I. INTRODUCTION

Quantum dot light-emitting diodes (QLEDs) have emerged as prominent candidates for next-generation displays and lighting due to their remarkable attributes including exceptional color purity, tunable emission wavelengths, high efficiency, and solution processability ^[1-6]. The continuous advancements in fabrication technology have led QLEDs to steadily approach the performance levels of organic light-emitting diodes (OLEDs) ^[1-11]. The efficiency and functionality of QLEDs are significantly influenced by the engineering and selection of functional layers. While most QLEDs adopt a bottom-emitting structure, top-emission (TE) structures have attracted increasing attention due to their adaptability. TE-QLEDs minimize the effects of the many thin-film transistors integrated onto the substrate, thus expanding the range of compatible substrates to include even opaque materials. ^[12,13]. These TE structures can be effectively fabricated onto silicon or metal foils, thereby enhancing their potential applications in displays. Several metal oxides such as ZnO nanoparticles (NPs) were solution-processable for QLED and OLED fabrications. ZnO nanoparticles (NPs) are widely used as electron transport layers (ETLs) in QLEDs due to their high mobility, transparency, and conduction band level compatibility with quantum dots (QDs) ^[9,11, ^14-19]. However, QLEDs employing ZnO NP ETLs often experience imbalanced charge injection, wherein electron injection and transport surpass hole injection and transport ^[19-21]. This imbalance can lead to the accumulation of excessive electrons at the interface between the QD emission layer (EML) and ZnO NP ETLs, potentially charging the QDs and increasing the risk of non-radiative Auger recombination ^[14,20,21]. Additionally, ZnO NPs often contain many defects arising from oxygen vacancies ^[21] and surfaces of ZnO NPs synthesized in air are known to be rich of hydroxide oxygen which act as exciton quenching sites affecting performance ^[22]. Direct contact between a QD emission layer (EML) and ZnO NP ETLs rich in defects and hydroxide oxygen results in spontaneous electron transfer at the interface, causing exciton quenching ^[23-27]. These phenomena are facilitated by interfacial charge transfer and/or intragap-assisted non-radiative recombination centers ^[20], significantly influencing device performance. Moreover, certain studies suggest that holes might also transfer from QDs to the intragap state of ZnO NPs ^[6], inducing exciton dissociation within the QD EML, negatively influencing device performance.

A straightforward way to suppress exciton quenching is to passivate on the surface of the ZnO using an insulating layer of poly(methyl methacrylate) (PMMA), as demonstrated by Dai et al. ^[15]. Another approach involves inserting a Mg-doped ZnO (ZnMgO) NP interlayer between the ZnO NP ETL and QD EML to suppress exciton quenching and reduce the electron current, thereby enhancing the charge balance in QLEDs ^[28,29]. Heo et al. fabricated QLEDs with ZnMgO interfacial layers between the QD EML and ZnO NP ETLs, resulting in an 86.9% increase in external quantum efficiency (EQE). The stepped interfacial electronic structure is preventing direct contact between the QD EML and ZnO NP ETLs ^[28]. The use of ZnMgO as ETLs in QLEDs allows tuning of mobility, conductivity, surface defects and conduction band levels, contributing to improved charge balance and reduced exciton quenching ^[30,31]. Some groups have developed highly efficient QLEDs using a ZnMgO core-shell structure as an ETL, passivating defects in ZnO NPs, attributed to Mg in the shell layer ^[32,33]. Currently, many researchers are focusing on direct contact between QDs and ZnMgO rather than between QDs and ZnO NPs.

