Figure 3-1 shows the PL spectra of ZnTe/ZnMnSe QD structures with different coverage
of
e-resolved photoluminescence (PL). Conventional PL was used to detect the ground state transition of the QD samples. To study the type-II band alignments, we also measured the excitation power dependence of spatially indirect transitions in ZnTe/ZnMnSe structures. The difference between two-dimensional (2D) layer structure and zero-dimensional (0D) QD structure is discussed. Time-resolved spectroscopy was used to understand the carrier recombination processes and also verify the dimensional changes of ZnTe/ZnMnSe structures with increasing coverage. Furthermore, circular polarization spectroscopy without an applied magnetic field was utilized to study the spin polarization.
thicknesses of ZnTe layer at low temperature. All of the spectra consist of three emission bands. The sharp emission peak near 2.8 eV is attributed to the near band edge transition of ZnMnSe layer. The other two emission bands observed in the PL spectra from 1.8 to 2.6 eV are associated with the ZnTe layer. The lower-energy emission (labeled by QD in Figure 3-1), which is more sensitive to the change of coverage thickness, is due to the recombination in QD or 2D layer before the QD formation. The peak energies of QD emission bands for 1.8, 2.2, 2.4, 2.7, and 3.0 ML coverage are 2.175, 2.005, 1.963, 1.917, and 1.888eV, respectively. In contrast to type-I semiconductor structures, the energy minima for electrons and holes lie in different layers in type-II structures. In Fig. 3-2, the band alignment of ZnTe/ZnMnSe QD is shown. We assume the band alignment of ZnTe/ZnMnSe QD with low Mn concentration is similar to that of ZnTe/ZnSe QD, which was carefully determined by
the AFM and PL study [29]. This spatially separated electrons and holes will greatly influence optical properties. All of the emission peak energies given above are smaller than that of the band gap energy of ZnTe epilayer. This implies a type-II band alignment for the ZnTe QDs grown in ZnMnSe matrix. The higher-energy PL band (denoted as H in Figure 3-1) near 2.3 eV is much broad and the intensity is smaller than that of the QD PL band. The H emission is ascribed to the emission from the ZnTe/ZnMnSe interface [30].
Figure 3-3 summaries the peak energy shifts of the lower-energy emission for ZnTe
xcitation power dependence of spatially indirect transitions of ZnTe 2D layer and 0D q
/ZnMnSe structures with different coverage thicknesses of ZnTe layer. The red-shift in energy with increasing ZnTe coverage is attributed to the decrease in the quantum confinement of the holes in the ZnTe QDs. Similar to the ZnMnTe /ZnSe QD structure grown in our laboratory [31], the change in red-shift energy can be roughly characterized by two different slopes, as observed in the Figure 3-3. The peak energy decreases promptly with increasing ZnTe coverage when the ZnTe coverage is less than 2.4 ML. However, when the coverage exceeds 2.7 ML, the decrease of peak energy becomes smooth. The critical thickness of the two slopes is between 2.4 and 2.7 ML. For the ZnTe coverage with thickness thinner than the critical thickness, the PL emission origins from the transitions in 2D layer, while 0D QD emission is expected for samples with thicker ZnTe coverage thickness. This result was also confirmed by the study of RHEED patterns. Figure 3-4(a), (b), and (c) show the RHEED patterns of samples with 1.8, 2.4, and 2.7 ML ZnTe coverage, respectively. The streaky RHEED pattern of 1.8 ML results from the two-dimensional layer. As the ZnTe coverage increases to 2.4 ML, the RHEED pattern starts to transfer from streaky to spotty.
This indicates a growth transition from a two-dimensional layer growth to a three-dimensional island growth and implies the formation of the ZnTe QDs at ZnTe coverage thickness above 2.4 ML.
The e
uantum dot were investigated. In Fig. 3-5, we show the relation between the PL peak
energy and the excitation power of three samples with coverage of 2.2 ML, 2.4 ML and 2.7 ML. When the excitation power was increased, blue-shifts for three samples were observed.
The lines fit the blue-shifts of three samples using the power function described by Y=a+bxs. The powers (s) for samples of coverage with 2.2 ML, 2.4ML, and 2.7 ML are 0.32 (dashed line), 0.3 (dot line), and 0.1(solid line), respectively. The cubic root (s=0.33) power dependence of PL peak shift caused by the band-bending effect is characteristic of type-II quantum wells and has been reported theoretically [32, 33]. The band-bending effect is caused by the spatial separation of holes confined in the ZnTe layer and the hole-attracted electrons localized in the nearby ZnMnSe regions. With the increasing excitation power, the strong band-bending effect (red-shift) accompanying with the carrier filling effect (blue-shift) results in a theoretical-predicted blue-shift of PL peak with the 1/3 of the excitation power [32, 33]. It corroborates the previous result that the 2.2 ML and 2.4 ML samples are 2D layer structures.
However, in the case of 2.7 ML QD structure, only a few holes are allowed to be confined in each QD. As a result, the band-bending effect does not play important role and the blue-shift of PL peak deviates from the cubic root. The carrier distribution effect is a possible mechanism responsible for the blue-shift of 0D QD sample. At the beginning, the carriers are excited to the barrier by the He-Cd laser of the wavelength with 325 nm and then quickly transfer to the QD. The transfer behavior is a selection effect which decides the carriers to transfer to the lower energy states (larger QDs) first. As the excitation power increases, the carriers will gradually fill in the smaller QDs. This will result in a blue-shift with increasing power intensity. Fig. 3-6(a) shows a simple schematic diagram for the carrier filling distribution of different QD sizes as excitation power increases. The estimated power-dependent blue-shift is shown in Fig. 3-6(b). The power of the estimated blue-shift for QD size filling effect is about 0.11 and is close to the experimental result of QD sample with coverage of 2.7 ML. It indicates the excitation power-dependent PL is greatly affected by the QD size distribution.