Improved Performance of Electron Blocking Layer Free AlGaN Deep Ultraviolet Light-Emitting Diodes Using Graded Staircase Barriers

To prevent electron leakage in deep ultraviolet (UV) AlGaN light-emitting diodes (LEDs), Al-rich p-type AlxGa(1−x)N electron blocking layer (EBL) has been utilized. However, the conventional EBL can mitigate the electron overflow only up to some extent and adversely, holes are depleted in the EBL due to the formation of positive sheet polarization charges at the heterointerface of the last quantum barrier (QB)/EBL. Subsequently, the hole injection efficiency of the LED is severely limited. In this regard, we propose an EBL-free AlGaN deep UV LED structure using graded staircase quantum barriers (GSQBs) instead of conventional QBs without affecting the hole injection efficiency. The reported structure exhibits significantly reduced thermal velocity and mean free path of electrons in the active region, thus greatly confines the electrons over there and tremendously decreases the electron leakage into the p-region. Moreover, such specially designed QBs reduce the quantum-confined Stark effect in the active region, thereby improves the electron and hole wavefunctions overlap. As a result, both the internal quantum efficiency and output power of the GSQB structure are ~2.13 times higher than the conventional structure at 60 mA. Importantly, our proposed structure exhibits only ~20.68% efficiency droop during 0–60 mA injection current, which is significantly lower compared to the regular structure.


Introduction
The AlGaN-based ultraviolet (UV) light-emitting diodes (LEDs) offer tremendous potential for a wide range of applications, including air/water purification, surface disinfection, biochemical sensing, cancer cell elimination, and many more [1]. These UV LEDs have the potential to replace the bulky and toxic conventional UV lamps due to advantages like environment-friendly material composition, longer life-time, low power consumption due to low DC drive voltage, compact in size, and tunable emission across the UV region from~200 nm to~365 nm [2]. Nevertheless, the external quantum efficiency (EQE) and light output power of AlGaN deep UV-LEDs are still low due to several challenges. For instance, strong induced polarization fields and quantum-confined Stark effect (QCSE) contribute significantly to the separation of electron and hole wave functions, leading to reduced carrier confinement and radiative recombination in the device active region. Subsequently, the electron overflow, which acts as one of the primary reasons for efficiency droop is increased [3].
To eliminate the electron overflow, a p-doped Al-rich electron blocking layer (EBL) has been introduced between the active region and p-region [4]. This could mitigate the electron leakage to only an extent. However, hole injection efficiency is affected owing to the

Device Structure and Parameters
Firstly, to validate our device model and parameters, we have considered the conventional EBL-based AlGaN deep UV LED structure grown on a c-plane AlN template with 284 nm wavelength emission as a reference structure and denoted with LED 1. This study was experimentally reported by Yan et al. [21]. LED 1 consists of a 3 µm n-Al 0.6 Ga 0.4 N layer (Si doping concentration: 5 × 10 18 cm −3 ), succeeded by an active region, followed by a 20 nm p-Al 0.65 Ga 0.35 N EBL (Mg doping concentration: 2 × 10 19 cm −3 ), then capped by a 50 nm p-Al 0.5 Ga 0.5 N hole injection layer (Mg doping concentration: 2 × 10 19 cm −3 ), and finally a 120 nm p-GaN contact layer (Mg doping concentration: 1 × 10 20 cm −3 ). The active region comprises of five intrinsic 3 nm Al 0.4 Ga 0.6 N QWs sandwiched between six intrinsic 12 nm Al 0.5 Ga 0.5 N QBs. The schematic diagram of LED 1 is presented in Figure 1a and the Al composition (%) profile related to the conduction band energy diagram of LED 1 is shown in Figure 1b. The mesa area of the deep UV LED chip is 400 µm × 400 µm. As illustrated in Figure 1c, LED 2 has the same structure as LED 1 except the QBs, where the Al composition of the QBs is gradually increasing from QB 2 to QB 6 as 0.51, 0.54, 0.57, 0.60, and 0.75, respectively. The proposed structure referred to as LED 3 is identical to LED 2 with the replacement of GSQBs instead of uniform Al composition QBs. As depicted in Figure 1d, 12 nm thick each QB consists of Al x Ga (1−x) N (4 nm)/Al (x + 0.5)/2 Ga 1−(x + 0.5)/2 N (4 nm)/Al 0.5 Ga 0.5 N (4 nm) step layers. The Al composition (x) in the last five QBs is 0.51, 0.54, 0.57, 0.60, and 0.75, respectively. The x values are chosen by carefully optimizing the structure, similar to our previous study [13].