In this study, we aimed to optimize the charge balance and reduce interfacial quenching in TE-QLEDs by solely adjusting the thickness of Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O NP ETLs. The goal was to effectively mitigate the concentration of hydroxide oxygen in Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O NPs varying Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O NP ETL, lowering exciton quenching and ultimately enhancing the radiative recombination efficiency within the QD layer. Through X-ray photoelectron spectroscopy (XPS) measurements, we investigated hydroxide oxygen states to explore interfacial electronic properties. Additionally, we examined oxygen vacancy states to analyze electron injection and electron transport in Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O NP ETLs. According to the results, QLEDs with a 30-nm-thick Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O ETL demonstrated a remarkable performance with a high level of efficiency due to an extremely low concentration of hydroxyl group, resulting in a low exciton quenching. We found that a TE-QLED with a 30-nm-thick Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O NP ETL demonstrated outstanding performance, with a maximum current efficiency of 91.92 cd/A and a maximum EQE of 21.66%. We believe that the use of Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O NP ETLs of appropriate thickness will allow the fabrication of high-performance TE-QLEDs for next-generation display and lighting applications.

II. MATERIALS AND METHODS

2.2 Device Fabrication

TE-QLED fabrication employed a solution-process approach. The glass substrate is underwent meticulous cleaning in acetone, isopropyl, alcohol, methanol, and deionized water and then subjected to oxygen plasma treatment. Poly (3,4-ethylenedioxythiophene):poly (styrene sulfonate) (PEDOT:PSS) (Al4083; Heraeus, Hanau, Germany) in isopropanol (Daejung Chemicals and Metals, Gyeonggi, Republic of Korea) was spin-coated onto the glass substrate, followed by baking at 60$^{\circ}$C for 10 m in air to eliminate residual water. The resulting 25-nm-thick PEDOT:PSS film served as the buffer layer between the glass substrate and Ag anode, enhancing adhesion during Ag evaporation onto the glass substrate and improving the electrical conductivity of the Ag anode. Subsequently, glass substrates with PEDOT:PSS buffer layers were placed in a high-vacuum deposition chamber (Cetus OL 100; Celcose, Gyeonggi, Republic of Korea) under a pressure of 6 ${\times}$ 10$^{-}$$^{7}$ Torr.

In the chamber, a 150-nm-thick Ag anode layer was deposited at an evaporation rate of 1.2 Å/s. The anode layer was patterned as desired using an in-situ shadow mask. A bank layer was photolithographically fabricated on the Ag anode to eliminate leakage current between the anode and cathode. SU-8 2002 (Kayaku Advanced Materials, MA, USA) was deposited onto the glass/PEDOT:PSS/Ag substrate via spin-coating at 500 rpm for 5 s at room temperature, followed by spin-coating at 3,000 rpm for 30 s at room temperature. The SU-8 thin film thus formed was soft-baked at 90℃ on a hot plate for 2 min to evaporate the solvent and create the thin film. The substrate with the SU-8 thin film was aligned with the mask of the bank layer pattern on a mask aligner (ABM/6/350/NUV/DDD/M; ABM, CA, USA) and exposed to UV light of 365 nm for 15.6 s. The SU-8 layer was thus polymerized and crosslinked, becoming insoluble. After UV exposure, baking proceeded at 95℃ on a hot plate for 2 min. The UV-exposed SU-8 was immersed in SU-8 developer for 1 min to remove unreacted materials, followed by a 10-s rinse in isopropyl alcohol (IPA) (Daejung Chemicals and Metals, Gyeonggi, Republic of Korea). A nitrogen jet removed residual moisture. Thus, all developer and cleaning solutions were eliminated. To enhance adhesion, a hard bake ran on a 150℃ hot plate for 30 min, yielding the bank layer. To create the hole injection layer (HIL), an additional PEDOT: PSS layer was spin-coated onto the glass/PEDOT: PSS/Ag substrate within the bank as described above. To create the hole transport layer (HTL), poly(9 vinylcarbazole) (PVK) dissolved in toluene was coated onto the pre-fabricated glass/PEDOT: PSS/Ag/PEDOT: PSS substrate within the bank, and the PVK film was annealed in air at 60℃ for 10 min. The PVK film thickness was 40 nm. To fabricate the QD EML, CdSe/ZnS QDs (Zeus, Gyeonggi, Republic of Korea) were dissolved to 5 mg/mL in heptane, and the solution was spin-coated onto the glass/PEDOT: PSS/Ag/PEDOT: PSS/PVK substrate within the bank; the QD film thickness was 10 nm. The Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O NP solution was then spin-coated onto the glass/PEDOT:PSS/Ag/PEDOT:PSS/PVK/QD substrate within the bank. Finally, a 15-nm-thick Ag cathode was deposited onto the glass/PEDOT:PSS/Ag/PEDOT:PSS/ PVK/QD/Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O NP substrate within the bank using the in-situ method employed for Ag anode deposition; the TE-QLEDs were now complete (Fig. 1).

