Time and temperature dependence of the drain current of PF-based OFETs

Abstract

We have studied the effects of bias temperature stress (BTS) on organic field-effect transistors (OFETs) in accumulation (negative stress bias) and depletion (positive stress bias) using both dc and ac stress biases. The device studied is an inverted, gate-planarized, co-planar thin-film transistor that has been previously described [1]. Indium tin oxide (ITO) was used for the source and drain contacts and benzocyclobutene (BCB) / amorphous silicon nitride was used for the gate-planarization / insulator. The organic semiconductor F8T2 (poly 9,9-dioctylfluorene-co-bithiophene) was deposited by spin-coating from xylenes solution. These devices exhibit typical p-type field-effect transistor behavior. Typical values of the linear regime field-effect mobility, threshold voltage, and subthreshold swing for these devices are: 5x10-3 cm2/Vs, -20 V, and 3.0 V/decade respectively. All measurements were performed in the dark and in air using an HP4156 connected to a Karl Suss PM-8 probe station with a temperature-controlled chuck. For the case of negative dc BTS over long time scales (>104sec), we have used both interrupted and noninterrupted stress methods measured over a range of temperatures (293K < T < 353K). The major observable effect is a shift of the threshold voltage to more negative values as the stress time accumulates, causing a decrease in the drain current at a specific applied gate bias. The observed dependence on stress temperature is analyzed in terms of the kinetics of the stress mechanism. This analysis is performed by unifying the threshold voltage shift curves through either the normalization of the accumulated stress time by a thermally activated time constant for the stress or by using the thermalization energy [2,3]. We note that the values of both the activation energy of the time constant and the thermalization energy are approximately 0.25eV. We propose that this energy corresponds to the peak of a density of trap states above the valence band/HOMO level of F8T2. The observed bias stress effects are reversible at room temperature in the dark. However, recovery of the device is accelerated at elevated temperatures and by illumination with strongly absorbed illumination, as has been observed by others [4], indicating charge trapping/de-trapping as the general stress/recovery mechanism. For the case of positive dc BTS, we observe an unexpected shift of the threshold voltage towards more negative values as well as a significant degradation of the subthreshold swing, while the field-effect mobility is left unchanged throughout the duration of the positive BTS. The effects of the positive BTS are also reversible and we have observed that the recovery of the threshold voltage lags the recovery of the subthreshold swing. This is a possible indication that there are at least two competing stress mechanisms occurring in this device for positive BTS. We propose that there is an additional threshold voltage shift due to the movement of charged species in the insulator (presumably in the BCB) under the influence of positive applied stress bias. We have also investigated the effects of pulsed (ac) BTS in both the accumulation and depletion regimes over a range of pulse frequencies (10 to 100Hz) with a base value of OV and a duty cyde of 50%. In each case, the observed effect is a threshold voltage shift, while the field-effect mobility remains constant. For negative ac BTS, the threshold voltage shifts are similar to that observed for dc BTS if the effective stress time is taken into account. For positive ac BTS, we observe that, for this device, there is a small negative threshold voltage shift which maximizes relatively quickly (compared to the negative BTS case). After a short time, the threshold voltage shift begins to swing back towards a positive threshold voltage shift. For the range of pulse frequencies used here, the threshold voltage shift does not seem to exhibit any discemable dependence on the pulse frequency. It is interesting to note that, while the threshold voltage shifts describe an undesirable instability of these devices, the subthreshold swing appears to decrease as the stress progresses for all of the ac BTS measurements, though further experiments are necessary to fully describe this effect and its dependence on the ac bias stress conditions. In condusion, we have performed and analyzed positive (depletion) and negative (accumulation) BTS using both ac and dc stress bias signals. For each case, the major observable effect is a threshold voltage shift caused by either the trapping of charge carriers in the organic semiconductor or, in the case of positive BTS, the movement of charged species in the organic insulator becomes suspect The results have been successfully analyzed using thermalization energy and kinetics concepts.