Ya-Hui
Chang
ab,
Yen-Shou
Lin
ab,
Konthoujam
James Singh
b,
Hsiang-Ting
Lin
a,
Chiao-Yun
Chang
ac,
Zheng-Zhe
Chen
ad,
Yu-Wei
Zhang
ae,
Shih-Yen
Lin
ae,
Hao-Chung
Kuo
ab and
Min-Hsiung
Shih
*abf
aResearch Center for Applied Sciences (RCAS), Academia Sinica, Taipei 11529, Taiwan. E-mail: mhshih@gate.sinica.edu.tw
bDepartment of Photonics and Institute of Electro-Optical Engineering, College of Electrical and Computer Engineering, National Yang Ming Chiao Tung University, Hsinchu 30010, Taiwan
cDepartment of Electrical Engineering, National Taiwan Ocean University, Keelung 202301, Taiwan
dDepartment of Physics, National Taiwan University, Taipei, Taiwan, Taipei 10617, Taiwan
eGraduate Institute of Electronics Engineering, National Taiwan University, Taipei 10617, Taiwan
fDepartment of Photonics, National Sun Yat-sen University, Kaohsiung 80424, Taiwan
First published on 9th December 2022
Light-emitting diodes (LEDs) are used widely, but when operated at a low-voltage direct current (DC), they consume unnecessary power because a converter must be used to convert it to an alternating current (AC). DC flow across devices also causes charge accumulation at a high current density, leading to lowered LED reliability. In contrast, gallium-nitride-based LEDs can be operated without an AC–DC converter being required, potentially leading to greater energy efficiency and reliability. In this study, we developed a multicolor AC-driven light-emitting device by integrating a WSe2 monolayer and AlGaInP–GaInP multiple quantum well (MQW) structures. The CVD-grown WSe2 monolayer was placed on the top of an AlGaInP-based light-emitting diode (LED) wafer to create a two-dimensional/three-dimensional heterostructure. The interfaces of these hybrid devices are characterized and verified through transmission electron microscopy and energy-dispersive X-ray spectroscopy techniques. More than 20% energy conversion from the AlGaInP MQWs to the WSe2 monolayer was observed to boost the WSe2 monolayer emissions. The voltage dependence of the electroluminescence intensity was characterized. Electroluminescence intensity–voltage characteristic curves indicated that thermionic emission was the mechanism underlying carrier injection across the potential barrier at the Ag–WSe2 monolayer interface at low voltage, whereas Fowler–Nordheim emission was the mechanism at voltages higher than approximately 8.0 V. These multi-color hybrid light-emitting devices both expand the wavelength range of 2-D TMDC-based light emitters and support their implementation in applications such as chip-scale optoelectronic integrated systems, broad-band LEDs, and quantum display systems.
TMDC-based light-emitting devices with various structures have been demonstrated by several groups.17–21 The structures of electroluminescent (EL) devices based on 2D TMDC materials can be mainly divided into two types in accordance with the driving source: (i) p–n junction structures and (ii) capacitor structures.22 The p–n junction structures, formed through lateral stacking of p- and n-type semiconductors, are generally operated from a direct current (DC) voltage source. In such structures, electrons and holes are injected simultaneously into a TMDC from two electrode terminals when a bias voltage is applied.23–29
TMDC materials with a p–n junction structure are typically produced using mechanical exfoliation, limiting their applications in large-scale devices and systems. The complex heterostructure and small size of exfoliated TMDC flakes make the p–n junction structures unsuitable for highly efficient industrial-scale production.30 Moreover, one of the major bottlenecks experienced in the field of 2D TMDC-based electronics is the ultrahigh metal–TMDC contact resistance (of the order of several kilo-ohms to megaohms), which is due to the formation of a Schottky barrier at the metal–TMDC interface. The large metal–TMDC contact resistance causes inefficiency in carrier injection.31,32 Many studies have reported that this contact resistance can be lowered through various strategies, such as the selection of an appropriate contact metal, doping, insertion of a space layer at the interface, and leveraging the contact geometry.33–37 However, these methods increase the complexity of manufacturing processes.
