Browse

• Abstract

SECTION 1

INTRODUCTION

In recent years, the AlGaN-based ultraviolet light-emitting diodes (UV-LEDs) have attracted much attention because of their applications in air, water, and surface sterilization, high density optical storage systems, full-color displays, chemical sensors, and medical applications [1], [2], [3]. Despite the attractive opportunities, the effects of huge built-in electric fields caused by the spontaneous and piezoelectric polarizations along the c-axis limit the performance of the quantum well active regions in the UV-LEDs [4]. The built-in electric fields lead to the spatial separation between electron and hole wave functions, which gives rise to the reduction of radiative recombination rate [5], [6]. Various approaches have been pursued for suppressing charge separation in III-Nitride QWs by using non-polar/semi-polar QWs [7] and novel QWs with large overlap designs [8], [9], [10]. Specifically, for AlGaN QWs, recent works have resulted in improved understanding on the physics of active regions for mid and deep UV high Al-content AlGaN QWs LEDs [11], [12], [13], [14], [15], [16], [17]. The identification valence subband crossover in high Al-content AlGaN QW has been reported [11], [12], [13], and the use of delta-based QWs has been pursued [14], [15], [16], [17]. Meanwhile, the emitting power of the AlGaN-based LEDs undergoes an unsatisfactory non-linear increment when elevating the current [19]. This phenomenon is known as the efficiency droop [19], [20], which is a significant challenge to the improvement of the output power of AlGaN-based LEDs. Over the past few years, this phenomenon has been studied widely [21], [22], [23], [24], and one of the most frequently cited explanation for the efficiency droop is electron leakage from multi-quantum wells (MQWs) into the p-type layers [24], resulting in the poor radiative recombination in the active region. Conventionally, p-AlGaN electron-blocking layers (EBLs) with high Al contents are used to suppress the electron overflow. But at the meantime, holes will be blocked and few can be injected into the active region through the EBL with large barrier height, which can also aggravate the escape of electrons into p-type layers. However, to our knowledge, few researchers have focused on this issue in AlGaN-based deep UV-LEDs [25], [26], [27], [28], [29].

In our previous work, we theoretically reported the study of AlGaN-based 310 nm UV LEDs with compositional graded AlGaN electron blocking layers (EBLs). The results showed that the spill-over of electrons out of the active region could be effectively suppressed by adopting graded AlGaN EBL, and thus could greatly improve the internal quantum efficiency [30]. Previously, the work by Piprek et al. had also demonstrated the importance of compositional graded EBLs for AlGaN-based UV LEDs [31]. In this letter, we have proposed a new structure to mitigate efficiency droop in AlGaN-based deep UV-LEDs by inserting a p-AlGaN layer between the EBL and last barrier. Furthermore, the Al composition of the inserted layer has been optimized by Crosslight APSYS (Advance Physical Model of Semiconductor Devices) programs [32]. To shed light on the mechanisms responsible for the improvement, the band diagram, optical and electrical properties of this structure were simulated by solving the Schrödinger equation, the Poisson's equation, the carrier transport equations and the current continuity equation self-consistently [33], [34].

SECTION 2

STRUCTURE AND PARAMETER

The structure of a conventional UV-LED (denoted as structure A), as shown in Fig. 1, is used as a reference in this study, and the proposed one is labeled as structure B. Three-dimension finite element analysis is used for simulation, and the sizes of electrodes are set to be $200 \times 200\ \mu\hbox{m}^{2}$. The structure A is grown on a c-plane sapphire substrate, followed by 3-$\mu\hbox{m}$-thick Si-doped n-type $\hbox{Al}_{0.55}\hbox{Ga}_{0.45}\hbox{N}$ layer $(\hbox{Si doping} = 5 \times 10^{18}\ \hbox{cm}^{-3})$. The active region includes five 2-nm-thick $\hbox{Al}_{0.45}\hbox{Ga}_{0.55}\hbox{N}$ QWs separated by six 10-nm-thick $\hbox{Al}_{0.55} \hbox{Ga}_{0.45}\hbox{N}$ barriers. Then a 10-nm-thick Mg-doped p-type $\hbox{Al}_{0.6}\hbox{Ga}_{0.4}\hbox{N}$ EBL $(\hbox{Mg doping} = 5 \times 10^{18}\ \hbox{cm}^{-3})$ is on the top of the active region, followed by a 10-nm-thick p-type $\hbox{Al}_{0.55}\hbox{Ga}_{0.45}\hbox{N}\ (\hbox{Mg doping} = 5 \times 10^{18}\ \hbox{cm}^{-3})$ and a 100-nm-thick p-type GaN contact layer $(\hbox{Mg doping} = 3 \times 10^{19}\ \hbox{cm}^{-3})$. While in structure B, a 5-nm-thick Mg-doped p-type $\hbox{Al}_{0.5}\hbox{Ga}_{0.5}\hbox{N} \ (\hbox{Mg doping} = 5 \times 10^{18}\ \hbox{cm}^{-3})$ is inserted between the EBL and last barrier of the MQWs, which can be realized in the process of growth in MOCVD system [35]. Mg dopant ionization efficiency in this program is set to be 1%.

