Semiconductor quantum dots (QDs) were originally predicted to provide high gain, improved temperature performance and high-frequency modulation characteristics. In the last few years, several demonstrations of QD lasers based on self-assembled growth in different material systems have pointed out the possibility of achieving very low threshold current densities and extending the wavelength range, but also the limitations due to low modal gain. Overall, it is tempting to describe QDs as an active material with a low and spectrally-broad density of states. However, the energy spectrum of each dot consists of a ladder of discrete energy states, and although this discreteness is masked by the inhomogeneous broadening in macroscopic devices, it strongly affects the device behaviour. In this presentation some of the consequences of the quantum nature of QD active regions will be addressed: On one hand, the effect of finite interlevel relaxation time on laser operation will be described. On the other hand, I will discuss the progress towards the realisation of devices incorporating single QDs in the active region, where emission can be controlled at the single photon level.
In-plane lasers incorporating 2 or 3 stacks of self-assembled InAs QDs grown on GaAs (areal density 3e10 cm-2) typically show ground state (GS) lasing around 1300 nm for cavity lengths >1.5 mm and excited state (ES) lasing (around 1200 nm) for shorter lengths, due to the higher modal gain available on the ES. We have recently found that GS and ES lasing can coexist when a device 1.5-2 mm-long is biased well above threshold. Lasing starts on the GS but a second lasing line appears on the ES as the bias is increased. The two lasing states compete for the same carriers, so that saturation of the GS power is observed at the ES threshold. The wavelength-resolved light-current characteristics are fully understood by a simple rate-equation model by assuming a intradot (GS-ES) relaxation time of 7 ps, close to the experimental 10 ps photoluminescence rise time. In fact, as bias is increased above threshold, carriers must be injected to the GS at a higher rate, which implies an increasing ES population. In short cavities, where GS lasing occurs with a nearly-saturated GS, this "relaxation bottleneck" effect is exacerbated by the very limited number of available states, so that the increase in ES population leads to ES lasing. The incomplete clamping of the ES population also affects the modulation characteristics of QD lasers, particularly in terms of the linewidth enhancement factor, which is seen to increase up to a value of a=10 as the bias is increased well above threshold, due to the increasingly asymmetric gain spectrum.
Carrier localisation is another intrinsic characteristic of QD active regions. We have fabricated ultrasmall (submicrometer) QD light-emitting diodes (LEDs). From the scaling behaviour of the current-voltage curves for devices with different areas, we deduce that carrier diffusion is largely suppressed in QD LEDs (diffusion length < 100 nm at room temperature), as opposed to QW LEDs (diffusion length 2.7 um). This shows the possibility of isolating single QDs in the active region of an LEDs, in order to achieve an electrically-pumped single-photon emitter. Recent results on integrating a high-Q 3D optical microcavity with the nanoscale current injection will also be described.
|