Calculating the Effect of AlGaN Dielectric Layers in a Polarization Tunnel Junction on the Performance of AlGaN-Based Deep-Ultraviolet Light-Emitting Diodes

Abstract

In this work, AlGaN-based deep-ultraviolet (DUV) light-emitting diodes (LEDs) with AlGaN as the dielectric layers in p⁺-Al_0.55Ga_0.45N/AlGaN/n⁺-Al_0.55Ga_0.45N polarization tunnel junctions (PTJs) were modeled to promote carrier tunneling, suppress current crowding, avoid optical absorption, and further enhance the performance of LEDs. AlGaN with different Al contents in PTJs were optimized by APSYS software to investigate the effect of a polarization-induced electric field (E_p) on hole tunneling in the PTJ. The results indicated that Al_0.7Ga_0.3N as a dielectric layer can realize a higher hole concentration and a higher radiative recombination rate in Multiple Quantum Wells (MQWs) than Al_0.4Ga_0.6N as the dielectric layer. In addition, Al_0.7Ga_0.3N as the dielectric layer has relatively high resistance, which can increase lateral current spreading and enhance the uniformity of the top emitting light of LEDs. However, the relatively high resistance of Al_0.7Ga_0.3N as the dielectric layer resulted in an increase in the forward voltage, so much higher biased voltage was required to enhance the hole tunneling efficiency of PTJ. Through the adoption of PTJs with Al_0.7Ga_0.3N as the dielectric layers, enhanced internal quantum efficiency (IQE) and optical output power will be possible.

1. Introduction

AlGaN-based deep-ultraviolet (DUV) light-emitting diodes (LEDs) have many advantages, including environmental protection, low power consumption, Hg-free material, compact size, controllable wavelength, and a long lifetime, and they can be applied in the fields of disinfection, sterilization, purification, biomedicine, gas sensing, optical data storage, non-line-of-sight communication, identification of hazardous biological agents and so on [1]. AlGaN-based DUV LEDs with a short wavelength of 210 nm [2] and an improved external quantum efficiency (EQE) of 20.3% (for 275 nm) [3] have been achieved. However, the EQE is still lower than that of GaN-based blue and green LEDs, and the EQE drops dramatically with decreasing wavelength [1]. As is well-known, the EQE is determined by the internal quantum efficiency (IQE) and the light extraction efficiency (LEE), where the IQE is determined by the material’s quality [4], the carrier injection [5], and the recombination [6]; and the LEE is determined by the total internal reflection (TIR) [7], the optical absorption [7], and the optical polarization [8,9].

In order to improve the EQE, high-efficiency carrier injection is required to realize high-performance AlGaN-based DUV LEDs. However, it is difficult to achieve high-efficiency doping, particularly p-type doping, in AlGaN with a high Al content [10,11]. In addition, AlGaN has much lower hole mobility than electron mobility (electron mobility of ~300 cm²/V·s and hole mobility of ~14 cm²/V·s at room temperature for unintentionally doped AlN [12,13]), which results in low hole injection efficiency and high electron leakage [5,14,15,16,17,18,19,20,21,22,23]. Moreover, poor ohmic contact between AlGaN and the electrode will increase the series resistance of an LED. Furthermore, the use of a horizontal electrode structure results in a very strong current crowding effect in the p-type side of an LED [24]. The vertical electrode structure is an alternative solution to suppress the current crowding effect [25,26]. However, there will be more difficulties and challenges in fabricating the vertical LEDs. In order to improve the current spreading and suppress the current crowding effect, some approaches, such as improving the design of the hole injection layer [27], optimization of the mesa area size to adjust the current distribution [28], enhancing the conductivity of the contact layer with the p-electrode [27], transparent p-side electrode fabrication [7,29], and use of a tunnel junction (TJ) or a polarization TJ (PTJ) [30,31,32,33,34,35,36,37,38,39,40,41] have been proposed.

