Interface control is the key for organic optoelectronic devices, as it directly impacts charge transport, recombination and overall device efficiency[1-12]. In particular, organic-light emitting diodes consist of multiple stacked semiconductor layers, where precise interface manipulation is essential[8, 13]. A critical factor for enhancing performance is minimizing the energy offset for charge injection (or extraction), which is an important parameter in optoelectronic devices[11, 12, 14-20].

In organic semiconductors, the widely accepted understanding is that the semiconductor’s Fermi level (EF) is determined primarily by its interface with adjacent materials due to the absence of dangling bonds[11]. Under these conditions, the EF pinning arises when the work function (Φ) of the electrode exceeds the semiconductor’s ionization energy (IE) or is lower than its electron affinity (EA), as a result of interfacial electron exchange. However, certain organic semiconductors may not fully adhere to these rules.

For example, 2,3,5,6-tetrafluoro-7,7,8,8-tetracyano quinodimethane (F4-TCNQ) and 1,4,5,8,9,12-hexaaza triphenylene-2,3,6,7,10,11-hexacarbonitrile (HAT-CN) increases the Φ by forming an interface dipole that effectively shifts the local vacuum level upward at the substrate surface[21-25]. Typically, these strong electron acceptors are introduced onto the substrate (e.g. indium tin oxide (ITO)) to increase the effective Φ by inducing the redistribution of electron density at the interface.

However, a fixed Φ attributed to EF pinning has also been observed, even when the substrate's Φ exceeded the EA of HAT-CN. This implies that, beyond the commonly assumed permanent dipole layer, gap states play a primary role in electron donation at the interface[21, 26]. This recent finding calls for a reconsideration of the conventional hole injection mechanism when strong electron acceptor materials (e.g. HAT-CN) are applied between the substrate and the hole transport layer.

To address the issue, we investigated the electrical characteristics—specifically current density-voltage (J-V) and impedance-voltage (Z-V) behaviors—using a hole-only device structured as Al/NPB/HAT-CN/ITO. Furthermore, the energy level alignment at the NPB/HAT-CN/ITO interface was examined using in situ ultraviolet photoelectron spectroscopy (UPS). Our findings suggest that the hole injection at the NPB/HAT-CN/ITO interfaces occurs via tunneling.

ITO substrates were cleaned with ultrasonication in deionized (DI) water, detergent, acetone, methanol, and DI water again. After wet-cleaning process, ITO substrates were exposed to UV-ozone for 20 minutes at 100 °C. organic semiconductors (100 nm NPB, 3 nm HAT-CN) are thermally evaporated in the ultrahigh vacuum (UHV) chamber at a base pressure below 1.0 × 10​-8 Torr at a deposition rate of 0.1 nm/s for the device fabrication. Finally, 100 nm Al cathode was deposited onto the organic layer. The rate of deposition and total thickness were obtained by calibrated quartz crystal microbalance. The device active area was 2×2 mm​2 and J-V were measured using a Keithley 2400 source-measure meter. Temperature dependent J-V measurements were conducted in 1.0 × 10​-3 Torr. Impedance measurements were performed using a Solartron 1260 impedance/gain-phase analyzer with a 25 mV oscillation.

In situ UPS measurements were conducted using PHI 5700 spectrometer with the deposition chamber connected directly to the analysis chamber. An ultraviolet discharge lamp was used as an excitation source (He Ⅰ, 21.22 eV), and sample bias of -10 V was applied to obtain the secondary electron cutoff (SECO) in normal emission geometry. For the in situ measurements, organic semiconductor deposited stepwise in a deposition chamber with a 0.01 nm/s.

To examine the hole injection behavior, we start by studying the current-voltage (J-V) characteristics of hole-only device with the structure Al/NPB (100 nm)/HAT-CN (3 nm)/ITO. As shown in Fig. 1(a), the device with HAT-CN layer exhibits higher current density than without HAT-CN layer, which is also higher current density compared to the CoPc device[3, 5].

