First, to prepare the SnO2-based gasistor, the Si substrate was cleaned using acetone, methanol, and deionized water for 15 min in the same sequence. The Pt bottom electrode (BE), which was 50 nm thick, was deposited on the cleaned Si substrate using a radio frequency (RF) sputtering system. The SnO2, which was 50 nm, was then deposited on the Pt/Si substrate using an RF sputtering system. The SnO2 deposition was performed at a 100 W sputtering power in an Ar ambient at a base pressure of ~ 20 mTorr and a working pressure of ~ 5 mTorr. Subsequently, the Ti top electrode (TE) was deposited on SnO2/Pt/Si substrates using RF sputtering system. In order to evaluate the resistive switching (RS) characteristic, we measured the electrical characteristics of the SnO2-based gasistor using a Keithley 4200-Semiconductor Characterization Systems (SCS) and a high-frequency capacitance–voltage (C–V) meter. To demonstrate the gas sensing capability of the SnO2-based gasistor at room temperature, the current of the SnO2-based gasistor was monitored while inducing the isopropyl alcohol gas. The SnO2-based gasistor was placed onto the ground plate inside of the gas sensing chamber and electrically connected to a Keithley 4200-SCS. The humidity was maintained under a 65%RH. The isopropyl alcohol gas was injected through a pipe to the gasistor. The concentration of isopropyl alcohol gas was controlled to 10, 20, and 30 ppm. Then, the current at a read voltage of 0.13 V was measured while inducing the isopropyl alcohol gas.
A schematic of the fabricated SnO2-based gasistor with a Ti/SnO2/Pt/Si structure is described in Fig. 1a. the SnO2-based gasistor basically had a metal–insulator–metal structure in this paper. As shown in Fig. 1b, the Ti, SnO2, and Pt layers with a 50 nm thick were identified. So as to evaluate the material properties, we measured the X-ray diffraction (XRD) peaks, as shown in Fig. 1c. The pristine SnO2 films had amorphous characteristics, so the diffraction peaks of SnO2 were observed to be near 30° and 55°. To analyze the diffraction peaks in detail, we used JCPDS card to match the measured diffraction peaks of SnO2. As a result, the measured diffraction peaks were in a good agreement with the JCPDS card 41-1445. The diffraction peaks near 30° consisted of (110) and (101) planes of SnO2 and the diffraction peak near 55° was (220) plane of SnO2. The wide diffraction peaks were generally attributed with small crystal sizes . The average crystal size of SnO2 was identified following the Debye–Sherrer equation, which is R = 0.9λ/βcosθ, where R = the average crystal size, λ = is the wavelength of the X-ray, β is the full width at half maximum of the peak, and θ is the Braggs angle. The average crystal sizes for the (110), (101), and (220) planes of SnO2 were estimated to be 0.51, 1.41, and 0.4 nm, respectively.
The pristine SnO2 was initially an insulator, whereby by following the conducting filaments (CFs) forming process with a positive bias sweep from 0 to 2.7 V to the TE of SnO2-based gasistor, in direct current (DC) mode, the oxygen ions migrated toward the TE and left the oxygen vacancies, which were manifested. As the concentration of oxygen vacancies around the BE of the SnO2-based gasistor was above the concentration of the oxygen vacancy accumulation, the oxygen vacancies were reconstructed, which showed the formation of the CFs between the BE and the TE of the SnO2-based gasistor. Therefore, according to the formation of CFs, abrupt increase in the current was observed at 2.67 V, as shown in the inset figure of Fig. 2a, and it represents that the CFs that consisted of oxygen vacancies were formed, which is consistent with the previous reports. [11, 13,14,15]. In order to investigate the RS characteristics of the SnO2-based gasistor that has a Ti/SnO2/Pt structure with a 50-nm-thick SnO2 layer, a DC bias sweep was applied in a sequence that followed: 0 → − 1.3 V → 0 V → 0.8 V → 0 V. As a result, the SnO2-based gasistor showed a typical current–voltage (I–V) characteristic, because a bipolar RS behavior was observed by the DC bias sweep, which is shown in Fig. 2a. As the DC bias swept to a negative direction to − 1.3 V, an abrupt decrease in the current was manifested close to − 1.24 V, because the CFs were partially disrupted, which indicated that the resistance was changed from a low resistance state (LRS) to a high resistance state (HRS), which is called the reset process, as shown in step 1 and 2 steps of Fig. 2a. Conversely, as the DC bias was swept in a positive voltage direction from − 1.3 V with a compliance current of 100 mA, the current was drastically increased, because the partially disrupted CFs were formed again at 0.72 V, which represented that the resistance was changed from the HRS to the LRS, which is called the CFs forming process, as shown in 3 step and 4 step of Fig. 2a. The current ratio between the HRS and the LRS was greater than 10.53 at a read voltage of 0.13 V. In order to evaluate the long-term stability of the resistance state on the SnO2-based gasistor, the resistance state was measured for 104 s at a read voltage of 0.13 V, as shown in Fig. 2b. As a result, LRS was stably maintained above 104 s, which indicated that the CFs were stably kept during the gas sensing without any degradation or reaction in the air at room temperature.