Under alternating current (AC) pulsed voltage operating conditions, TMDC-based EL devices are typically fabricated with capacitor structures and consist of an insulating layer placed between the emission layer and the substrate. AC-driven EL devices have advantages over DC-driven devices; for instance, the vertically stacked capacitor structure is simpler than the p–n junction structure, and TMDC monolayers can be synthesized on a substrate directly through chemical vapor deposition (CVD), which is suitable for high-efficiency, low-cost, and large-scale light-emitting applications.20,27,38,39 Most crucially, the carrier injection performance at the metal–TMDC interface when AC pulsed voltage is applied is almost independent of the Schottky barrier's height. Thus, efficient carrier injection can be achieved.40
The capacitor structure originated from strong-field EL devices, and were first developed in 1936 by George Destriau, who demonstrated the light emission of ZnS powder when a large electric field was applied.41 Many studies have detailed construction of AC-driven EL devices with versatile emitting materials, ranging from inorganic thick semiconductors to organic atom-thin semiconductors, and such devices are considered promising alternatives to DC-driven EL devices.42–50
In addition to some scholars investigating AC-driven EL devices based on 2D-TMDCs, other groups have fabricated traditional III-nitride semiconductor-based light-emitting diodes (LEDs) driven by an AC voltage. LEDs have existed since 1962 and have been used in numerous fields, such as displays for electronics, lighting, and optical communication. However, LEDs and most electronic devices are operated with low-voltage DC, which is obtained using an AC–DC converter to convert 110/220 V AC voltage. This process of conversion causes unnecessary power consumption at magnitudes of 15%–30%. Moreover, a DC flow across devices causes charge accumulation at high current density, and such an accumulation influences the reliability of LEDs. Some of the latest research has demonstrated that gallium nitride (GaN)-based LEDs can be operated without an AC–DC converter being required, revealing their potential for application in energy-efficient and reliable LED lighting systems.51–56 Although AC-driven TMDC monolayer LEDs have been demonstrated recently, it is worth noting that several important issues, such as device upscaling difficulty and a small emission area, still remain unresolved at the current stage.
In this paper, we present a two-color AC-driven light-emitting device with a capacitor structure composed of a CVD-grown WSe2 monolayer on top of an AlGaInP-based LED wafer to form a 2D/3D heterostructure. An AlOx insulating layer is inserted between these two materials to prevent carrier leakage.
To verify the optical properties of WSe2 on sapphire, a Raman spectrum was obtained with a 488 nm continuous-wave laser (Fig. 1(c)). The Raman spectrum shows two distinct peaks at 250.3 and 260 cm−1, respectively, which are labelled E12g and A1g. The E12g peak was generated by the in-plane vibration of tungsten and sulfur atoms, whereas the A1g peak corresponded to the out-of-plane vibration of selenium atoms. The observed peaks in the Raman spectrum are consistent with previous reports. A B12g peak at approximately 310 cm−1, which represents an interlayer interaction between the different WSe2 atomic layers, was not observed. This is indicative of WSe2 monolayers grown on a sapphire substrate. The AFM images and Raman spectrum were used to determine the number of WSe2 layers. The WSe2 monolayer grown on sapphire was then transferred onto an AlGaInP quantum well LED epilayer covered by an AlOx layer of thickness 30 nm. The optical microscope image of the transferred CVD-grown WSe2 monolayer, with the transfer being achieved using the PMMA-assisted transfer method, is displayed in Fig. 1(d). Fig. 1(e) shows the PL spectrum of the hybrid WSe2 monolayer/GaP-based structure by using a 532 nm wavelength laser. Two strong emission peaks associated with the AlGaInP quantum wells and the WSe2 monolayer are observed. To verify the emission properties of the WSe2 monolayer before and after the transfer process, the PL spectra of the transferred and as-grown WSe2 monolayers are characterized, and shown in Fig. 1(f). Compared with the spectrum of the as-grown WSe2 monolayer sample, that of the transferred WSe2 monolayer shows a peak that is blue-shifted. This small blue shift is attributed to the release of strain during the transfer process, where strain was caused by the difference in the thermal expansion coefficients of the CVD-grown WSe2 monolayer and sapphire substrate. The small oscillations in the PL spectrum are due to Fabry–Perot oscillations from the bottom AlGaInP MQW structure.