Micromachines 2021, 12, x 3 of 11 LED 2 with the replacement of GSQBs instead of uniform Al composition QBs. As depicted in Figure 1d, 12 nm thick each QB consists of AlxGa(1−x)N (4 nm)/Al(x + 0.5)/2Ga1-(x + 0.5)/2N (4 nm)/Al0.5Ga0.5N (4 nm) step layers. The Al composition (x) in the last five QBs is 0.51, 0.54, 0.57, 0.60, and 0.75, respectively. The x values are chosen by carefully optimizing the structure, similar to our previous study [13]. In this study, the above-mentioned LED structures are numerically studied using the Advanced Physical Models of Semiconductor Devices (APSYS) tool. The energy bandgap of GaN and AlN are estimated using the Varshni formula [22] where Eg(T) and Eg(0) are the energy bandgap at temperatures T and 0 K, respectively. a and b are material constants. The values of a, b, and Eg (0) for GaN are 0.909 meV/K, 830 K, and 3.507 eV [23]. The corresponding values for AlN are 1.799 meV/K, 1462 K, and 6.23 eV, respectively [23]. The band offset ratio and bowing parameter for AlGaN are taken as 0.67/0.33 and 0.94 eV, respectively [24]. The carrier mobility is estimated using the Cauchy-Thomas approximation [25] and the energy band diagrams of LED structures are calculated by using 6 × 6 k.p model [26]. Additionally, the Mg activation energy of AlxGa(1−x)N alloy for 0 < x < 1 is set to scale linearly from 170 meV to 510 meV [6]. The Shockley-Read-Hall (SRH) recombination life-time, radiative recombination coefficient, Auger recombination coefficient, and light extraction efficiency are set as 15 ns, 2.13 × 10 −11 cm 3 /s, 2.88 × 10 −30 cm 6 /s, and 15%, respectively [27]. Moreover, the built-in polarization due to spontaneous and piezoelectric polarization is estimated using the methods proposed by Fiorentini et al. [28] and considered as 50% of the theoretical value. All simulations are performed at room temperature and other band parameters can be found elsewhere [29].

Results
The numerical device model and parameters implemented in this study are optimized based on the experimentally measured data of LED 1 published by Yan et al. [21]. Figure 2 shows the numerically calculated light-current-voltage curves of LED 1 closely matching with the experimentally obtained curves that validate our device model.
To investigate the performance of the proposed structure, we have performed a numerical study on three LEDs, and the results are carefully analyzed. As a part of this study, In this study, the above-mentioned LED structures are numerically studied using the Advanced Physical Models of Semiconductor Devices (APSYS) tool. The energy bandgap of GaN and AlN are estimated using the Varshni formula [22] where E g (T) and E g (0) are the energy bandgap at temperatures T and 0 K, respectively. a and b are material constants. The values of a, b, and E g (0) for GaN are 0.909 meV/K, 830 K, and 3.507 eV [23]. The corresponding values for AlN are 1.799 meV/K, 1462 K, and 6.23 eV, respectively [23]. The band offset ratio and bowing parameter for AlGaN are taken as 0.67/0.33 and 0.94 eV, respectively [24]. The carrier mobility is estimated using the Cauchy-Thomas approximation [25] and the energy band diagrams of LED structures are calculated by using 6 × 6 k.p model [26]. Additionally, the Mg activation energy of Al x Ga (1−x) N alloy for 0 < x < 1 is set to scale linearly from 170 meV to 510 meV [6]. The Shockley-Read-Hall (SRH) recombination life-time, radiative recombination coefficient, Auger recombination coefficient, and light extraction efficiency are set as 15 ns, 2.13 × 10 −11 cm 3 /s, 2.88 × 10 −30 cm 6 /s, and 15%, respectively [27]. Moreover, the built-in polarization due to spontaneous and piezoelectric polarization is estimated using the methods proposed by Fiorentini et al. [28] and considered as 50% of the theoretical value. All simulations are performed at room temperature and other band parameters can be found elsewhere [29].

