Investigation and Characterization of Low Density InAs and InGaAs Quantum Dots

• Main Authors: Chun-Yuan Huang, 黃俊元
• Other Authors: Meng-Chyi Wu
• Format: Others
• Language: en_US
• Published: 2006
• Online Access: http://ndltd.ncl.edu.tw/handle/04497696768743422533

博士 === 國立清華大學 === 電子工程研究所 === 94 === The main propose of this dissertation is fabricating and investigating the low density self-assembled InAs/GaAs and InGaAs/GaAs quantum dots (QDs) grown by molecular beam epitaxy. The surface morphology and optical characteristics of quantum dots are investigated. Several measurement techniques have been employed: atomic force microscopy (AFM), cross-section transmission electron microscopy (TEM), photoluminescence (PL), micro-photoluminescence (m-PL), and spectral response. In principle, the AFM is used to analyze the density, shape, size and height of uncapped QDs while the cross-section TEM is for the analysis of the interrelation of embedded ones and the beneath wetting layer (WL). From the statistic results of AFM images, the uniformity of QDs can also be derived. Secondly, the general features of the QD photoluminescence, including the state-filling effect and its interplay with carrier dynamics, and the temperature effects on carrier distributions are comprehensively discussed. Furthermore, the m-PL measurement is prepared for the analysis of few even sigle QD and the ultimate single photon emission at low temperature. The spectral response is an essential measurement technique for the analysis of detection region of the infrared photodetectors as well. First of all, we investigated the InAs QDs grown by Stranski-Krastanow (SK) growth mode under different growth parameters, including the growth temperature and growth rate. In growth temperature tuning experiment, improper temperature quenching procedure can be responsible for the two groups of size distribution of QDs. The larger QDs, so called relaxed islands, are associated with the dislocations release and can greatly degrade the PL emission efficiency. Besides, the two stair-like PL intensities with increasing temperature in growth rate tuning experiment is resulted from the decrease of state-filling effect of the ground state and the thermally activated repopulation of electrons to nearby dots. In second part, we try to use different growth modes to fabricate the low density InAs and InGaAs QDs. We choose the self-assembled InAs QD growth via postgrowth annealing. The 1.5 monolayer of InAs is deposited since the dot density would significantly increase with increasing InAs layer thickness. The results exhibit the optimal annealing time and annealing temperature are 60 sec and 515 ℃, respectively. For the InGaAs QD growth, we choose the simplest method-extremely reduce the growth rate. A particular phenomenon of one-dimensional QD ordering along [110] can be gradually apparent after the growth rate is reduced beyond 0.054 mm/hr. It is never observed for InGaAs QDs from previous reports. It should be originated from the enhancement of the difference of anisotropic In adatom migration lengths by low growth rate. Finally, we fabricated the quantum dot infrared photodetectors (QDIPs) with 30 periods of InAs/GaAs QD layer and evaluated the relationship of PL and responsivity spectra. The weak PL signal from WL can be attributed to the short average distance between the QDs, as compared to the mean free path of carriers. Our QDIPs have a detection peak at 5.5 mm at the temperature ranging from 120 to 220 K. The mechanism of high-temperature operation should be from the substantial reduction of electron relaxation time when the inter-level spacing is larger than the phonon energy. Consequently, the longer electron relaxation time from the excited states to the ground state in quantum dots allows the photoexcited carriers to escape from the dot and contribute to the photocurrent before relaxing back to the ground state. On the other hand, the normalized spectral response is almost identical to the peak of shifted m-PL spectra at the same temperature, which leads us to conclude that the detection range of a QDIP can be predicted by the m-PL spectra and therefore the complicated process can be omitted.