A TJ is a heavily doped p⁺–n⁺ junction with a thickness of only some 10 nm. When an LED is operated at a forward biased voltage, the top TJ is operated at a reverse biased voltage, so the electrons can tunnel from the p⁺-side valence band into the n⁺-side conduction band. In order to reduce the resistance of the TJ, the heavy doping electron and hole concentrations are commonly over 1 × 10²⁰ cm⁻³. A PTJ is a p⁺–i–n⁺ heterojunction, in which the built-in electric field (E_b) is modulated by polarization induced electric field (E_p) due to the existence of spontaneous polarization and piezoelectric polarization in the p⁺–i–n⁺ heterojunction. Therefore, the coupled electric field (E_c) comes from the contributions of not only E_b but also E_p. The integration of a TJ or a PTJ in an LED can effectively enhance the hole injection efficiency, and PTJs can further promote hole tunneling by introducing E_p. In addition, it can suppress the current crowding effect, improve lateral current spreading, and further enhance the uniformity of the top emitting light of the LED, which is possible through the contribution of its relatively high resistance. Moreover, it uses a low-resistivity n-type material instead of a high-resistivity p-type material as the top contact with the anode of the LED, which is very advantageous to ohmic contact. Moveover, it can be applied in multi-color LEDs [42,43], micro LEDs [44] and stacked laser diodes (LDs) [45]. Although there are many advantages, fabricating TJs or PTJs with heavy doping electron and hole concentrations (over 1 × 10²⁰ cm⁻³) is very challenging.

In 2001, an InGaN-based blue LED with a p⁺-GaN/n⁺-GaN TJ was reported, with an optical output power about twice that of conventional LEDs without a TJ [32]. However, the forward voltage of the LED was accordingly increased from 3.9 to 4.9 V, which was caused by the relatively high resistance in TJ. A TJ has a positive impact on the lateral current spreading but has a negative impact on the forward voltage. Therefore, it is crucial to improve the tunneling efficiency by adopting PTJs. In 2013, an InGaN-based blue LED with a p⁺-GaN/In_0.15Ga_0.85N/n⁺-GaN PTJ was reported, and improved performance was achieved [35]. Polarization can induce high densities of positive and negative sheet charges at the p⁺-GaN/InGaN and InGaN/n⁺-GaN interfaces, which originate from spontaneous polarization and piezoelectric polarization in InGaN along the c-orientation [46]. The high density of polarization induced sheet charges at the p⁺-GaN/InGaN and InGaN/n⁺-GaN interfaces can build up E_p. E_p has a direction along the [000-1] orientation, in the same direction as E_b. Therefore, E_c comes from the contributions of not only E_b but also E_p, which can be enhanced by optimizing E_p in the PTJ to improve the tunneling efficiency. The intensity of E_c in a PTJ can be formulated as follows:

$$E=e\times \left|\frac{N_{dopant}\times L_{depletion}\pm {\sigma }_p}{{\epsilon }_r\times {\epsilon }_0}\right|$$

(1)

where E is the intensity of E_c in PTJ, e is the value of a unit of electronic charge, ε₀ is the absolute dielectric constant, ε_r is the average relative dielectric constant for the PTJ, N_dopant is the ionized dopant concentration in the space charge region, L_depletion is the depletion region’s width, and σ_p is the polarization-induced sheet charge density. The symbol “±” represents the direction of E_p. The symbol “+” represents the same direction as E_b, and the symbol “–” represents the reverse direction to E_b. According to the expression in Equation (1), E can be determined by the factors N_dopant, L_depletion, σ_p, and ε_r.

It is necessary to point out that ε_r is the inherent constant of a material. For InGaN, there is a linear relationship between ε_r and the In content of InGaN, and ε_r will become larger with an increase in the In content of InGaN. In addition, there is strong optical absorption in the PTJ when the In content of InGaN in the PTJ is higher than that in the active region, as it is quite disadvantageous to the LEE of an LED. AlGaN has a lower ε_r than InGaN, and the ε_r becomes small with an increase in the Al content of AlGaN. This means that AlGaN as the dielectric layer of a PTJ is superior to InGaN as the dielectric layer of a PTJ for enhancing the intensity of E_c, promoting hole tunneling and avoiding optical absorption, as is very advantageous to the IQE and LEE of an AlGaN-based DUV LED.

