Supporting Information Spatially-resolved imaging on photocarrier generations and band alignments at perovskite/pbi2 hetero-interfaces of perovskite solar cells by light-modulated scanning tunneling microscopy Min-Chuan Shih, Shao-Sian Li, Cheng-Hua Hsieh, Ying-Chiao Wang, Hung-Duen Yang, Ya-Ping Chiu,,,,#, * Chia-Seng Chang, and Chun-Wei Chen,#,* Department of Physics, National Sun Yat-sen University, Kaohsiung 804, Taiwan Department of Materials Science and Engineering, National Taiwan University, Taipei 106, Taiwan Department of Physics, National Taiwan University, Taipei 106, Taiwan Institute of Physics, Academia Sinica, Nankang, Taipei 115, Taiwan # Taiwan Consortium of Emergent Crystalline Materials (TCECM), Ministry of Science and Technology, Taiwan Corresponding Author * Ya-Ping Chiu, ypchiu@phys.ntu.edu.tw * Chun-Wei Chen, chunwei@ntu.edu.tw 1
Supporting information 1 1. Electrical modulations of light-modulated STM To reduce the thermal effect in our STS measurements with light illumination, all the work was measured in the ultra-high vacuum (UHV) condition and at the temperature of 100 K. In addition, we used the electronic control system to replace the conventional mechanical chopper. During the acquisition of current-voltage (I-V) spectra, the laser was electrically modulated by the STS control signal. Detailed parameters for the electrical modulations in STS measurements with light illumination are addressed in the in the Figure S1. Figure S1: (a) The electronics control system with light modulation for STS measurements. (b) The experimental tunneling current (I) as a function of the sample bias (V) when the electronics control system was combined with the laser modulation. Since the laser modulation was turned on after the feedback mechanism of the STS system was closed, the current-voltage spectra with and without illumination were obtained at the same tip-sample separation, which was fixed by the set voltage and current without illumination (see Figure S1(a)). In the present STS measurement, the voltage was ramped from -4V to +4V, recording 161 steps (points) spectroscopic data. Thus, the resolution of the electronic spectroscopic data was 0.05 V, and each voltage step was lasting 1000 µs (T-Raster). In addition, the period of the laser which was switched on and off was 5000 µs (T-Laser) when the voltage was ramped in STS measurements. Therefore, when the electronics control system was combined with the laser modulation, the experimental tunneling current (I) as a function of the sample bias (V) was shown in Figure S1(b). In Figure S1(b), the tunneling current (colored by blue) was varied with the laser on/off. The dashed green curve in Figure S1(b) indicates the I- V curve in dark, and the solid green curve shows that with light illumination. The results of I-V curves in dark and with light illumination therefore provide us for investigating the light illumination effect on the photocarriers generation behavior in perovskites. 2
Supporting information 2 2. Material characterizations Figure S2 showed the x-ray diffraction pattern of PbI2 thin film and perovskite CH3NH3PbI3 thin films deposited using sequential deposition process. The compositional nonuniformity in perovskite thin films containing CH3NH3PbI3 and PbI2 was clearly observed with the strong reflections at 14.2 and 28.5 assigned to the (110) and (220) planes of the tetragonal perovskite lattice and the reflection at 12 assigned to the (001) plane of the PbI2 lattice, indicating the presence of PbI2 in the thermally annealed CH3NH3PbI3 perovskite thin film. Figure S2: X-ray diffraction (XRD) characterization of the perovskite thin film after thermal annealing. Figure S3 presented the optical absorption spectra of PbI2 and CH3NH3PbI3 thin films. A very pronounced absorbance difference at wavelength of 532 nm was observed which indicated the excitation at 532 nm on the sequentially deposited CH3NH3PbI3 thin films was mostly absorbed by perovskite region rather than by PbI2 parts. Figure S3: Absorption spectra of CH3NH3PbI3 and PbI2. 3
Figure S4 showed the cross-section image of CH3NH3PbI3 thin films deposited on Si substrate used for LM-STM measurement. An image contrast corresponding to 300 nm SiO2 was clearly observed below TiO2 mesoporous structure and TiO2 compact layer. This insulating ~ 300 nm-thick SiO2 layer was used to prevent the photocurrent contribution from Si substrate to the STM tunneling current. Figure S4: Cross-sectional scanning electron microscope image of CH3NH3PbI3 perovskite solar cell with the architecture consisting of Si substrate/sio2/ch3nh3pbi3 fabricated using the conventional sequential deposition processes. 4
