Photoluminescence (PL) is a non-destructive optical technique for semiconductor characterisation. In this process, a substance absorbs high-energy photons and then re-radiates photons. Quantum-mechanically, this can be described by the excitation of carriers to a high-energy state and then their return to a low-energy state accompanied by photon emission. PL in semiconductors is dependent on the carrier lifetimes and recombination through different mechanisms. For a bulk material with a bandgap energyEg, the pump light must have a higher energy such thathν > Eg; where ν is the pump frequency andhis the Planck constant. In this case, absorption can generate electron-hole pairs, which will eventually recombine through one of three main processes. In the case of radiative recombination, the
carriers return to their ground states with the emission of photons, and these can be detected as PL signal. On the other hand, there are non-radiative recombination mechanisms that do not yield a PL signal. In Shockley-Read-Hall recombination, the carriers recombine through deep-level traps on the surface or in the bulk, and the energy is dissipated as phonons. Alternatively, the energy liberated during recombination may be directly absorbed by another carrier and this non-radiative process is named Auger recombination [23]. The latter is a property of the material itself and dominates at high carrier densities. Shockley-Read-Hall recombination, however, is limited by material quality and purity, and can be significant even for low carrier densities. In order to maximise the sensitivity of PL measurements, these non-radiative mechanisms should be minimised.
The energy of the photon emitted during radiative recombination can vary depending on the specific transition process. In the most simple case, band-to-band recombination will emit light with energy EPL=Eg and this can be significant at room temperature. Alternatively, excitonic recombination will yield PL with a lower energy, EPL = Eg − Ex, where Ex is the excitonic binding energy. Free excitons can occur in very pure materials, whereas bound excitons dominate in the presence of impurities. Impurities can also lead to donor-acceptor recombination, where the electrostatic interaction can influence the PL wavelengths, and EPL = Eg − Ea −Ed +q2/(s0r). In this expression, q is the carrier charge, s is the dielectric constant of the semiconductor, 0 is the permittivity of free space, r is the distance between the impurity atoms and Ea and Ed are the energies of the acceptor and donor levels, respectively [23]. So the wavelength of the emitted light can provide information on the quality and composition of bulk semiconductors, including compound semiconductors where Eg varies with stoichiometry.
In addition, quantum wells (QWs) and quantum dots (QDs) have discrete energy levels that vary with the composition and dimensions. In general, PL emission from QWs will have energy EPL=Eg+Ec+Eh, where Eg is the bandgap energy of the QW material,Ec is the ground-state energy of an electron in the conduction band and Eh is the ground-state energy of a heavy hole. The latter parameters are particularly sensitive to the QW width and composition, so PL measurements across a two-inch wafer can detect non-uniformities in the MOCVD growth of these heterostructures. Whereas QWs often vary monotonically towards the edge of a wafer due to the boundary effect during growth, PL indicates that QD size and composition is less predictable. The intermixing processes discussed in chapters 7
and8lead to diffused interfaces and hence modified potential profiles. Interdiffusion occurs on the atomic scale and is difficult to measure directly, however PL can easily
Figure 2.10: The optical setup used for photoluminescence measurements, consisting of a solid-state laser (SSL), adjustable neutral-density filter (ND), optical modulator (OM), band-pass (BP) and long-wavelength pass (LP) filters, mirrors (M), focussing (FL) and collimating (CL) lenses, monochromator (MC), InGaAs detector (D), lock-in amplifier (SR830) and personal computer (PC). The sample (S) was usually mounted in a liquid- nitrogen-cooled cryostat (LNC) for 77 K measurements.
detect the change in energy levels. PL is an invaluable tool for characterising these systems, particularly given its benign nature.
The optical setup used for PL measurements in this research is shown in figure2.10. A 532 nm frequency-doubled solid-state laser was used as the excitation source. A bandpass filter (350–650 nm) ensured that the infrared light was excluded and the signal was modulated by a chopper rotating at 330 Hz. Two mirrors were used for alignment and a lens was used to focus an intense spot on the sample. The emitted light was collected by a collimating lens, and another lens focussed the PL into a 50 cm spectrometer. A long-wavelength pass filter (> 620 nm) was placed in front of the entrance slit to remove any scattered laser light. The collected PL was dispersed in a Princeton Instruments SpectraPro 2500i monochromator. For most measurements, a blazed grating with 1200 grooves/mm and a 1400 nm cutoff wavelength was selected. The entrance and exit slit widths were usually set to 1 mm, as a compromise between the signal intensity and spectral resolution. The detector was a 2 mm Electro-Optical Systems InGaAs photodiode that was thermo-electrically cooled. The signal was fed into a Stanford Research Systems SR830 DSP lock-in amplifier and the chopper signal was used as the reference. PL is often performed at a low temperature, in order to decrease thermally- activated non-radiative recombination and to minimise thermal broadening of emission line-widths. Although the as-grown QD samples generally emitted a
strong PL signal at room temperature, all of the PL measurements presented in this thesis were measured from samples at 77 K. The samples were mounted in a portable liquid-nitrogen-cooled cryostat, which was evacuated to about 3×10−5
Torr before cooling. The signal gain and integration times were optimised for each sample, depending on the intensity of the PL.