlin, Germany Supplementary Information S.1 Relationship between Electric Field Profile and EBIC Profile The magnitude of the EBIC signal depends on the efficiency of charge separation and collection. Wherever an internal field is present in the material the collection efficiency should be high, because the field assists charge separation. If there is no field to direct the charge separation then the collection is limited by the diffusion length of the free carriers. An illustration of the vacuum level in a p-n and p-i-n junction is shown in figure S1 along with the expected (schematic) EBIC profiles. Maximum current collection appears where field assisted charge separation exists. The signal decays exponentially in zero-field conditions as the point of generation moves further from the selective contact/space charge region. The slope of the signal decay depends on the minority carrier diffusion length, which is normally the limiting factor in current collection in p-n junction type solar cells. For a p-i-n cell structure, such as that of n-hbl/i-mapbi 3 (Cl)/p-HTL, collection should be uniform over the entire absorber length. HBL and HTL refer to hole blocking layer and hole transport layer respectively. If the absorber is sufficiently thick and the electric field across it is not constant, then a double peak profile will evolve 1. Figure S1: Illustration of the E vac profile (in blue) and the corresponding expected EBIC signal (in black). The position of the contacts is marked in red. A change in E vac indicates an internal field, which promotes charge separation. The middle figure shows a p-i-n junction with the field distributed evenly across the entire width of the semiconductor while the right figure shows a p-i-n junction with a zero field region in the semiconductor. 1

2 S.2 Additional Data, Experimental Details, and Comments Hysteresis in J-V curves is unlikely to Affect EBIC Results: The effect of hysteresis in our devices is mainly in the V OC and FF of the cells and the hysteresis is not expected to affect the major EBIC results (light effect) as they were performed at short-circuit conditions, where hysteresis is negligible. Normally in order to see hysteresis in perovskite cells biasing to V > V OC is required, including in our case. As we did not bias the sample above V OC hysteresis was not considered for the EBIC results. It is also possible that there could be a hysteretic effect of applied bias on the beam damage observed in the EBIC signal as discussed in section S.9. We did not find any correlation between beam damage and applied bias on the sample, beam damage can be seen no matter how long or short one waits after the first measurement to take the second. Generation of EBIC Signal in Contact Materials: Generation of carriers occurs in the contact materials (electron- and hole- transport layers) from e-beam irradiation and this can result in an EBIC signal. This is seen best in the FTO/TiO 2 layer, where holes generated in this layer diffuse to the space charge region and are separated from electrons. This is part of the reasoning behind attributing the observed EBIC peak to a p-n junction since the exponential decay of signal characteristic of carrier diffusion can be observed in this layer. We note also that signal can occasionally be seen on the hole collection side of the device and also above the gold layer (i.e., in free space), because the cross-section is not perfectly aligned or perfectly flat so that a high signal is sometimes generated due to morphological defects. For that reason we limit our analysis to areas in the image where the cross section is reasonably flat and the back contact is minimally exposed. Alternative figure 4 showing location of contacts: Fig. S2 shows Fig. 4 from the main text, but with the positions of the interfaces between the FTO/TiO 2 and perovskite layer and between the HTL and the perovskite layer indicated in the EBIC images. In the EBIC profiles the HTL/perovskite interface is not shown as there are only clear features at the FTO/TiO 2 electron contact / MAPbBr 3 (Cl) interface and in the MAPbBr 3 (Cl) layer. 2

3 Figure S2: Alternative version of figure 4 in main text, showing the location of the perovskite contacts with the FTO/TiO 2 and with the HTL in the EBIC images and that of the FTO/TiO 2 with the perovskite layer in the EBIC profiles (the HTL contact is not present in the profile). Comparison of EBIC images in the dark and under illumination generating ~20 na photocurrent. The shown line profiles (taken from the lines shown in the images) were selected as they show signal decay corresponding to the average electron diffusion lengths for this sample. EBIC Profiles of Biased and Illuminated Sample The diffusion length of illuminated samples does not change with biasing the sample. 3

