The InGaAs/InGaAlAs MQW wafers used in this study were grown in a Riber 21T MBE system. Fig. A.1(a) and 1(b) show a schematic cross section of the epitaxial layer structure (samples 50d and C071). The wafers were designed for the applications in PICs including semiconductor optical amplifiers and ring resonators. The experimental procedure involves placing a small piece of the wafer into an inductively-coupled-plasma reactive ion etching (ICP-RIE) system for Ar+ plasma bombardment. The experimental conditions include an Ar flow rate of 100 sccm, a working pressure of 80 mTorr, an RF power of 480 watt, and an ICP power of 250 watt. Three different time durations of 5, 10 and 14 minutes are used. To measure the sputtering rate, the sample 50d is first coated with a 250-nm thick layer of SiO2 by plasma-enhanced chemical vapor deposition (PECVD). A photoresist pattern consisting of multiple 3-μm-wide stripes is then printed by photolithography. The stripe pattern is transferred onto the SiO2 layer through CF4 reactive ion etching. The wafer is then cut into many small pieces for Ar+ plasma bombardment. After the ICP-RIE process, the stripes are removed by dilute HF and the sputter depth is measured by a surface profiler.
The sputter rate of the sample is found to be 6.5 nm/min. The maximum sputtering time of 14 minutes is chosen to avoid further bombardment on the InAlAs layer, which can cause serious damage on the surface.
After Ar+ plasma exposure, the wafers are annealed in a RTA chamber at a temperature of 620°C for 120 sec in 15%H2/85%N2 gas at a flow rate of 100 sccm. The samples are sandwiched between two fresh pieces of GaAs proximity caps [A.2, A.3, A.5-A.9]. After the completion of the above procedure, PL spectra of the wafers are measured.
For both the treated and untreated samples, the samples are etched repeatedly in many steps and the PL measurement is performed after each etching step. The maximum PL signals of different samples obtained during this etch down procedure are used for comparison.
A.2.1 Experimental Results
Figure A.2 shows room-temperature and post-anneal PL spectra of as-grown sample 50d and samples bombarded for 5, 10, and 14 minutes. The numbers in parentheses refer to the peak wavelength (left) and peak intensity (right) for the PL signal. The results show that the PL signal is strongly enhanced by more than one order of magnitude by the Ar-plasma bombardment and annealing procedure, and the blue shift of the wavelength is 15 nm.
1300 1350 1400 1450 1500 1550 1600 1650 1700 0.000 exposure times with Ar gas at 100 sccm flow rate, 80-mtorr working pressure, 480
watt RF power and 250 watt ICP power, and followed by RTA at a temperature of 6200C for 120 sec in 15%H2/85%N2 gas at 100 sccm flow rate. The numbers in parentheses refer to the peak wavelength (left) and peak intensity (right) for the PL signal. The PL signal is enhanced by more than one order of magnitude with a maximum blue shift of 15 nm.
Similar results are obtained when the same experimental procedure is repeated by using sample C071 as shown in Figure A.3. It is thus quite evident that Ar ion bombardment plays an importance role for PL signal enhancement. Comparing the data in Figs. A.2 and A.3, the PL intensity shows monotonically increase against the Ar-plasma bombardment time for
most cases. However, the PL result for sample 50d of 10 mins bombardment shows weaker intensity than the case of 5 mins bombardment. The reason is related to the sputtering-exposed InAlAs surface and the etch-down procedure for PL measurement, which may damage the surface and reduce the PL intensity.
1350 1400 1450 1500 1550 1600 1650 1700 -0.005
(1470, 38.1x10-3) At room temperature
PL I n te nsity( a.u .)
Fig. A.3 PL spectra at room temperature of sample C071 for different Ar+ plasma exposure times with Ar gas at 100 sccm flow rate, 90 mTorr working pressure, 550 watt RF power and 550 watt ICP power, and followed by RTA at a temperature of 6200C for 120 sec in 15%H2/85%N2 gas at 100 sccm flow rate. The numbers in parentheses refer to the peak wavelength (left) and peak intensity (right) for the PL signal.
The PL signal is enhanced by up to 23 times with a maximum blue shift of 15 nm.
The effects of different Ar pressures (80 and 90 mTorr) and different RF powers (480 and 550 watt) are investigated by using sample C071. Figure A.4(a) and 4(b) show respectively the dependence of the PL intensity and the peak wavelength on the annealing time (5, 10, 14 min.) for three different combinations of Ar pressure and RF power. All of the four cases use the same 100 sccm Ar flow rate, 250 watt ICP power and annealing conditions at a temperature of 6200C for 120 sec in15%H2/85%N2 gas at 100 sccm flow rate. Figure A.4(a) shows that the PL intensity generally increases with the time of exposure to the Ar
plasma with the strongest PL enhancement observed at the higher Ar pressure (90 mTorr) and the higher RF power (550 watt). Figure A.4(b) shows that as the exposure time to Ar plasma is increased, the PL peak wavelength shows an initial red shift of 10 to 15 nm and then settles down to a constant red shift of 10 nm except for the 90 mTorr, 550 watt case which shows an eventual blue shift of 15 nm.
0 2 4 6 8 1 0 1 2 1 4
Fig. A.4 Experimental results for sample C071 showing the variations of (a) the relative room-temperature PL intensity, and (b) the PL peak wavelength. Both are plotted as functions of the Ar plasma exposure time, working pressure and the RF power.
Other common experimental parameters include: Ar flow rate of 100 sccm, ICP power of 250 watt and RTA at 6200C for 120 sec in 15%H2/85%N2 gas at 100 sccm flow rate.
From the experimental conditions, the Ar+ plasma bombardment presumably creates crystal vacancies near the sample surface. A depth of 55-75 nm was estimated for the surface damage by similar conditions [A.1]. When sufficient number of these vacancies diffuses through the quantum wells during the RTA cycle, intermixing of materials in the wells and the barriers can occur, and results in blue shift of the PL peak wavelength [A.2-A.4]. The present results indicate that the surface vacancies are located much more than 1 μm away from the MQWs. The RTA condition is not sufficient to activate the In-Ga-vacancy disordering for the QWI effect [A.10], while the described procedure is sufficient to facilitate the out-diffusion of non-radiative recombination centers and significantly increase the PL intensity for most cases. Even in the only case where a moderate amount of blue shift is observed, the shift is limited to a maximum of 15 nm. On the other hand, the red-shift of 15 nm is attributed to the modification of the quantum well shape and composition by the In-Ga inter-diffusion between the quantum wells and barriers.