Effect of Bias Stress in Inkjet-Printed Polymer Thin-Film Transistors: Generation and Annihilation of Subgap Density of States

1. INTRODUCTION

Organic printed electronic devices, which are made by directly printing functional organic semiconductor inks onto flexible substrates, have been investigated extensively for next-generation flexible electronics because of their mechanical flexibility and the potentially low cost of manufacturing them using graphic-art printing techniques [1-6]. Among the various printed electronic devices based on π-conjugated polymer semiconducting materials, polymer thin-film transistors (PTFTs) have shown considerable potential for widespread use as crucial building blocks for logic, arithmetic, and memory circuits for future electronic systems [2-4]. However, despite their considerable potential, the electrical instability of PTFTs caused by bias stress remains a major problem that needs to be addressed and solved to enable further developments of PTFT technology [3,7–9]. Here, we report an investigation of the effects of bias stress on the density of states (DOS) in PTFTs with electron donor–acceptor type conjugated copolymers of poly(tetryldodecyloctathiophene-alt-didodecylbithiazole)−P8T2Z−C12[the chemical structure is shown in Fig. 1(a)]−which were inkjet-printed to form an active channel layer [10]. The threshold voltage (V_T) of the inkjet-printed PTFT shifted in the positive direction when the gate of the PTFT was subjected to positive bias stress (PBS), whereas it shifted in the opposite (negative) direction when the gate bias (V_G) was removed. We analyzed the effect of the bias stress manifested by the V_T shift by monitoring the device parameters and the DOS extracted from a combination of generation–recombination current spectroscopy, multi-frequency capacitance–voltage spectroscopy [11,12], and current–voltage (I–V) modeling [13–15]. The extracted subgap DOS revealed that the subgap states of the acceptors in the active channel strongly affect the V_T shift of the inkjet-printed PTFTs during operation. We also dis cuss the physical mechanism of the electrical instability induced by bias stress in terms of the subgap DOS.

Fig. 1. (a) Chemical structure of the P8T2Z-C12 semiconducting copolymer, (b) schematic of the coplanar PTFT fabricated via inkjet printing, (c) output curves of the inkjet-printed P8T2Z-C12 PTFT, (d) transfer curve and gate-voltage-dependent mobility plots of the PTFT at V_D=−10 V.

Download Original Figure

2. METHODS

We measured the electrical properties of P8T2Z-C12-based inkjet-printed PTFTs using a bottom-gate, coplanar, structured device without any passivation to accelerate the electrical degradation, as shown in Fig. 1(b). Except for the passivation process, we fabricated the inkjet-printed PTFT as described previously [3]. We performed the electrical measurements under ambient conditions (relative humidity 30%) using an Agilent 4284A LCR meter and a 4156C semiconductor analyzer.

3. RESULTS AND DISCUSSION

Fig. 1(c) and Fig. 1(d) show representative output and transfer curves, respectively, of the inkjet-printed P8T2Z-C12 PTFT. The output characteristics show reasonable saturation with low hysteresis. As the drain bias (V_D) increases, the output current, I_D, becomes larger than ~25 μA at V_D=−40 V and V_G=−20 V. The transfer characteristics exhibit p-type behavior, with on/off ratios exceeding ~107. The curves of the mobilities with respect to V_G show that the mobilities in the linear region (V_D ≤ −10 V) increase gradually with V_G and reach 0.13 cm²V⁻¹s⁻¹ at V_G=−20 V.

Fig. 2(a) and Fig. 2(b) show semilogarithmic-scale transfer curves at V_D=−4 V and the corresponding generation–recombination (I_G–R) currents as a function of V_G for the inkjet-printed PTFT during PBS at V_G=20 V and V_D=−10 V together with the recovery obtained by releasing the V_G from the PTFT. The symbols in the graph are from the electrical measurements, and the lines are from calculations based on an I–V model [13-15]. As shown in Fig. 2(c), when the PTFT is subject to PBS at V_G=20 V and V_D=−10 V, V_T moves toward more positive values. This means that charge carriers are trapped either in the P8T2Z-C12 polymer layer or at the interface during operation, which consequently shields the gate potential, thereby inducing the V_T shift. The trapping of negative electrons results in the positive shift of V_T. However, V_T recovered (i.e., moved in the negative direction) as the recovery time increased. In addition, the subthreshold swing (S.S.) increased as the duration of the bias stress increased, and it then decreased slightly with increasing recovery time.

Fig. 2. (a) V_T shifts during PBS and during the recovery process of the P8T2Z-C12 TFT in semilogarithmic-scale transfer curves at V_D=−4 V, (b) the corresponding generation–recombination (I_G–R) currents as a function of V_G during PBS and during the recovery process. The symbols are from electrical measurements and the lines are from modeling, (c) extracted electrical parameters from the stress-induced I–V characteristics of the inkjet-printed PTFTs.

Download Original Figure

The dependence of the changes in these characteristics of the inkjet-printed PTFT upon the duration of the PBS and of the recovery process can be analyzed quantitatively through a physical model based on the full-range DOS. To investigate the effects of bias stress in the PTFTs, we therefore obtained the profiles of the full-range DOS by applying various extraction techniques.

We modeled the DOS function [g(E)] in the bandgap of the P8T2Z-C12 polymer channel as a combination of a donor-like DOS [g_D(E)], which includes donor-like tail states [g_TD(E)] and donor-like deep states [g_DD(E)] near the valence band (E_V), together with shallow acceptor states [g_SA(E)] and acceptor-like DOS [g_TA(E)] near the conduction band (E_C). We express this model in terms of the following equations [13],[16]:

where N_TD, kT_TD, N_DD, and kT_DD are, respectively, the density of donor-like tail states at E=E_V, the characteristic slope of the donor-like tail states, the density of donor-like deep states at E=E_V, and the characteristic slope of the donor-like deep states. Similarly, we defined N_SA, kT_SA, E_SA, N_TA, and kT_TA, respectively, as the peak density of the shallow acceptor states, the characteristic slope of the shallow acceptor states, the position of the peak density of the shallow acceptor states, the density of the acceptor-like tail states at E=E_C, and the characteristic slope of the acceptor-like tail states.

