Figure 1a shows the transfer characteristics of the TFT using a sputtered 200 nm-thick SiO2 layer as gate dielectric. The current on/off ratio and subthreshold swing are found to be ~105 and 83 mV dec−1, respectively. In the saturation region, the threshold voltage is found to be very close to zero at 0.06 V. An anticlockwise hysteresis with a small threshold voltage shift of −0.02 V is observed at a sweep rate of 15 mV s−1. The total leakage current is found to be less than 0.1 nA and the leakage current density is around 6.0 × 10−8 A cm−2. Figure 1b shows the output characteristics of the TFT. A typical linear region is observed at low drain voltages. At a gate voltage, V G, of 1 V, and a drain voltage, V D, of 1 V, a drain current, I D, higher than 2 μA is obtained.
Figure 1Characteristics of EDL TFTs. (a) Transfer characteristics of the TFT gated with SiO2 deposited at 85 W and an Ar pressure of 5 × 10−3 mbar on a glass substrate. The schematic cross-sectional view of the device is shown in the inset. The channel width and the channel length are 1.5 mm and 80 μm, respectively. (b) Output characteristics of the TFT. (c) Specific capacitance of the sputtered SiO2 layer as a function of frequency and the phase angles during the measurement.
Full size image
To demonstrate the stability and repeatability of device properties, the performance of ten TFTs fabricated in two batches at different times but under the same conditions were measured. Since the capacitance of electrolyte gate dielectric is a function of frequency (unlike conventional TFTs), we have chosen to analyse the transconductance g m rather than mobility itself. A detailed statistical analysis of on/off ratio, subthreshold swing, threshold voltage and transconductance at V G = 1 V and V D = 2 V with the average value and standard deviation bar is shown in Supplementary Figure S1. Devices numbered 1 to 5 were fabricated in one batch and devices 6 to 10 were fabricated in another batch. The electrolyte dielectrics were deposited with an RF power of 85 W and an Ar pressure of 5 × 10−3 mbar. According to the statistical results, the devices fabricated in different batches showed similar performance. The on/off ratios of the devices are all around 3 × 105 with a minimum value still higher than 1 × 105. The magnitude of subthreshold swing is always better than 100 mV dec−1 with an average value of 85 mV dec−1. The transconductance values of the devices are also close to each other, indicating good uniformity and reproducibility.
It should be noted that our sputtered SiO2 electrolyte layer is a new type of dielectric to enable an extremely low operating voltage. Conventional SiO2 insulator has a relative dielectric constant of 3.9, meaning that a layer of 200-nm-thick SiO2 insulator provides a specific capacitance of around 17 nF cm−2. However, our 200 nm SiO2 electrolyte exhibited a specific capacitance of about 300 nF cm−2 which is more than one order of magnitude higher. Indeed, our IGZO TFTs showed an operating voltage of 1 V, which is a drastic improvement from more than 10 V operating voltage of TFTs gated with conventional SiO2 dielectric24,25,26. This is also significantly better than that of IGZO TFTs gated with 200-nm-thick high-κ dielectrics, such as Ta2O5 (3 V)27, HfO2 (5 V)28 and ZrO2 (6 V)28.
Figure 1c shows the specific capacitance of the 200 nm-thick SiO2 film at frequencies ranging from 20 Hz to 100 kHz using an Al/SiO2/Al sandwich structure. The low phase angle, smaller than −45°, indicates the sputtered SiO2 layer remains capacitive up to 100 kHz (ref. 29). A maximum capacitance of 0.45 µF cm−2 is obtained at 20 Hz, which is 26 times larger than that of thermally oxidised 200 nm-thick SiO2 (17.3 nF cm−2). Such a high capacitance enables the ultra-low operating voltage of the TFT in this work. Moreover, unlike thermally oxidised SiO2, the capacitance of the sputtered SiO2 shows a strong frequency dependence which is similar to the gate capacitance of other EDL transistors17, 29, 30.