However, there is still an inadequate view that AlGaN as the dielectric layer of PTJ cannot enhance the intensity of E_c, due to the reverse direction of E_p compared with E_b. As a result, there are few reports on AlGaN as the dielectric layer of PTJs in LEDs. In 2017, InGaN-based near-ultraviolet (NUV) LEDs with p⁺-GaN/AlGaN/n⁺-GaN PTJs were optimized, and enhanced LED performance was achieved. This was the first report on the use of AlGaN as a dielectric material of a PTJ [36,37].

Additionally, the difficulty of heavy doping (a carrier concentration of over 1 × 10²⁰ cm⁻³) in a PTJ has slowed down the progress of research into AlGaN-based DUV LEDs with PTJ. However, the research into AlGaN-based DUV LEDs with PTJs never stopped. In 2018, an AlGaN-based DUV LED with a p⁺-Al_0.65Ga_0.35N/In_0.2Ga_0.8N/graded n⁺-AlGaN PTJ (p⁺ = 5 × 10¹⁹ cm⁻³ and n⁺ = 1 × 10²⁰ cm⁻³) was grown and fabricated, with a wavelength of 287 nm, a forward voltage of 10.5 V, and an optical output power of 54.4 W/cm² at a current density of 1 kA/cm² [34,35,36]. In 2021, an AlGaN-based DUV LED with a p⁺-GaN/n⁺-AlGaN PTJ was grown and fabricated, with a wavelength of 245 nm, a maximum EQE of 0.35% and a wall-plug efficiencies (WPE) of 0.21% [36]. It was proven that high-performance AlGaN-based DUV LEDs with PTJs can be achieved. However, InGaN or GaN as the dielectric layer of a PTJ is quite unsuitable for AlGaN-based DUV LED due to strong optical absorption in PTJ. AlGaN has a lower ε_r and a lower optical absorption than InGaN. This means that AlGaN is superior to InGaN as the dielectric layer of PTJ for promoting hole tunneling, suppressing current crowding, increasing current spreading, and avoiding optical absorption of the PTJ in AlGaN-based DUV LEDs.

In this work, AlGaN-based DUV LEDs with AlGaN as the dielectric layers in p⁺-Al_0.55Ga_0.45N/AlGaN/n⁺-Al_0.55Ga_0.45N PTJs were proposed. AlGaN with different Al content in PTJs was optimized to investigate the effect of p⁺-Al_0.55Ga_0.45N/Al_xGaN/n⁺-Al_0.55Ga_0.45N PTJs on the performance of AlGaN-based DUV LEDs, where x = 0.4, 0.55, and 0.7. The output characteristics of the AlGaN-based DUV LEDs including the forward voltage, optical output power, IQE and WPE were characterized. APSYS software was used to conduct a finite element analysis of the electrical, optical, and thermal properties of the LEDs.

2. Research Methods and Physical Models

AlGaN-based DUV LEDs for Structures A, B, and C were designed, as shown in Figure 1. The basic structures include a c-plane sapphire substrate, a 1 μm AlN buffer layer, a 2 μm n-Al_0.55Ga_0.45N layer (n = 5 × 10¹⁹ cm⁻³), a 50 nm n-AlGaN grading layer with the Al content ranging from 0.55 to 0.85 (n = 5 × 10¹⁹ cm⁻³), a 10 nm n-Al_0.85Ga_0.15N hole blocking layer (HBL) (n = 5 × 10¹⁹ cm⁻³), five pairs of multiple quantum wells (MQWs) with a 5 nm undoped Al_0.55Ga_0.45N quantum barrier (QB) and a 1.5 nm undoped Al_0.4Ga_0.6N quantum well (QW), a 10 nm p-Al_0.85Ga_0.15N electron blocking layer (EBL) (p = 5 × 10¹⁹ cm⁻³), a 50 nm p-AlGaN grading layer with the Al content ranging from 0.85 to 0.55 (p = 5 × 10¹⁹ cm⁻³), and a total of 322 nm for the top layers. For all structures, the bottom layers were the same and the 322 nm top layers were different. For Structure A (conventional TJ, no polarization layer), the top layers were an 11 nm p⁺-Al_0.55Ga_0.45N (p⁺ = 1 × 10²⁰ cm⁻³)/11 nm n⁺-Al_0.55Ga_0.45N TJ (n⁺ = 1 × 10²⁰ cm⁻³) and a 300 nm n-Al_0.55Ga_0.45N contact layer (n = 5 × 10¹⁹ cm⁻³). For Structure B (PTJ), the top layers were a 10 nm p⁺-Al_0.55Ga_0.45N (p⁺ = 1 × 10²⁰ cm⁻³)/2 nm Al_0.4Ga_0.6N/10 nm n⁺-Al_0.55Ga_0.45N (n⁺ = 1 × 10²⁰ cm⁻³) PTJ and a 300 nm n-Al_0.55Ga_0.45N contact layer (n = 5 × 10¹⁹ cm⁻³). For Structure C (PTJ), the top layers were a 10 nm p⁺-Al_0.55Ga_0.45N (p⁺ = 1 × 10²⁰ cm⁻³)/2nm Al_0.7Ga_0.3N/10 nm n⁺-Al_0.55Ga_0.45N (n⁺ = 1 × 10²⁰ cm⁻³) PTJ and a 300 nm n-Al_0.55Ga_0.45N contact layer (n = 5 × 10¹⁹ cm⁻³). Si and Mg were used to supply the electrons and holes in the n- and p-layers and the TJ. The anode of the LED was the ohmic contact with the top layer, and the mesa size was set to 300 × 300 μm², with an anode size of 300 × 50 μm² and a cathode size of 300 × 70 μm².