Supporting information 3 3. Surface roughness effect of perovskite films in STM measurements Figure S5(a) and S5(c) show the STM topography image and the corresponding analysis of the height profile across several perovskite grains indicated by the white dashed line in Figure S5(a), respectively. The numbers 1, 2, and 3 indicate the valley positions due to the height difference among grains when we analyzed the height profile across grains. The corresponding electronic di/dv image and the analysis of the electronic di/dv signals across grains are shown in Figure S5(b) and S5(d), respectively. According to the results in Figure S5(d), those valley positions among grains which are numbered 1, 2, and 3 give rise to the undetectable electronic signal in STS measurements, which were regarded as the signal from the background, as displayed by black in Figure S5(b). To identify the distribution between CH3NH3PbI3 and PbI2 in perovskites is of great importance in the work. In the present STM measurements, we determined the distribution between CH3NH3PbI3 and PbI2 based on their individually distinct electronic properties. According to the result shown in Figure S5(d), the significant discrepancy of electronic characteristics between CH3NH3PbI3 (colored by red) and PbI2 (colored by blue) in grain A and grain B overwhelm the surface roughness effect of grains. The variation of the electronic di/dv signal due to the surface roughness in CH3NH3PbI3 is much less significant compared to those between CH3NH3PbI3 (colored by red) and PbI2 (colored by blue). Thus, the distribution of the respective constituents at the outer part of grains and grain interiors can be clearly distinguished based on their individually distinct di/dv electronic signal in the present work. Figure S5: (a) A typical STM topography image in perovskites. (b) The corresponding normalized di/dv image. Analysis of (c) the height profile and (d) the normalized di/dv profile across the perovskite grains. 5
Supporting information 4 4. Electronic properties of perovskites under light illumination Figure S6 shows the experimental di/dv tunneling spectroscopy on perovskite surface in dark (dashed curve) and under illumination (solid curve) situations. The symbol ED is denoted as the separation between the conduction band edge and the Fermi level in the bulk in dark. In addition, it is worth noting that the tunneling current at the negative sample bias was significantly increased with light illumination, indicating an increased carrier density of photoinduced holes in the CH3NH3PbI3 parts of the perovskite grains. More detailed discussions are addressed in the following section to explain the increased carrier density of photoinduced holes. Figure S6: Representative di/dv curves of perovskites in dark (dashed curve) and under light illumination (solid curve) condition. Figure S7 shows the conduction (Ec) and valence (Ev) band-edge as a function of the distance from the perovskite s surface when the whole perovskite-based system is applied at the negative sample bias (Vs= -1.5 V). The n-type perovskite surface exhibits a Fermi level in the bulk of the perovskite system (EF) is close to the conduction band edge as shown in Figure S7. In addition, the tip with its Fermi energy at (EF + evs) is shown on the left side when the sample bias (Vs= -1.5 V) is applied. The dark (light) gray areas represent the filled (empty) states of the system. Dashed lines show the system s band edges without illumination, while solid lines correspond to those under illuminated situation. Under illumination, the surface band bending is reduced. 6
Figure S7: When the sample is applied at the negative sample bias (Vs= -1.5 V), the tunneling current is mainly from the valence band of the sample into the tip. When the perovskite surface was illuminated, the band bending is reduced by the generation of photoinduced carriers. The positions of the band edges are shown by solid lines in Figure S7. Due to the reduction of the band bending under illumination, the tunneling current from Iv is increased when the sample bias is at the negative sample. The result therefore suggests that the increased tunneling current at the negative sample bias is a result of an increased carrier density of photoinduced holes in the CH3NH3PbI3 parts of the perovskite grains. 7
Supporting information 5 5. Stability of LM-STM measurements 1) We discuss first the consistence and the repeatability of STS data in LM-STM measurements. The consistence and the repeatability of the data can be examined by consecutive images which are shown in Figure S8. Figure S8 shows the consecutive distribution images of the E D (E D = E C - EF) derived for the perovskite grains under illumination. The consecutive distribution images of the E D(1) and E D(2) in Figure S8 (a) and S8 (b) are images with the time interval of ~2 hours. Figure S8 (c) shows the energy differences between E D(1) and E D(2). The average of the energy differences (E D(1) - E D(2)) in Figure S8 (c) is ~ 0.04 ev, which is around the resolution of the electronic spectroscopic. It suggests that the light intensity doesn t induce critical migration in the present work. Due to the stability of the measurement, the energy band edge positions are not significantly affected under light illumination. Figure S8: Consecutive images of the spatial distribution of E D values derived for the perovskite grains at (a) t = t 0 and (b) t = t 0 + 2 hours under illumination. (c) A mapping image of the energy differences between E D(1) and E D(2). 8
