Effects of thermal annealing on the emission properties

Figure 2.7(a) shows the PL spectra for the as-grown and the annealed QD samples measured at 10 K under a low excitation power of 0.5 mW. The PL spectrum for the type-I InAs/GaAs QDs is also shown for comparison. With the increasing TA, a blueshift and a

22

narrowing of the QD’s emission peak were observed [62-67]. The energy blueshift is caused by the alloy intermixing between the QDs and barrier materials, which shallows the QD’s confining potential due to the incorporation of more Ga atoms into the InAs QDs [62-67]. For the as-grown sample, it has been confirmed in our previous study (Sec. 1 in this chapter) that the InAs/GaAs_0.84Sb_0.16 QDs exhibited a type-II band lineup, where the radiative recombination occurs between the electrons confined in the QDs and the holes in the GaAsSb layer. Annealing induced alloy intermixing also tends to smooth the VB discontinuity at the InAs–GaAsSb interface. Therefore, the electron-hole wave function overlaps and hence the radiative recombination rate is expected to be enhanced or even changed gradually to type-I transitions after thermal annealing.

To understand the effects of thermal annealing on the recombination dynamics in the type-II QDs, TRPL measurements have been performed and the results are shown in Fig.

2.7(b). The PL decay time was found to decrease with the increasing T_A, indicative of a more penetrated hole wavefunction into the InAs QDs due to the reduced VB offset caused by alloy intermixing at the InAs–GaAsSb interface. In particular, we found that the decay transient for the 900 °C annealed QDs became as fast as the InAs/GaAs type-I QDs. This implies that the QD structure has been changed to a type-I band alignment after high-temperature annealing.

In principle, the recombination lifetime can be quantitatively deduced from the measured PL decay transients. However, the determination of recombination lifetime is not straightforward for a type-II system. One complication arises from the effect of nonequilibrium carriers: as the carriers recombine continuously, the PL shifts to lower energies due to the reduced VB bending surrounding the QDs. Therefore, the decay transient recorded at a given wavelength does not reflect the true lifetime since the PL decay arises not only from the carrier recombination but also from the temporal PL shift. For the InAs/GaAsSb material system, the carrier dynamics is further complicated by the presence of

23

Fig. 2.8: Energy dependent carrier lifetimes and time evolution of the PL spectra for the samples annealed at (a) 700 °C and at (b) 800 °C.

localized hole states in the GaAsSb layer due to alloy fluctuations and/or Sb clustering. The redistribution of holes among these localized states also significantly influences the overall carrier dynamics, particularly at low temperatures. In order to clarify the role of both effects (i.e., the nonequilibrium carriers and the localized states) in the PL transients, energy dependent TRPL has been performed. The results for the 700 and 800 °C annealed QDs are shown in Figs. 2.8(a) and 2.8(b), respectively. For the 700 °C annealed QDs, a clear temporal PL redshift can be observed. Since the excitation power was kept low, the effect of

24

nonequilibrium carriers is expected to be less significant. In fact, an even more pronounced temporal PL shift was observed for the as-grown sample. Thus we ascribe the temporal PL redshift to the effect of hole localizations in the GaAsSb layer. On the contrary, the temporal PL redshift was absent for those QDs annealed at TA≥800 °C. This indicates that the localized hole states have been removed by the annealing induced alloy intermixing. The temporal PL redshift is closely related to the nonsingle exponential decay observed in PL transients shown in Fig. 2.7(b). Investigations of all samples revealed that the temporal PL redshift can be observed only when the decay transient is nonsingle exponential. By using a double exponential function ( ) to fit the PL decay recorded at each wavelength, the decay time constants and for the faster and the slower decay components can be obtained. The fitted results are shown in the upper panel of Fig. 2.8, along with its time integrated PL spectrum. For the 700 °C annealed QDs, both and show strong energy dependences, indicative of hole transfers among localized states in the GaAsSb layer. For the 800 °C annealed QDs, we found that is almost unchanged across the ground state emission band.

25

Fig. 2.9: The schematic illustration of hole redistribution among fluctuant potentials and the corresponding energy dependent lifetime.

The energy dependent lifetime can be described by

( )

, where is the radiative lifetime, E0 describes the localization depth, and Eme is the energy similar to the mobility edge [68]. The schematic illustration of hole redistribution among fluctuant potentials and the corresponding energy dependent lifetime are shown in Fig.

2.9. As shown in the figure, carriers in shallower potentials are with higher probability of transfers, thus, resulting in shorter lifetimes. On the other hand, carriers in deeper potentials tend to recombine and the measured lifetimes are closer to radiative lifetimes.

26

Fig. 2.10: Radiative recombination rate as a function of the annealing temperature. The data from the as-grown sample are plotted at 500 °C.

Figure 2.10 shows the radiative recombination rate ( ) deduced from the energy dependent lifetime for all annealed samples. The measured increases slightly from 0.095 ns⁻¹ for the as-grown sample to 0.15 ns⁻¹ after annealing at T_A=800 °C, but still about an order of magnitude lower than the InAs/GaAs type-I QDs (1.25 ns⁻¹). The low recombination rates indicate that their band lineups remain type-II, with an electron hole wave function overlap of only ~28%–35% of the type-I QDs. As TA was increased to 900 °C, the recombination rate increases dramatically to =1.2 ns⁻¹, which is very close to that of the type-I QDs, and confirms the band alignment change after annealing.

27