Efficient energy-level alignment is essential for organic light-emitting diodes (OLEDs) to minimize the energy difference between the charge-transport level of the charge transport layer and the Fermi level of the electrode [1–4]. Therefore, high/low work-function anodes and cathodes are required. However, conventional electrode materials do not have sufficiently high or low work functions. Thus, charge-injection layers that modify the work function have been employed. MoO₃ is a popular hole injection layer (HIL) for efficient anode systems because of its high work function [5]. However, the accurate charge injection and transport mechanisms of MoO₃ HIL are still under investigation. Kröger et al. suggested that the electrons in the highest occupied molecular orbital (HOMO) of an adjacent organic hole transport layer (HTL) are withdrawn by an applied bias, resulting in effective hole injection [6]. Conversely, Yi et al. [7] suggested that the gap states of the MoO₃ HIL close to the Fermi level can assist in hole injection from the anode to the organic HTL. Importantly, a HOMO level close to the Fermi level was required in both cases.

However, the work function of the MoO₃ HIL varied significantly with its thickness [7–12]. Although a previous device study found that the highest efficiency of OLEDs was achieved with a 5 nm thick MoO₃ HIL [13], this thickness may not be sufficient to saturate the increase in the work function of the anode system. The HOMO level of the HTL can be further decreased by increasing the thickness of the MoO₃ HIL. A recent study reported that a sufficiently thick layer is necessary to decrease the HOMO level of an HTL with a 1,4,5,8,9,11- hexaazatriphenylene hexacarbonitrile (HAT-CN) HIL, which has an electronic structure similar to that of the MoO₃ HIL [14]. This highlights the importance of understanding the energy-level alignment with different thicknesses of the MoO₃ HIL in formulating a strategy to further enhance OLED performance.

In this study, the energy-level alignment at indium tin oxide (ITO)/MoO₃ (20 nm)/N,N′-di(1-naphthyl)-N,N′-diphenyl-(1,1′- biphenyl)-4,4′-diamine (NPB) interfaces was analyzed using in situ X-ray and ultraviolet photoelectron spectroscopy (XPS and UPS, respectively). NPB was selected as the HTL material because of its widespread use in OLEDs, and a 20 nm thickness of MoO₃ layer was chosen to saturate the work-function increase. The measured HOMO level of the NPB HTL was compared with that of ITO/MoO₃ (5 nm)/NPB interfaces reported previously [5]. The change in the HOMO level of the NPB HTL provides useful information for designing device architectures.

The details of our in situ photoelectron spectroscopy system were described elsewhere [15]. The system consisted of deposition and analysis chambers that are connected via a gate valve, allowing for the sample to be transported between them without breaking the vacuum. The deposition chamber was equipped with Knudsen cells for the thermal evaporation of materials and a quartz crystal microbalance (QCM) for monitoring the deposition rate and total thickness. The analysis chamber was equipped with a PHI 5700 spectrometer (Physical Electronics Inc., USA), an Al K_α X-ray source (hv = 1486.7 eV), a He Iα discharge lamp (hv = 21.22 eV), and an Ar⁺ ion gun. The Fermi level of the spectrometer was calibrated using clean Au. The base pressures of deposition and analysis chambers were 5 × 10⁻⁸ and 1 × 10⁻⁹ Torr, respectively.

The ITO-coated substrate was cleaned via ultrasonication in deionized (DI) water, detergent, acetone, ethanol, and DI water. The cleaned ITO was dried under a N₂ gas flow and loaded into the entry chamber. The ITO was then sputter cleaned with an Ar⁺ ion gun in the analysis chamber to remove surface hydrocarbon contaminants [16]. MoO₃ (99.97%, Sigma-Aldrich Inc., USA) and NPB (> 99.0%, Tokyo Chemical Industry Co., Ltd. Japan) were evaporated at rates of 0.1 and 0.01 nm s⁻¹, respectively, onto the ITO in a stepwise manner. At each deposition step, the sample was transported to the analysis chamber and the XPS and UPS spectra were recorded. During the UPS measurements, a sample bias of −15 V was applied to determine the secondary electron cutoff (SEC).

Figure 1 shows the XPS core-level spectra of the (a) Mo 3d, (b) C 1s, (c) N 1s, and (d) O 1s regions of the ITO/MoO₃ (20 nm)/NPB (0.3, 0.5, and 2.0 nm). The In 3d signal was not detected (not shown here) because the MoO₃ layer was thicker than the XPS probing depth. Mo 3d_5/2 and 3d_3/2 doublet peaks were observed in Fig. 1(a) at 232.9 and 236.0 eV, respectively. Each peak could be fitted with a single Gaussian/ Lorentzian mixing curve with a ratio of 0.3 and full width at half maximum (FWHM) of 1.4 eV (fitting not shown here). The intensity ratio of Mo 3d_5/2 to 3d_3/2 was 3:2, which is consistent with the spinorbit splitting factor. Moreover, no reduced Mo 3d peaks (e.g., Mo⁵⁺) because of oxygen deficiency were detected. Reduced Mo 3d peaks are typically observed in ultrathin MoO₃ layers on metal substrates because of the electron transfer between the metal and MoO₃ [9,17]; only single Mo⁶⁺ peaks were observed in this study owing to the sufficiently thick thickness of the deposited MoO₃ layer. The NPB deposition increased the FWHM to 1.7 eV, implying electron transfer from NPB to MoO₃. The peak intensity gradually decreased with increasing NPB layer thickness. The thickness of the NPB layer was estimated by measuring the decrease in the Mo 3d peak intensity upon the NPB deposition [18]. The NPB thickness d was calculated using the following equation:

Figure 1. XPS core-level spectra of (a) Mo 3d, (b) C 1s, (c) N 1s, and (d) O 1s regions of ITO/MoO₃ (20 nm)/NPB (0.3, 0.5, and 2.0 nm).

where λ is the electron attenuation length, φ is the take-off angle (45° in the current measurement), I0 is the Mo 3d intensity of the 20 nm thick MoO₃ sample, and I is the Mo 3d intensity of NPB on the 20 nm thick MoO₃ sample. The λ values were obtained from the NIST database [19]. The calculation yielded a thickness of 1.8 nm for the 2.0 nm thick NPB layer, verifying the reliability of the thickness estimation using QCM. In Fig. 1(b), no C 1s peak is observed in the 20 nm thick MoO₃ layer, indicating minimal hydrocarbon contamination. With the NPB deposition, two C 1s peaks originating from the C–C and C–N bonds appeared at 284.3 and 285.6 eV, respectively, and their intensities increased until a 2.0 nm thickness. In Fig. 1(c), the Mo 3p peak was observed at 398.6 eV in the 0.3 nm thick NPB layer. With NPB deposition, the spectral features were separated into two peaks owing to the emergence of the N 1s peak of the amine moiety in the NPB molecule. In Fig. 1(d), the O 1s peak was observed at 530.5 eV in the 20 nm thick MoO₃ layer and could be fitted with a single Gaussian/Lorentzian mixing curve with a ratio of 0.3 and FWHM of 1.6 eV. This indicates that the oxygen deficiency was not easily detectable by XPS measurements. With the deposition of the 2.0 nm thick NPB layer, the peak width changed marginally, and the peak position shifted slightly toward a high binding energy (0.1 eV), which could be related to electron transfer. The presence of the Mo 3d, N 1s, and O 1s peaks was attributed to the NPB layer thickness being lower than the XPS probing depth.

Figure 2 shows the UPS spectra of the (a) SEC, (b) valence band (VB), and (c) magnified VB regions of the ITO/MoO₃ (20 nm)/NPB (0.3, 0.5, and 2.0 nm). In determining the SEC and HOMO onset energies, system broadening (0.1 eV), evaluated using the broadening of the Fermi step of clean Au, was considered [20]. The SEC region spectra were normalized to clearly show the SEC shift and drawn with respect to the kinetic energy scale to indicate the work function. The Shirley-type background was removed from the VB spectrum to accurately determine the HOMO onset. In the SEC region [Fig. 2(a)], the work function of ITO was measured at 3.96 eV. The deposition of a 20 nm thick MoO₃ layer resulted in a significant increase in the work function which reached 6.48 eV. This indicates that the 20 nm thick MoO₃ layer increases the work function by 2.52 eV. Upon deposition of the NPB layer, the SEC gradually decreased to 5.48, 5.17, and 5.03 eV, for the 0.3, 0.5, and 2.0 nm thicknesses, respectively. In the VB region [Fig. 2(b)], the O 2p valence band of the bottommost ITO spectrum with a range of 3.2–7.2 eV is shown. Upon deposition of the 20 nm thick MoO₃ layer, the spectral shape was changed to that of the MoO₃, and its O 2sp valence band was observed with a range of 2.5– 5.5 eV [9,17]. The NPB layer deposition led to the characteristic NPB peaks observed at 0.9, 1.3, 2.3, 3.0 eV, and so forth. The 2.0 nm thick NPB layer spectrum showed almost only NPB features, indicating that a sufficiently thick film was deposited. To determine the HOMO onset of NPB, the magnified VB region spectra are shown in Fig. 2(c). In the ITO spectrum, a Fermi step of 0 eV is observed owing to its metallic properties. In the 20 nm thick MoO₃ layer, the gap state of MoO₃ in the range of 0–2.3 eV was observed. This state originated from its oxygen-deficient nature [9,17,21,22], which could not be detected by XPS analysis (Fig. 1). In the 0.3 nm thick NPB HTL, NPB HOMO features appeared and significantly overlapped with the MoO₃ gap state. To reliably determine energy-level alignment, the energy levels must be measured layer-by-layer. Therefore, the HOMO of the NPB layer was evaluated at a thickness of 0.5 nm. In the 0.5 nm thick NPB layer, the HOMO level was observed at 0.37 eV. At an NPB layer thickness of 2.0 nm, the NPB HOMO level shifted to 0.47 eV. This indicates the band bending of 0.10 eV at the NPB HTL interface. Meanwhile, spectral features observed with NPB deposition below the Fermi level are attributed to the He Iβ excitation and should not be considered when analyzing hole injection properties.

