SUPPORTING INFORMATION Surface-Guided CsPbBr 3 Perovskite Nanowires on Flat and Faceted Sapphire with Size-Dependent Photoluminescence and Fast Photoconductive Response Eitan Oksenberg, Ella Sanders, Ronit Popovitz-Biro, Lothar Houben and Ernesto Joselevich * Department of Materials and Interfaces and Chemical Research Support, Weizmann Institute of Science, Rehovot, 76100, Israel. *E-mail: ernesto.joselevich@weizmann.ac.il Contents: 1. Methods and synthesis scheme (Figure S1). 2. HRTEM of guided CsPbBr 3 on C-plane sapphire (Figure S2) S1
3. Powder XRD measurements (figure S3) 4. Nanowires grow along different numbers of nanogrooves along their growth axis (Figure S4) 5. Half unit cell peaks in selected area FFT (Figure S5) 6. Intensity-dependent PL measurements (figure S6) 7. Optical images of photodetectors base on guided CsPbBr 3 NWs (figure S7) 8. FETs based on guided CsPbBr 3 NWs (figure S8) 1. Methods CsPbBr 3 guided nanowires synthesis. The synthesis system is illustrated in the figure S1 in the Supporting Information. The non-catalyzed growth was carried out in a three-zone horizontal-tube furnace. The quartz tube reactor was purged with a N 2 (99.999%, Gordon Gas) and H 2 (99.99995%, Parker Dominic Hunter H 2 -generator) 7:1 mixture and kept 300 mbar with a constant 400 sccm flow of the N 2 /H 2 mixture. The flat sapphire substrates (Roditi int.) did not undergo any preparation process. For the precursor CsBr and PbBr 2 powders (both purchased from Sigma-Aldrich) were mixed in a stoichiometric ratio and heated at 390 C for 20 min in the same N 2 /H 2 atmosphere. This precursor was held at 550 C in the first heating zone of the furnace, while the samples were placed downstream in the second heating zone and held at 350-390 C. After a 15 min growth period, the furnace was moved away and the source and sample could be rapidly cooled down to room temperature. S2
Figure S1. Synthesis system illustration: A quartz tube within a 3 zone furnace the samples are placed 20 cm downstream to the CsPbBr 3 precursor. Structural and elemental characterization. The nanowires were imaged with a scanning electron microscope (Supra 55VP FEG LEO Zeiss). In order to analyze the crystallographic structure and orientations of the nanowires and substrate, a focused-ion beam (FIB, FEI Helios 600 Dual Beam microscope) was used to cut thin (50-100 nm) lamellae across or along the nanowire, which were later inspected under a high-resolution transmission electron microscope (HRTEM, FEI Tecnai F30). The HRTEM images were analyzed using fast Fourier transform (FFT) from selected areas across the nanowire, and the FFT peaks were fitted to the crystallographic tables of CsPbBr 3 and sapphire. To perform a non-destructive compositional analysis we used STEM EDS measurements on a FEI Tecnai F-20 microscope with 120 kev acceleration voltage and 1nA beam current. 90 EDS spectra that were taken from an area of about 70 nm 2 in the center of the NW. The spectra were summed in order to improve the signal to noise ratio and to achieve a quality quantification. The composition quantification was done using the L lines of the elements and a Gatan tem imaging and analysis software. Optical characterization. Photoluminescence (PL) measurements were done using a micro- Raman/micro-PL system (Horiba LabRAM HR Evolution). A 325 nm He-Cd laser was S3
focused on the nanowire through a reflective 70X objective lens. The PL signal was collected using the same objective and sent to a 300 lines/mm grating and an EMCCD camera. 2. HRTEM of guided CsPbBr3 on C-plane sapphire Figure S2. Guided CsPbBr3 NWs on C-plane sapphire. (A) SEM image of Guided CsPbBr3 NWs that grow in triangular network (B) HRTEM images that showcase lattices of the lead halide perovskite and the sapphire substrate. 3. Powder XRD measurements X-ray diffraction (XRD) pattern for phase identification was obtained using TTRAX III (Rigaku, Japan) theta-theta diffractometer equipped with a rotating anode Cu anode operating at 50 kv and 200 ma and with a scintillation counter as detector. The measurements were carried out in reflection geometry using Parallel Beam optics based on a multilayered mirror (CBO attachment, Rigaku). XRD spectra were collected without diffracted beam collimating slits in the asymmetric 2-theta scanning mode with the incident angle fixed at 1 degree, and with a step size of 0.025 degrees. The scan rate was 0.5 deg/min. The divergence of the S4
diffracted X-ray beam was limited in the diffraction plane by a 0.114 Soller slit. The divergences of both the incident and diffracted beams were limited in the perpendicular direction by 5 Soller slits. A Ni filter was used to reduce the intensity of the Kβ line and of the white X-ray spectrum. Qualitative phase analysis was made using the Jade 2010 software (Materials Data, Inc.) and PDF-4+ 2016 database (ICDD). We find that the cubic and orthorhombic phases coexist in our sample (figure S3). This is not surprising because the orthorhombic phase is the most stable one at room temperature in the bulk, and our samples do not only contain NWs, but also other CsPbBr 3 crystal shapes, such as platelets, pyramids and a few larger crystals without well-defined shapes, which are also produced during the growth process of the NWs. Since this measurement samples large areas that contain multiple crystals and shapes, we can see different coexisting phases. There are some added uncertainties in our measurement compared to regular powder XRD. Our substrate is a single crystal so some random substrate reflections can appear in the spectrum, and at least some of the CsPbBr 3 crystals (the NWs) are oriented in a non-random manner on the substrate, which can change the peak intensities that are expected in such a diffractogram. Since this measurement samples large areas that contain multiple crystals with varying shapes and sized, at this point this experiment cannot provide additional data on the crystal phase of the NWs. S5
