Unraveling Structure and Device Operation of Organic Permeable Base Transistors

Organic permeable base transistors (OPBTs) are of great interest for flexible electronic circuits, as they offer very large on‐current density and a record‐high transition frequency. They rely on a vertical device architecture with current transport through native pinholes in a central base electrode. This study investigates the impact of pinhole density and pinhole diameter on the DC device performance in OPBTs based on experimental data and TCAD simulation results. A pinhole density of NPin = 54 µm−2 and pinhole diameters around LPin = 15 nm are found in the devices. Simulations show that a variation of pinhole diameter and density around these numbers has only a minor impact on the DC device characteristics. A variation of the pinhole diameter and density by up to 100% lead to a deviation of less than 4% in threshold voltage, on/off current ratio, and sub‐threshold slope. Hence, the fabrication of OPBTs with reliable device characteristics is possible regardless of statistical deviations in thin film formation.

the same time, it can also be employed as a four-terminal device for built-in logical operations. [14] Identical or similarly structured transistor devices are also published under the names Metal-Base Organic Transistor (MBOT) [15] or Space-Charge Limited Transistor (SCLT). [16,17] The formation of a native oxide layer and of nano-sized pinholes in the thin metallic base electrode are utilized, providing very small structure sizes to define the transistor channel. The short channel length leads to large current densities [18] and, consequently, high transition frequencies. Additionally, the limiting influence of contact resistance is reduced in OPBTs, because of their ability to inject current over the entire device area. [19] The central base electrode in an OPBT device is permeable to electrons via nano-sized intrinsically grown pinholes in the metal film, [20] allowing for current transport through the device, as indicated in Figure 1. A voltage applied between base and emitter is capable of modulating the current between emitter and collector over many orders of magnitude. The pinholes in the metallic base layer appear due to strain in this layer during the oxidation. This intrinsic and self-limiting process though is expected to show statistical variations in pinhole size and density which might affect the transistor's performance. In this work, we evaluate the size and density of these pinholes by experiment and evaluate the impact of variations on the device performance based on TCAD (Technology Computer Aided Design) simulations.

Device Fabrication and TCAD Simulation
Organic Permeable Base Transistors are fabricated by consecutive vacuum evaporation of the electrode and semiconductor Organic permeable base transistors (OPBTs) are of great interest for flexible electronic circuits, as they offer very large on-current density and a record-high transition frequency. They rely on a vertical device architecture with current transport through native pinholes in a central base electrode. This study investigates the impact of pinhole density and pinhole diameter on the DC device performance in OPBTs based on experimental data and TCAD simulation results. A pinhole density of N Pin = 54 µm −2 and pinhole diameters around L Pin = 15 nm are found in the devices. Simulations show that a variation of pinhole diameter and density around these numbers has only a minor impact on the DC device characteristics. A variation of the pinhole diameter and density by up to 100% lead to a deviation of less than 4% in threshold voltage, on/off current ratio, and sub-threshold slope. Hence, the fabrication of OPBTs with reliable device characteristics is possible regardless of statistical deviations in thin film formation.