Results
The numerical device model and parameters implemented in this study are optimized based on the experimentally measured data of LED 1 published by Yan et al. [21]. Figure 2 shows the numerically calculated light-current-voltage curves of LED 1 closely matching with the experimentally obtained curves that validate our device model.
we have calculated the energy-band diagrams of LED 1, LED 2, and LED 3 at 60 mA injection current, as shown in Figure 3. The effective CBBH at the corresponding barrier (n) and EBL layer are denoted as фen and фEBL, respectively.  In the same way, фhn denotes the effective valence band barrier heights (VBBH) at the corresponding barrier (n). The values for each of CBBH are extracted from the energy band diagrams and listed in Table 1. The value of фEBL is 235 meV in the case of LED 1 due to EBL, which is the maximum CBBH to block the electron overflow in LED 1. This value is comparatively low in contrast to LED 2 and LED 3 without EBL. In LED 2 and LED 3, the value of фen is progressively increasing with each QB and effectively blocking the electrons overflow by preventing the electrons from jumping out of the QWs. Further, the value of maximum CBBH i.e., фe6 in LED 3, is even higher than that of LED 2 as listed in To investigate the performance of the proposed structure, we have performed a numerical study on three LEDs, and the results are carefully analyzed. As a part of this study, we have calculated the energy-band diagrams of LED 1, LED 2, and LED 3 at 60 mA injection current, as shown in Figure 3. The effective CBBH at the corresponding barrier (n) and EBL layer are denoted as ф en and ф EBL , respectively. we have calculated the energy-band diagrams of LED 1, LED 2, and LED 3 at 60 mA injection current, as shown in Figure 3. The effective CBBH at the corresponding barrier (n) and EBL layer are denoted as фen and фEBL, respectively.  In the same way, фhn denotes the effective valence band barrier heights (VBBH) at the corresponding barrier (n). The values for each of CBBH are extracted from the energy band diagrams and listed in Table 1. The value of фEBL is 235 meV in the case of LED 1 due to EBL, which is the maximum CBBH to block the electron overflow in LED 1. This value is comparatively low in contrast to LED 2 and LED 3 without EBL. In LED 2 and LED 3, the value of фen is progressively increasing with each QB and effectively blocking the electrons overflow by preventing the electrons from jumping out of the QWs. Further, the value of maximum CBBH i.e., фe6 in LED 3, is even higher than that of LED 2 as listed in In the same way, ф hn denotes the effective valence band barrier heights (VBBH) at the corresponding barrier (n). The values for each of CBBH are extracted from the energy band diagrams and listed in Table 1. The value of ф EBL is 235 meV in the case of LED 1 due to EBL, which is the maximum CBBH to block the electron overflow in LED 1. This value is comparatively low in contrast to LED 2 and LED 3 without EBL. In LED 2 and LED 3, the value of ф en is progressively increasing with each QB and effectively blocking the electrons overflow by preventing the electrons from jumping out of the QWs. Further, the value of maximum CBBH i.e., ф e6 in LED 3, is even higher than that of LED 2 as listed in Table 1, demonstrating LED 3 is the optimal choice to confine the electrons in the active region. As a result, in comparison with other LEDs, improved and maximum electron concentration in the active region for LED 3 was observed and is shown in Figure 4a. Though LED 2 has boosted electron concentration as compared to LED 1, but it is lower than LED 3. It is also noticed that due to improved electron confinement in the active region, electron leakage into the p-region is significantly reduced in LED 3, as shown in Figure 4b. Subsequently, this would reduce the non-radiative recombination of the overflowed electrons, and incoming holes in the p-region thereby contribute to better hole injection efficiency into the active region. However, LED 2 has even higher electron leakage as compared to LED 1. Due to this, the non-radiative recombination in the p-region of LED 2 would increase and reduce the hole injection efficiency into the active region, irrespective of the creation of negative sheet polarization charges at the last QB and p-Al 0.5 Ga 0.5 N interface.  Table 1, demonstrating LED 3 is the optimal choice to confine the electrons in the active region. As a result, in comparison with other LEDs, improved and maximum electron concentration in the active region for LED 3 was observed and is shown in Figure 4a. Though LED 2 has boosted electron concentration as compared to LED 1, but it is lower than LED 3. It is also noticed that due to improved electron confinement in the active region, electron leakage into the p-region is significantly reduced in LED 3, as shown in Figure 4b. Subsequently, this would reduce the non-radiative recombination of the overflowed electrons, and incoming holes in the p-region thereby contribute to better hole injection efficiency into the active region. However, LED 2 has even higher electron leakage as compared to LED 1. Due to this, the non-radiative recombination in the p-region of LED 2 would increase and reduce the hole injection efficiency into the active region, irrespective of the creation of negative sheet polarization charges at the last QB and p-Al0.5Ga0.5N interface.  