Supporting Information for Flexible nonvolatile polymer memory array on plastic substrate via initiated chemical vapor deposition Byung Chul Jang, #a Hyejeong Seong, #b Sung Kyu Kim, c Jong Yun Kim, a Beom Jun Koo, a Junhwan Choi, b Sang Yoon Yang, a Sung Gap Im, *b and Sung-Yool Choi *a a School of Electrical Engineering, Graphene Research Center, Korea Advanced Institute of Science and Technology (KAIST), Daejeon 34141, Korea, b Department of Chemical and Biomolecular Engineering, Graphene Research Center, Korea Advanced Institute of Science and Technology (KAIST), Daejeon 34141, Korea, c Department of Materials Science and Engineering, Korea Advanced Institute of Science and Technology (KAIST), Daejeon 34141, Korea, # These authors contributed equally to this work. * Address correspondence to sungyool.choi@kaist.ac.kr, sgim@kaist.ac.kr S-1
1. Process of initiated chemical vapor deposition (icvd) based on gas-phase polymerization Figure S1 shows a schematic of the icvd process mechanism and the chemical structure of pv3d3. The gas-phase polymerization occurs as follows: 1) The vaporized initiator and monomer are injected into the vacuum reactor, 2) the initiators are pyrolyzed by means of the hot filament to form radicals for a radical polymerization reaction, 3) the monomer is adsorbed onto the cooled substrate, and 4) the free radicals are polymerized by collisions between the radicals and the adsorbed monomers. In this work, pv3d3 was polymerized from a monomer of 1,3,5-trimethyl-1,3,5-trivinyl cyclotrisiloxane (V3D3) for a polymer-based RRAM. The advantageous characteristics of the icvd process, such as low process temperature, surface-growth properties, and solvent-free processing, result in pv3d3 thin films that are highly crosslinked, uniform, and pure. Owing to these benefits, the pv3d3- RRAM is an ideal polymer-based RRAM that is free from the problems regarding uniformity and long-term stability of existing polymer-based memory. Figure S1. Schematic of the icvd process mechanism and the chemical structure of pv3d3. S-2
2. Bipolar resistive switching behavior of pv3d3-rram Figure S2. I-V characteristics of pv3d3-rram with BRS behavior. S-3
3. HRS and LRS as functions of the pulsewidth Figure S3. LRS and HRS as functions of the pulse width with amplitudes of 5 V for V SET and 2 V for V RESET. S-4
4. Long-term stability of pv3d3-rram under air environment Figure S4. I-V characteristics of pv3d3-rram after 18 and 27 days in an air environment. S-5
5. The origin of device failure after severe bending test To reveal the cause of device failure under a bending radius of 3 mm, we analyzed the failed device using a scanning electron microscope (SEM). As shown in Figure S5, we observed cracks on the metal electrodes of the failed device. The number of cracks on the Al bottom electrode was much higher than that on the Cu top electrode, as the bending radius was decreased. To investigate the effect of the fracture in the Al bottom electrode, we subjected the Al electrode deposited on a PES substrate to bending tests under different bending radii from 4.5 mm to 3 mm. Figure S6 shows that the crack density of the Al electrode increased as the bending radius decreased, and the formation of transverse cracks across the Al electrode occurred at a bending radius of 3 mm. Furthermore, as shown in Figure S5d, the resistance of the Al line electrode did not show significant degradation until the bending radius reached 3.8 mm. This result indicates that the device failure is determined not by the crack density but by the formation of transverse cracks across the Al bottom electrode. Figure S5. SEM images of the pv3d3-rram under bending radius of (a) 4.5 mm, (b) 3.8 mm, and (c) 3 mm. S-6
Figure S6. SEM images of the Al electrode under bending radius of (a) 4.5 mm, (b) 3.8 mm, and (c) 3 mm. (d) I-V characteristics of Al line electrode as a function of bending radius. S-7
6. Scheme for condunging C-AFM measurment Our C-AFM facility can apply voltage to chuck only. (not a tip) For conducting C- AFM measurement, pv3d3-rram with dot pattern with diameter of 200 µm were fabricated. After SET and RESET process, Cu electrodes were etched by Cu etchant and then silver paste was used to ensure contact between C-AFM metallic chuck and Al bottom electrode of pv3d3-rram. As shown in Figure S8, the scanning of current onto the pv3d3 films was executed with a scanning tip. Figure S7. Scheme for conducting C-AFM measurement with pv3d3-rram. S-8
7. Comparison of key parameters for polymer-based RRAM Device structure Endurance Retention Mechanism Reference Cu/pV3D3/Al 10 5 cycles 10 5 s Cu filament This work Ag/Chitosan/Pt 100 cycles 10 4 s Ag filament [1] Cu/pEGDMA/ITO 500 cycles 3 10 6 s Carbon filament [2] Ag/GQD/PVP/Ag 500 cycles 2.6 10 6 s Carbon filament [3] Cu/PEDOT:PSS/Al 150 cycles 3 10 6 s Cu filament [4] Ag/WPF-BT- FEO/highly doped Si 1000 cycles 10 4 s Ag filament [5] Table S1. Comparison of memory device performance based on various reported polymerbased RRAM devices. References 1. Hosseini, N. R.; Lee, J. S. Resistive Switching Memory Based on Bioinspired Natural Solid Polymer Electrolytes. ACS Nano 2015, 9, 419-426. 2. Lee, B. H.; Bae, H.; Seong, H.; Lee, D.-I.; Park, H.; Choi, Y. J.; Im, S. G.; Kim, S. O.; Choi, Y.-K. Direct Observation of a Carbon Filament in Water-Resistant Organic Memory ACS Nano 2015, 9, 7306-7313. 3. Ali, S.; Bae, J.; Lee, C. H.; Choi, K. H.; Doh, Y. H. All-Printed and Highly Stable Organic Resistive Switching Device Based on Graphene Quantum Dots and Polyvinylpyrrolidone Composite Org. Electron. 2015, 25, 225-231. 4. Wang, Z.; Zeng, F.; Yang, J.; Chen, C.; Pan, F. Resistive Switching Induced by Metallic Filametns Formation through Poly(3,4-ethylene-dioxythiophene):Poly(styrenesulfonate) ACS Appl. Mater. Interfaces 2011, 4, 447-453. 5. Cho, B.; Yun, J.-M.; Song, S.; Ji, Y.; Kim, D.-Y.; Lee, T. Direct Observation of Ag Filamentary Paths in Organic Resistive Memory Devices Adv. Funct. Mater. 2011, 21, 3976-3981. S-9