To investigate the interactions between the WSe2 monolayer and the AlGaInP QWs, we transferred WSe2 monolayers to the AlGaInP QW substrate and the AlOx substrate and characterized the PL spectra of the two WSe2 devices and the AlGaInP QW device. Fig. 1(g) shows the PL spectra of the AlGaInP QW (the blue solid line), the WSe2 monolayer (the black solid line) on AlOx substrate, and the hybrid WSe2 monolayer/AlGaInP QW (the red line) devices. The color conversion phenomena associated with the AlGaInP QWs and the WSe2 monolayer were observed. If the QWs and WSe2 monolayer were integrated into a single device, the corresponding PL intensities were reduced in the AlGaInP QW emission region but enhanced in the WSe2 emission region compared with the individual gain spectra. This result can be attributed to the energy transfer from the high energy medium (i.e. the AlGaInP QWs in the study) to the low energy medium (the WSe2 monolayer in the study).57,58 The color conversion efficiency η can be calculated as follows:
(1) |
Electrical measurements were conducted at room temperature and under ambient conditions by using arbitrary function generators to generate AC pulse square waves for driving the devices. The devices were connected to two probe stations, which contained tungsten probe tips for electrical testing. The signal was observed using a microscope objective lens. The signal passed through the 20 μm-wide entrance slit of the spectrometer and was then focused on a 600 g mm−1 grating to separate the light signal into spatial spectra.
A 3D schematic and the vertical structure of the WSe2 monolayer–AlGaInP-based light-emitting device are shown in Fig. 2(a) and (b), respectively. The device consists of four components: two silver electrodes, emitter layers, and an insulating layer. The emitter layers comprise a WSe2 monolayer and the AlGaInP–GaInP multiple quantum well (MQW) active region. One of the electrodes serves as the bottom gate, while the other electrode serves as the source contact. Carriers are injected at the interface between the source contact and the semiconductor. To introduce an electric field into the insulating layer, a high-dielectric-constant AlOx film is selected as the insulating layer. The WSe2 monolayer is transferred onto the AlGaInP-based LED epilayer with an AlOx film inserted between the WSe2 monolayer and the LED epilayer to serve as a capacitive component. Electrodes are deposited on the top of the WSe2 layer and the surface of the bare LED epilayer to act as the source and gate electrodes, respectively. The left side of Fig. 2(c) shows the top-view scanning electron microscopy (SEM) image of the device at a magnification of 7000×. The geometry of the top electrode is a simple square window. The dark area enclosed by the dashed white line is the transferred WSe2 monolayer. Images of light emission from the device, captured using a charge-coupled device (CCD) camera, are shown on the right side of Fig. 2(c). The spatial distribution of the intensity of light emitted by the WSe2 monolayer and AlGaInP-based MQWs was nonuniform, which resulted from current crowding effects. In the WSe2 monolayer, this was because carrier injection and recombination occurred near the edge of the electrode. The thickness and composition of the layers in the device were determined through transmission electron microscopy (TEM) and energy-dispersive X-ray spectroscopy (EDX). Checking the thickness of WSe2 was extremely important because its properties strongly depend on the number of layers. Fig. 2(d) shows a bright-field scanning transmission electron microscopy (BF-STEM) image of the WSe2–AlOx–AlGaInP layered structure. Fig. 2(e) and (f) show the higher-magnification TEM images of the square area within the dotted blue line in Fig. 2(d) and of the GaInP–AlGaInP MQWs, respectively. The cross-sectional TEM image of WSe2 reveals that the thickness of single-layered WSe2 was approximately 0.7 nm, which perfectly matches the previous report of a monolayered WSe2 thickness. To further characterize the elemental composition of the layered structure, EDX was performed. The EDX elemental mapping images of WSe2–AlOx–AlGaInP shown in Fig. 2(g)–(j) indicate the spatial distribution of various elements over the structure. The distributions of W and Se were uniform. In addition, the presence of Al and Ga was expected due to the presence of the AlOx insulating layer and substrate of the AlGaInP-based LED.