4 Figure S3: EBIC profiles of biased and illuminated sample S.3 Light Absorption and Collection Efficiency in MAPbBr 3 (Cl) PV Devices The theoretical maximum current that can be expected from this material is 10 ma/cm 2 (Shockley-Queisser limit), more than twice the J SC of our devices. The highest J SC value reported for MAPbBr 3 -based devices, (excluding cases where hysteresis was not mentioned, or values increased by hysteresis) is ~ 9 ma/cm 2. 2,3 The J SC is mostly governed by the EQE of the cells, which includes reflection losses, incomplete coverage of the absorber, recombination at interfaces etc. As the cell performance is not the focus of this ms., we do not discuss these issues in detail. To assess the effect of diffusion length on collection efficiency, we consider generated photocurrent as a function of distance. The photocurrent at a given point is the product of the photons absorbed there (to generate carriers) and the carrier collection efficiency there. Integrating this photocurrent over the whole active layer and dividing by the incident photon density will give the carrier collection efficiency. The generated photocurrent as a function of position in the MAPbBr 3 (Cl) active layer is shown for various conditions in figure S.4. The high extinction coefficient of MAPbBr 3 (Cl) (60,000-70,000 cm -1 ) results in 70-90% of the total absorbed light being absorbed in the SCR, depending on its width ( nm). Separation and charge collection efficiency of carriers reaching the SCR is high (assumed to be 100% here). The remaining light is absorbed in the quasi-neutral region, where the diffusion length limits collection efficiency. The effect of SCR width and diffusion length on generated photocurrent can be seen for the parameter extremes we measured in EBIC. The minimum SCR width (50 nm) and minimum diffusion length (50 nm) result in collection of 73% of the generated carriers. At the other extreme (SCR width = 150 nm, diffusion length = 350 nm), carrier collection efficiency will be as high as 98%. The expected collection efficiencies are well above the ~50% we observe in our devices. Most of the loss in our devices is likely due to incomplete coverage of the perovskite, a known issue for MAPbBr 3 (Cl) films. From this analysis we conclude that the J SC values we obtain are in agreement with the diffusion lengths we obtained via EBIC. Figure S4: Illustration of carrier collection as a function of position in the MAPbBr 3 (Cl) active layer. The curves are a product of light absorption and carrier collection efficiency. A fully depleted active layer (w = full layer) will give 100% collection efficiency (ignoring other possible losses). 4

5 S.4 How to Estimate Carrier Concentration Generated by Electron Beam Following the excellent review by Leamy 4 we estimate the carrier density generated by the electron beam from the known beam current and electron energy. In our case the beam current is 7-9 pa and the electron energy 3 kev. The physical process that occurs upon e-beam exposure is as follows: the incident electron beam generates free carriers which diffuse outward from the point of generation according to their diffusion length, eventually reaching a steady state distribution that depends on beam current, incident electron energy, and carrier diffusion length (L). The carrier generation is estimated from the beam current and electron energy. For a semiconductor of bandgap ~2 ev, ~6 ev are required to generate an electron-hole pair by electron beam irradiation 4. Therefore each electron in a 3 kv beam can generate ~500 electron-hole pairs. Multiplying this number, the generation factor, by the beam current gives the total carrier generation. All these carriers can be estimated to diffuse into a hemisphere with radius of ~2L (i.e., we consider a sphere defined by a border where the carrier concentration has decayed to 1/10 its original value). By dividing the total carrier generation by the volume of this sphere an estimate of the steady-state carrier generation per unit volume is obtained. To obtain the steady-state carrier concentration, the generation (per unit volume) must be multiplied by the lifetime, in this case assumed to be ns, based on the lifetime obtained for MAPbBr3(Cl) in refs. 3,5 S.5 Analysis of the Dielectric Constant and Carrier Density of MAPbBr 3 (Cl) We estimate the dielectric constant using our experimental EBIC and CV data on our samples, because, unfortunately, there is still no consensus in the literature on what the value of the dielectric constant (both static and AC) is for any of the CH 3 NH 3 PbX 3 (X = Cl, Br, I) compounds, including the much more heavily studied iodide. The reported values for the iodide and bromide perovskite range from ε r = 5 to >100 depending on the frequency range under consideration and the method of measurement We begin by interpreting the impedance response of our MAPbBr 3 (Cl) perovskite solar cells. The Nyquist plots showed a parallel RC arc in the frequency range 2 khz 1 MHz that we attribute to the response of the SCR and used to determine the dielectric constant. At low frequencies (< 500 Hz) a strong deviation from the simple RC model is evident, which is similar to features commonly observed in planar iodide perovskite solar cells The capacitance at the relevant frequency range was used for the following analysis. Taking 1/C 2 vs. bias between -0.3 and 0.8 V, a Mott-Schottky plot is obtained for all frequencies. The carrier concentration, 8 x x cm -3, is calculated from the derivative d(1/c 2 )/dv of the Mott-Schottky plot and the dielectric constant determined below according to N A = -2(eε 0 ε r d(1/c 2 )/dv) -1 (where N A is the acceptor density). 5