In order to explore the origin of such high capacitance, a series of sputtering conditions have been experimented during the deposition of the SiO2 layer. Figure 2a shows the transfer characteristics of three IGZO TFTs gated with 200 nm-thick SiO2 dielectrics sputtered at different Ar pressures of 1 × 10−2, 5 × 10−3, and 1 × 10−3 mbar, respectively. The sputtering power is fixed at 85 W. The TFT with the SiO2 layer sputtered at 1 × 10−3 mbar shows a much lower on-current compared with the values obtained by the other two TFTs. An anticlockwise hysteresis is observed for all three devices, indicting mobile ions in the dielectric layer31. The TFT with the SiO2 layer sputtered at 1 × 10−2 mbar also shows a small region of clockwise hysteresis at gate voltages higher than −0.5 V, indicating electron trapping at the dielectric/channel interface31. Previous studies indicate that protons are common mobile ions in solid-state EDL transistors29, 30, 32. As the deposition process in this work does not involve any noticeable source of protons, it is plausible that they are introduced by moisture in the ambient air. It has been reported that water might be able to diffuse into the sputtered IGZO layer33, 34. Since the size of a water molecule is only around 2.5 Å (ref. 35), it is reasonable to assume that there might be some water molecules at the surface of, or inside, the SiO2 layer if the structure of SiO2 is porous.
Figure 2Effects of difference sputtering conditions. (a) Transfer characteristics of the IGZO TFTs based on the SiO2 gate dielectric sputtered at 85 W with Ar pressures of 1 × 10−2, 5 × 10−3, and 1 × 10−3 mbar, respectively. (b) High resolution cross-section SEM images of the TFTs with the SiO2 gate dielectric sputtered at Ar pressures of 1 × 10−2 (left, Sample A), 5 × 10−3 (middle, Sample B), and 1 × 10−3 mbar (right, Sample C), respectively. (c) Cross-sectional HRTEM bright field micrograph of the SiO2 layer. The SAED image of the SiO2 film is shown in the inset. (d) Rutherford backscattering spectrum of the SiO2 electrolyte deposited on Al/Si substrate at 85 W with an Ar pressure of 5 × 10−3 mbar. (e) Transfer characteristics of the TFTs using 45 W-sputtered and 85 W-sputtered SiO2 as gate dielectric after one month in atmospheric conditions.
Full size image
With competitive price and timely delivery, Acetron sincerely hope to be your supplier and partner.
According to Thornton’s model, increasing deposition pressure shall result in a reduction of deposition rate and even a porous film structure36, 37. Thus it is important to analyse the microstructures of the sputtered SiO2 films. Figure 2b shows the cross-sectional SEM images of these three TFTs, corresponding to the three different sputtering pressures, 1 × 10−2 (Sample A), 5 × 10−3 (Sample B), and 1 × 10−3 mbar (Sample C), respectively (Full images can be found in Supplementary Figures S2–4). It is found that there are more granular-like structures in Sample A than those in Sample B. There are hardly any granular-like structures in Sample C when the sputtering pressure is the lowest which is as expected. A highly granular-like microstructure is capable of absorbing water molecules and desirable for proton conduction38, 39. The absorbed water molecules may be ionised into H+ and OH−. When a positive bias is applied to the gate electrode, protons are repelled to the dielectric/channel interface32. These repelled protons will induce a large number of electrons in the channel, thus forming an EDL at the dielectric/channel interface. Because of the lack of granular-like microstructure in Sample C, the SiO2 layer sputtered at the lowest Ar pressure cannot absorb a significant amount of water to form an EDL, resulting in a much lower TFT current as shown in Fig. 2a.
Figure 2c shows an HRTEM image of the SiO2 layer sputtered at 5 × 10−3 mbar, which is obtained in bright field mode where a darker region indicates a higher material density. The SiO2 film exhibits an amorphous structure with an inhomogeneous density distribution. The selected area electron diffraction (SAED) image in the inset of Fig. 2c only shows a diffuse halo without clear rings or spots, confirming the amorphous structure of the material. These results suggest that the sputtered SiO2 has a low-density network structure surrounding higher density regions. Such low-density structure may promote proton hopping between oxygen atoms in the SiO2 layer32, 40, 41.