Fig. 1. Schematic diagrams of the (a) conventional structure A and (b) the proposed structure B.

The Shockley-Read-Hall (SRH) recombination time is set to be 1.5 ns for all layers except the p-type inserted layer as 1 ns because the SRH lifetime is dependent upon the doping level [36]. The internal loss is 2000 $\hbox{m}^{-1}$ [37]. In general, the bandgap energy of AlGaN is governed by following formulas: TeX Source $$E_{g} (Al_{x}Ga_{1 - x}N) = xE_{g, AlN} + (1 - x) E_{g, GaN} - bx (1 - x)\eqno{\hbox{(1)}}$$ where $E_{g, AlN}$ and $E_{g, GaN}$ are the bandgap energies of GaN and AlN, which equal to 3.42 eV and 6.25 eV, respectively. The bowing parameter b is 1 eV, and the band-offset ratio is assumed to be 0.7/0.3 for AlGaN materials [38]. The Auger recombination coefficient is set to be $1 \times 10^{-30}\ \hbox{cm}^{6}/ \hbox{s}$ to fit the experiment [39], and the operating temperature is assumed to be 300 K [40]. In this simulation, the built-in interface charges due to the spontaneous and piezoelectric polarization are calculated based on the method proposed by Fiorentini et al. [41]. Furthermore, taking the screening by defects into consideration, the surface charges densities are assumed to be 40% of the calculated values [42]. Other material parameters of the semiconductors used in the simulation can be found in Ref. [43].

SECTION 3

SIMULATION RESULTS AND DISCUSSION

Regarding the advantages of inserting a p-AlGaN layer with a proper Al mole composition between the EBL and last barrier on the performance of UV-LEDs, optical power and the internal quantum efficiency (IQE) of both structures have been investigated. As shown in Fig. 2, the power of structure B increases with the current sub-linearly and the slope of increase is sharper than that of structure A. Furthermore, compared to structure A, the power of structure B is increased about 37.7% at 150 mA, and even much higher with the increase of current. At the bias current of 250 mA, the output powers are 52.86 mW and 75.43 mW for the conventional structure and the proposed one, respectively. The higher light total power may result from more carriers confined in the quantum wells and higher radiative recombination. In order to explain this, the IQE has been investigated.

Fig. 2. Total output powers as a function of current for the conventional structure A (blue line) and the proposed structure B (red line).

As shown in Fig. 3, the IQE of structure B is higher than that of structure A. What is more, the efficiency droop in the structure B is improved apparently. In the whole injection current range, the efficiency droop in the structure B is 19.7%, when the efficiency is defined as TeX Source $$(IQE_{\max} - IQE_{250\, mA})/IQE_{\max},\eqno{\hbox{(2)}}$$ smaller than structure A $(\hbox{efficiency droop} = 28.2\%)$. Therefore, the optical performance of UV-LED can be improved significantly by inserting a p-type layer between the EBL and last barrier. The performance of the proposed structure can be improved further by optimizing the Al mole composition of this inserted layer, which will be discussed in detail below.

Fig. 3. IQE as a function of current for the conventional structure (blue line) and the proposed one (red line).

To shed light on the mechanism responsible for this improvement, the band diagrams of structure A and B under the forward current of 150 mA are given in this study. As shown in Fig. 4, the effective barrier height of structure B against the escape of electrons is 592 meV, higher than that of structure A (560 meV). The higher barrier height prevents electrons from leaking to the p-GaN contact layer more efficiently, which can mitigate the IQE droop at large current injection. Furthermore, the effective barrier height for holes injection is decreased from 150 meV to 73 meV, which can promote the holes injection efficiency and thus improve the IQE. Because the Al mole composition of the inserted layer is lower than the last barrier, the opposite piezoelectric polarization will be introduced in this layer, resulting in the increase of the last barrier in the conduction. Then the electron blocking is enhanced. This is different from the work reported by S. J. Lee et al., who inserted an undoped AlGaN layer between the MQWs and EBL of InGaN/GaN LEDs to improve the hole injection efficiency [35]. However, this structure may not enhance the electron blocking.

