Stability of organic permeable base transistors

A wide range of applications are enabled by the inherent flexibility of organic transistors, and bendable or even foldable displays have moved into the focus of research.¹ Organic Permeable Base Transistors (OPBTs) are a relatively young but highly promising type of organic transistor. OPBTs have a unique vertical structure, shown in Fig. 1. They consist of three electrodes (emitter, base, and collector), which are separated by a layer of an organic semiconductor. The base electrode, often consisting of aluminum, is very thin and not fully closed, i.e., small openings remain in the electrode. Furthermore, the base electrode is exposed to ambient air after processing to grow a thin oxide layer on its surface.

The vertical structure of OPBTs allows for a switching mechanism distinctively different from conventional field-effect transistors. Current injected at the upper emitter electrode is transmitted through the small holes in the base electrode and finally collected by the bottom electrode. This vertical current can be controlled by the base potential, i.e., the current can be switched “OFF” or “ON” if the potential difference between the injecting contact and the base electrode is varied.

The OPBT structure allows us to tune the device thickness with nanometer precision, which leads to high performance.^2–5 A high switching ratio (10⁸), large on-state current densities (75 A/cm²) at low voltages (2 V), and a large transient frequency up to 40 MHz have been reported.⁶ Numerical modeling has shown that OPBTs have the potential to reach even switching frequencies in the range of 100 MHz.⁷ Furthermore, current amplification ratios as large as 10⁵ were reported.

This excellent performance, in particular, the large current amplification reached recently, makes OPBTs a potential alternative to other thin film transistor technologies used as driving transistors in active matrix organic light-emitting diode (OLED) displays.⁸ However, this application poses strict requirements on the stability of the device characteristics, in particular, the threshold voltage, under electrical stress. The driving transistor in a backplane of an OLED display has to continuously drive a preset current through the OLED. Any shift in the threshold voltage during operation will therefore result in an unwanted dimming of the pixel. Furthermore, a different driving history of different pixels will lead to a brightness variation across the display, which is easily noticeable.

In this paper, the operational stability of OPBTs is studied. In particular, the shift in the transfer characteristic under continuous cycling and under constant base bias stress is discussed. It is shown that the transfer characteristic of these devices shifts constantly while cycling, and akin to gate bias stress of organic field-effect transistors, under a constant bias applied to the base electrode. With the help of a two-dimensional (2D) device model, these effects are discussed in terms of trapping of charges at the base electrode.

All devices were prepared by thermal evaporation under high vacuum conditions with the pressure between 10^–7 and 10^–8 Torr, similar to the devices discussed in Ref. 9. The devices were structured by laser cut metallic shadow masks to define an active area of 0.04 mm², 0.025 mm², and 0.01 mm². Optical images of the structure of these devices can be found in the supplemental material of Ref. 9. Glass substrates were cleaned by sonication in acetone, methanol, and iso-propanol, and the cleaned substrates were dried by blowing pure air.

The substrates were loaded in a vacuum chamber followed by the deposition of the different layers shown in Fig. 1. First, 100 nm of Al was deposited at a rate of 0.5 Å/s, followed by the deposition of 20 nm of Cr at a rate of 0.1 Å/s to form a bottom contact used as a collector electrode. 100 nm of an intrinsic, i.e., undoped, layer of the organic semiconductor C₆₀ (i-C₆₀) was deposited on the top of Cr at a rate of 0.5 Å/s. A thin layer of Al (15 nm) was deposited on the top of i-C₆₀ to form a base electrode. An extra 50 nm thick layer of Al is deposited outside the active area of the device to improve the contact layer of a base as shown in Fig. 1.

After deposition of the base electrode, the overall Al/Cr/i-C₆₀/Al stack (also termed the bottom diode) was exposed to pure air for 30–90 min placed close to a reservoir of distilled water to form a thin insulating layer of Al₂O₃(⁠$≈$ 2–3 nm) on the base Al. The variation in exposure time has a small influence on the ON/OFF ratio of the transistor, as discussed in Ref. 9. Afterward, the devices were annealed at 60 °C for four hours before they were transferred back to the deposition chamber. To define the active area of the device, a bank structure of 200 nm of Spiro-TTB (2,2$′$⁠,7,7$′$-Tetra(N,N-di-p-tolyl)amino-9,9-spirobifluoreneSpiro-TTB, Lumtec Corp.) was deposited at a rate of 0.5 Å/s. Overall, controlling the device area by this additional Spiro-TTB layer was shown to increase the performance of OPBTs.⁹ After the deposition of the Spiro-TTB layer, 100 nm of i-C₆₀ was deposited. To improve the charge injection in the devices, 20 nm of n-doped C₆₀ (n-dopant W₂hpp₄, 2 wt.%) was deposited on top of i-C₆₀. This deposition step was followed by deposition of 20 nm Cr and 100 nm Al to form the top emitter contact and to complete the OPBT structure.