APSYS software (APSYS-2017, Crosslight Software Inc. Vancouver, Canada) was used to simulate the LEDs. In the simulation, the current continuity equations, Poisson equations, and Schrödinger equations would be solved with proper boundary conditions. The following parameters were set: an Auger recombination coefficient of 1 × 10⁻⁴² m⁶/s [35,47], a Shockley–Read–Hall (SRH) lifetime of 1 × 10⁻⁸ s [48,49], an energy band offset ratio (ΔE_C/ΔE_V) of 0.5 [50,51], a [0001] polarization level of 40% [46,52,53], effective masses of the electrons and holes for GaN and AlN [54], and an ambient temperature of 300 K. The effective masses of the tunneling particles in the TJs were 0.230 m₀, 0.225 m₀, and 0.234 m₀ for Structures A, B, and C, respectively; m₀ is the free electron mass. Other parameters of III-nitride based semiconductors can be found elsewhere [55].

3. Results and Analysis

Figure 2 shows the distributions of the electric fields (Figure 2a–c), the charged state densities of the space charges (Figure 2d), and the charged state densities of the fixed space charges at the interfaces (Figure 2e) in the TJs for Structures A, B, and C at a bias voltage of 0 V. N is the charged state density of the space charge.

For Structure A, the TJ is a p⁺–n⁺ homojunction and E_c comes from the contribution of only E_b, as shown in Figure 2a. E_b comes from the ionized dopant in the space charge region; has a direction along the [000-1] orientation, beginning with N₁₁ in the n⁺-Al_0.55Ga_0.45N layer and ending with N₁₂ in the p⁺-Al_0.55Ga_0.45N layer; and can be calculated as N₁₁ = 1.0 × 10²⁰ cm⁻³ and N₁₂ = −1.0 × 10²⁰ cm⁻³, as shown in Figure 2d.

For Structure B, the PTJ is a p⁺–i–n⁺ heterojunction, and E_c comes from the contributions of E_b and E_p, as shown in Figure 2b. E_b has a direction along the [000-1] orientation, beginning with N₂₁ in the n⁺-Al_0.55Ga_0.45N layer and ending with N₂₂ in the p⁺-Al_0.55Ga_0.45N layer, and it can be calculated as N₂₁ = 1.0 × 10²⁰ cm⁻³ and N₂₂ = −1.0 × 10²⁰ cm⁻³, as shown in Figure 2d. E_p is built up from the differences between N₃ at the Al_0.4Ga_0.6N/n⁺-Al_0.55Ga_0.45N interface and N₄ at the p⁺-Al_0.55Ga_0.45N/Al_0.4Ga_0.6N interface, where N₃ > 0 and N₄ < 0 under compressive strain of Al_0.4Ga_0.6N, and can be calculated as N₃ = 2.9 × 10²⁰ cm⁻³ and N₄ = −3.1 × 10²⁰ cm⁻³, as shown in Figure 2e. E_p has a direction along the [000-1] orientation in the Al_0.4Ga_0.6N layer, in the same direction as E_b, and a direction along the [0001] orientation at both sides of the Al_0.4Ga_0.6N layer, in the reverse direction to E_b, as shown in Figure 2b.