Fig. 3(a) shows the room-temperature EL spectra of the WSe2–AlGaInP-based light-emitting device driven by various voltages at a fixed frequency. For both the AlGaInP-based LED and WSe2, the EL intensity increased with the voltage. The inset of Fig. 3(a) shows the EL spectra for WSe2 emission at 1 MHz with increasing voltage. The spectra contain two distinguishable emission peaks at approximately 627 and 755 nm, corresponding to the AlGaInP-based MQWs and WSe2 monolayer, respectively. Because of the poor emission efficiency of the WSe2 monolayer, the EL peak intensity of WSe2 has been multiplied by a factor of 10 to better enable visualization. The peak at a wavelength of 627 nm in the red emission band with a full width at half maximum linewidth of 10 nm was attributed to the AlGaInP–InGaP MQWs, whereas the peak at a longer wavelength was attributed to the direct intralayer exciton emission of monolayered WSe2, in good agreement with previous works.59,60 The small oscillation due to Fabry–Perot interferences in the QW layer was also observed in the EL spectrum.
Fig. 3(b) shows the voltage-dependent peak EL intensity of WSe2 at constant frequencies. The EL intensity increases clearly as the operating voltage becomes larger than the threshold voltage. The threshold voltage (Vth) of the WSe2 monolayer, defined as the x-intercept of the linear portion of the EL intensity–voltage curve, was approximately 10 V (green line in Fig. 3(b)). As the operating frequency increases, the threshold voltage of the WSe2 monolayer emission decreases. No light emission was observed at voltages below Vth due to insufficient carrier injection from the Ag electrode. The EL intensity of WSe2 was proportional to the voltage above Vth. Recent studies have proposed the following empirical model of the relationship between the EL intensity and the voltage: IEL ∝ exp(ακV),61 where κ is the constant of the injected carrier concentration in the 2D semiconductor and α is the constant related to the characteristics of excitons. A semi-log plot of the EL intensity as a function of voltage is displayed in Fig. 3(c). This figure indicates a linear relationship between the logarithm of the EL intensity and the voltage, which conforms to the aforementioned model.
Research has revealed that thermionic emission and Fowler–Nordheim (FN) emission are the main mechanisms underlying charge transport through the Schottky contact at a metal–2D-material interface.62,63 The carrier injection efficiency is severely limited by the Schottky barrier at the metal–TMDC interface. To examine the mechanism of charge injection through the interface between the electrode and WSe2 under external electric field and room temperature conditions, we plotted ln(IEL) versus V1/2 and ln(IEL/V2) versus V−1. These curves represent thermionic-emission-like and FN-like behaviors, respectively. Thermionic emission is usually considered to occur at high temperatures; charges can acquire sufficient energy to overcome the Schottky barrier. When thermionic emission occurs without the presence of an external electric field, the current density can be described as a function of temperature, as in the Richardson–Dushman equation.64 However, if the effect of an electric field is considered, the current density can be expressed by a Schottky equation:65
(2) |
According to the Schottky equation, the logarithmic EL intensity is proportional to the square root of the voltage. The thermionic emission plot, displayed in Fig. 3d, shows that the relationship between ln(IEL) and V1/2 is positive and linear in the low-voltage range (V < Vth; left-hand side of the red dotted line). Therefore, thermionic emission was considered to be the mechanism underlying carrier injection across the potential barrier at the Ag–WSe2 interface at a low voltage.
