Organic semiconductors have recently been used in various optoelectronic devices, including organic photovoltaics (OPVs) and organic light-emitting diodes (OLEDs) [1,2]. In organic semiconductors, molecules interact via van der Waals forces; thus, free charge carriers cannot be generated using thermal energy. Therefore, charge carriers must be injected from electrodes or organic layers to operate the device. This implies that energy level alignment plays an important role in device performance [3–6]. Understanding and engineering this energy-level alignment requires accurate measurement of the electronic structure.

Among the various combinations of organic semiconductors, organic p-n junctions are the most widely used. For example, in OPVs, organic p-n junctions are employed to separate the excitons generated by light absorption. The energy difference between the highest occupied molecular orbital (HOMO) level of a p-type organic semiconductor and the lowest unoccupied molecular orbital (LUMO) level of an n-type organic semiconductor (hereafter HOMO_p−LUMO_n) typically determines the open-circuit voltage (VOC) because the maximum quasi-Fermi level splitting reaches the HOMO_p−LUMO_n [7–11]. To increase VOC, the HOMO_p−LUMO_n should be maximized by selecting a high ionization energy (IE) of a p-type organic semiconductor and a low electron affinity of an n-type organic semiconductor. In addition, organic p-n junctions have been used as charge-generation layers in tandem OLEDs. In the charge-generation layer, an electron from the HOMO level of a p-type organic semiconductor is injected into the LUMO level of an n-type organic semiconductor by an applied bias, creating a hole (a process known as ‘charge generation’) [12–14]. The remaining holes and injected electrons moved in opposite directions in their respective layers, generating charge currents. Here, HOMO_p−LUMO_n represents the energy barrier for charge generation, which must be minimized to achieve efficient device performance. Consequently, the determination of HOMO_p−LUMO_n and its impact on device performance have been extensively studied.

Some studies have reported that the work function of an electrode can affect the HOMO_p−LUMO_n in organic p-n heterojunctions. For example, Zou et al. [15] reported a significant change in the HOMO_p− LUMO_n of a pentacene/C₆₀ interface, from 0.86 eV on an indium tin oxide (ITO) substrate to 1.54 eV on an ITO/MoO₃ substrate. Similarly, Zhou et al. [16] reported that the HOMO_p−LUMO_n of a copper phthalocyanine/C₆₀ interface varied from 1.03 eV on a Mg substrate to 0.66 eV on an ITO substrate. Such inconsistencies complicate the prediction of HOMO_p−LUMO_n and pose a challenge to the design of high-performance organic devices. Therefore, further studies on energy-level alignment are required to elucidate the principles of HOMO_p−LUMO_n formation.

In this study, the energy-level alignment of pentacene/C₆₀ on an ITO substrate was investigated. The electronic structures at the valence and core levels were determined by in-situ ultraviolet photoelectron spectroscopy (UPS) and X-ray photoelectron spectroscopy (XPS) measurements. UPS is the most accurate technique for determining charge transport levels. In addition, the in-situ method allows for the accurate probing of energy levels without unwanted external effects from contamination. The HOMO_p−LUMO_n of pentacene/C₆₀ was estimated and compared with literature values to understand the origin of these discrepancies.

The ITO substrate was cleaned via ultrasonication in deionized (DI) water, detergent, methanol, acetone, and DI water baths for 10 min each. After wet cleaning, the ITO substrate was dried with a flow of N₂ gas. The ITO substrate was transferred to an entry chamber. The ITO surface was further cleaned by Ar-ion sputtering until the XPS C 1s signal attributed to hydrocarbon contamination was eliminated (not shown here) [17]. In-situ UPS and XPS measurements were performed using a PHI 5700 spectrometer (Physical Electronics Inc., United States) equipped with a He I_α (h_v = 21.22 eV) discharge lamp and an Al K_α (h_v = 1486.7 eV) X-ray source. Pentacene (purity: 99.999 %, Tokyo Chemical Industry Co. Ltd., Japan) and C₆₀ (purity: 99.9 %, Sigma-Aldrich Inc., United States) were deposited stepwise in a preparation chamber using a Knudsen cell at a deposition rate of 0.01 nm s⁻¹. The deposition rate and total thickness were monitored using a quartz-crystal microbalance. After each deposition step, the sample was transferred to an analysis chamber without breaking the vacuum, and the UPS spectra were recorded [18]. This process was repeated until the thickness was sufficient to achieve no spectral shift. The base pressures of the analysis and preparation chambers were 5 × 10⁻⁹ and 1 × 10⁻⁷ Torr, respectively. A sample bias of −15 V was applied to obtain the secondary electron cutoff (SEC).

