Resistive Switching Mechanisms on TaO x and SrRuO 3 Thin Film Surfaces Probed by Scanning Tunneling Microscopy Marco Moors, 1# Kiran Kumar Adepalli, 2,3# Qiyang Lu, 3 Anja Wedig, 1 Christoph Bäumer, 1 Katharina Skaja, 1 Benedikt Arndt, 1 Harry Louis Tuller, 3 Regina Dittmann, 1 Rainer Waser, 1,4 Bilge Yildiz, 2,3 * and Ilia Valov 1,4 * 1 Peter Grünberg Institut, Forschungszentrum Jülich, Wilhelm-Johnen-Str., 52425 Jülich, Germany 2 Department of Nuclear Science and Engineering, and 3 Department of Materials Science and Engineering, Massachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge, Massachusetts 02139, United States 4 Institut für Werkstoffe der Elektrotechnik 2, RWTH Aachen University, Sommerfeldstr. 24, 52074 Aachen, Germany *corresponding authors, e-mail: byildiz@mit.edu, i.valov@fz-juelich.de # equally contributing authors Supplementary STM measurements on as-deposited TaO x films: The as deposited TaO x films exhibit poor electronic conductance in as-received state after DC sputtering. Higher negative tip voltages of at least -5 V are necessary for performing STM measurements. Significant charging effects are observed on as-received samples (as shown in Fig. S1) after scanning a small 50 x 50 nm 2 area with -8 V. I-V characterization of as-received samples exhibited a typical capacitive behavior. Annealing the sample under UHV conditions leads to a partial reduction of the TaO x layer (see XPS results), which eliminates charging effects and also reduces the minimum bias voltages for imaging by a factor of approx. 2.
Fig. S1: (a) STM topography image (500 nm x 500 nm, U Tip = -3.0 V, I T = 0.5 na) of an as deposited 3 nm thick TaO x film after scanning the inner area with U Tip = - 8.0 V and I T = 0.5 na; (b) STS at different positions as marked in (a) shows capacitive-like behavior; (c) STS I-V sweep on the TaO x surface after annealing in UHV. The tunneling current increases under negative tip bias leading to LRS (ON). Under reverse polarity a HRS (OFF) is observed. The typical hysteresis in tunneling current (or inversely, the tunneling resistance) is a characteristic of resistance switching. Fig. S1(c) shows the tunneling current versus voltage (I-V) sweep obtained by applying bias between the STM tip and the TaOx surface (x < 2.5). The curve shows an increase in tunneling current at negative tip bias values, and a decrease in current on reversal of the polarity (positive tip bias voltage), corresponding to the switching between LRS and HRS, respectively. While the collected I-V data are sensitive to the changes in the near-surface region of the TaOx films, the typical hysteresis in tunneling current (or inversely, the tunneling resistance) is a characteristic of resistance switching as observed for macroscopic I-V measurements in ReRAM devices.
XPS measurements on TaO x films: Fig. S2: Peak fitting results of the Ta 4f spectra of TaO x / Ta / SiO 2 under different conditions: (a) RT, as-received, (b) 200 C UHV annealing, (c) 400 C UHV annealing, (d) 500 C UHV annealing, (e) cooled down to RT, after 500 C UHV annealing, (f) after aging in air atmosphere for 2 weeks and (g) after UHV annealing and exposed to air for 10 min. Fig. S3: Quantitative results on the concentration of Ta(V), Ta(IV), Ta(II) and Ta(0) under different conditions from peak fittings of Ta 4f XPS spectra. Usually when a metal in a metal oxide is chemically reduced, the binding energy shifts towards a lower energy. The XPS peaks of Ta 5+ in Fig. 3 and Fig. S2
appears to be shifting towards higher energies after being annealed in UHV. However, a careful consideration of Ta 5+ peak positions of the raw data shows the following BE (binding energies) for different conditions: conditions position/ev 1 26.81 2 26.99 3 27.07 4 27.03 5 26.99 6 26.91 7 27 We observe the peak positions are very close, it is only +-0.2eV (from conditions 1 to 5) and the instrument uncertainty is about 0.1 ev. The change in Ta 4f is rather small, within the error bar and one cannot draw any valid conclusion from it. However, the big changes are evident in the lower binding energy side, namely Ta4+, 2+ and 0 (binding energies between 22 ev and 25 ev), which is a large contribution and compelling reason that stoichiometry of the film is changed. Fig. S4: O 1s XPS spectra under different UHV conditions.
Fig. S5: Peak fitting results of the O 1s spectra of TaO x / Ta / SiO 2 under different conditions: (a) RT, as-received, (b) 200 C UHV annealing, (c) 400 C UHV annealing, (d) 500 C UHV annealing, (e) cooled down to RT, after 500 C UHV annealing, (f) after aging in air atmosphere for 2 weeks, and (g) after UHV annealing and exposed to air for 10 mins. A shoulder peak at a higher binding energy was observed with decreasing intensity upon UHV annealing but not fully recovered on exposing to atmosphere.
Fig. S6: Quantitative results on the concentration of the O 1s spectra after fitting with two peaks for low and high binding energy (BE). The low binding energy is the known energy for TaO x lattice oxygen. STS measurements on SrRuO 3 films under vacuum conditions: The application of a static electric field as it is obligatory for tunneling spectroscopy measurements in absence of oxygen has a significant influence on the surface topography of the SrRuO 3 film as shown in Fig. S7. Performing a negative bias voltage ramp (potential polarity with respect to the sample) at a fixed tip position results in the formation of an irreversible crater structure, which might be due to a decomposition of the sensitive SrRuO 3 layer as a result of field-induced material evaporation. Bias ramping to positive sample voltages shows qualitatively the same result but in this case the effect is much weaker. Fig. S7: STM topography image (180 nm x 180 nm, U Tip = -0.5 V, I T = 1.0 na) of the SrRuO 3 surface after performing STS measurements under UHV conditions at three different positions using bipolar (1), negative (2) and positive (3) bias voltage ramps, respectively. The shown current values are not limited by a current compliance but outside the detection limit of the measurement setup. Area-wise STM switching on SrRuO 3 films at small scale:
Fig. S8: STM images (100 x 100 nm 2, V Tip = -0.5 V, I T = 50 pa) of a SrRuO 3 film measured (a,c) before and (b,d) after scanning the marked area (typically 5x5 nm 2 and 2x2 nm 2 area) with V Tip = +2.5 V and I T = 1.0 na. PEEM data of SrRuO 3 thin films Fig. S9 shows O 1s and Ru 3p 3/2 core level spectra obtained inside and outside of STM-switched regions on a 10 nm thick SrRuO 3 film. Unlike the single Ru signal around 463 ev the O 1s spectra exhibits two clear signals, a broad one around 532 ev and a sharper one at 529 ev. While the Ru signal remains identical after switching, the intensity of the O 1s signals obviously decrease. Hereby, the broad signal in the O 1s spectra is typical for hydroxide based adsorbates indicating a cleaning effect of the STM tip, while the sharper signal is probably related to the perowskite oxygen.
(a) (b) Fig. S9: Oxygen 1s (a) and Ruthenium 3p 3/2 (b) signal intensity obtained by PEEM measurements inside and outside the STM-switched areas at the surface of the SrRuO 3 film shown in Fig 7. TaO x sample preparation and electrode configuration: Fig. S10 shows the top and side view of the investigated TaO x device. Due to the insulating nature of the SiO 2 substrate a Pt pad has been deposited on one side to provide a metallic contact between the Ta bottom electrode and the STM sample holder. Fig. S10: Schematic drawing of the investigated TaO x samples: a) intersection, b) top view.