Channels Poly-Si TFTs

The static hot carrier stress condition is determined at kink-effect occurrence, Vd

= 6 V and Vg = 3V, and the source potential is common. The dynamic pulse train stress is defined at constant Vd = 6 V and dynamic Vg = 3V (high), 0V (low) with the duty cycle of 50%, and the source potential is common. The waveform of the pulse train is shown in Fig.4.11. In this section, we discuss the deceive reliability after a series of stress frequency (f), rising time (tr), falling time (tf) and substrate temperature conditions.

Fig. 4-12 (a) depicts typical M10 poly-Si TFT Id

-V

-V

curves before and after DC hot carrier stress at 10000 second.

Figure 4-13 depicts trans-conductance degradation of S1 and M10 TFT as a function of the stress time with different frequencies (DC, f = 1K Hz, and f = 1M Hz).

The S1 and M10 TFT show the similar Gm degradation. These results reveal that the S1 and M10 have similar tail state generation.

Figure 4-14 and 4-15 depict the threshold voltage and subthreshold swing

variation of S1 and M10 TFT as a function of the stress time with different frequencies, respectively. For M10 TFT the Vth and SS variation is much lower than the S1 TFT. The results indicate that the M10 TFT has less deep states. Firstly, the M10 TFT has more effective NH3 plasma passivation than that of S1 TFT due to the ten split nanowire channels of M10 TFT has wide NH3 plasma passivation area.

Secondly, M10 TFT has robust tri-gate control, thus the additional two side-gate surface scattering (Fig. 3b) reduce the hot carrier effect. Therefore, the deep states generation of M10 TFT by the hot carrier impaction is lower than which of S1 TFT.

Notably, for S1 TFT, the Vth and SS variation increase with the frequency increasing from 1 K Hz to 1 MHz. These results reveal that the device reliability is strongly dependent on the transient current Id (displace current).

voltage.

Thus, the transient current induced hot carrier is dependent on the frequency.

Figure 4-16 and 4-17 depict the ON current (ION) variation and ON/OFF ratio of S1 and M10 TFT as a function of the stress time with different frequencies, respectively. The M10 TFT shows lower ION variation than S1 TFT. These results reveal that the S1 TFT has high Vth variation and increase with the frequency increasing. Thus, the ION of S1 TFT is much lowering than that of M10 TFT. For the

the ON/OFF ratio of M10 still remains exceed 10⁸, under different frequencies.

Figure 4-18 depicts Gm degradation of S1 and M10 TFT as a function of the stress time with different rising time (Tr) and falling time (Tf) under the frequency of 1 KHz. For S1 TFT (dash-line) with the same falling time of 100 ns, the S1 TFT has the similar Gm degradation. However, for the same rising time of 100 ns, as the falling time increasing from 100 ns to 1 us, the Gm degradation is reduced from 40%

to 20% at the stress time at 1000 second. These results reveal that the device reliability is strongly dependent on the transient current. According to the Uraoka et al.

report [4-2], only the transient current induced by the falling period would cause more damage near the drain. Figure indicates clearly that the amount of the hot carriers generated depends on the pulse falling time. Therefore, using the dynamic stress with longer falling time will be helpful for the reliability improvement in the poly-Si TFTs.

On the other hand, for M10 TFT, however the Gm degradation dependent on falling time is not significant. These results indicate that M10 TFT has highly effective NH3

plasma passivation and robust tri-gate control to screen the transient current hot carrier effect, which induced by the falling period.

Figure 4-20 and 4-21 depicts Vth variation and ON current (Ion) degradation of S1 and M10 TFT as a function of the stress time with different rising time (Tr) and falling time (Tf) under the frequency of 1 KHz. For the same reason, the Vth variation

and ON current (Ion) degradation of S1 TFT (dash-line) is highly dependent on the falling time rather than rising time. As the falling time increasing from 100 ns to 1 us, the Vth variation and ON current (Ion) degradation are reduced. Again, M10 TFT has highly effective NH3 plasma passivation and robust tri-gate control to screen the transient current hot carrier effect, which induced by the falling period.

Figure 4-22 depicts Gm degradation of S1 and M10 TFT as a function of the stress time with different subtract temperature with 25⁰C, and 75⁰C under the same frequency of 1 KHz, and the same Tr = Tf =100 ns. For both S1 and M10 TFT, the Gm degradation is reduced with the temperature increasing from 25⁰C to 75⁰C. These results reveal that hot carrier effect is reduced with the temperature increasing. As the temperature increasing, the mean free path (λ) [4-3] is decreasing.

2 ) tanh(

0 kT

E_P

⋅

=λ λ

Thus, that hot carrier energy is reduced as the mean free path (λ) is decreasing.

Figure 4-23 and 4-24 depicts Vth variation and ON current (Ion) degradation of S1 and M10 TFT as a function of the stress time with 25⁰C, and 75⁰C under the same frequency of 1 KHz, and the same Tr = Tf =100 ns. For both S1 and M10 TFT, the Vth variation and ON current (Ion) degradation is reduced with the temperature increasing from 25⁰C to 75⁰C. For the same reason, hot carrier effect is reduced with the temperature increasing as the mean free path (λ) decreasing.

Wavenumber (cm-1)

400 600 800 1000 1200 1400 1600 1800 2000

Ab so rba nce (a .u )

-0.05 0.00 0.05 0.10 0.15 0.20 0.25 0.30

TEOS

FSG (CF4=600sccm)

Si-O Si-F

Figure4.1 FTIR Spectra of FSG film and TEOS film between 2000cm-1 and 400 cm-1

Figure4.2 (a) Comparsion of Id-Vg characteristics of FSG spacer TFT and TEOS spacer TFT for Vd=0.1 V (W/L=10um/5um)

V

(V)

Figure 4.2 (b) Comparsion of output characteristics for FSG spacer and TEOS