Introduction
Organic thin film transistors have great potential in flexible electronic applications like electronic displays [1] as well as in low-cost devices like RFID tags. [2] The performance of organic transistors has greatly improved in recent years due to optimization layers. A passivating oxide film of Al 2 O 3 around the aluminum base electrode is formed by native oxidation in air or with an electrochemically controlled process. [21][22][23] The devices consist of an organic semiconductor layer sandwiched between the collector and emitter electrodes and the thin base electrode, which is permeable to electrons, in the center. Typical layer thicknesses are 100 nm for the collector and emitter electrodes and the semiconductor layers and 15 nm for the base electrode. Electrical DC measurements on fabricated OPBTs with C 60 as organic semiconductor (OSC) have been taken as a reference for the Sentaurus TCAD simulation setup. [24] For investigation of the properties of the base layer in a transmission electron microscope (TEM), a base-only device is prepared: A TEM-grid is vacuum-coated with a 50 nm thick C 60 layer to provide the same surface and morphology as in an OPBT device. Then, a 15 nm film of aluminum is deposited and exposed to ambient air in the dark for 15 min to ensure native oxidation of the metal. This oxidation process is also employed for all devices used for electrical testing. High-angle annular dark-field scanning transmission electron microscopy and spectrum imaging based on energy-dispersive X-ray spectroscopy (HAADF-STEM and EDXS) are conducted with a Talos F200X (Thermo Fischer Scientific/FEI, USA) operated at 200 kV and equipped with a Super-X EDX detector. The latter provides detailed element maps (C, O, Al) of the base layer. The pinholes are manifested by decreased X-ray counts in the aluminum map.
For TCAD simulations, the device is modeled as illustrated in Figure 2a. The data shown in Figure 2b reveals a quite reasonable agreement of the DC characteristics between simulations (solid lines) and the measurements (symbols) for the parameter values denoted in Table 1 (device dimensions for Figure 2b are W = 250 µm and L E = 250 µm). In order to remove an initial hysteresis in the device from the measurement, the voltage is swept up and down for several times. [25] A Poole-Frenkel mobility model with a square-root dependence on the electric field, as represented by Equation (1) and a Gaussian density of state (DOS), as in Equation (2), are used with the default values. [24] ( ) where μ 0 is the low-field mobility, E 0 is the effective activation energy, k is the Boltzmann constant, T is the temperature, β and γ are the Poole-Frenkel coefficients, and F is the applied electric field. The Gaussian DOS is described by where N C is the effective density of states, σ DOS is the width of the DOS distribution, and E c is the energy center.
In the framework of hopping transport, charge carriers are localized on individual molecules due to the weak overlap of intermolecular orbitals and only occasionally hop to other favorable molecular sites. Thus, charges cannot be treated as plane waves, and hence, a quantum confinement due to the pinhole size is not expected.
Furthermore, it should be emphasized that the base leakage current I B is quite small and negligible compared to the collector current I C (see Figure 2b). Hence, the base leakage current has no influence on the device DC characteristics including the threshold voltage V th , on/off current ratio, subthreshold slope SS, etc.

Pinhole Size Distribution Analyzed by TEM
To analyze the morphology and element distribution of the Al base layer as well as to deduce the pinhole size distribution, EDXS-based chemical mapping was performed in scanning TEM mode. The distribution of aluminum is shown in Figure 3a. The 15 nm thin film is percolated, providing a conductive electrode that can distribute an applied base potential across the entire device. Between the conductive paths, voids are visible where no aluminum is present. These voids are pinholes in the film and will be filled by C 60 , hence allowing for vertical electron transport through the device. By adding the information about the distribution of oxygen atoms in the layer, Figure 3b is obtained. It shows that the base electrode layer is evenly oxidized. At the rims of the pinholes, a stronger oxygen signal is visible, representing the oxidation of the vertical sidewalls of each pinhole. Since the thickness of the sidewall-oxide is uniform throughout the sample, every void in the aluminum film that is above a certain size has a remaining opening in its center. The structure is analyzed using a grey-scale threshold value to identify voids. Noticeable voids of at least 25 nm 2 are considered as pinholes. The density of the so-defined pinholes is found to be 54 per µm 2 . Their size distribution can be seen in Figure 4. Typical dimensions of the mostly elongated pinholes are between 5 and 25 nm.
Based on these findings and the insight that thin-film formation is always a statistical process, the questions arise, which pinhole density and dimension would be ideal for the OPBT and how deviations in the microstructure will influence the device characteristics. It has previously been shown that the pinhole size and density is not limiting the maximum current level in an OPBT, but only determines the device behavior at low base potential. [12] Nevertheless, it is of interest to investigate how fluctuations in the pinhole formation will influence the device performance and hence affect the reproducibility in a potential large-scale OPBT production. In the following, answers to these questions will be given based on TCAD simulations.