It is worthwhile to note that in the last QB of LED 1, a sharp bending in the conduction band is formed due to induced positive polarization sheet charges at the heterointerface of the last QB and EBL. This area accumulates a large number of electrons i.e., 3.66 × 10 16 cm −3 , which eventually contributes to non-radiative recombination [30]. In addition, due to this induced positive polarization sheet charges in LED 1, a hole depletion region is formed at the heterointerface of the last QB and EBL, as shown in Figure 3a, which reduces the hole injection efficiency [5]. The formation of the hole depletion region problem is eliminated in the case of LED 2 and LED 3 by removing the EBL. In the case of LED 2, a hole accumulation region is formed at the interface of the last QB and p-region, which generally should improve the hole injection efficiency, whereas in LED 3 the hole injection efficiency even should further improve due to the formation of two-hole accumulation regions as shown in Figure 3c. The boosted hole injection efficiency in LED 3 can be seen from Figure 4c. This is also because of the reduced electron overflow in LED 3 due to improved electron confinement in the active region. Moreover, the effective VBBH, ф hn due to each QB, are calculated and listed in Table 2. As ф hn increases with the increase in Al composition in the QBs, the values of ф hn are found to be high in LED 2 and LED 3 compared to LED 1. This supports the improved hole confinement and increased hole concentration in the active region. However, a very high ф hn can also affect the hole transportation in the active region at the same time, which is in the case of LED 2. Moreover, already the hole injection efficiency is poor in LED 2, altogether the hole concentration is very low in LED 2, as shown in Figure 4c. In this regard, LED 3 has a smaller value of ф hn as compared to LED 2 due to again GSQB structures. Altogether, due to effective hole injection efficiency along with a comparable height of ф hn in the active region, hole concentration in LED 3 is relatively evenly distributed as compared to other LEDs. Overall, the hole concentration in the active region of all three LEDs is 7.2 × 10 16 cm −3 , 4.8 × 10 15 cm −3 , 7.8 × 10 16 cm −3 , respectively. Importantly, the overlap level of electron and hole wave functions in the active region for LED 1 and LED 3 are summarized in Table 3. It is seen that even though the hole concentration of LED 3 is close to LED 1, the proposed structure in LED 3 improves the electron and hole wavefunctions overlap level as compared to LED 1. As a result, the radiative recombination is significantly increased in LED 3, as depicted in Figure 4d. Finally, the IQE and output power of LED 1, LED 2, and LED 3 as a function of injection current are illustrated in Figure 5a,b, respectively. Figure 5c depicts the electroluminescence (EL) spectra of the three LEDs. As shown in Figure 5a, LED 3 exhibits the maximum IQE of 44.34%, whereas it is only 35.69% and 29.46% in the case of LED 1 and LED 2, respectively. In addition, the droop in the IQE during 0 mA-60 mA injection current is remarkably reduced to 20.68% in the proposed structure as compared to 53.68% and 94.7% in LED 1 and LED 2, respectively. This is due to the enhanced carrier transportation and confinement in the active region, thereby reduced electron overflow into the p-region because of GSQBs in the proposed structure. As depicted in Figure 5b, the output power of LED 3 is 2.13 times higher than LED 1 and 22.56 times higher than LED 2. As shown in Figure 5c, LED 3 depicts higher EL intensity as compared to LED 1 and LED 2 at the emission wavelength of~284 nm due to improved radiative recombination in the active region. EL intensity of LED 3 is~2.12 times higher than LED 1 and~22.24 times higher than LED 2. Different parameters related to IQE and output power of three LED structures are summarized in Table 4. remarkably reduced to 20.68% in the proposed structure as compared to 53.68% and 94.7% in LED 1 and LED 2, respectively. This is due to the enhanced carrier transportation and confinement in the active region, thereby reduced electron overflow into the p-region because of GSQBs in the proposed structure. As depicted in Figure 5b, the output power of LED 3 is 2.13 times higher than LED 1 and 22.56 times higher than LED 2. As shown in Figure 5c, LED 3 depicts higher EL intensity as compared to LED 1 and LED 2 at the emission wavelength of ~284 nm due to improved radiative recombination in the active region. EL intensity of LED 3 is ~2.12 times higher than LED 1 and ~22.24 times higher than LED 2. Different parameters related to IQE and output power of three LED structures are summarized in Table 4.  To better understand the role of GSQB structure in LED 3, the schematic model for transportation of electrons in LED 1 and LED 3 is depicted in Figure 6. In this study, the total number of injected electrons into the n-Al0.6Ga0.4N region is considered as N0 for LED 1 and LED 3. For the simplicity of the model, electron loss through non-radiative recombination in n-Al0.6Ga0.4N region is neglected. The captured electrons in the quantum well (Ncapture) are correlated with the electron mean free path (lMFP) as expressed in Equation (2) [31].  To better understand the role of GSQB structure in LED 3, the schematic model for transportation of electrons in LED 1 and LED 3 is depicted in Figure 6. In this study, the total number of injected electrons into the n-Al 0.6 Ga 0.4 N region is considered as N 0 for LED 1 and LED 3. For the simplicity of the model, electron loss through non-radiative recombination in n-Al 0.6 Ga 0.4 N region is neglected. The captured electrons in the quantum well (N capture ) are correlated with the electron mean free path (l MFP ) as expressed in Equation (2) [31].