For thermally oxidised SiO2 films, the Si/O atom ratio and mass density are around 1:2.1 and 2.25 g cm−3, respectively42, 43. However, the RBS spectrum, as shown in Fig. 2d, indicates that the Si/O atom ratio and mass density for the SiO2 films sputtered at 85 W with an Ar pressure of 5 × 10−3 mbar are 1:2.7 and 1.87 g cm−3, respectively. The low mass density confirms that the sputtered SiO2 film at high Ar pressures has a porous structure. The small value of the Si/O ratio indicates the existence of excess oxygen atoms, suggesting that there are a large number of hydroxyl groups or water molecules at the surface of, or inside, the SiO2 layer.
The power dependence of the SiO2 electrolyte has also been studied. Figure 2e shows the transfer characteristics of two TFTs gated with dielectrics sputtered at the same Ar pressure of 5 × 10−3 mbar but different RF powers of 45 W (low power) and 85 W (high power), respectively. According to Messier’s model, the kinetic energy of sputtered particles from the target is positively correlated to the RF power36, 37. As such, at the lower RF power (45 W), the sputtered particles will have lower energy to self-organise to form a denser film on the substrate. The turn-on voltage of the lower-power-sputtered device is indeed lower than the value obtained for the higher-power-sputtered device as shown in Fig. 2e.
As protons in the sputtered SiO2 film may be generated by ionised water molecules, testing the air stability of the SiO2 electrolyte TFTs may offer a deeper understanding of the EDL formation mechanism. As shown in Fig. 2e, the performance of the TFT with SiO2 electrolyte sputtered at the higher power (85 W) remains almost the same after one-month storage in ambient atmosphere. On the contrary, the transfer characteristic of the TFT with SiO2 electrolyte sputtered at the lower power (45 W) shows a significant degradation. However, the performance of the lower-power-sputtered device can be recovered after annealing for 1 h in N2 at 100 °C (shown in Supplementary Figure S5). The annealing treatment may remove the water molecules inside the electrolyte and restore the device performance. However, the exact mechanism that causes the very different ambient stabilities of the 45 W and 85 W devices is not clear and needs further studies. Figure 3a shows the transfer characteristics of the TFT with higher-power-sputtered SiO2 gate dielectric before and after treatment in dry N2 for 12 h. The drain current at a gate voltage of 2 V is 4.6 µA before the N2 treatment and slightly drops to 1.2 µA immediately after taking the device out of the N2 ambient. The drain current increases continuously after leaving the device in ambient atmosphere until the device regains its high performance after about 20 min. This confirms that the sputtered SiO2 electrolyte functions by absorbing water and suggests that protons are the mobile ions that are responsible for the formation of EDL layer.
Figure 3Stability of EDL TFTs. (a) Transfer characteristics of the TFT gated with 200-nm thick SiO2 sputtered at 85 W before (dashed line) and recovering from (solid lines) a treatment in dry N2 for 12 hours. (b) Transfer characteristics of IGZO EDL TFTs with and without a capping layer of 400 nm PMMA, before and after a dry N2 treatment for 12 hours.
Full size image
It is common for electrolyte-based TFTs to be sensitive to the environment. Such a property itself can be useful in developing environmental sensors44, 45. Furthermore, it is possible to apply a capping layer to control the environmental stability of the devices. Here, we deposited a layer of 400 nm PMMA as a capping layer on some of the devices, and compared their environmental stability before and after PMMA capping. As shown in Fig. 3b, a device without PMMA capping shows a much lower on current after a dry N2 treatment for 12 hours due to the reduction of protons or hydrogen ions in the electrolyte. The TFT was then placed in ambient atmosphere for one hour, which resulted in a recovery in device performance as shown by the green curve. After capping the TFT by spin coating a 400 nm-thick PMMA layer, the device was placed in the dry N2 chamber for 12 hours and measured again. No clear degradation was observed as indicated by the dashed green curve. Such an experiment confirms that adding a capping layer can improve and control the stability of our TFTs, and the useful method is most likely applicable also to other types of EDL devices.
For more sio2 sputteringinformation, please contact us. We will provide professional answers.