Figure 1 shows the in-situ UPS spectra of the (a) SEC, (b) HOMO, and (c) magnified HOMO regions of pentacene at different thicknesses (0.3, 0.5, 1.0, 2.0, 4.0, 8.0, and 16.0 nm) on ITO. In the bottommost spectrum, the work function of ITO was measured to be 4.02 eV. As the pentacene layer was deposited, the work function gradually decreased. The work functions for the pentacene layers of 0.3, 0.5, 1.0, 2.0, 4.0, 8.0, and 16.0 nm thickness were 3.77, 3.74, 3.68, 3.66, 3.67, 3.66, and 3.67 eV, respectively. Therefore, the work function shift was saturated at the 1.0 nm thickness and the values remained consistent (3.66–3.68 eV) for thicker layers. In Figs. 1(b) and 1(c), the He I_β was removed from the measured spectrum to accurately determine the HOMO onset. In the bottommost spectrum, the valence band of ITO, mainly derived from the O 2p orbitals, was observed at 3–8 eV [19], with a valence band maximum at 3.24 eV. The emission features around 2 eV can be attributed to the defect states generated by Ar-ion sputtering. As the pentacene layer was deposited, the HOMO peak of pentacene emerged at approximately 1.7 eV, with other characteristic emission features at 2.9, 3.4, 4.0, 4.7 eV, etc. [20]. The intensities of these peaks increased with the thickness of the pentacene layer. As shown in Fig. 1(c), the Fermi step of the ITO was observed at 0 eV, which served as the energy reference. The HOMO onset of pentacene was determined to be 1.19 eV at a thickness of 2.0 nm, which is higher than the estimated molecular size of pentacene. Based on the density functional theory (DFT) calculations, the molecular size of pentacene was estimated to be 1.4 nm (data not shown). With increasing pentacene thickness, the HOMO onset gradually shifts toward a higher binding energy. For the 4.0, 8.0, and 16.0 nm-thick pentacene layers, the HOMO onsets were observed at 1.23, 1.25, and 1.30 eV, respectively. This shift in the HOMO level indicates that the band bending of the pentacene layer at the interface with ITO was 0.11 eV.

Figure 1. In-situ UPS spectra of the (a) SEC, (b) HOMO, and (c) magnified HOMO regions of pentacene (0.3, 0.5, 1.0, 2.0, 4.0, 8.0, and 16.0 nm) on ITO (inset: chemical structure of pentacene).