6 Figure S5: Mott-Schottkey plot obtained at 1 MHz. The active layer of the solar cell is considered to be composed of two capacitors in parallel: the capacitor of interest, composed of the MAPbBr 3 (Cl) crystallites and their interface with TiO 2 and the HTL, and a capacitor composed of the HTL filling in the space between the crystallites and its interface with TiO 2 and Au. The capacitor composed of the HTL and no perovskite, regardless of its area, will contribute negligible capacitance to the total because of: (1) the low dielectric constant of the organic HTL compared to that of the perovskite, (2) the nm thickness of this capacitor compared to the space charge region width in the perovskite, and (3) vacuum alignment at the interfaces, which is characteristic of organic semiconductors (including CBP) and prevents formation of a space charge region at either interface 15. Thus, the capacitance of the MAPbBr 3 (Cl) layer is dominated by the capacitance of the space charge region at the MAPbBr 3 (Cl)/TiO 2 interface. The width of this region at 0 V bias ranges between nm (see main text). The area of this capacitor is the area of the solar cell times the coverage fraction of the MAPbBr 3 (Cl) which is estimated to be 50% 75%. Using the corrected area a relative dielectric constant of r = 5 12 is estimated at 2 khz 1 MHz range according to the relationship C/A= 0 r /w (where 0 is the vacuum permittivity, r is the relative dielectric constant of MAPbBr 3 (Cl), w is the space charge region width, and A is the corrected area of the capacitor). S.6 Device Fabrication Procedure F-doped tin oxide (FTO) transparent conducting substrates (Xinyan Technology TCO- XY15) were cut and cleaned by sequential 15 min sonication in warm aqueous alconox solution, deionized water, acetone, and ethanol, followed by drying in a N 2 stream. A compact 60 nm thin TiO 2 layer was then applied to the clean substrate by spray pyrolysis of a 30 mm titanium diisopropoxide bis(acetylacetonate) solution in isopropanol using air as the carrier gas on a hot plate set to 450 C, followed by two step annealing at 160 o C and 450 C, each for 1 h in air. A CH 3 NH 3 PbBr 3-x Cl x solution was prepared as described elsewhere. In short, CH 3 NH 3 Br was prepared by mixing methyl amine (40% in methanol) with hydrobromic acid (47% in water) in a 1:1 molar ratio in a 6