Devices were then annealed at 150 °C in ambient air, which was shown to improve the performance of OPBTs.¹⁰ Finally, the devices were tested using a semiconductor parameter analyzer (Keithley 4200).

The stability of OPBTs is tested in two distinct modes of operation. In the first mode, the device is cycled for an extended time, i.e., the transfer characteristics are measured consecutively for several minutes. In a second mode, the stability is tested when a constant voltage is applied to the base electrode, which is closer to a potential application of OPBTs as driving transistors in display backplanes.

Figure 2(a) shows the transfer characteristic of an OPBT cycled for 27 min. The collector-emitter voltage V_CE is kept at 2 V. A small shift in the transfer characteristics is clearly visible. Overall, the transfer characteristic of the device shifts toward more positive base-emitter voltages V_BE for the first 10 min, after which the shift saturates. The shift in the transfer characteristic is correlated with a shift in the threshold voltage, which is plotted as shown in Fig. 2(b). As seen in Fig. 2(b), the voltage shift depends exponentially on the cycling time. It can be fitted by a stretched exponential relation given by

where $(ΔVth)∞$ is the equilibrium shift in threshold voltage (i.e., for $t→∞$⁠), τ is the relaxation time, and β is the stretching parameter. We find a time constant of τ = 245 s and a stretching parameter β of 0.9, i.e., relatively close to 1. Overall, the shift in the threshold voltage is small (around $ΔVth$ = 0.2 V) and saturates quickly.

To study the effect of constant stress on the transistor performance, the base of the OPBT is biased at V_BE = 2 V, while the collector is biased at three different voltages (V_CE = 1 V, 1.5 V, and 2 V). After stressing the device, the transfer characteristics are measured and the shift in threshold voltage is calculated. Furthermore, the shift in the collector current at the particular base and collector potential is monitored.

The results are shown in Fig. 3(a). Overall, the shift in the transfer characteristics is stronger compared to the cycling experiment shown in Fig. 2(b). The threshold voltage shifts by approximately $ΔVth$ = 0.7 V for a stress time of 1200 s, which does not depend on the particular collector bias V_CE [cf. Fig. 3(b)]. Interestingly, the shift in threshold voltage does not fully saturate, in contrast to the shift in threshold voltage observed in the cycling experiment.

In contrast to the shift in threshold voltage, the decrease in collector current strongly depends on the applied collector bias V_CE and increases from approximately $ΔIC$ = 1.5 mA at V_CE = 1 V to $ΔIC$ = 3 mA at V_CE = 2 V [cf. Fig. 3(c)].

This qualitative difference between the shift in the threshold voltage and in collector current can be understood by the nonideal output characteristic shown in Fig. S1 in the supplementary material. The different collector potentials used during stressing the device correspond to different operation regimes, i.e., the devices are operated in the linear region at V_CE = 1 V and in a weak saturation regime at V_CE = 1.5 V and 2 V. Consequently, the transconductance,

is different for the different collector potentials, leading to a different change in collector currents even though the shift in the effective base-emitter voltage $(VBE)eff$ = $VBE−Vth$ is identical for all collector potentials.

The base bias stress effect is partially reversible, shown in Fig. 3(d). After removing the base bias, the collector current is monitored for several hours. It is found that I_C almost returns to its initial value after 27 h.

The base bias stress shown in Fig. 3 resembles gate bias stress effects in conventional Organic Field-Effect Transistors (OFETs).^11,12 Often, it is argued that the shift in threshold voltage of OFETs is due to trapping of charges in different operation regimes. The trapping of charges in OFETs may occur (i) between the interface of dielectric material and the semiconductor material,¹³ (ii) within the bulk material,¹⁴ (iii) around the defects of the semiconductor material,¹⁵ or the observed shift might be due to (iv) formation of bipolarons in the semiconductor.^16,17 Increasing the stress time increases the number of trapped charges, thus resulting in a more pronounced threshold voltage shift.