2) We discuss the influence of the height variations in grains on the resulting spectra in STS measurements. To avoid the possible confusion due to the comparison of large height variations in grains, we discuss the issue in terms of grains with the same height, such as the grain A (~ 37nm in height) and A (~ 36nm in height) in Figure S9. The comparison of the same height gradient in grains is suitable to discuss the influence of the height variations on the resulting spectra in STS measurements. Figure S9 (a) and S9 (b) show the STM topography and electronic di/dv images, respectively. The corresponding analyses of the signal profile across several perovskite grains indicated by the white dashed line in Figure S9 (a) and S9 (b) are shown Figure S9 (c) and S9 (d), respectively. Those valley positions among grains which are numbered 1, 2, and 3 give rise to the undetectable electronic signal in STS measurements, which were regarded as the signal from the background. Figure S9: (a) A typical STM topography image in perovskites. (b) The corresponding normalized di/dv image. Analysis of (c) the height profile and (d) the normalized di/dv profile across the perovskite grains A and A. The grain A and A have the same topographic height in Figure S9 (a). However, referring the di/dv image in Figure S9 (b), the large height gradient of grain A didn t cause a thick PbI2 layer as observed in grain A. The observation implies that a larger height roll-off at the edges doesn t benefit a sufficient amount to affect the STS data at the edges. The geometrical height variation in grains is not a critical factor to impact on the resulting STS spectra. The individually distinct electronic property in grains is the main factor to distinguish the distribution between perovskite and PbI2. The significant discrepancy of electronic characteristics between CH3NH3PbI3 and PbI2 in grains overwhelm the height gradient in grains. 9
3) Next, we discuss if the light-induced thermal expansion causes a relatively large change in STS data for a grain with a sharp height gradient. To discuss if the light-induced thermal expansion causes a relatively large change in STS data for a grain with a sharp height gradient, we refer the comparison of the energy difference ED (ED = EC - EF) in perovskite grain A and A in dark and with light illumination, as shown in Figure S10 (a) and S10 (b), respectively. Figure S10: Imaging of the spatial distribution of the E D and E D values derived for the perovskite grains across the grain A and A (a) in the dark and (b) under illumination. The grain A and A have the same height in the topographic image (refer Figure S9 (a)). The thickness of the PbI2 layer in grain A and A is ~ 20 nm and 60 nm, respectively. Under light illumination, a grain with a sharp height gradient doesn t benefit to a sufficient amount to affect the STS data. The thickness of the PbI2 layer in grain A and A keeps constant in STM measurements even though the light is illuminated. The result fairly suggests that the lightinduced thermal expansion doesn t cause a relatively larger change in STS data for a grain with a sharper height gradient. The electronic control system efficiently reduces the thermal effect in our STS measurements with light illumination. The light-induced thermal effect of the tip can also be estimated based on the general relationship between the thermal expansion and the resulting tip movement, where, ln 2. (,,,,, denote the coefficient of thermal expansion, the laser power, absorption coefficient, heat conductivity of the tungsten tip, the effective thickness of the heated spot, and the spot radius of the light.) [Ref.: Grafström, S.; Schuller, P.; Kowalski, J.; Neumann, R. J. Appl. Phys. 1998, 83, 3453-3460.] Based on the numerical evaluation, the estimation of the thermal expansion of the tip is ~ 0.1 (nm/mw). 10
Supporting information 6 6. Surface photovoltage (SPV) information of the perovskite solar cells In the LM-STM measurement, we illuminated the sample with light of the intensity indicated by the black dashed line in Figure S11. Figure S11: Light intensity dependence of SPV measured on the perovskite region (V S = +3.0 V). Figure S12 shows the I-V spectra in dark (dashed red curve) and under light illumination (solid red curve). The surface photovoltage (SPV) information can be obtained from the illuminated/dark I-V curves in Figure S12 (a) and S12 (b) shows the SPV as a function of the bias voltage in the present perovskite system. Figure S12: (a) The experimental tunneling current (I) as a function of the sample bias (V) when the electronics control system was combined with the laser modulation. (b) The SPV spectrum derived from (b). 11