Then, in order to evaluate the isopropyl alcohol gas sensing capability of the SnO2-based gasistor at room temperature, the current of the SnO2-based gasistor was monitored at a read voltage of 0.13 V while inducing the isopropyl alcohol gas. Figure 3a shows the change of the current density on the SnO2-based gasistor before the CFs forming process when 30 ppm of the isopropyl alcohol gas was induced into the gasistor. As a result, the resistance was not changed by the isopropyl alcohol gas. However, as CFs were formed in SnO2-based gasistor, the resistance change according to the concentration of isopropyl alcohol gas was observed. Figure 3b shows the transient response current against concentration of isopropyl alcohol. The pristine current in the air was about 4.05 mA, whereas the currents were decreased to 3.47, 2.4, and 2 mA for the concentration of 10, 20, and 30 ppm, respectively. To evaluate the gas sensing characteristics in detail, the sensitivity (S) was calculated as S = Rgas/Rair, where Rgas is the resistance of the SnO2-based gasistor in the target gas, and Rair is the resistance of the SnO2-based gasistor in the air. As shown in Fig. 3c, the sensitivities were 1.17, 1.69, and 2.02 for 10, 20, and 30 ppm of isopropyl alcohol, respectively.
On the other hand, the fast response/recovery time for the high concentration of isopropyl alcohol gas was essential for the gas sensor. The proposed SnO2-based gasistor represented a relatively fast response time that was below 4 s against all concentrations of the isopropyl gas sensor, whereas the response time for 30 ppm was relatively slower than the others. As shown in Fig. 3d, the recovery time for 30 ppm was above 6 s.
The decrease in the current with an increase in the concentration of isopropyl alcohol gas was caused by the reaction between the CFs consisted of the oxygen vacancies and the adsorbed isopropyl alcohol gas . Before inducing isopropyl alcohol gas into the SnO2-based gasistor, the resistance state of the SnO2-based gasistor showed LRS, because the CFs provided conductive paths for the electrons to flow, which is described in Fig. 4a. As 10 ppm of isopropyl alcohol gas was induced into the SnO2-based gasistor, the isopropyl alcohol gas was adsorbed at the surface of the SnO2 particles and the adsorbed isopropyl alcohol gas had negative charge because the adsorbed isopropyl alcohol was a reducing gas [16,17,18]. The adsorbed isopropyl alcohol gas, which had a negative charge, was then reacted with the oxygen vacancies, Vo2+, which resulted in the broken CFs that are shown in Fig. 4b. Therefore, the reaction was attributed to the broken CFs, so a decrease of the current was observed as isopropyl alcohol was induced. As the concentration of isopropyl alcohol was increased to 30 ppm, the current of the SnO2-based gasistor decreased more than 10 ppm of the isopropyl alcohol gas. A current decrease against the increased concentration of isopropyl alcohol was due to the destructed CFs being increased as the quantity of the adsorbed isopropyl gas was increased, which is shown in Fig. 4c. When the isopropyl gas flow was stopped, the adsorbed isopropyl alcohol that occupied the CFs was automatically removed, which resulted in a recovery to the original state. In particular, the proposed SnO2-based gasistor showed fast recovery/response time because the nano-sized CFs were changed even with a minor change in concentration of isopropyl gas, which resulting in a change of the gasistor.