A numerical 2D simulation is used in order to study the origin of the base bias stress observed here. The numerical model, discussed in detail in Refs. 9 and 18, is sketched in Fig. 4(a). A cylindrical symmetry with a single opening/hole in the base electrode at r = 0 nm is assumed. Following the results of Yutani et al.¹⁹ and Kaschura et al.,²⁰ the base electrode (thickness of d_base = 12 nm) is covered with a thin Al₂O₃ layer (d_ox = 2 nm), which suppresses injection at the base. The C₆₀ layer (total thickness between the emitter and the collector $dC60$ = 200 nm) is assumed to show an electron mobility of μ_n = 0.1 cm² V^–1 s⁻¹ and a permittivity of ϵ_r = 3.

Following the results of Kaschura et al.,²⁰ OPBTs function very much like a vertical OFET instead of an organic BJT. A horizontal channel is formed that transports charge toward the pore in the base electrode at r = 0, qualitatively sketched in Fig. 4(a). This effect, similar to the gating effect seen in OFETs, leads to the current distribution shown in Fig. S2 in the supplementary material (plotted for V_BE = 1.2 V and V_CE = 2 V), where most of the current is concentrated at the oxide/C₆₀ interface.

However, the oxide covering the base electrode is grown by exposing the base to ambient air. Usually, hydroxyl (-OH) groups are formed, which trap electrons.²¹ To include this effect in the model, a thin layer (2 nm thick) of traps is included at the interface between the Al₂O₃ insulator on top of the base electrode and the C₆₀ layer. Different trap concentrations ranging from $(2−20)×1019$ cm⁻³ are assumed to model the consecutive shift in the transfer characteristics with the increasing stress time. These values correspond to an interfacial trap density of $(4−40)×1012$ cm⁻².

The results of the simulation are shown in Fig. 4(b). Indeed, the transfer characteristics of the transistor shift continuously toward higher threshold voltages for a higher concentration of traps. This shift results from a decrease in the density of free (i.e., not trapped) charge carriers inside the channel, which leads to a shift in the threshold voltage by $ΔVth=entrapdC$⁠, where e is an elementary charge, n_trap is the density of trapped charges, d represents the thickness of the layer of traps, and $C=ϵ0ϵrdox$ represents the specific capacitance of the oxide layer on top of the base electrode).

This mechanism can explain the trends observed in Figs. 2 and 3. Cycling the transistors (Fig. 2) allows us to partially trap free electrons at the Al₂O₃ surface during the negative part of the cycle, which in accordance with the experiment is expected to result in a weaker shift in threshold voltage as for a constant positive bias stress.

Furthermore, Al₂O₃/C₆₀ responsible for trapping of electrons is shielded from the collector potential, which explains that the shift in threshold voltage does not depend on the emitter-collector voltage V_CE as observed in Fig. 3(b). Finally, the trapped electrons will be thermally released after releasing electrical stress, and the original performance is recovered, which is indeed observed in Fig. 3(d).

A high operational stability is essential for the success of vertical organic permeable-base transistors. OPBTs have to show a small drift in the threshold voltage during continuous stress, in particular, if they are used as driving transistors in active matrix displays.

As shown here, OPBTs are subject to similar limitations as standard organic field-effect transistors, i.e., they are susceptible to base-bias stress. Overall, the threshold voltage of OPBTs shifts by approximately 0.7 V during the first 20 min of stressing, which corresponds to a decrease in their on-state current by 1–3 mA.

The origin of the shift in threshold voltage due to an extended bias applied to the base is discussed with the help of a numerical device model. It is shown that the observed trends can be explained by increasing trapping at the oxide/semiconductor interface on top of the base electrode.

Understanding the mechanism of base bias stress can be used to improve the stability of OPBTs. Most importantly, the trap states at the oxide surface have to be passivated. To passivate these trap states, a conformal and insulating layer has to be found that is not blocking holes in the base electrode. Self-assembled monolayers, such as phosphonic acid base monolayers used for OFETs,²² are therefore promising candidates for optimizing the stability of OPBTs.

See the supplementary material for a more detailed experimental description, a representative output characteristic of the OPBTs discussed here, and more simulation results.

Funding from the National Science Foundation (Grant No. 1639073) and from the Bi-national Science Foundation (Grant No. 2014396) is greatly acknowledged. Characterization of samples was partially done at the Characterization Facility of the Liquid Crystal Institute, Kent State University. A.A. was supported by The Higher Committee For Education Development in Iraq.