Hysteresis of Transistor Characteristics of Amorphous IGZO TFTs Studied by Controlling Measurement Speed

The a-IGZO TFTs in this work are based on the bottom-gate TFT devices with symmetrical source/drain (S/D) fabricated on the glass substrate. The shaped Ti/Al/Ti (50/200/50 nm) gate electrodes were capped with 400-nm-thick SiNx gate dielectric. The active layer of the TFT is 60 nm in thickness. The width and length are 20 μm and 5 μm, respectively. The detailed fabrication process is described elsewhere.¹⁴ Same TFT was used for all measurement. Before each measurement, TFT was put in vacuum for 24 hr and the transfer curve was checked to make sure its initial characteristics in dark is restored. Keithley 4200 semiconductor parametric analyzer is used to measure the electrical properties of the TFTs with controlled sweeping voltage waveform for the gate bias. The wavelength and the intensity of the light used for illumination are around 465 nm and 29000 lux (or 56.2 mW/cm²), respectively.

Fig. 1a shows the transfer curves of the a-IGZO TFT measured by SMU. The Vg is swept from −20 V to +20 V and then swept reversely from +20 V to −20 V with step of 0.5 V and drain voltage Vd of 10 V. The corresponding curves of simultaneously recorded time versus Vg are also plotted in the right axis of Fig. 1a. For the two time curves, the time taken in the measurement is automatically determined by SMU. It can be seen that, the sweeping speed in the on region for the two measurements is the same. It is further observed that, the step time in the off region is calculated to be 56 ms/0.5 V for the current range limit set at 100 nA, while it is 1003 ms/0.5 V for the current range limit of 100 pA. It takes longer time to measure the low drain current Id in the off region for the integral operation, even though it provides better resolution.

Zoom In Zoom Out Reset image size

We would rather correlate the hysteresis to the sweeping step time in the off region than current range limit. By setting the current range limit, we can manipulate the sweeping speed from 1003 ms/0.5 V to 56 ms/0.5 V. For each measurement, the duration time of forward Vg sweeping in the off region (−20 V to −5 V) T_off is recorded and the voltage difference ΔV at drain current of 1 nA is extracted from the Id-Vg curves to represent the degree of hysteresis. The curve of ΔV versus step time in off region is plotted the inset of Fig. 1a. It is obviously seen that the clockwise hysteresis ΔV positively relates to the sweeping time. By extrapolating the curve, we can expect ΔV to vanish when step time is shorter than 21 ms/0.5 V. Such a fast measurement cannot be implemented by SMU. We need to use Parameter Measure Unit (PMU) to do the fast sweeping. Using PMU, we have not only fast acquisition time but also the freedom to set the Vg waveform. Fig. 1a can be replicated in substance by using PMU to set the exact step times to be 1000 ms/0.5 V and 50 ms/0.5 V for Vg < −5 V and the step time of 25 ms/0.5 V for Vg > −5 V, as shown in Fig. 1b. We observed the similar behaviors in hysteresis. Thus, we take these advantages in the following experiments.

Fig. 2 shows the Id-Vg curves of the a-IGZO TFT obtained by two different modes of sweeping Vg along with their corresponding Vg waveforms in dark and under illumination. In order to increase the sweeping speed, we sacrifice the resolution for the off current by selecting the higher current range limit of 1 mA. In both modes of sweeping Vg, the speed in the range from −10 V to +10 V is identical to be 0.01 ms/0.5 V. As for the ranges of Vg < −10 V and Vg > +10 V, the sweeping speeds in Mode A and B are 0.01 ms/0.5 V and 1 s/0.5 V, respectively.

Zoom In Zoom Out Reset image size

As shown in Fig. 2, the time duration in the on region for Mode B is 40 s, much longer than that for SMU with slower sweeping. However, no significant positive V_TH shift is observed. It means that the effect of positive bias is less effective, which is consistent with the previous reports.⁷ Therefore, in the following discussion, we focus on the negative bias region.

The duration T_off in Mode A is 0.3 s. In the dark case, the absence of the hysteresis in Mode A is as expected. Under illumination, the hysteresis is significant even the sweeping speed is in the fast Mode A. It becomes more obvious in Mode B. The enhancement of hysteresis by illumination was also observed previously.¹¹

The longer T_off can possibly leads to both more charges trapped and more V_O²⁺ created under illumination. We wonder which one dominates the ΔV behavior. We know that the created V_O²⁺ cannot recover to be uncharged immediately.^9,11 If we do the successive measurements with very short interval, we can monitor how the accumulated V_O²⁺ affect the V_TH shift and ΔV. In the meantime, we can set different sweeping speed for each measurement to alter the ΔV for comparison. Therefore, a series of Id-Vg measurement with changing sweeping speed are arranged into sequence and conducted automatically by a preset program.

The results measured in the dark and under illumination are plotted in accordance to the sequence in Fig. 3. The step time in each measurement is the same and it is correspondingly indicated on the top of the figure. The series is designed to be in three stages. For the first 7 measurements, the step time is increased from 1 μs/0.5 V to 500 ms/0.5 V, which corresponds to the decrease in sweeping speed. Following them, the fast and slow ones are alternately arranged. Then, the step time is decreased back to 1 μs/0.5 V. For simplicity, the Id-Vg curves are not shown in this paper. Only the V_TH extracted at Id = 1 μA as well as the respective difference ΔV between the V_TH in forward and reverse sweeping to represent the hysteresis are plotted against the measuring sequence.

Zoom In Zoom Out Reset image size

It is observed that V_TH keeps shifting negatively along the measure sequence. We further observe that the V_TH shift in a zigzag way for both sweeping in the dark case. It reveals the opposite hole and electron trapping induced by the negative and positive bias during the measurement, respectively.^6,7

On the other hand, V_TH shift under illumination is enhanced, because light can induce much more V_O²⁺ to make V_TH negatively shift and enhance hole trapping efficiency.⁷ The reverse V_TH shifts more smoothly than the forward V_TH. It can be explained by V_O²⁺ overwhelm the positive V_TH shift owing to electron trapping.⁷ As for the forward V_TH, the shift get more apparent in the measurement with step time of 500 ms/0.5 V. It can be attributed to the hole trapping. To our surprise, the shift diminishes even in the contiguous measurement with the step time of 1 μs/0.5 V. This phenomenon cannot be explained by the accumulation of V_O²⁺. It suggests that the dominant mechanism for the V_TH shift in the forward sweep is either the hole trapping⁸ or the slow response of the donor-like states from V_O²⁺.¹³

As shown in the lower part of Fig. 3, for both dark and illuminated cases, ΔV is determined by the step time only, no matter how the sequence is arranged and how V_TH shifts. It further excludes the role of accumulation of V_O²⁺ in ΔV. The dependences of ΔV on the step time in the dark and under illumination are shown in the inset of Fig. 3, wherein the ΔV acquired in one-shot and sequence of measurement are both plotted. The step time for ΔV to become negligible is 10 ms/0.5 V in dark, which is faster than the best of 56 ms/0.5 V provided by SMU. Under illumination, for the same step time, ΔV is raised because of more holes to be trapped or more generated donor-like states.

In this paper, we prove the Vg sweeping speed plays an important role in the hysteresis, mainly due to the duration in the off region. Based on this new discovery, the dependency of the hysteresis on the step time is characterized. It is found that the response time of charge traps or donor-like states is in the range that overlaps with the conventional time step in the measurement. The effect of sweeping speed must be considered in the study of the hysteresis for IGZO TFT.

This work was support in part by the Ministry of Science and Technology under MOST 103-2221-E-009-061.