Supplementary Information All Nanocellulose Nonvolatile Resistive Memory Umberto Celano, Kazuki Nagashima, Hirotaka Koga, Masaya Nogi, Fuwei Zhuge, Gang Meng, Yong He, Jo De Boeck, Malgorzata Jurczak, Wilfried Vandervorst and Takeshi Yanagida 1
Supplementary Information S1 a ITO 0.02wt% cellulose nanofibers solution (10µl) b Thickness (nm) 500 400 300 200 100 Si substrate 0 0 5 10 15 20 25 30 Drop cycles (times) (a) Schematic illustration of drop casting method for fabricating the resistive-switching layer. 10 μl of cellulose nanofiber solution (0.02 wt%) was dropped onto ITO coated Si substrate, followed by drying at 40 C for 15 min. The drop casting was repeated until the desired thickness was obtained. (b) Obtained thickness of the resistive-switching layer as a function of number cycles of drop casting. 100 nm thickness was achieved by 3 cycles. 2
Supplementary Information S2 a as-is (0.02wt%) 10 times diluted 50 times diluted 100 times diluted 1µm 1µm 1µm 1µm 500 times diluted 1000 times diluted 5000 times diluted 10000 times diluted 1µm 1µm 1µm 1µm b Density of CNFs (µm -2 ) 200 150 100 50 0 1 0.5 0 10 2 10 3 10 4 10 5 densely packed 10-1 10 0 10 1 10 2 10 3 10 4 10 5 Dilution ratio (1:x) Density control of cellulose nanofibers. (a) Field emission scanning electron microscopy (FESEM) images of cellulose nanofibers onto Si substrate with various dilution ratio. Cr with 40 nm was deposited prior to the observation for avoiding the charging effect. Accelerating voltage of 3 kv was utilized for observation. (b) Number density of cellulose nanofibers as a function of dilution ratio. A single cellulose nanofiber was obtained over 1:1000 dilution ratio. 3
Supplementary Information S3 a b 20nm c 1nA 0nm 0A (a) Topographic atomic force microscopy (AFM) image of single cellulose nanofiber. (b) Magnified AFM image of projected area in (a). (c) conductive-afm (C-AFM) image of single cellulose nanofiber. All the measurement was performed onto Ag coated Si substrate. 4
Supplementary Information S4 a Single cellulose nanofiber device Before metal contact Support pad 2µm Cellulose nanofiber SiO 2 /Si Pt 1µm 500nm b Current (x10-13 A) 1 0.5 0-0.5 Under light Dark -1-2 -1 0 1 2 Voltage (V) (a) Left FESEM image of the nanogap test device consist of a single cellulose nanofiber. Pt was utilized for contacting with cellulose nanofiber to avoid the possible oxidization at electrode-cellulose nanofiber interface. Upper right image shows the cellulose nanofiber prior to the electron beam (EB) lithography. Lower right image shows the magnified nanogap device. The gap size is 100 nm. (b) Current-voltage (I-V) curves of single cellulose nanofiber device taken in 1x10-3 Pa at room temperature. The conductions under dark condition (blue) and light irradiation (red) were almost identical, indicating the negligible photo carriers in cellulose nanofiber. The resistance was over 100 TΩ, indicating the highly insulative property of cellulose nanofiber. 5
Supplementary Information S5 10-2 10-4 slope ~ 1.01 LRS Current (A) 10-6 10-8 10-10 Measured data Fitting curve slope ~ 1.02 HRS slope ~ 1.75 10-12 10-2 10-1 10 0 10 1 Voltage (V) Here we discuss the switching mechanism of the nanopaper memory. Among various models, the one based on the formation and the rupture of Ag conducting filaments [1,2] via the electrochemical redox process can appropriately explain our the resistive switching results in Ag/nanopaper/ITO structure. In this model, Ag are ionized near the positively biased electrode, and then the Ag + ions migrate into the nanopaper layer toward negatively charged counter electrode under the electric field. When the Ag + ions reach to the counter electrode, the Ag + ions are reduced to be Ag by receiving the electrons from the electrode. Such successive metallization process creates the Ag conductive filaments between the electrodes. The rupture of Ag filaments occurs through the ionization and the migration of Ag + ions under the opposite electric field polarity. To examine the conduction mechanism, we analyzed the I-V curves of HRS and LRS taken from 200 x 200 μm 2 sample. The data fitting indicated that the electrical conduction in LRS follows Ohmic law. This is 6