Fig. 4. Profiles of potential of (a) the conventional structure and (b) the proposed one.

This explanation can be confirmed by the increase of carrier concentration in Fig. 5. Note that the horizontal position of the conventional LED layer has been shifted slightly for better observation and similar operation has been taken to several figures below. The electron concentration is almost twice of the proposed structure with an inserted layer than the conventional one, and so does the hole concentration. The enhancement in carrier concentration evidently demonstrates that a mitigated leakage overflow and an increased hole injection can be obtained by inserting a p-AlGaN layer between the EBL and last barrier. Meanwhile, the p-type AlGaN inserted layer can act as a hole reservoir which can offer holes to inject into the active region without jumping through the EBL.

Fig. 5. (a) Electron concentrations for the conventional structure (blue line) and the proposed one (red line) at 150 mA. (b) Hole concentrations for the conventional structure (blue line) and the proposed one (red line) at 150 mA.

As mentioned above, the inserted layer will act as a hole reservoir between the last barrier and EBL. If the recombination of the electrons and holes happens in this shallow well significantly, the luminescent spectra may also be broadened, which is adverse to the UV-LED performance. In order to make it clear whether the recombination in the inserted layer contributes to the light emitting, the radiative recombination rates in both structures are studied. As shown in Fig. 6, the recombination is only observed in the active region, i.e., no recombination takes place in the inserted layer, illustrating that the improvement of optical power comes from the increase of hole injection and enhancement of electron blocking.

Fig. 6. Radiative recombination rates for the conventional structure (blue line) and the proposed structure (red line) at 150 mA.

The advantages of the proposed AlGaN-based deep UV-LED derive from that the inserted layer can change the profile of potential on one hand, which can enhance the electrons blocking and holes injection. On the other hand, it can work as a hole reservoir, which will increase the hole concentrations in the active region and affect the property of luminescent spectra. However, these two aspects are both related to the Al mole composition of p-AlGaN inserted layer. Therefore, the effects of Al content of the p-AlGaN inserted layer on the performance of UV-LED are also studied by varying the Al mole composition from 0.3 to 0.55.

It must be mentioned that the AlGaN material with Al composition less than 0.45 (the Al composition of the well in MQWs) may absorb the light emitted from the designed LED in this study. Therefore, the absorption of the AlGaN interlayer should be taken into account. The absorption coefficient of the AlGaN interlayer with lower Al content can be expressed by the following formulas [44]: TeX Source $$\alpha = \alpha_{0}\sqrt{(E - E_{g, AlGaN})}\eqno{\hbox{(3)}}$$ where $E$ is the energy of the emitted light and $E_{g, AlGaN}$ is the bandgap of the interlayer which can be obtained from the band diagrams. Here, the $\alpha_{0} =7.32\ \mu\hbox{m}^{-1} \hbox{eV}^{-1/2}$ is used and $e^{-\alpha L}$ is accounted as the factor of the absorption loss of the p-AlGaN interlayer [44]. Therefore, the absorption losses of p-AlGaN interlayers with Al content 0.3, 0.35, 0.4 are calculated to be 2.24%, 1.83%, 1.27%, respectively.

As shown in Fig. 7, the output power is increased when the Al mole composition is varied from 0.3 to 0.4 and decreased from 0.4 to 0.55. Furthermore, it is also observed that the powers of structures with Al compositions below 0.4 are generally larger than those with Al compositions above 0.4. This is consistent with the IQE shown in Fig. 8.

Fig. 7. (a) Total light power as a function of current for the conventional LED and the proposed LEDs with different Al mole compositions of the p-AlGaN inserted layers. (b) Total light power as a function of Al composition when the current is 250 mA.
Fig. 8. (a) IQE as a function of current for the conventional LED and the proposed LEDs with different Al mole compositions of the p-AlGaN layers. (b) IQE as a function of Al composition when the current is 250 mA.

