Prospectively of Carbon-Doped Indium-Tungsten-Oxide Channel TFT for Bias Stress Instability

Abstract

Introduction InOx-based thin film transistors (TFTs) such as In-Ga-Zn-O, In-Ti-O, In-W-O, and In-Si-O have been widely investigated. We also reported carbon doped In-Si-O (In1-xSixOC) as channel material because of the highest bond dissociation energy of 1076 kJ/mol between carbon and oxygen. The turn on voltage shift (ΔV on) of In0.80Si0.20OC (C: 0.9 at. %) and In0.79Si0.21O TFT were 0 V and -0.2 V, respectively. The ΔV on of the In0.80Si0.20OC (C: 0.9 at. %) TFT was improved 0.2 V compared to that of In0.79Si0.21O TFT under positive gate bias stress (PBS) due to excess electron trapping reduction. Although In0.80Si0.20OC showed a good stability, the base In0.79Si0.21O TFT shows a low mobility of 7.7 cm2/Vs. To archive both high mobility and reliability, we investigated carbon doped In-W-O (In1-xWxOC) TFT because the In0.98W0.02O has a high mobility (> 39.0 cm2/Vs). In this paper, we studied transistor properties of In1-xWxOC TFT. We also discuss effect of carbon in In-W-O by comparing the stability between In1-xWxO and In1-xWxOC TFTs. Experiment The In1-xWxOC TFTs were fabricated on p++-Si (0.00099 Ω·cm) with 300-nm-thick thermal-oxide layers. 10-nm-thick In1-xWxOC films were formed on the substrates through a stencil shadow mask at room temperature by co-sputtering using In2O3 and WC targets under an Ar/O2 atmosphere at 0.2 Pa with PO 2 = 0.08 Pa. The XW of In1-xWxOC films were controlled in the range of 0.02 ~ 0.06 by changing the sputtering power for each target. Then, the post deposition annealing (PDA) was performed at 300 °C for 60 min in air. Finally, Au (100 nm)/Ti (10 nm) layer as source and drain electrodes were deposited by thermal evaporation method. The channel length and width of the TFT used in this study were 350 and 1000 μm, respectively. Results and Discussion The carbon content and chemical bond state of In1-xWxOCfilms were examined by X-ray photoelectron spectroscopy (XPS) measurements. The atomic ratio of carbon in In0.98W0.02OC, In0.96W0.04OC, and In0.94W0.06OC films were 0.3, 0.4, and 1.2 at. %, respectively. Moreover, the C1s peaks at 285 and 290 eV are observed after PDA, and attribute to W-C and O-C=O bonds in the In1-xWxOC films, respectively. Figure 1 (a) and (b) show comparison of the ΔV on of the In0.96W0.04OC and the In0.98W0.02O TFTs as a function of stress time under PBS and NBS conditions, respectively. The subthreshold swing (S.S.), on/off current ratio (I on/I off), threshold voltage (V th) and saturation field-effect-mobility (μ sat) values of In0.96W0.04OC TFT were 0.52 V/decade, 1.0×109, 0.4 V and 5.6 cm2/Vs, respectively at the initial state. The ΔV on in the In0.96W0.04OC TFT was 4.4 V after the PBS application of 10800 s (Fig. 1(a)). The ΔV on of the In0.96W0.04OC TFT was always smaller than that of In0. 98W0. 02O film under PBS. This indicates that trap state in In1-xWxOC channel were suppressed by introducing carbon. On the other hand, it is clear that the ΔV on (~ 0 V) of In0.96W0.04OC TFT was improved significantly compared to that (-0.5 V) of In0.98W0.02O TFT under the NBS of 5000 s. It was reported that the large ΔV on under NBS is due to the abnormal VO formation in oxide channel for GIZO TFT [7]. Therefore, this indicates that excess abnormal VO formation in In1-xWxOC channel is suppressed by introducing carbon. Conclusions We succeeded to introduce carbon (0.3-1.2 at. %) into In-W-O by co-sputtering WC and In2O3 targets. The S.S., I on/I off, V th and μ sat values of In0.96W0.04OC TFT were 0.52 V/decade, 1.0×109, 0.4 V and 5.6 cm2/Vs, respectively. The ΔV on of In0.96W0.04OC TFT was suppressed compared to that of In0. 98W0. 02O film under PBS. The ΔV on of In0.96W0.04OC TFT was not changed stress time up to 5000 s. The In0.96W0.04OC TFT is useful to reduce ΔV on under both the PBS and the NBS.