Theoretical analysis…

In order to investigate the physical origin of efficiency droop in these UV LEDs, we investigated the above structures by using the APSYS simulation software. Commonly accepted Shockley-Read-Hall recombination lifetime (several nanoseconds) and Auger recombination coefficient (about ~10^-30 cm⁶s^-1) are used in the simulations. In addition, because of lattice match in barrier between AlGaN and InAlGaN, we can exclude the effect of total polarization fields including spontaneous and piezoelectric polarizations. The total polarization fields can be obtained through the calculation of the In0.025Ga0.975N, Al0.08Ga0.92N and In0.0085Al0.1112Ga0.8803N for the -0.0305, -0.0391 and -0.0398 (Cm^-2), respectively.[63]

Therefore, we use the same factor of 50% for charge screening effect. However, the preliminary simulation results cannot fit in with experiment. Thus, it must has some reasons for this outstanding UV LED with InAlGaN barrier, and here we intend to consider carrier mobility and band offset ratio as factor on droop behavior for these UV LEDs.

5.5.1 Carrier mobility issue

It’s difficult to calculate minority carrier hole mobility in semiconductor material because of the degenerate valence bands. On the other hand, as mentioned before in Sec 3.2.4, the majority carrier electron mobility of Ga1-x-yAlxInyN can be calculated by Caughey Thomas approximation. In our simulation, the calculated electron mobility is 354 cm²/V^-1s^-1 for

Al0.08Ga0.92N and 642 cm²/V^-1s^-1 for In0.0085Al0.1112Ga0.8803N, respectively. Hence, to investigate the efficiency droop in these two samples, we assume that InGaN-based UV LED with InAlGaN barrier has relatively high carrier mobility.

To prove above hypothesis, we vary the carrier mobility of InAlGaN depending on the value of AlGaN. These simulation results are shown in Fig 5.5.1. It can clearly be seen that the droop behavior is dominated by hole mobility, and we find the efficiency curve will nearest to the experimental result when hole mobility of InAlGaN is about 5 times the value of AlGaN. However, this value of hole mobility for InAlGaN compared with AlGaN is seemed unreasonable.

Fig. 5.5.1 Simulation results of normalized IQE under different carrier mobility.

5.5.2 Band offset ration issue

Besides, a different band-offset ratio is also considered in our simulation. Former researches indicated the band offset ratio is between 6:4 and 7:3 for InGaN/(Al)InGaN heterojunction. [64]

For UV LED with InAlGaN barrier, after simulating with band offset ratio from 5:5 to 7:3, both the efficiency droop behavior can be elevated with higher band offset ratio, as shown in Fig. 5.5.2. Therefore, the band offset ratio from 6:4 to 7:3 is used in this simulation for introducing of indium in AlGaN. We can know that under the same energy bandgap of barrier, the band-offset ratio from 6:4 to 7:3 will lead to higher conduction-band offset and lower valence-band offset between well and barrier. This is useful for electron confinement and hole distribution in low indium content InGaN-based UV LEDs.

Fig. 5.5.2 Simulation results of normalized IQE under different band offset ratio.

5.5.3 Conclusion

Finally, we performed the numerical simulation with different parameters in band-offset ratio and carrier mobility, listed in Table 5.1. The results of the EQE droop simulation of both different structures are in good agreement with the experimental data as shown in Fig. 5.5.3.

Furthermore, we investigated the carrier distribution in our simulation to reveal the physical situation behind these results. Fig. 5.5.4 shows the calculated carrier distribution in these UV LEDs structure under a high forward current density of 100 A/cm² (1000 mA).

When we adjust the band-offset ratio and increase the carrier mobility in InGaN/InAlGaN MQWs, the carrier distribution becomes uniform. Comparing to electrons, hole distribution shows more non-uniform due to holes have larger effective mass and lower mobility. Thus, the adjustment in hole mobility and band-offset ratio can reduce the carrier leakage and increase the chance of electron-hole pair radiatively recombination.

Table 5.1 Simulation parameters in band-offset ratio and carrier mobility Band offset

Fig. 5.5.3 Normalized Efficiency curves of experimental and simulated.

Fig. 5.5.4 Distribution of (a) Electron (b) Hole concentrations, and (c) Radiative recombination rates concentrations of the LEDs with AlGaN and InAlGaN barrier under a high forward current den sity of 100 A/cm².

5.6 Summery

In summary, we have compared InGaN-based UV MQWs active region with ternary AlGaN and quaternary InAlGaN barrier layers. HRXRD and TEM measurements show the two barriers are consistent with the lattice, and smooth morphology of quaternary InAlGaN layer can be observed in AFM. Under a particular investigation, the electroluminescence results indicate that the light performance of the InGaN-based UV LEDs can be enhanced effectively when the conventional LT AlGaN barrier are replaced by the InAlGaN barrier.

Furthermore, simulation results show that InGaN-based UV LEDs with quaternary InAlGaN barrier exhibit higher radiatively recombination rate and lower efficiency droop at a high injection current. We attribute this change to a drastic improvement in the light output and efficiency droop from the higher band-offset ratio and higher carrier mobility within quantum barriers, substantially higher hole mobility leads to the superior redistribution of holes and reduction of scatterings due to better morphology in the transverse carrier transport through the InGaN/InAlGaN MQWs.