Figure 1. (Color online) (a) J-V characteristics of hole-only device with HAT-CN (red line) and without HAT-CN (dotted gray line). The inset depicts the device structure and voltage direction, with holes injected from the ITO side. The organic active layers are 100 nm thick for NPB and 3 nm thick for HAT-CN. (b) temperature-dependent J-V characteristics of hole-only device with HAT-CN from 100 K to 300 K in 20 K intervals. (c) impedance magnitude (|Z|)-V, phase angle (arg(Z))-V, and capacitance (C)-V curves of the hole-only device with HAT-CN.

In Fig. 1(a), injection-limited current behavior is observed in the low bias regime (below 1 V) in the device with HAT-CN, indicating a low energy barrier that allows holes to begin injecting into the active layer. As the voltage increases, the injected charges at the low bias regime transition from injection-limited current to bulk-limited current. In the intermediate regime (1 V–2 V), the current rises steeply due to trap state filling, transitioning into a space-charge-limited current (SCLC) regime with a characteristic slope of 2. As shown in Fig. 1(b), bulk conduction behavior diminishes at low temperature since the weak thermal energy at low temperature is insufficient to fill the trap states and achieve SCLC behavior. Typically, organic semiconductor devices exhibit temperature-dependent current behavior in low bias regime, as thermal energy is required to inject charge carriers[14]. However, the current behavior observed in the device with HAT-CN layer is independent of temperature, which is an unusual characteristic in organic semiconductor devices and suggests direct tunneling injection[3].

It should be noted that this result contradicts the widely accepted view that the HAT-CN layer is described as a charge-generation layer (CGL), which enables hole injection into the semiconductor by extracting electrons from the semiconductor to the metal, typically requiring the overcoming of an energy barrier to inject the holes. However, our results differ from that explanation because the CGL mechanism is inherently temperature dependent, whereas our findings indicate temperature-independent current injection. To gain a further insight into this phenomenon, we examined the impedance–voltage (Z-V) characteristics as shown in Fig. 1(c). We observed that the impedance magnitude (|Z|) decreases and the phase (arg(Z)) approaches 0° as the bias voltage increases, indicating that holes are conducted along the organic semiconductor layers. In this context, capacitance (C) increases in the intermediate voltage regime (1 V–2 V), where injected hole begin filling the trap states, and then decreases as conduction becomes fully established along the bulk organic layers.

Notably, the phase is already -60° at zero bias, which is hard to explain with CGL mechanism. This suggests two possible conduction mechanisms at the interface: trap-assisted conduction or tunneling. Since trap-assisted conduction is strongly influenced by thermal activation energy[14], we rule it out in this scenario based on temperature-dependent J-V results. Consequently, we conclude that tunneling dominates the conduction in the low-bias regime in hole-only device with HAT-CN.

To further understand the conduction mechanism, we investigated the electronic properties of HAT-CN as its thickness increased on the ITO substrate. As shown in Fig. 2, the Φ of bare ITO was determined to be 4.87 eV, derived from the SECO region. With the deposition of a 3 nm-thick HAT-CN layer, the Φ increased to 5.17 eV. Interestingly, the highest occupied molecular orbital (HOMO) onset was observed at 3.47 eV below the EF for the 3 nm-thick HAT-CN layers, with no significant shift detected in the valence region. We note that the overall valence band peak of 3 nm-thick HAT-CN spectrum shifts by 0.2 eV to higher binding energy and becomes broader compared to 20 nm-thick HAT-CN spectrums[4]. This observation may suggest weaker intermolecular screening effects in the thinner HAT-CN layer, with interface screening effects potentially having a significant influence. These results collectively indicate that the electronic properties of the 3 nm-thick HAT-CN layers are dominated by both interfacial effects and change the charge screening behavior. This suggests that while the HOMO level increased with HAT-CN deposition, the valence region remained stationary. The linear shift of the vacuum level as a function of HAT-CN thickness can be explained by the presence of an interface dipole, as shown in Fig. 2(b). As depicted in Fig. 2(c), a gap state near EF was identified, which we attribute to negatively charged HAT-CN molecules[21]. This charged state of HAT-CN likely contributes to the formation of a strong interfacial dipole, providing a clear explanation for the significant increase in Φ.

