We have studied the phase separation and ordering phenomenon of InAlAs epilayer grown by low pressure MOCVD with DCXRD, PL, TEM measurements. To understand the phase separation and ordering, each samples are grown at 565℃(A), 590℃(B), 615℃(C), 700℃(E) in the SI-InP substrate.
First, we used DCXRD to analyze the structural quality of epilayer. The value of FWHM shows that C has the smallest and then A, B in order. E shows the lattice-matching peak with substrate and FWHM really has much smaller value than DCXRD. The result of intensity shows the inverse order. Therefore, the structural quality of epilayer appears the best condition in E, C and then A, B in order. Especially, both A and C undergo the compressive stress but the structural quality of C with high lattice misfit shows better than that of A with low lattice misfit. This result indicates the other effect except dislocation influences in the epilayer. We confirm that this result is originated from the phase separation and ordering by using TEM.
Secondly, we used PL to analyze the optical quality. We observed the same result with DCXRD in the intensity and band gap shift between expected value in DCXRD and observed value in PL. The band gap shifts of A,B,C,E show 291,397,246,28 meV, respectively. These values are originated from carrier captures by phase separation and ordering in the epilayer.
All samples grown at beyond 650℃ show the streak pattern into [220] direction and the phase separation images in the [110] cross and plan view TED and TEM. The distinct difference is that while C shows the discontinuous and unclear phase boundaries, A shows the planar and clear. In the TEM image photograph, the bright region is AlAs-rich phase and the dark region InAl-rich phase. The thickness of each region in A and C is 120Å, 170Å and 140Å, 400Å, that is the wavelength of phase separation increases and InAs-rich region is thicker than AlAs-rich region as the growth temperature increases. We confirm that in the lower growth temperature the band gap reduction increases due to many phase boundaries. On the other hand, E sample grown at higher temperature(700℃) doesn't show phase separation. This result agrees with theoretical critical temperature(about 650℃) on the disappearance of phase separation.
We observed ordering in [110] cross section TED of both A and C. The intensity of ordered spot shows the strong extra spot in the 615℃ comparison with 565℃(A). Therefore, the ordering effect is strong in the medium growth temperature and in the lower and higher temperature show weak. With all TEM and PL results we can know that because band gap reduction is large in the lower growth temperature, the most of band gap shift value is generated from phase separation.
We have studied experiment to control band gap with thermal treatment of phase separation and ordering. The result shows the increase of band gap of 77meV by 880℃, 3 min annealing in the P overpressure atmosphere. We observed with TEM that the ordered structure disappeared through short time and high temperature thermal annealing but phase separation doesn't change. Because ordering has wavelength of a few Å and phase separation wavelength of hundreds Å, ordering first disappears by thermal annealing. We can know that band gap reduction by ordering is about 80meV in annealing experiment and the three fourth of reduction is generated from phase separation.
When the thermal annealing of 1 hour is performed at 650℃ with $ZnAs_2$ for surface protection and easy atomic movement, phase separation completely disappears. However, a lot of stacking faults are generated at the interface due to Zn atoms.