Fig. 1. Schematic of TE-QLED fabrication.

III. RESULTS AND DISCUSSION

Fig. 2 shows the characteristic 2${\theta}$ XRD patterns of Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O NPs, elucidating their crystal structure. The XRD analysis of Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O NPs revealed a hexagonal wurtzite lattice, with reflections across the (100), (002), (101), (102), (110), (103), and (112) planes. These reflections closed matched those of ZnO, confirmed by comparison with the Joint Committee on Powder Diffraction Standards (JCPDS) (Card Number 1-1136). Minor deviations in peak positions and alterations in interatomic distances were observed. Notably, when the XRD patterns of the Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O NPs were compared with the those of MgO on the JCPDS (Card Number 1-1235), entirely different patterns emerged. This disparity is attributed to the rocksalt structure of MgO. Thus, the presence of the identical hexagonal wurtzite structure in the Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O NPs as in ZnO, distinct from the rocksalt structure of MgO, signifies successful incorporation of Mg into ZnO.

Fig. 2. X-ray diffraction patterns XRD of Zn0.9Mg0.1O nanoparticles (NPs) over a 2θ range of 20-80°.

Fig. 3. Field-emission transmission electron micrographs of Zn0.9Mg0.1O NPs.

Fig. 4. X-ray photoelectron O 1s spectra of Zn0.9Mg0.1O NP thin films of varying thickness deposited onto a PEDOT:PSS/Ag/PEDOT:PSS/PVK/QD substrate: (a) 10-nm-thick films; (b) 30-nm-thick films; (c) 50-nm-thick films; (d) 70-nm-thick films; (e) Area ratios of the three peaks changes according to thickness

Fig. 5. Current density versus applied voltage of electron-only devices (EODs) with Zn0.9Mg0.1O NPs of various thicknesses (10, 30, 50, and 70 nm) and the hole-only device (HOD).

Fig. 6. Electroluminescence (EL) characteristics of TE-QLEDs with Zn0.9Mg0.1O NP ETLs of different thicknesses: 10, 30, 50, and 70 nm: (a) Current density-voltage curves as a function of voltage; (b) luminance curves as a function of voltage; (c) current efficiency and external quantum efficiency curves as a function of current density.

Fig. 7. Normalized photoluminescence spectrum of the CdSe/ZnS quantum dot (QD) and the EL spectra of TE-QLEDs with 10-, 30-, 50-, and 70-nm-thick Zn0.9Mg0.1O NP ETLs.

Fig. 3 shows a typical FE-TEM image of Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O NPs, the average diameter of which was 4.61 nm.

XPS measurements were conducted to examine the oxygen configuration of Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O NP thin film as a function of varying thickness, providing information about the quality of the NPs, as shown in Fig. 4. This figure illustrates the normalized O 1s spectra of the Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O NPs deposited onto the PEDOT: PSS (25 nm)/Ag (150 nm)/PEDOT:PSS (25 nm)/QD (10 nm) substrate. By employing deconvolution, three peaks were extracted, each corresponding to specific components, including the metal oxide lattice bond (OM), oxygen vacancies O2- ions (OV), and hydroxide oxide (-OH) ^[21,34]. The area of each peak exhibited variations depending on the thickness of the Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O NP thin film. OT represents the overall O 1s peak. Fig. 4(e) includes a distribution graph illustrating the ratio of the areas of the three peaks areas concerning the thickness of Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O NP thin film. The ratios of the peak area (OV/OT) of 10-nm-thick Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O NP thin film, 30-nm-thick Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O NP thin film, 50-nm-thick Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O NP thin film, and 70-nm-thick Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O NP thin film were calculated to be 35.46%, 27.20%, 25.26%, and 21.70%, respectively. On the other hand, the ratios of the peak area (-OH/OT) of 10-nm-thick Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O NP thin film, 30-nm-thick Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O NP thin film, 50-nm-thick Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O NP thin film, and 70-nm-thick Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O NP thin film were estimated to be 9.36%, 0.57%, 3.67%, and 5.10%, respectively. As the thickness increased, there was a clear tendency toward a greater concentration of metal oxide lattice bond, while the concentration of oxygen vacancies decreased. These vacancies act as trapping sites, and their increase leads to higher carrier concentration, improved conductivity, and reduced resistance of the Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O NP thin film ^[23,28,35]. Conversely, hydroxyl oxygen species can act as an exciton quenching sites, affecting the device efficiency.