Figure 2. (Color online) (a) UPS spectra of HAT-CN layer (0.2, 0.6, 1.0, 1.5, 2.0, 3.0 nm) on ITO substrate. (b) variation of work function (Φ) as function of the HAT-CN thickness. (c) valence region on a logarithmic scale. Broad gap states (red shaded region) are identified below the HOMO level of the 3 nm HAT-CN layer.

To obtain further insight, we investigated the electronic properties of NPB on the 3 nm HAT-CN substrate. As shown in Fig. 3, with the deposition of a 2.8 nm-thick NPB layer, the Φ decreased by 0.23 eV and saturated at 4.94 eV. In the HOMO region, the HOMO features of NPB gradually increased and were identified near 1 eV as the layer thickened. The HOMO onset of NPB was observed at 0.49 eV below the EF for the 3.8 nm-thick HAT-CN layers, with no significant shift detected in the valence region, unlike the SECO change, which implies that NPB did not fully cover the surface of the HAT-CN layer below 2.8 nm. Notably, this value aligns with monolayer thickness of NPB, which is approximately 2.5–3 nm.

Figure 3. (Color online) UPS spectra of NPB layer (0.3, 0.6, 1.2, 1.8, 2.8, 3.8 nm) on the 3 nm HAT-CN substrate.

By combining all UPS spectra (Figs. 2 and 3), we constructed the energy level diagram of the ITO/HAT-CN/NPB interface, is illustrated in Fig. 4. The IE of the 3 nm-thick HAT-CN and NPB layer was determined to be 8.64 eV and 5.43 eV, which is slightly smaller than previously reported values[4, 23]. This discrepancy can be attributed to the effects of the less screening effect in extremely thin layers such as sub-monolayer thickness of the HAT-CN layer. The lowest unoccupied molecular orbital (LUMO) position in the diagram was taken from previous values for both molecules via inverse photoemission spectroscopy; however, the 3nm-thick HAT-CN molecule might exhibit a different energy position, which require further investigation. The hole injection barrier (Φh) is calculated to be 0.49 eV, if we consider the CGL mechanism at the HAT-CN/NPB interface electrons should overcome an energy barrier of approximately 0.69 eV (Φe = 0.2 eV)[26]. However, our experimental results suggest that the presence of a charged HOMO state and ultrathin HAT-CN layer enable efficient hole injection via direct tunneling conduction to the NPB layer.

Figure 4. (Color online) Schematic energy level diagram of the ITO/HAT-CN/NPB interface (unit: eV).

This mechanism enables hole injections without requiring thermal activation energy, despite the large bandgap of HAT-CN. Furthermore, due to the dipole effect of the thin HAT-CN layer, the HOMO level of NPB is strongly pinned near the EF, facilitating efficient hole injection at the HAT-CN/NPB interface. Notably, the proposed direct conduction mechanism addresses the hole injection process through the thin HAT-CN layer without involving the CGL mechanism.

We demonstrated that the thin HAT-CN layer at the ITO/NPB interface plays a crucial role in enabling efficient hole injection through direct tunneling conduction. This mechanism contrasts with the conventional charge generation layer (CGL) model, which relies on thermal activation energy. Our experimental results reveal that the presence of gap states near the Fermi level (EF) is attributed to sub-monolayer thickness and reduced intermolecular screening effects. These findings provide a novel perspective on the charge injection mechanisms and offer future insights for optimizing organic optoelectronics devices using the HAT-CN layer.

This research was supported by Global-Learning & Academic research institution for Master’s · PhD students, and Postdocs(LAMP) Program of the National Research Foundation of Korea(NRF) grant funded by the Ministry of Education(No. RS-2024-00442775).