Pinhole Diameter L Pin
Using TCAD simulations, we discuss the effect of a variation of the pinhole diameter L Pin on the device currents, threshold voltage, and subthreshold slope. We vary L Pin from 2 to 100 nm in order to cover the entire range of what is experimentally observed. Figure 5 shows the simulated transfer characteristics of an OPBT for different diameters of the pinholes L Pin at V CE = 1.5 V. The on-current is increasing slightly with the diameter of the pinholes, while the off-current remains almost constant up to L Pin = 50 nm and drastically increases with L Pin = 100 nm.
The reasons for an increased off-current for large L Pin are the reduced base control over the channel and increased short-channel effects.   Figure 6 shows the on-(V BE = 1.5 V) and off-(V BE = 0 V) current in more detail as a function of the pinhole diameter at V CE = 1.5 V. As can be seen in Figure 6b, the on-current increases from 500 up to 510 µA (2%) for a pinhole diameter variation of 100% from 6 to 12 nm. Therefore, the pinhole diameter is not the main limiting factor of the OPBT current I CE . This is in agreement with earlier findings that show the total device current in the on-state to reach a space-chargelimited current in the OSC. [18] A very small deviation of the threshold voltage (extracted as the intersection of the tangent at the maximum conductance) and subthreshold slope is shown as a function of pinhole diameter in Figure 6c. The threshold voltage is about V th = 0.72 V for a pinhole diameter of L Pin = 6 nm and V th = 0.70 V for L Pin = 12 nm. This means a shift in the threshold voltage of about 3% is expected with an increase of 100% in pinhole diameter. The subthreshold slope is about SS = 115 mV dec −1 for a pinhole diameter of L Pin = 6 nm and SS = 126 mV dec −1 for L Pin = 12 nm, resulting in a slope degradation of up to 10%. Figure 7 displays the charge and current density profiles of the OPBT with one single pinhole at V CE = V BE = 3 V. The whole device current I C has to pass through the pinhole, thus there is a high current density in this opening and the opening size might be a current-limiting factor. The charge and current densities have their maximum at the semiconductor/oxide interface due to the electrostatic and applied base bias V BE .

Number of Pinholes N Pin
The influence of a variation of the pinhole density from the targeted value of about 50 µm −2 in the fabricated real devices has been investigated to show the current flow deviation with   the number of pinholes per µm 2 . The deviation of DC characteristics (on-and off-current, threshold voltage V th , and subthreshold slope SS) of OPBTs is shown as a function of the pinhole density in Figure 8. Figure 8a shows that the on-current is increasing with the number of pinholes per µm 2 . It is, however, not linearly dependent on the number of pinholes. The current I C increases slightly with an additional pinhole. Hence, the on-current (V BE = 1.5 V) must saturate at a certain number of pinholes per µm 2 . Figure 8b shows that a variation of pinhole density from the target value of 50 µm −2 up to 100 µm −2 (an increase of 100%) has a negligible (almost 2%) impact on the device on-current. Thus, the OPBTs on-current does not significantly depend on the number of pinholes within a large range. The off-current I OFF (V BE = 0 V) remains almost constant for different pinhole densities (see Figure 8b), and also the impact on the threshold voltage and subthreshold slope is negligible (Figure 8c). Therefore, a very small deviation in the DC characteristics of OPBTs is expected with the variation of the number of pinholes per µm 2 .

Conclusion
In this work, it has been shown that even significant variations of the pinhole diameter and pinhole density from the targeted values have a negligible impact on OPBT DC characteristics. For a wide parameter range, the OPBT performance is not limited by the pinholes, but rather by the vertical current transport in the OSC.
Stable and reproducible DC characteristics can be expected and achieved regardless of statistical variations in the fabrication of the base layer in OPBTs in terms of a deviation of pinhole diameter and density.