Electron-only devices (EODs) and hole-only devices (HOD) were fabricated to investigate hole injection, electron injection, and electron transport. Four EODs that differed in Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O NP thickness (10, 30, 50, and 70 nm) were developed (refer to Fig. 6), along with an HOD. The EOD structures comprised Ag (150 nm)/Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O NPs (10, 30, 50, 70 nm)/QD (10 nm)/Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O NPs (10, 30, 50, 70 nm)/Ag (150 nm). The HOD structure involved Ag (150 nm)/PEDOT: PSS (25 nm)/PVK (40 nm)/QD (10 nm)/Ag (150 nm). The thickness of the Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O NP layers was identical on both sides of the EOD QDs. The additional layer of Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O NPs on the anode blocked the arrival of holes from the anode and prevented excessive current flow. Electrons were injected from the cathode into the conduction band minimum (CBM) level of the QDs via the Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O NPs. The EOD measurements revealed that as the thickness of Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O NPs increased, the current density decreased, consistent with the XPS results, which revealed fewer oxygen vacancies as the thickness of Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O NPs increased. This reduced conductivity and increased the resistance of the Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O NP ETL. The HOD current density fell to within the range of the current densities associated with EODs having thicknesses 30 and 50 nm. This implies that the optimal charge balance of the QD EML can be achieved by a TE-QLED with a Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O ETL thickness of 30 or 50 nm.

Fig. 6 illustrates the EL characteristics of the TE-QLEDs with Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O NP ETLs of varying thickness (10, 30, 50, and 70 nm). In Fig. 6(a), the curves for current density are presented as a function of applied voltage. Note that as the thickness of the Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O NP ETLs increases, the current density decreases markedly, attributable to a reduction in conductivity that reflects the fewer oxygen vacancies within the Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O NP ETL, as supported by the XPS results. This decrease is also associated with a reduced electric field attributable to the increased ETL thickness. Fig. 6(b) shows luminance versus voltage curves. The turn-on voltages at 1 cd/m2 for TE-QLEDs with 10-, 30-, 50-, and 70-nm-thick Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O NP ETLs were extrapolated to be 4.15 V, 4.06 V, 4.57 V, and 4.52 V, respectively. Thus, the TE-QLED with a 30-nm-thick Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O NP ETL exhibited the lowest turn-on voltage; exciton quenching was minimized at the interface between the QD and Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O NP ETL, attributable to a decrease in hydroxide oxygen species (as confirmed by XPS analysis) and attainment of an optimal charge balance in the QD EML. Fig. 6(b) shows the maximum luminance values of TE-QLEDs with 10-, 30-, 50-, and 70-nm-thick Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O NP ETLs; these were estimated to be 30,560.1 cd/m$^{2}$, 257,307.4 cd/m$^{2}$, 157,337.4 cd/m$^{2}$, and 59,782.06 cd/m$^{2}$, respectively. The TE-QLED with a 30-nm-thick Zn $_{\mathrm{0.9}}$ Mg$_{\mathrm{0.1}}$O NP ETL exhibited the highest luminance. This correlated with the better-balance charge carriers of electrons and holes within the QD EML in TE-QLEDs with 30- or 50-nm-thick Zn $_{\mathrm{0.9}}$ Mg$_{\mathrm{0.1}}$O NP ETLs, as confirmed by analysis of the EODs and HOD (refer to Fig. 3). Additionally, the reduced concentration of hydroxide oxygen species in the Zn $_{\mathrm{0.9}}$ Mg$_{\mathrm{0.1}}$O NPs (as shown by the XPS data in Fig. 4) minimized exciton quenching at the interface between the QD EML and Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O NP ETL, thus enhancing luminance. Fig. 6(c) shows the current efficiencies and EQEs of TE-QLEDs with Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O NP ETLs with thicknesses of 10, 30, 50, and 70 nm; these were 9.68 cd/A, 90.92 cd/A, 55.33 cd/A, and 20.98 cd/A, respectively. The maximum EQEs for TE-QLEDs with 10-, 30-, 50-, and 70-nm-thick Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O NP ETLs were 3.28%, 21.66%, 13.22%, and 6.08%, respectively. Remarkably, the TE-QLED with a 30-nm-thick Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O NP ETL exhibited an outstanding maximum current efficiency and a maximum EQE of 90.92 cd/A and 21.66%, respectively, in line with the above discussion. It is essential to choose an appropriate Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O NP ETL thickness to balance the charge carriers of electrons and holes and minimize the concentration of hydroxide oxygen species that engage in quenching at the interface between the QD and Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O NP ETLs. Table 1 summarizes the key parameters of TE-QLEDs with 10-, 30-, 50-, and 70-nm-thick Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O NP ETLs.