For Structure C, the PTJ is also a p⁺–i–n⁺ heterojunction, and E_c also comes from the contributions of E_b and E_p, as shown in Figure 2c. E_b has a direction along the [000-1] orientation, beginning with N₃₁ in the n⁺-Al_0.55Ga_0.45N layer and ending with N₃₂ in the p⁺-Al_0.55Ga_0.45N layer, and can be calculated as N₃₁ = 1.0 × 10²⁰ cm⁻³ and N₃₂ = −1.0 × 10²⁰ cm⁻³ as shown in Figure 2d. E_p is built up from the differences between N₅ at the p⁺-Al_0.55Ga_0.45N/Al_0.7Ga_0.3N interface and N₆ at the Al_0.7Ga_0.3N/n⁺-Al_0.55Ga_0.45N interface, where N₅ > 0 and N₆ < 0 under tensile strain of Al_0.7Ga_0.3N, and can be calculated as N₅ = 3.7 × 10²⁰ cm⁻³ and N₆ = −3.4 × 10²⁰ cm⁻³, as shown in Figure 2e. E_p has a direction along the [0001] orientation in the Al_0.7Ga_0.3N layer, in the reverse direction to E_b, and a direction along the [000-1] orientation at both sides of the Al_0.7Ga_0.3N layer, in the same direction as E_b, as shown in Figure 2c.

Figure 3 compares the electric field profiles at a relative horizontal position of 100 μm in the TJs of LEDs (300 × 300 μm²) for Structures A, B, and C at a current of 180 mA. For all three structures, the intensities of E_c were very strong in the TJ regions. For Structure A, E_c comes from the contribution of only E_b with a direction along the [000-1] orientation, as shown in Figure 2a. For Structure B, E_c comes from the contributions of E_b and E_p. E_b has a direction along the [000-1] orientation. E_p has a direction along the [000-1] orientation in the Al_0.4Ga_0.6N layer, in the same direction as E_b, and a direction along the [0001] orientation in the p⁺-Al_0.55Ga_0.45N and n⁺-Al_0.55Ga_0.45N layers, in the reverse direction to E_b, as shown in Figure 2b. As a result, the intensity of E_c is enhanced by E_p in the Al_0.4Ga_0.6N layer, and degraded by E_p in the p⁺-Al_0.55Ga_0.45N and n⁺-Al_0.55Ga_0.45N layers compared with Structure A, as shown in Figure 3. For Structure C, E_c comes from the contributions of E_b and E_p. E_b has a direction along the [000-1] orientation. E_p has a direction along the [0001] orientation in the Al_0.7Ga_0.3N layer, in the reverse direction to E_b, and a direction along the [000-1] orientation in the p⁺-Al_0.55Ga_0.45N and n⁺-Al_0.55Ga_0.45N layers, in the same direction as E_b, as shown in Figure 2c. As a result, the intensity of E_c is degraded by E_p in the Al_0.7Ga_0.3N layer, and enhanced by E_p in the p⁺-Al_0.55Ga_0.45N and n⁺-Al_0.55Ga_0.45N layers, as compared with Structure A, as shown in Figure 3.

Moreover, ε_r has a role in controlling the intensity of E_c, according to the expression in Equation (1). For AlGaN, ε_r becomes small with an increase in the Al content of AlGaN. For Structure B, Al_0.4Ga_0.6N as the dielectric layer has a higher ε_r than Structure A with Al_0.55Ga_0.45N, which should have decreased the intensity of E_c. Nevertheless, an increased electric field is still obtained because of the contribution of a polarization-induced electric field in the center. For Structure C, Al_0.7Ga_0.3N as the dielectric layer has a lower ε_r than Structure A with Al_0.55Ga_0.45N, which enhances the intensity of E_c because of the contribution of ε_r in the center. Thus, according to the final results regarding the contributions of E_b, E_p, and ε_r, the intensities of E_c are in the following order: Structure B > Structure C > Structure A in the center peak, and Structure C > Structure A > Structure B at both sides of the center, as shown in Figure 3.