Figure S3. Powder XRD measurements: experimental results (black) along with calculated spectra of the Orthorhombic (blue) and cubic (Red) phases. 4. Nanowires grow along a different number of nanogrooves as they grow. During the structural characterization of CsPbBr 3 NWs that grow along nanogrooves on annealed M-plane we noticed that some of the NWs do not keep a constant width and the expand or reduce their width by changing the number of nanogrooves they grow along in a discrete manner as can be seen in figure S4. We attribute this intriguing behavior to the dynamic nature of CsPbBr 3 S6
Figure S4. Guided CsPbBr3 NWs that follow a varying number of nanogrooves. (A) SEM image of Guided CsPbBr3 NWs that grow along nanogrooves on annealed M-plane sapphire and (B) higher magnification images that showcase the change in the width of the NWs as the grow along a different number of nanogrooves. 5. Half unit cell peaks in selected area FFT. We note that for almost all of the NWs that were analyzed, both epitaxial and graphoepitaxial, we found peaks in the selected area FFT that correspond to lattice distances that match half a unit cell (Figure S5). The origin of the half unit cell peaks is most likely either phase domain boundaries or Ruddlesden Popper domains within the CsPbBr3 NWs. Both phenomena are well known for oxide perovskites, and both were observed at least once in CsPbBr3 structures.27-28 In order to determine which structural phenomenon is present in our CsPbBr3 NWs, we intend to perform an atomic level aberration-corrected STEM analysis.29 S7
Figure S5. Half unit cell peaks in selected area FFT. (A) TEM image of a cross-section of a guided CsPbBr 3 NW that grow along nanogrooves on annealed M-plane. (B) Higher magnification image its expected FFT (C) and the actual FFT (D) with lower intensity peaks that match half a unit cell distance. S8
6. Intensity-dependent PL Figure S6. Intensity-dependent PL on a small (91 nm-nw (c) in figure 3c-b) and a large (997 nm-nw (i) in figure 3c-b) CsPbBr 3 guided NW. Photoluminescence spectra were taken under different illumination intensities presented as percentages of 10 5 W/cm 2 irradiance using a 325 nm UV laser. 7. Optical images of photodetectors base on guided CsPbBr 3 NWs We exploited the deterministic control over the growth direction of the CsPbBr 3 NWs (figure S7A) and fabricated multiple devices using only a single fabrication step (figure S7B-C). We used a custom made shadow mask (Suron) with a 10 µm channel and used electron-beam deposition to fabricate C/Ti electrodes (10 nm/100 nm). Amorphous carbon was evaporated using a carbon evaporator fiber in an Edward evaporator. The number of NWs that cross the channel between the electrodes are determined by the density of the guided NWs, their length and the channel location imposed by the shadow mask. A typical photodetector has between S9
1 to 10 relatively large CsPbBr 3 guided NWs that cross a 10-20 µm gap between the electrodes (figure S7C). Figure S7. Optical images of photodetectors based on CsPbBr 3 NWs on annealed M plane sapphire. (A) Optical image of green photoluminescence of arrays of aligned NWs. (B) Multiple devices fabricated randomly with metal deposition through a shadow mask, and (C) An optical image of a single 8. FETs based on guided CsPbBr 3 FET were fabricated on the basis of the photodetectors using atomic layer deposition (ALD) to deposit a 50 nm dielectric layer of Al 2 O 3, followed by fabrication of Cr/Au (5/50 nm) gate electrode using standard photolithography and e-beam evaporation (figure S8 A-B). We find that the devices are gate responsive and that they qualitatively exhibit a p-type behavior (figure S8C) However, after the ALD deposition we observed substantial hysteresis and temporal changes in the source drain current under constant biases with some devices (figure S10
S8D). In order to assess the behavior of such FETs in a quantitative manner, we must acquire a deeper understanding of the hysteresis phenomenon Figure S8. Electrical properties of guided CsPbBr 3 NWs on annealed M plane sapphire. (A) Optical image of a typical FET. (B) Illustration of the FET fabrications process: (1) randomly dispersed NWs on the annealed M plane substrate, (2) source-drain electrodes evaporation through a shadow mask, (3) deposition of a dielectric layer (Al 2 O 3 ) via ALD, (4) deposition of a gate electrode via photolithography. (C) Source-drain current vs. gate voltage for different gate voltages. (D) A hysteresis loop of a FET device (V gate =0). S11