Essential calibration: Post-bake optical test structure

Faebian Bastiman

After a bake out, thorough cell outgas and growth rate calibration the material quality of the MBE system must be determined and assessed. On a III-As system you would want to assess the quality of GaAs and InAs and/or InGaAs and AlGaAs. A 6nm GaAs QW with RT PL of 820 nm is a good optical test layer since it is easy to differentiate between the substrate and the well. An In15Ga85As QW has an RT PL of ~950nm and is sufficiently longer in wavelength that it does not interfere with the GaAs peaks. 2.5 ML (i.e a total of 15.645 In atoms/nm2 on GaAs) InAs QDs can be tailored to give a variety of wavelengths. If grown at ~490-500°C with a growth rate of 0.017 ML/s (i.e. 0.106 In atoms/nm2/s) the RT PL wavelength should be around 1250nm. In fact, these 3 wavelengths are sufficiently distant such that with a little AlGaAs confinement you can put them all in one structure. An example structure is shown in figure 1 below:

post bake jpg 2

The total growth time is 80 minutes including temperature ramps and As flux changes. The GaAs should  be grown around 570°C, the InGaAs around 540°C and the InAs around 500°C. Once grown the sample surface should have a slightly purple hue and of course should be shiny. The RT PL should be taken from 800 – 1300 nm and should resemble figure 2 below. Remember this is the first sample after bake out, but after a thorough cell outgas the quality should already be very good.

 otc PL

Figure 2

The peak at ~1250nm is the InAs QDs and the peak at 955 is the InGaAs QWs. Both of these are excellent intensities, around 33% of the best values obtained at these wavelengths on this system. This is an indirect indication that the Al40Ga60As is good quality, since poor AlGaAs cladding around the QDs would introduce non-radiative recombination centres. The 3 x InGaAs QW are expected to be brighter than the single InAs QD layer. But  what has happened to the GaAs QWs?

Well the insert shows the scan limited from 750 to 885nm, and the GaAs wells are there. The structure is excited with a 650nm diode laser, which excites carriers and stimulates radiative recombination in all structures. The 820nm GaAs QW PL light must first travel up through the structure and escape before it can be detected. Since it is shorter in wavelength (higher in energy) than the InGaAs and InAs above it, a fraction is reabsorbed. In this case that fraction is around 99%. Thus the GaAs does not seem to emit at all compared to the InGaAs. Hence to actually achieve comparable intensities from all the active layers the InGaAs QW must be reduced from 3 to 1, or the GaAs QW must be placed above the InGaAs in the structure.