According to Figure 3, the calculated peak intensities of E_c are as follows: E_A = 7.40 × 10⁶ V/cm, E_B = 9.07 × 10⁶ V/cm, and E_C = 8.24 × 10⁶ V/cm for Structures A, B, and C in the center peak, respectively. This means that very strong electric field intensities are generated in the TJ regions of the LEDs in Structures A, B, and C. Driven by E_c, the electrons in the valence band of the p⁺-Al_0.55Ga_0.45N layer can tunnel through the TJ and inject themselves into the conduction band of the n⁺-Al_0.55Ga_0.45N layer. The holes are simultaneously generated in the valence band of the p⁺-Al_0.55Ga_0.45N layer, and then injected into the MQWs of the LEDs.

Figure 4 shows a comparison of the energy-band profiles at a relative horizontal position of 100 μm in the TJs of LEDs (300 × 300 μm²) for Structures A, B, and C, at a current of 180 mA. For Structures A, B, and C, the conduction bands of n⁺-Al_0.55Ga_0.45N (n⁺ = 1 × 10²⁰ cm⁻³) align well below the valence bands of p⁺-Al_0.55Ga_0.45N (p⁺ = 1 × 10²⁰ cm⁻³). When the LEDs are operated at a forward biased voltage, the top TJs are operated at a reverse biased voltage, so the electrons can tunnel from the p⁺-side valence band into the n⁺-side conduction band, and the holes are simultaneously generated in the p-type layer. Under the reverse biased voltage, the conduction band levels in the TJs are totally different. The n⁺-side conduction band levels are in the following order: Structure B > Structure A > Structure C. The widths of the TJs in Structures A, B, and C are the distances between Positions O and A, Positions O and B, and Positions O and C, respectively. The calculated widths of the TJs are as follows: |OA| = 8.9 nm, |OB| = 9.5 nm, and |OC| = 8.0 nm. The widths of the TJs for Structures A, B, and C are in the following order: Structure B > Structure A > Structure C. The short width of the PTJ in Structure C can counteract the large energy barrier, and may help to enhance the tunneling probabilities of the electrons in the PTJ.

Figure 5a shows a comparison of the hole concentration distributions along the vertical direction at a relative horizontal position of 100 μm in the MQWs of the LEDs for Structures A, B, and C, at a current of 180 mA. The hole concentrations along the vertical direction in the MQWs are in the following order: Structure C > Structure A > Structure B. The high hole concentration in the MQWs in Structure C comes from the enhanced lateral current spreading in the PTJ.

Figure 5b shows a comparison of the radiative recombination rates along the vertical direction at a relative horizontal position of 100 μm in the MQWs of the LEDs for Structures A, B, and C at a current of 180 mA. The radiative recombination rates along the vertical direction in the MQWs are in the following order: Structure C > Structure A > Structure B. The radiative recombination rates come from the contributions of the high electron and hole concentrations in the MQWs, as is consistent with the hole-concentration distributions in the MQWs, as shown in Figure 5a.

In order to investigate the effects of the TJs on the current spreading in LEDs, the lateral distributions of the hole concentrations and the radiative recombination rates along the horizontal direction (x-axis) in the LEDs were characterized. Figure 5c shows a comparison of the lateral distributions of the hole concentrations and the radiative recombination rates along the horizontal direction (x-axis) and at a relative vertical position of the fifth QW (c-axis) in LEDs for Structures A, B, and C. For Structure B, the lateral distributions of the hole concentrations and the radiative recombination rates along the horizontal direction in the LEDs are very non-uniform: high at the horizontal positions of 0~50 μm, which is just below the anode, and low at the positions of 50~200 μm. This means that the positions under the anode are the main current flow paths going through the MQWs, which will cause current crowding and Joule heat. For Structures A, B, and C, the lateral distribution uniformities of the hole concentrations and the radiative recombination rates along the horizontal direction (x-axis) in the LEDs are in the following order: Structure C > Structure A > Structure B. The uniformities come from the contributions of high resistance in the TJs. If we compare Structures A, B, and C, the tunneling probability of electrons and the Al content of AlGaN in the TJs determine the resistance. Structure C has a lower electron tunneling probability and a higher Al content of AlGaN in the PTJ than Structure B, thus resulting in the higher resistance of the PTJ in Structure C. Therefore, Structure C with Al_0.7Ga_0.3N as the dielectric layer has higher uniformity than Structure B with Al_0.4Ga_0.6N as the dielectric layer, which means that relatively higher resistance of the TJ has a meaningful impact on the current spreading of an LED.