where t QW is the quantum well thickness. Illustrated in Figure 6a,b, the incoming electrons (N 0 ) are scattered and fall into the quantum wells, denoted by process 1. Some of those fallen electrons recombine with the holes radiatively as well as with the crystal defects as depicted by process 2, while remaining electrons escape from the QWs as illustrated by process 3. In addition, some electrons with longer l MFP travel to a remote position without being captured by the QWs as indicated by process 4. To increase N capture , the l MFP of these electrons needs to be reduced so that the electron concentration in the QWs would be increased that would favor the higher radiative recombination rate in the active region. At the same time, l MFP depends on thermal velocity (v th ) and the scattering time (τ sc ) as shown in Equation (3). For LED 1, v th can be further expressed as illustrated in Equation (4) [31].
where E is the excess kinetic energy in the n−Al 0.6 Ga 0.4 N layer, qV 1 is the work done to the electrons by the induced polarization electric field in QBs of LED 1, and m e is the effective mass of electrons. +∆E c denotes the conduction band offset between QB n and QW n , while −∆E c represents the conduction band offset between QW n and QB n + 1 . On the other hand, the GSQB structure in LED3 forms discontinuity in the conduction band of QB layers due to which the probability of the electrons to be scattered increases.
where tQW is the quantum well thickness. Illustrated in Figure 6a,b, the incoming electrons (N0) are scattered and fall into the quantum wells, denoted by process 1. Some of those fallen electrons recombine with the holes radiatively as well as with the crystal defects as depicted by process 2, while remaining electrons escape from the QWs as illustrated by process 3. In addition, some electrons with longer lMFP travel to a remote position without being captured by the QWs as indicated by process 4. To increase Ncapture, the lMFP of these electrons needs to be reduced so that the electron concentration in the QWs would be increased that would favor the higher radiative recombination rate in the active region. At the same time, lMFP depends on thermal velocity (vth) and the scattering time (τsc) as shown in Equation (3). For LED 1, vth can be further expressed as illustrated in Equation (4) [31].