The interfacial electronic structure of C₆₀ on the ITO/pentacene was also investigated. Figure 2 shows the in-situ UPS spectra of the (a) SEC, (b) HOMO, and (c) magnified HOMO regions of C₆₀ at different thicknesses (0.5, 1.0, 2.0, 4.0, 8.0, 16.0, and 32.0 nm) on ITO/pentacene (16.0 nm). As shown in Fig. 2(a), the work function gradually increased as the C₆₀ layer was deposited. For the 0.5, 1.0, 2.0, 4.0, 8.0, 16.0, and 32.0 nm-thick C₆₀ layers, the work functions were 3.88, 3.97, 4.16, 4.27, 4.33, 4.38, and 4.40 eV, respectively. In Fig. 2(b), the HOMO peak of C₆₀ is observed at approximately 2.6 eV, with other characteristic emission features at 3.9, 5.8, 6.1, 7.4 eV, etc. [21]. In Fig. 2(c), the HOMO onset of pentacene shifted by 0.30 eV toward a lower binding energy until the 1.0 nm-thick C₆₀ layer was deposited, indicating band bending in the pentacene layer. Pentacene features were undetectable with C₆₀ layers thicker than 1.0 nm (the weak features between the C₆₀ HOMO and the Fermi level are attributed to the residual He I_β signal). Meanwhile, the intensities of the C₆₀ peaks were high compared to those of the pentacene peaks because of the high symmetry of C₆₀. According to the DFT results, the molecular size of C₆₀ was estimated to be 0.7 nm; thus, the HOMO onset was determined at higher thicknesses. At a thickness of 1.0 nm, the HOMO feature of C₆₀ is observable. For the thicknesses of 1.0, 2.0, 4.0, 8.0, 16.0, and 32.0 nm, the HOMO onsets of C₆₀ were observed at 2.33, 2.28, 2.23, 2.17, 2.12, and 2.08 eV, respectively. Therefore, the HOMO shift of C₆₀ was 0.25 eV, indicating band bending in the C₆₀ layer.

Figure 2. In-situ UPS spectra of the (a) SEC, (b) HOMO, and (c) magnified HOMO regions of C₆₀ (0.5, 1.0, 2.0, 4.0, 8.0, 16.0, and 32.0 nm) on ITO/pentacene (16.0 nm) (inset: chemical structure of C₆₀).

The energy-level shifts of pentacene and C₆₀ were further investigated by XPS. Figure 3 shows the XPS C 1s spectra of C₆₀ films with different thicknesses (0.5, 1.0, 2.0, 4.0, 8.0, 16.0, and 32.0 nm) on ITO/pentacene (16.0 nm). The XPS C 1s peak of pentacene is observed at 284.0 eV. Upon the deposition of the C₆₀ layer, the C 1s peak of C₆₀ appeared at a higher binding energy. Concurrently, the pentacene peak shifts toward a lower binding energy. Although both pentacene and C₆₀ are composed solely of C atoms, their C 1s peaks are located at different binding energies, allowing their spectral features to be distinguished [22]. As the C₆₀ layer became thicker, both the C 1s peaks gradually shifted toward lower binding energies. At the 4.0 nm-thick C₆₀ layer, the pentacene C 1s peak was undetectable. The discrepancy in the C₆₀ thicknesses at which the pentacene features disappeared between the UPS and XPS measurements was attributed to their different probing depths. The C₆₀ peak continued to shift toward a lower binding energy, reaching 284.3 eV for the 32.0 nm-thick C₆₀ layer. These peak shifts of both pentacene and C₆₀ toward lower binding energies support the presence of significant band bending at the pentacene/C₆₀ interface.

Figure 3. In-situ XPS C 1s spectra of C₆₀ (0.5, 1.0, 2.0, 4.0, 8.0, 16.0, and 32.0 nm) on ITO/pentacene (16.0 nm).

Based on the UPS results, an energy-level diagram for ITO/pentacene/C₆₀ was constructed, as shown in Fig. 4. The transport gaps of pentacene and C₆₀ (2.20 and 2.30 eV, respectively) were obtained from [14]. The hole- and electron-injection barriers from ITO to pentacene were estimated to be 1.19 and 0.91 eV, respectively. At the ITO/pentacene interface, the band bending estimated from the HOMO level shift was 0.11 eV. The interface dipole was calculated using the equation of eD = ΔSEC − V_b, where eD is the interface dipole, ΔSEC is the SEC shift, and V_b is the band bending [23]. Therefore, the dipole at the ITO/pentacene interface was 0.24 eV. The IE of pentacene was calculated to be 4.97 eV. Similarly, at the pentacene/C₆₀ interface, the band bending of pentacene and C₆₀ was 0.30 and 0.25 eV, respectively, and the interface dipole was 0.18 eV. The IE of C₆₀ was determined to be 6.48 eV. It was observed that p-type pentacene and n-type C₆₀ formed a type-II alignment, which was favorable for exciton separation and charge generation. The upward band bending of pentacene and the downward band bending of C₆₀ at the interface were attributed to charge transfer for thermal equilibrium. Consequently, HOMO_p−LUMO_n was estimated to be 0.97 eV. The position of the LUMO of C₆₀ slightly below the Fermi level, is probably owing to the uncertainty in the transport gap caused by the large spectral broadening in inverse photoelectron spectroscopy.