The maximum IQE and the efficiency droop are both optimal when the Al composition of the inserted layer is 0.4, and they drop monotonously when the Al composition of the inserted layer is increased or decreased from 0.4. While the Al compositions vary from 0.45 to 0.55, the IQEs are nearly the same (52.6%, 52.7%, 52.1%, respectively), but the IQEs differ from each other at the current of 250 mA (32.5%, 28.2%, 23.5%, respectively), which means aggravated efficiency droop with the increasing of the Al mole composition from 0.45 to 0.55. Furthermore, the IQEs decrease with the Al contents varying from 0.4 to 0.3 and the deepened inserted well may be the reason. The deeper well gives rise to more carriers confined in the interlayer. However, nonradiative recombination is more serious in this p-type interlayer than the undoped quantum wells [36], which results in lower IQE when the Al content is less than 0.4. In order to study the influences of the Al mole compositions on the output power and IQE, the diagrams of radiative recombination rates at 150 mA are plotted.

Fig. 9 shows the radiative recombination rates for LEDs with p-AlGaN inserted layer whose Al composition is varied from 0.3 to 0.45. The recombination in the active region increases with the increased Al composition from 0.3 to 0.45. Meanwhile the recombination in the inserted layer is very dramatic when Al mole composition is less than or equal to 0.4, and even nearly fivefold higher than that in the MQWs when the Al mole composition is 0.3. Therefore, the increase of the power and the efficiency may be attributed to the recombination in the p-AlGaN inserted layer with the Al composition below 0.45. However, this strong recombination in the inserted layer may broaden the luminescent spectra dramatically, degrading the overall performance of the UV-LEDs as mentioned above. On the contrary, when the Al composition is above 0.45, the radiative recombination does not happen in the inserted layer. Furthermore, according to Fig. 9(b), the recombination in the quantum wells is the most dramatic when the Al composition is 0.45. In addition, the radiative recombination rates in the wells of MQWs decrease overall with the increase of Al composition from 0.45 to 0.55 as is shown in Fig. 10, which may result from the shallowed hole reservoir when the Al composition of p-AlGaN is increased.

Fig. 9. (a) Radiative recombination rates in the proposed LED and (b) the radiative recombination rates in active region (right), when Al mole compositions in the p-AlGaN layer are 0.3, 0.35, 0.4, and 0.45, respectively.
Fig. 10. Radiative recombination rates of the active region for LEDs with the inserted layer with Al compositions are 0.45, 0.5, and 0.55, respectively.

In general, when the Al mole composition of the inserted layer is below 0.4, the recombination between the electrons and holes in the inserted layer is very intensive. The recombination can contribute to the light power and the IQE undoubtedly. Nonetheless, it may also increase the broadening of the luminescent spectra. Therefore, the best Al component should be around 0.45.

SECTION 4

CONCLUSION

The advantages of inserting a p-AlGaN layer between the EBL and the last barrier of the MQWs are investigated numerically. The results show that the carrier concentrations in the active region can be increased, which is attributed to two main reasons. First, the inserted layer with relative narrow bandgap could not only increase the value of the effective barrier height in the conduction band but also decrease the counterpart in the valence band, resulting in a diminished leakage overflow and an increased hole injection rate. Secondly, holes from the p-type AlGaN inserted layer can contribute to the hole injection efficiency because these holes can inject directly into the active region not via the EBL. As a result, the power of the proposed LED inserted with an $\hbox{p-Al}_{0.5}\hbox{Ga}_{0.5}\hbox{N}$ is improved about 37.7% and IQE efficiency droop decreases about 8.5% under the injection current of 150 mA, in comparison with the conventional LED. The effects of the Al composition in the inserted layer are also investigated and the best Al composition is found to be around 0.45. In general, the designed LED shows larger light output power and higher IQE, which is promising for high efficiency sterilization and other optoelectronic application. However, this approach requires further growth optimization in realizing the potential advantages.

Footnotes

This work was supported in part by the National Basic Research Program of China under Grants 2012CB619302 and 2010CB923204; by the National Natural Science Foundation of China under Grants 10990103, 51002058, and 11104150; and by China Postdoctoral Science Foundation under Grant 20100480064. Corresponding author: C. Chen (e-mail: cqchen@hust.edu.cn).

References

No Data Available

Cited By

No Data Available

None

Multimedia

No Data Available
This paper appears in:
No Data Available
Issue Date:
No Data Available
On page(s):
No Data Available
ISSN:
None
INSPEC Accession Number:
None
Digital Object Identifier:
None
Date of Current Version:
No Data Available
Date of Original Publication:
No Data Available