7 100 ml flask under continuous stirring at 0 C for 30 min. CH 3 NH 3 Br was then crystallized by removing the solvent in rotary evaporator, washing three times in diethyl ether for 30 min, and filtering the precipitate. The material, in the form of white crystals, was dried overnight in vacuum at 65 C. It was then kept in a dark, dry environment until further use. A 40 wt % solution of CH 3 NH 3 PbBr 3 was prepared by mixing PbCl 2 and CH 3 NH 3 Br in a 3:1 mol ratio in DMF. To coat the substrate, the solution was spin-coated on a substrate at 80 o C in two stages, 3 s at 500 rpm and then at 6000 rpm for 40 s. The substrate was then heated on a hot plate set at 155 C for 45 min. All procedures were carried out in an ambient atmosphere. To finish the device fabrication, 125 μl of a hot (ca. 60 C) hole conductor solution (40 mg of CBP in 1 ml of chlorobenzene, mixed with 3.5 μl of tert-butylpyridine and 7 μl of 170 mg/ml LiTFSI, bis(trifluoromethane)sulfonamide (in acetonitrile)), was applied by spin-coating 3 s at 500 rpm and then at 2000 rpm for 40 s. The samples were left overnight in the dark in dry air before 100 nm gold contacts were thermally evaporated on the back through a shadow mask with 0.24 cm 2 rectangular holes. S.7 Complete Details of EBIC Apparatus Two separate EBIC setups were used to complete the measurements in the manuscript. Setup 1 consisted of a Zeiss Supra scanning electron microscope (SEM), together with a Stanford Instruments pre-amplifier model SR570. The EBIC signal was directed into an input channel of the SEM computer, and recorded using the standard Zeiss SEM software. Setup 2, which allowed lock-in amplification via beam-blanking, consisted of a Zeiss UltraPlus SEM with a commercial EBIC setup by Point Electronic GmbH, a lock-in amplifier by Stanford Instruments, and a beam blanker and controller from Raith. The beam blanker was operated at 5 khz frequency and the lock-in was done using 3 ms time constant for EBIC imaging. In order to limit exposure a 25 x 250 pixel image was taken corresponding to 550 x 5500 nm area. In both setups a 3kV accelerating voltage, 10 µm in diameter aperture, and 8 10 mm working distance was used which resulted in a 5-7 pa (setup 1) or 2-5 pa (setup 2) beam current. The experimental parameters put a lower limit on resolution of ~22 nm (pixel resolution); following Leamy 4 (ref. 15 in the main text) the resolution of the beam is ~25 nm. This resolution limits our accuracy in reporting the lower range of diffusion lengths, but it will have a smaller effect for the mid-high values >100 nm. More importantly, as the resolution is fixed, any light or bias dependent changes are not affected by it. 7

8 S.8 EBIC profile analysis The EBIC profile was converted to a numeric format from the EBIC-SEM image using Image-J software. The profile was then fitted to a single exponential decay, with offsets to account for the onset of the decay and the background signal, in order to find the diffusion length. For setup 1, 5 EBIC profiles were randomly selected from each image. The diffusion length was then calculated for each profile and an average value for the image was calculated. For setup 2, 25 lines (i.e. all the image lines) were combined into a single pixel-averaged profile from which the average diffusion length for the image was extracted. For each illumination condition at least 5 EBIC images from different areas in the sample were taken. The reported diffusion length is a result of averaging these 5 image diffusion lengths. The error estimate in both cases is standard deviation of the image-to-image variation of values (since each image usually captures a single grain, this is the grain-to-grain variation). S.9 Avoiding Beam Damage in EBIC Measurements The organo-metal-halide perovskite materials are highly susceptible to beam damage, much more so than other inorganic materials (i.e. Si or CIGS) in our experience. Furthermore, the Br-based perovskites appear to be more sensitive than their I-based counterpart. This beam damage has a very significant impact on EBIC results and must be taken into account when deciding the best practice for EBIC imaging of perovskites. Figure S2 shows the results of multiple scans of the same area, shown chronologically. The impact on collection efficiency is clear, each scan lowers the charge collection efficiency and diffusion length of this region of the device. 8