Fig. 3 shows the evolution of the full-range DOS function g(E) during the PBS and the recovery process. We note that there is little change in g_D(E) and only a slight change in g_TA(E) as the duration of the PBS and the recovery time increase, whereas g_SA(E) increases with increasing PBS duration and then decreases with increasing recovery time. Furthermore, the position of the Fermi level (E_F) gradually shifts closer to E_V over time during PBS, leading to a positive shift in V_T, and upon recovery the position of EF moves away from E_V, which results in a negative shift in V_T.

Fig. 3. Evolution of the full-range DOS in the active channel layer of the inkjet-printed P8T2Z-C12 during PBS and the recovery process.

Download Original Figure

Fig. 4(a) shows the portions of the total V_T occupied by each parameter extracted from the I–V modeling and the DOS profile to visualize the effect of each parameter on V_T. The quantity ΔV_OX corresponds to the interaction between the charge carriers and the subgap states within the gate insulator. The quantities ΔV_GSA(E) and ΔV_gD(E) are associated with the interactions with the electron-acceptor states and the hole-donor states, respectively, within the active layer. As shown in Fig. 4(a), ΔV_OX and ΔV_GSA(E) mainly affect the shifts of V_T during the PBS and the recovery time. However, the effect of ΔV_gD(E), which corresponds to the interaction with hole-donor states, is negligible. Fig. 4(b) shows the changes in ΔV_T as a function of time in the inkjet-printed PTFT, which was first stressed with V_G= −20 V and V_D=−1 V for 1,500 s, after which the stress was removed. We monitored the changes in ΔV_T by collecting transfer characteristics over the range from V_G=20 V to –20 V, with V_D=–10 V, at time intervals determined more frequently than in Fig. 2(c). As the duration of the PBS increases, the V_T of the PTFT shifts in the positive direction. At the beginning of the recovery process, V_T first shifts rapidly in the negative direction, but the shift gradually slows down and eventually remains at an irreversible point with increasing recovery time. We fitted ΔV_T for the PBS and recovery process to the respective functions ΔV_T(t)=ΔV₀ {1 – exp [–(t/τ)^β]} and ΔV_T(t)=ΔV_max exp [−(t/τ)^β] [17,18]. Here, ΔV₀=V_G–V_T⁰, where V_T⁰ is the initial threshold voltage, ΔV_max is the value of ΔV_T right before the recovery begins, β is the dispersion parameter, and τ is the charge trapping/ detrapping time constant. The fitted values of β and τ are 0.88 and 631 s for PBS and 0.17 and 2,990 s for the recovery process, respectively.

Fig. 4. (a) Dependence of the extracted parameter-dependent portion of ΔV_T on the duration of the PBS and on the recovery time, (b) the changes in ΔV_T as a function of time in the PTFT. The curves are fits using stretched exponential functions, and the symbols are the measured data. The physical mechanisms of the effect of bias stress are represented schematically as energy-band diagrams for (c) PBS and (d) the recovery process.

Download Original Figure

Based on these observations, the physical mechanisms for the changes in the PTFT ΔV_T during the PBS and the recovery process can be expressed schematically as shown in Fig. 4(c) and Fig. 4(d), respectively. First, the positive direction shifts of ΔV_T observed in the PTFT under PBS conditions can be explained by four processes: (i) Electrons are induced at the interface between the insulator and the channel by the positive bias applied to the gate, and most of the induced electrons are injected into the insulator by the electric field and are drawn toward the active layer. (ii) Some of the induced electrons at the interface diffuse to and are trapped in the shallow states near the E_C. (iii) Some anions (OH⁻), which decompose from the moisture (H₂O) in the air, are absorbed by the polymer layer and are attracted by the positive V_G, resulting in the generation of defect states [g_SA(E)] near E_V. (iv) Some OH⁻ ions migrate to the insulator under the influence of the electric field. Conversely, the recovery of ΔV_T caused by releasing the V_G applied to the PTFT is divided into a fast recovery step and a slow recovery step, each of which can be described as two sequential processes: (v) Electrons trapped in the shallow trapping states near EC are immediately detrapped by the electric field applied to the insulator by the floating gate. (vi) When the PBS is removed, most of the OH⁻ ions that had been attracted to the g_SA(E) within the active layer are pushed away from the interface and migrate to the back channel. During this process, the g_SA(E) generated by the OH− ions are annihilated as those ions move away. (vii) Most of the OH⁻ ions trapped inside the insulator by PBS gradually out-diffuse toward the channel through the insulator/active interface. (viii) Some trapped OH⁻ ions in the insulator remain even after the PBS is removed, leading to the irreversible component of ΔV_T. In addition, some OH⁻ ions that migrated to the polymer film also can react with the conjugated polymer chain and induce bipolaron states [19].

4. CONCLUSION

We have investigated the effects of bias stress in an inkjet-printed P8T2Z-C12 PTFT in terms of the subgap DOS. Under PBS conditions, the V_T of the PTFT shifts in the positive direction while generating defect states in the subgap DOS of the active semiconducting polymer layers. However, this shift in V_T is mostly recovered when the PBS is removed. We also confirmed experimentally that the defect states in the subgap DOS annihilated during the recovery process.