Fig. 7 shows the normalized photoluminescence (PL) spectrum of the CdSe/ZnS QD and the EL spectra of TE-QLEDs with 10-nm-thick, 30-nm-thick, 50-nm-thick, and 70-nm-thick Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O NP ETLs. A summary of the characteristic parameters is given in Table 2. The EL peak of the TE-QLED with a 10-nm-thick Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O NP ETL is centered at 524 nm, matching the PL peak of the CdSe/ZnS QD. In contrast, the EL peaks of TE-QLEDs with 30-nm-thick, 50-nm-thick, and 70-nm-thick Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O NP ETLs are centered at 532 nm, i.e., are red-shifted compared to the PL peak of the CdSe/ZnS QD. Such red shifts can be attributed to Föster energy transfer, microcavity effect and the Stark effect, which develops at high voltage and current ^[36,37].

Table 2 shows that the spectral FWHM increases with the thickness of the Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O NP ETLs. However, a thickness of 10 nm is too low to exert a significant impact. The TE-QLED with a 10-nm-thick Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O NP ETL exhibited the narrowest FWHM of 32.8 nm. The absence of parasitic PVK emission confirms that device emission was primarily attributable to electron-hole recombination within the CdSe/ZnS QD EML, in turn attributable to the good charge balance between holes and electrons within that layer.

IV. CONCLUSIONS

TE-QLEDs were fabricated inside an SU-8 bank formed on a PEDOT:PSS/Ag anode. XPS measurements revealed that the concentrations of oxygen vacancies and hydroxyl species changed as the thickness of the Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O NP ETLs varied. As the thickness increased, the number of oxygen vacancies decreased; this reduced conductivity, as confirmed by EOD. Notably, HOD current density remained within the range of that of EODs with 30- and 50-nm-thick Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O layers. A good QD EML charge balance is ensured by using such TE-QLEDs. When a 30-nm-thick Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O NP ETL was employed, the hydroxide oxygen concentration was lowest, minimizing the exciton quenching at the interface of the QD and Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O NP ETL and thus improving the charge balance within the QD EML, significantly enhancing QLED efficiency. The maximum TE-QLED current efficiency and EQE were 90.92 cd/A and 21.66%, respectively. Our results suggest that Zn$_{\mathrm{0.9}}$Mg$_{\mathrm{0.1}}$O NP ETL layers of appropriate thickness will enhance the performance of the TE-QLEDs required for next-generation display and lighting applications.