consistent with the Ag filament based conduction model. On the other hand, the conduction in high resistance state (HRS) follows a space charge limited current (SCLC) law, which is consisting of three parts, (i) the Ohmic region (I V) at lower electric field, (ii) the Child s law region (I V 2 ) at middle electric field range and (iii) the steep current increase region at higher electric field [3]. Ohmic conduction is derived from the thermally generated carrier in cellulose nanopaper memory. The Child s law and the steep current increase regions are associated trap-unfilled and trap-filled SCLC transport, respectively. Therefore, the main conduction mechanism in HRS is associated with the electric field and/or thermally driven electron hopping through the trap states in nanopaper layer [4]. 1. Yang, Y., Gao, P., Gaba, S., Chang, T., Pan, X. & Lu, W. Observation of conducting filament growth in nanoscale resistive memories Nat. Commun. 3, 732 (2012). DOI: 10.1038/ncomms1737 2. Valov, I. et al. Nanobatteries in redox-based resistive switches require extension of memristor theory Nat. Commun. 4, 1771 (2013). DOI: 10.1038/ncomms2784 3. Lampert, M. A. & Mark, P. Current Injection in Solids Academic Press: New York, NY, USA, (1970). ISBN: 978-0124353503 4. Lim, E. W. & Ismail, R. Conduction Mechanism of Valence Change Resistive Switching Memory: A Survey Electronics 4, 586-613 (2015). DOI:10.3390/electronics4030586 7
Supplementary Information S6 nanopaper memory a b c Ag electrode d DI water 1cm 1cm 1cm 1cm e Fresh sample After sonication 28kHz 5min propanol acetone ethanol DMF water Metal electrode Nanopaper f Electrode residual ratio (%) 120 100 80 60 40 20 acetone propanol ethanol DMF water 0 4 5 6 7 8 9 10 Polar index of solvent g electrode nanopaper polar solvent cellulose (a-d) Removal of Ag electrodes from nanopaper memory device by applying the sonication into DI water. (a) Prior to immerse the nanopaper memory device into DI water, (b) just after immersion, (c) 1 min after immersing without sonication, and (d) after 5 min sonication and dried up. (e) Examination of electrode sorting in various solution (propanol, acetone, ethanol, N,N-dimethylformamide (DMF) and water). The sonication was applied for 5 min. In water and DMF, the electrode was fully removed while it partially or fully remained in the other solvents. (f) Residual ratio of electrode (%) as a function of polar index of solvent. The electrode tended to be removed as the polarity of solvent increases. This trend can be understood by taking into account the 8
surface hydroxyl groups of cellulose nanofibers. Since the cellulose nanofibers are constructed by compiling the pyranose ring of cellulose through the hydrophobic interaction, many hydroxyl groups are exposed to the surface of cellulose nanofibers. The polar surface of cellulose nanofiber is familiar with the polar solvent, which allows the penetration of solvent underneath the electrode and/or into the nanopaper. (g) Schematic illustration for the mechanism of electrode removal in polar solution. 9
Supplementary Information S7 Natural soil 4L sampling in September at 34:49:30 north latitude 135:31:28 east longitude Temperature 22±2ºC Humidity 92±3% Plastic bag For the biodegradation experiment, we utilized the natural soil. The 4L of soil was collected in September at 34:49:30 north latitude and 135:31:28 east longitude and kept in a plastic bag. The temperature and the humidity of soil were controlled to be 22 ±2 C and 92 ±3 %, respectively, during the experiment. 10
Supplementary Information S8 a Day 0 c Day 36 Room air Humid-air Immersion into water 1cm 25ºC b 5ml 10ml 90ºC Not examined 1cm Since the biodegradation test in natural soil has several factors for the degradation of nanopaper such as humidity, temperature and bacterial (enzyme) effect, we conducted the biodegradation test without soil (bacterial effect). Figure S8(a) shows the nanopaper memory prior to the biodegradation test. In this study, we kept the samples (1) in room-air-condition (humidity 50-60%), (2) in humid-air-condition (humidity 100%) and (3) in water. Figure S8(b) shows the experimental set-up for the humid-air and water-immersion system. For the sample in humid-air-condition, 5 ml DI water was prepared in the 20 ml glass bottle and the nanopaper memory was set out of the water. For the sample in water, the nanopaper memory was immersed in 10 ml DI water. Figure S8(c) shows the results of biodegradation test obtained after 35 days. In water, the resistive-switching layer was fully removed with the electrode in several minutes, while the nanopaper substrate was fully maintained after 35 days. On the other hand, in humid-air-condition and in room-air-condition, both 11
of the resistive-switching layer and the nanopaper substrate were partially (humid-air-condition) and fully (room-air-condition) remained after 35 days, respectively. We further examined temperature effect in both humid-air and water condition. At 90ºC, there were no significant differences in biodegradation effect compared with the same test performed at room temperature. These clearly indicate that the biodegradation of our nanopaper memory was caused by a bacterial effect in natural soil. 12