9 Figure S6: Degradation of EBIC signal of MAPbBr 3 (Cl)-based solar cell during repeated imaging of the same spot. Imaging was done under mild conditions to minimize beam damage, but degradation is observed anyway. The EBIC line profiles show that the intensity degrades with scan number (bottom left) as does the diffusion length (bottom right, normalized profiles). Our best practices are outlined in the manuscript itself. More detail is given here. The EBIC image should be obtained as fast as possible upon exposure of the sample to a highly focused electron beam (i.e. at high magnification). To this end, image improvements should be completed on a sacrificial area. Then, the focus should be moved to a fresh spot and EBIC image should be taken upon the first exposure. The image should be taken using pixel averaging, and the pixel integration time should be kept to a minimum while avoiding a loss in signal-to-noise ratio. We stress also that conclusions made from imaging the same spot repeatedly are highly suspect. Instead, the best practice is to determine trends from statistics gathered over many spots imaged only once. (1) Araújo, D.; Romero, M. J.; Morier-Genoud, F.; Garcıá, R. Multiple Quantum Well GaAs/AlGaAs Solar Cells: Transport and Recombination Properties by Means of EBIC and Cathodoluminescence. Mater. Sci. Eng. B 1999, (2) Heo, J. H.; Song, D. H.; Im, S. H. Planar CH 3 NH 3 PbBr 3 Hybrid Solar Cells with 10.4% Power Conversion Efficiency, Fabricated by Controlled Crystallization in the Spin-Coating Process. Adv. Mater. 2014, (3) Sheng, R.; Ho-Baillie, A.; Huang, S.; Chen, S.; Wen, X.; Hao, X.; Green, M. A. Methylammonium Lead Bromide Perovskite-Based Solar Cells by Vapor-Assisted Deposition. J. Phys. Chem. C 2015, (4) Leamy, H. J. Charge Collection Scanning Electron Microscopy. J. Appl. Phys. 1982, R51 R80. (5) Zhang, M.; Yu, H.; Lyu, M.; Wang, Q.; Yun, J.-H.; Wang, L. Composition- Dependent Photoluminescence Intensity and Prolonged Recombination Lifetime of Perovskite CH 3 NH 3 PbBr 3 x Cl x Films. Chem. Commun. 2014, (6) Tanaka, K.; Takahashi, T.; Ban, T.; Kondo, T.; Uchida, K.; Miura, N. Comparative Study on the Excitons in Lead-Halide-Based Perovskite-Type Crystals CH 3 NH 3 PbBr 3 CH 3 NH 3 PbI 3. Solid State Commun. 2003, (7) Poglitsch, A.; Weber, D. Dynamic Disorder in Methylammoniumtrihalogenoplumbates (II) Observed by Millimeter-wave Spectroscopy. J. Chem. Phys. 1987, (8) Maeda, M.; Hattori, M.; Hotta, A.; Suzuki, I. Dielectric Studies on CH 3 NH 3 PbX 3 (X = Cl and Br) Single Cystals. J. Phys. Soc. Jpn. 1997, (9) Samiee, M.; Konduri, S.; Ganapathy, B.; Kottokkaran, R.; Abbas, H. A.; Kitahara, A.; Joshi, P.; Zhang, L.; Noack, M.; Dalal, V. Defect Density and Dielectric Constant in Perovskite Solar Cells. Appl. Phys. Lett. 2014, (10) Lin, Q.; Armin, A.; Nagiri, R. C. R.; Burn, P. L.; Meredith, P. Electro-Optics of Perovskite Solar Cells. Nat. Photonics 2015,

10 (11) Duan, H.-S.; Zhou, H.; Chen, Q.; Sun, P.; Luo, S.; Song, T.-B.; Bob, B.; Yang, Y. The Identification and Characterization of Defect States in Hybrid Organic inorganic Perovskite Photovoltaics. Phys. Chem. Chem. Phys. 2014, (12) Miyano, K.; Yanagida, M.; Tripathi, N.; Shirai, Y. Simple Characterization of Electronic Processes in Perovskite Photovoltaic Cells. Appl. Phys. Lett. 2015, 106 (9), (13) Gonzalez-Pedro, V.; Juarez-Perez, E. J.; Arsyad, W.-S.; Barea, E. M.; Fabregat- Santiago, F.; Mora-Sero, I.; Bisquert, J. General Working Principles of CH 3 NH 3 PbX 3 Perovskite Solar Cells. Nano Lett. 2014, (14) Kim, H.-S.; Mora-Sero, I.; Gonzalez-Pedro, V.; Fabregat-Santiago, F.; Juarez-Perez, E. J.; Park, N.-G.; Bisquert, J. Mechanism of Carrier Accumulation in Perovskite Thin-Absorber Solar Cells. Nat. Commun. 2013, 4. (15) Greiner, M. T.; Helander, M. G.; Tang, W.-M.; Wang, Z.-B.; Qiu, J.; Lu, Z.-H. Universal Energy-Level Alignment of Molecules on Metal Oxides. Nat. Mater. 2012,