Figure 4. Energy-level diagram of ITO/pentacene/C₆₀ derived from the UPS results.

Finally, the HOMO_p−LUMO_n values of the pentacene/C₆₀ interface measured in this study were compared with those reported in the literature. Figure 5 shows the HOMO_p−LUMO_n values as functions of the substrate work function [14,15,22,24]. For a reliable comparison, the transport gap of C₆₀ was assumed to be constant at 2.30 eV. The error bar was estimated to be 0.1 eV based on the spectral broadening of the Fermi level of clean Au. Although a slight deviation was observed, possibly owing to the method used to determine the band bending, most of the HOMO_p−LUMO_n values are close to 0.90 eV, regardless of the substrate work function. However, one case stands out with a significantly higher value (1.54 eV for MoO₃) [15]. This can be attributed to the thinness of the pentacene layer (7.5 nm). On the high-work-function MoO₃ substrate, the pentacene HOMO level was pinned, and significant charge transfer occurred between MoO₃ and pentacene. In addition, pentacene grows on MoO₃ in 3D growth mode, resulting in thinner regions within the pentacene layer [25]. In this case, some of the pentacene molecules in contact with C₆₀ may become charged, changing their charge neutrality level, and subsequently affecting the energy-level alignment with the adjacent C₆₀. However, although the pentacene layer was deposited on MoO₃, the HOMO_p−LUMO_n showed 0.83 eV when the thickness of the pentacene layer is sufficiently thick (20 nm) [14]. Most of the charged molecules are located near the substrate, whereas there are no charged molecules on the surface. In this case, the energy-level alignment did not differ significantly from those in the other cases. Therefore, if the p-type organic layer is thin, the substrate work function can significantly influence HOMO_p−LUMO_n. However, if the p-type organic layer is thick, the substrate work function does not significantly affect the HOMO_p−LUMO_n.

Figure 5. HOMO_p−LUMO_n values of the pentacene/C₆₀ interface as a function of the substrate work function. Reference number, substrate, and pentacene thickness are also presented.

The energy-level alignment of the pentacene/C₆₀ organic p-n junction was investigated using in-situ UPS and XPS measurements. At the ITO/pentacene interface, the hole and electron injection barriers were determined to be 1.19 and 0.91 eV, respectively. In addition, the interface dipole and band bending were measured to be 0.24 and 0.11 eV, respectively. At the pentacene/C₆₀ interface, significant band bending resulting from charge transfer was observed, 0.30 eV in the pentacene layer and 0.25 eV in the C₆₀ layer. The interface dipole was estimated to be 0.18 eV. This type-II alignment is favorable for exciton dissociation and charge generation. On ITO (with a work function of 4.02 eV), the HOMO_p−LUMO_n of pentacene/C₆₀ was measured to be 0.97 eV, which is similar to the reported value. However, when the thickness of the pentacene layer is low, the substrate work function can affect HOMO_p−LUMO_n. Further studies on various organic p-n junctions with different substrate work functions will provide deeper insights into the detailed mechanisms underlying HOMO_p−LUMO_n formation.

This study was supported by the National Research Foundation of Korea (Grant No. NRF-2021R1A2C1009324 and 2018R1A6A1A0302-5582), and the Semiconductor R&D Support Project through the Gangwon Technopark (GWTP) funded by Gangwon Province (No. GWTP 2023-027).

The authors declare no conflicts of interest.