In this study, we demonstrate that appropriate post-annealing could make a-IGZO TFTs adequately stable. The stability of a-IGZO TFTs were graded by the threshold voltage shift during sequent ID-VG curve measurement.
Here, we defined the parameter, △VTH, to reflect the instability of a-IGZO TFTs.
∆VTH ≡ VTH 7th − VTH 1st (3-1)
△VTH is the difference of threshold voltages of the first and seventh transfer curve. In
this study, we roughly judged the adequate annealing by threshold voltage of less than
29
0.5 V. Smaller △VTH means more stable characteristic of IGZO TFT.
In this section, we compare five a-IGZO TFTs with various annealing conditions.
The various annealing conditions are un-annealing, furnace annealing at 300ºC in nitrogen atmosphere, furnace annealing at 350ºC in nitrogen atmosphere (commonly used in papers [45-49]), pulse laser annealing and UV lamp annealing. Figures 3.7, 3.8, 3.9, 3.10 and 3.11 show the seven times sequent transfer characteristics of five
various annealing respectively. As shown in Fig. 3.7, un-annealed a-IGZO TFT was quite unstable (ΔVTH=10.8V). The as-deposited a-IGZO TFT may has a lot of defect
states caused by disorder structure on the channel/dielectric interface, and the defect states are trapping centers for mobile electrons. The applied positive gate voltage will induce the trapping process. The reduction of mobile electron at one gate voltage results in lower drain current. Therefore a larger positive voltage is required to turn on the transistor [44]. The threshold voltage will achieve saturation after some times sequent ID-VG curve measurement. In fact, the saturated transfer curve could shit again by applying higher gate bias. The measurement result demonstrate that the as-deposited a-IGZO TFT is quite unstable.
Then two thermal annealing were compared, Fig. 3.8 and 3.9 show the effect of furnace annealing at different temperature in nitrogenatmosphere. In Fig. 3.8, the annealing condition involves temperature of 300 ºC, nitrogen flow rate of 10L/min
30
and period of one hour. Although the criteria like field effective mobility (9.54 cm2/Vs), threshold voltage (2.8V) and subthreshold swing (0.1 V/dec.) as shown in Table 3.1 are satisfied, the stability was still inadequate (ΔVTH=1.6V). Therefore, furnace annealing at 300 ºC is not an appropriate condition. In chapter 1, we have concluded that the temperature of more than 300 ºC was required to have stable a-IGZO TFT. To improve the stability, a higher temperature of 350 ºC was used to get more stable a-IGZO TFTs. As shown in Fig. 3.9, The threshold voltage shift of 350 ºC annealed IGZO TFT decreased in depth (ΔVTH=0.5V) as compared with the one treated with 300 ºC thermal annealing. The temperature of 350 ºC was used generally in published papers [45-49]. The parameters of a 350 ºC furnace annealed a-IGZO TFT is listed in table 3.1. All parameters presented in this study were extracted from the seventh ID-VG curve in sequent measurement to make objectively comparison.
In Figs. 3.10 and 3.11, the effect of pulse laser and UV lamp annealing was presented respectively. Sequentially probed transfer curves of pulse laser and UV lamp annealing present adequate repeatability within seven times measurements (ΔVTH=0.5V for pulse laser and ΔVTH=0 V for UV lamp). Other parameters such as mobility, subthreshold swing and on/off ration are accepted listed in Table 3.1. The stability of light annealed IGZO TFTs are comparable to the one of 350 ºC thermal annealing. Therefore, pulse laser and UV lamp annealing are valid post-annealing
31
methods. As shown in Fig. 3.12(a), the estimated penetration depth of light at 266nm wavelength in a-IGZO film was about 95 nm. At a wavelength of 266nm, the absorptance was roughly 30%, which shows that the pulse laser light was efficiently and uniformly absorbed throughout the entire a-IGZO film thickness, shown in Fig.
3.12(b). The pulse laser with wavelength at 266 nm could provide enough energy to realign the atoms in a-IGZO film and make the structure more order. The laser annealing process could keep low temperature for substrate. The UV lamp at wavelength of 172nm is also a feasible light source for annealing. Because the photon energy of 7.2 eV is larger than the bandgap of a-IGZO film. Pulse laser annealing is relatively quick process for single TFT as compared with furnace annealing. The laser exposure time on one single TFT is just few seconds. The furnace annealing requires the temperature of more than 300 ºC and a period of one hour to get comparably stable device. It can be extrapolated that the low temperature fabrication process will be demanded for future industry and the laser or UV lamp annealing is a potential instrument.
3-3 I
D-V
DDiagram
Output characteristic (ID-VD) diagram is an important tool to present the characteristics of a transistor. Figure 3.13 show four ID-VD diagrams for four various annealing methods by cyclic sweeping measurement with VD ranging from 0V to 20V.
32
Figure 3.13 (a) present the ID-VD diagram of as-fabricated a-IGZO TFT. Figures 3.13 (b), (c) and (d) belongs to the a-IGZO TFT treated with furnace annealing at 350ºC in nitrogen atmosphere, pulse laser annealing and UV lamp annealing, respectively. The ID-VD curves probed by cyclic sweeping could reflect both the adjustability of current and stability. The stability could be graded by the deviation between two cyclic curves.
As shown in Fig. 3.13 (a), the saturation current was relatively low as compared with the one of 350 ºC furnace annealed IGZO TFT that could be explained by large threshold voltage (VTH=11.1V) attained after sequent measurement. In Fig. 3.13 (b), 3.13 (c) and 3.13 (d), the curve slop in linear region is relatively slow maybe caused by non-negligible contact resistance. Although the threshold voltages of pulse laser and UV lamp annealed a-IGZO TFTs are small (＜0.3V) and stable, the saturation
currents in Figs. 3.13(c) and 3.13(d) are still relatively small as compared with the one in Fig. 3.13(b). That could be explained by the smaller field effective mobility and un-annealed contact. Because the metal electrode of source/drain contact could shadow the irradiation during annealing process, the region under electrode was un-illuminated that may cause non-negligible contact resistance. In fact, a-IGZO TFTs with light annealing include both pulse laser and UV lamp has degraded mobility compared with as-fabricated a-IGZO TFTs although the stability was improved. The mechanism of light annealing is underdeveloped.
33