When a strong field is considered, carriers can tunnel through the barrier by a quantum mechanical effect. In quantum mechanical theory, the possibility of an electron penetrating a potential barrier is described by:66,67
(3) |
The tunneling probability depends on the width of the barrier. According to the barrier width equation x = Φ/eE, the width of barrier x is narrow when a strong electric field is applied. The tunneling of carriers occurs when the magnitude of the electric field is sufficiently high to change the barrier from being a flat band to having a triangular shape with a narrow effective width. This type of tunneling mechanism is known as FN tunneling.
The FN tunneling current equation is:
(4) |
The AC-voltage-driven electroluminescence mechanism of 2D materials has been investigated previously.20,27,40Fig. 3(f) shows a schematic explaining the mechanism of electroluminescence of WSe2. In brief, by applying a steady negative bias to the back-gate electrode, holes accumulate at WSe2 and the Schottky barrier blocks the flow of holes from WSe2 to the metal electrode (Fig. 3(f), part marked 1). When the AC square-wave pulse signal changes from a negative to a positive polarity, the capacitive elements (i.e., AlOx) induce a potential difference between the top source and the back gate. The potential drop across the WSe2 layer causes conduction band bending and then leads to the tunneling of electrons from the electrode to WSe2. Subsequently, the holes generated in the previous step recombine with the injected electrons, which causes the release of a photon (Fig. 3(f), part marked 2). The same behavior is observed when the voltage switches from a positive to a negative polarity (Fig. 3(f), parts marked 3 and 4).
The peak values in the spectra presented in Fig. 3(a) were recorded, and the relationship between the EL intensity of the AlGaInP-based MQWs and the peak-to-peak voltage was discovered (Fig. 4(a) and (b)). By extrapolating the EL intensity–voltage curves, a threshold voltage of approximately 5.5 V was obtained. The EL intensity of the AlGaInP-based MQWs follows an empirical equation:
L = L0exp(−β/V1/2) | (5) |
The empirical equation fitted the experimental data favorably and indicated that the EL intensity increased superlinearly with an increase in the voltage.71,72 The EL peak intensity of both the AlGaInP-based LED and WSe2 was strongly dependent on voltage rather than current due to the capacitance characteristics of the device.73
Ag–AlOx–AlGaInP has a metal–insulator–semiconductor (MIS) structure. It comprises an insulating layer (AlOx) that separates the metal electrode and semiconductor. The process of radiative recombination in an AlGaInP-based LED53,74 is given as follows. When a positive AC bias voltage is applied to the gate electrode, the majority of carriers (holes) from the p-type region of the AlGaInP LED accumulate at the AlOx–p-GaP interface. The accumulated carriers cannot be injected into the Ag metal from the semiconductor because of the tunnel barrier, which is an insulating layer of AlOx with a thickness of a few tens of nanometers. The electrons can tunnel from the electrode to the semiconductor when a high positive voltage is applied because this reduces the effective barrier width (FN tunneling). Finally, the electrons recombine with holes and release energy in the form of a photon (as illustrated in Fig. 4(c)).
By estimating the internal quantum efficiency and the extraction efficiency of the devices, the external quantum efficiency (EQE) of the WSe2 monolayer and AlGaInP QWs devices are approximately 0.018% and 0.056%, as given in Fig. S2 and S3 of the ESI.† The device is operated under ambient conditions and no significant drop in efficiency was observed. It is worth noting that because the external quantum efficiency of the AlGaInP QW LED is larger than that of a WSe2 monolayer, the thickness of the AlOx layer, deposited through atomic layer deposition, can be controlled to manipulate the ratio of the EL intensity of the AlGaInP QWs to that of the WSe2 monolayer by controlling the light extraction efficiency and the threshold voltage. This is an important step to counterbalance the EL intensities of the AlGaInP QWs and the WSe2 monolayer. However, two-color emission was still achieved at a voltage greater than 10 V in the study. The AC driving conditions, and the device structure, can be further optimized in the future for better multi-color emission from the hybrid LED device.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d2nr03725d |
This journal is © The Royal Society of Chemistry 2023 |