In order to quantitatively compare the lateral distribution uniformities of the hole concentrations and radiative recombination rates on the output characteristics of the LEDs for Structures A, B, and C, the integrated intensities of the lateral distributions of the hole concentrations and radiative recombination rates at the relative horizontal positions between 0 and 200 μm were characterized. Figure 5d shows a comparison of the integrated intensities of the lateral distributions of the hole concentrations and radiative recombination rates along the horizontal direction (x-axis) and at a relative vertical position of the fifth QW (c-axis) in LEDs for Structures A, B, and C. The integrated intensities for both the hole concentrations and radiative recombination rates are in the following order: Structure C > Structure A > Structure B. Compared with the lateral distributions of the hole concentrations and radiative recombination rates along the horizontal direction (x-axis), the integrated intensities more accurately reflected the output characteristics of LEDs with the different structures.

Figure 6 shows a comparison of the output current–voltage (I–V) characteristics of the LEDs (300 × 300 μm²) for Structures A, B, and C. The forward voltages of the LEDs are in the following order: Structure C > Structure A > Structure B. When an LED is operated at a forward biased voltage, the top TJ is operated at a reverse biased voltage. A TJ operating at a reverse biased voltage can be treated as a resistor, and its resistance determines the forward voltage of the LED, so that the forward voltage of an LED increases with an increase in the resistance of the JT. If we compare Structures A, B, and C, the tunneling probability of electrons and the Al content of AlGaN in the TJs determine the resistance, and the resistance determines the forward voltage of the LED. The resistances of the TJs are in the following order: Structure C > Structure A > Structure B, so the forward voltages of the LEDs are in the following order: Structure C > Structure A > Structure B. This indicates that the TJ has a nonnegligible impact on the forward voltage of DUV LEDs, and much higher biased voltage is required to realize higher tunneling efficiency for Structure C with a higher Al content of AlGaN.

Figure 7a shows a comparison of the IQE and optical output power of the LEDs (300 × 300 μm²) for Structures A, B, and C. Both the IQE and the optical output power of the LEDs are in the following order: Structure C > Structure A > Structure B, consistent with the results shown in Figure 5d, and inditcating that the enhanced IQE and optical output power of the LED for Structure C were achieved. Figure 7b shows a comparison of the WPE of LEDs for Structures A, B, and C. The WPE of the LEDs is in the following order: Structure B > Structure A > Structure C. The WPE is inversely proportional to the forward voltage, so the LED for Structure C had the lowest WPE due to the highest forward voltage.

4. Conclusions

AlGaN-based DUV LEDs with AlGaN as the dielectric layers in p⁺-Al_0.55Ga_0.45N/AlGaN/n⁺-Al_0.55Ga_0.45N PTJs were proposed for promoting carrier tunneling, suppressing current crowding, avoiding optical absorption, and further enhancing the performance of LEDs. Al_0.7Ga_0.3N with a lower dielectric constant can realize a higher hole concentration and a higher radiative recombination rate in MQWs than Al_0.4Ga_0.6N with a higher dielectric constant. The high hole concentration and high radiative recombination rate in MQWs come from enhanced lateral current spreading in the LED. Moreover, besides the high electrically conductive n⁺-Al_0.55Ga_0.45N layer in the PTJ, Al_0.7Ga_0.3N as the dielectric layer has relatively high resistance, which can increase the lateral current spreading and enhance the uniformity of the top emitting light of AlGaN-based DUV LEDs. As a result, the IQE and optical output power can be enhanced. However, its relatively high resistance results in an increase in the forward voltage, which is disadvantageous to the WPE. Through the adoption of PTJ with Al_0.7Ga_0.3N as the dielectric layers, enhanced IQE and optical output power can be achieved. It is strongly believed that the proposed structure is promising for further improving the IQEs of DUV LEDs, and the physics of the reported device is also useful for better understanding the carrier tunnel, transport and recombination mechanisms of DUV LEDs.