where E is the excess kinetic energy in the n−Al0.6Ga0.4N layer, qV1 is the work done to the electrons by the induced polarization electric field in QBs of LED 1, and me is the effective mass of electrons. +ΔEc denotes the conduction band offset between QBn and QWn, while −ΔEc represents the conduction band offset between QWn and QBn + 1. On the other hand, the GSQB structure in LED3 forms discontinuity in the conduction band of QB layers due to which the probability of the electrons to be scattered increases. Therefore, electrons would be thermalized more efficiently by interacting with longitudinal optical (LO) phonons, thereby reducing the vth and lMFP, as a result, electron confinement in the active region increases [19]. Hence, the vth in LED3 can be expressed as follows, ω υ Therefore, electrons would be thermalized more efficiently by interacting with longitudinal optical (LO) phonons, thereby reducing the v th and l MFP , as a result, electron confinement in the active region increases [19]. Hence, the v th in LED3 can be expressed as follows, where +∆E c1 represents the conduction band offset between QB n and QW n whereas −∆E c2 is the conduction band offset between QW n and QB n + 1 . As the QB heights are varying in LED 3 along the growth direction, ∆E c1 − ∆E c2 can therefore not be eliminated. qV 2 is the work done to the electrons by the induced polarization electric field in GSQBs of LED 3. The −hω LO denotes the total energy loss by phonon emissions due to each step layer in GSQBs. The values of qV due to each QB in LED1 and LED3 are listed in Table 5. Further, the values ofhω LO in each step of GSQBs for LED3 are calculated [32] and presented in Figure 7. where +∆Ec1 represents the conduction band offset between QBn and QWn whereas −∆Ec2 is the conduction band offset between QWn and QBn + 1. As the QB heights are varying in LED 3 along the growth direction, ΔEc1−ΔEc2 can therefore not be eliminated. qV2 is the work done to the electrons by the induced polarization electric field in GSQBs of LED 3. The −ℏꞷLO denotes the total energy loss by phonon emissions due to each step layer in GSQBs. The values of qV due to each QB in LED1 and LED3 are listed in Table 5. Further, the values of ℏꞷLO in each step of GSQBs for LED3 are calculated [32] and presented in Figure 7.   (4) and (5), it is understood that (E + ΔEc1 + qV2−ΔEc2−ℏꞷLO) < (E + qV1). Consequently, vth for LED3 would be less as compared to LED1. As a result, lMFP would be reduced, which improves the electrons capture (Ncapture) ability of the QWs in LED3. In addition, the electron overflow happening due to process 4 can also be reduced by increasing the barrier height as in the proposed structure shown in Figure 6b. Here, the QB heights before and after the QWs are not at the same level, rather it is progressively increasing along the growth direction due to which some of the electrons from process 3 and 4 would bounce back denoted as process 5, which can also aid to improved electron concentration in the QWs in comparison with LED 1 as shown in Figure 4a. The proposed AlGaN deep UV LEDs using graded staircase barriers can also be realized by experimentation due to a simple device architecture. As different AlGaN based UV LEDs with thinner epilayers than our proposed structure have already been grown by metal-organic chemical vapor deposition (MOCVD) [33][34][35] and molecular beam epitaxy (MBE) [36,37]. Therefore, it is anticipated that the proposed structure can also be grown by both MBE and MOCVD.

Conclusion
We have numerically demonstrated and investigated the performance of EBL-free AlGaN UV LEDs emitting light at ~284 nm wavelength with the incorporation of GSQB structures. The reduced thermal velocity and mean free path of electrons improved the electron capture efficiency in the multi QWs, thus electron overflow was suppressed eminently. In addition, carefully engineered GSQBs promoted the hole injection by forming negative sheet polarization charges and improved the spatial overlap of the electron-hole From Equations (4) and (5), it is understood that (E + ∆E c1 + qV 2 − ∆E c2 −hω LO ) < (E + qV 1 ). Consequently, v th for LED3 would be less as compared to LED1. As a result, l MFP would be reduced, which improves the electrons capture (N capture ) ability of the QWs in LED3. In addition, the electron overflow happening due to process 4 can also be reduced by increasing the barrier height as in the proposed structure shown in Figure 6b. Here, the QB heights before and after the QWs are not at the same level, rather it is progressively increasing along the growth direction due to which some of the electrons from process 3 and 4 would bounce back denoted as process 5, which can also aid to improved electron concentration in the QWs in comparison with LED 1 as shown in Figure 4a. The proposed AlGaN deep UV LEDs using graded staircase barriers can also be realized by experimentation due to a simple device architecture. As different AlGaN based UV LEDs with thinner epilayers than our proposed structure have already been grown by metal-organic chemical vapor deposition (MOCVD) [33][34][35] and molecular beam epitaxy (MBE) [36,37]. Therefore, it is anticipated that the proposed structure can also be grown by both MBE and MOCVD.

Conclusions
We have numerically demonstrated and investigated the performance of EBL-free AlGaN UV LEDs emitting light at~284 nm wavelength with the incorporation of GSQB structures. The reduced thermal velocity and mean free path of electrons improved the electron capture efficiency in the multi QWs, thus electron overflow was suppressed eminently. In addition, carefully engineered GSQBs promoted the hole injection by forming negative sheet polarization charges and improved the spatial overlap of the electron-hole wavefunction. Therefore, the proposed structure exhibited higher radiative recombination and recorded output power of 13.9 mW at 60 mA injection current, which is 2.13 times higher than the conventional structure. Hence, the reported structure shows incredible potential to develop high-efficiency UV light emitters for real-world applications.

Conflicts of Interest:
The authors declare no conflict of interest.