Figure 2. UPS spectra of (a) SEC, (b) VB, and (c) magnified VB regions of ITO/MoO₃ (20 nm)/NPB (0.3, 0.5, and 2.0 nm).

Figure 3 shows the energy-level diagrams of (a) ITO/MoO₃ (20 nm)/NPB and (b) ITO/MoO₃ (5 nm)/NPB. The energy-level alignment of ITO/MoO₃ (20 nm)/NPB was determined based on the results of this study, whereas that of ITO/MoO₃ (5 nm)/NPB was obtained from a previous report [5]. The interface dipole was calculated using the equation of eD = ΔSEC − Vb, where eD, ΔSEC, and Vb represent the interface dipole, SEC shift, and band bending, respectively [23]. The lowest unoccupied molecular orbital (LUMO) level of NPB was estimated using the transport gap measured by inverse photoelectron spectroscopy [24]. The ionization energies of NPB were almost the same (~5.5 eV) in both cases. The work functions of the ITO were almost identical because the same cleaning methods were used. Therefore, the changes in the energy-level alignment were solely owing to the MoO₃ layer thickness. Differences in the thicknesses of the MoO₃ layers resulted in different work functions. The 20 nm thick MoO₃ layer formed an interface dipole of 2.52 eV, resulting in a work function of 6.48 eV. In contrast, the work function of the 5 nm thick MoO₃ layer was 6.33 eV, which is 0.15 eV lower than that of the 20 nm thick MoO₃ layer. This demonstrates that a thick MoO₃ layer results in a high work function. Following NPB layer deposition, an interface dipole of 1.35 eV was formed on the 20 nm thick MoO₃ layer, and the HOMO level of NPB was observed at a binding energy of 0.37 eV. The formation of an interface dipole was attributed to electron transfer from NPB to MoO₃, as confirmed by the broadening of the XPS Mo 3d peaks [Fig. 1(a)]. However, for the NPB layer on the 5 nm thick MoO₃ layer, an interface dipole of 1.75 eV was formed, and the HOMO level of NPB was observed at a binding energy of 0.89 eV. That is, by increasing the thickness of the MoO₃ layer from 5 to 20 nm, the HOMO level of NPB decreased by 0.52 eV because of the increased work function. This can improve hole injection. Therefore, it is crucial to ensure that the thickness of the MoO₃ HIL is sufficient to saturate the increase in work function and enhance hole injection in the anode system. The difference in the interface dipoles (0.40 eV) was larger than that in the work function (0.15 eV). One possible explanation for this result is that morphological changes occur in the MoO₃ layer with increasing thickness [25]. The roughness of the 5 nm thick MoO₃ layer was expected to be higher than that of the 20 nm thick MoO₃ layer, resulting in a large interfacial area with the NPB. A large interface area may result in a high degree of electron transfer, resulting in a large interface dipole. However, further research is needed to gain a deep understanding of the relationship between the interface dipole and the work function.

Figure 3. Energy-level diagrams of (a) ITO/MoO₃ (20 nm)/NPB and (b) ITO/MoO₃ (5 nm)/NPB. E_vac, E_F, Ψ, eD, IE, E_g, V_b, LUMO, and HOMO denote vacuum level, Fermi level, work function, interface dipole, ionization energy, band gap, band bending, lowest unoccupied molecular orbital, and highest occupied molecular orbital, respectively (unit: eV).

This study investigated the influence of the MoO₃ HIL thickness on the energy-level alignment at the ITO anode and NPB HTL interfaces through in situ XPS and UPS analyses. The ITO/MoO₃ (20 nm) anode system exhibited a work function of 6.48 eV, indicating the formation of an interface dipole of 2.52 eV. These values are higher than those reported for ITO/MoO₃ (5 nm) system, which had a work function of 6.33 eV and an interface dipole of 2.35 eV. The HOMO level of NPB on the ITO/MoO₃ (20 nm) was found to be 0.37 eV below the Fermi level, which is 0.52 eV lower than that on the ITO/MoO₃ (5 nm). These findings suggest that a sufficiently thick MoO₃ HIL is necessary to minimize the energy barrier for hole injection in terms of energy-level alignment, as observed in a previous study on the HATCN HIL. However, in practical device applications, it is also important to understand the hole transport properties inside the MoO₃ HIL, because increasing the thickness results in increased series resistance. Therefore, achieving optimal OLED performance requires balancing the reduced energy barrier and increasing series resistance.

This study was supported by the National Research Foundation of Korea (NRF- 2021R1A2C100932 and 2018R1A6A1A03025582).