3.1.1 Sample Preparation and Structure
First step, we designed the sample structure for our demonstration, which was a combination of GaN and Al2O3 with an atomically-thin interfacial layer. A detailed structure of the sample used in this thesis is sketched in figure 3.1. It is basically a 3-nm-thick [0001]-oriented Wurtzite In0.14Ga0.86N embedded in-between a 3-mm-thick GaN buffer layer and a ~55-nm-thick GaN cap layer. This single quantum well (SQW) structure was for generation and detection of femtosecond acoustic phonon pulses, as introduced in Chapter 2. All of these three layers were grown by metal-organic chemical vapor deposition (MOCVD) on a double-polished c-plane sapphire substrate. Then, an 20.2-nm thick aluminum oxide (Al2O3) was grown on top of the GaN cap layer by a commonly-used technique called thermal atomic layer deposition (ALD)[33]. Water vapor and trimethylaluminum (TMA) were used as the two precursors for oxygen and aluminum [34]. Under a temperature of 200˚C, a pressure of 0.5torr and pulse duration of 0.012s for each cycle, the average growth rate of is about 1.07 angstrom/cycle. After the ALD process, a 1.85nm-thick interfacial layer (IL) was inevitably formed amid GaN cap layer and the deposited Al2O3 film, which was later confirmed by high-resolution transmission electron microscopy (HRTEM), as shown in figure 3.2. Similar ALD procedure of exactly same substrate and grown material (GaN and Al2O3, respectively) was also reported with an interfacial layer [11]. These top-most three layers, i.e.
GaN/IL/Al2O3, composed our model system. The sample was fabricated with the help of
Huan-Yu Shih from Prof. Miin-Jang Chen’s laboratory in the Department of Materials Science and Engineering, National Taiwan University.
Fig. 3.1 Schematic diagram of the sample structure.
Fig. 3.2 Bright-field high-resolution transmission microscopy (HRTEM) image of our
sample with different magnification, showing the existence of an interfacial layer between GaN and Al2O3. (a) 360 kx magnified (b) 930 kx magnified.
3.1.2 Setup and Measurement Details
We built up a typical one-color femtosecond pump-probe transient transmission setup at room temperature to perform our investigation, as depicted in Fig.3.3. A commercialized Kerr-lens mode-locked Ti:sapphire femtosecond laser with a repetition rate of 76 MHz and center wavelength at 800 nm was utilized as the light source for the measurement. Next, a 300-m-thick beta barium borate (BBO) crystal was applied for the second harmonic generation (SHG) process to crease frequency-doubled ultraviolet (UV) light pulse that matched the bandgap energy of InGaN SQW. Behind a UV-color
employed to divide the laser beam into an s-polarized pump beam and a p-polarized probe beam. In-between the UV-color filter and the beam splitter, we placed a half-wave plate for convenient modification of the pump/probe ratio. A high-resolution motorized linear translation stage equipped with a retro-reflector was introduced on the pump arm in an attempt to accurately control the optical path difference (OPD) between two arms, which corresponds to the time delay between pump and probe pulses. On the other arm, we set up a telescope for further improvement of the spatial overlapping of two beams and a neutral density (ND) filter wheel to ensure that the probe power was low enough not to affect the transient state pumped. The combined utilization of the half-wave plate and the ND filter wheel enabled us to individually adjust the power of pump beam and the probe beam. A pump power of 14 milliwatt (mW) and a probe power of 1.4mW were adopted during our experiment.
Fig. 3.3 The femtosecond pump-probe transient transmission measurement setup.
In the last part of the setup, two laser beams were aligned parallel before focused on the sample by an aspheric objective lens with a focal length of 0.5 cm. The position of the
sample was optimized for the condition that the overlapped portion of pump and probe light were located right at the SQW inside the sample, which efficiently fulfilled the generation and detection of femtosecond acoustic pulses, as introduced in Chapter 2.
Behind the sample, an iris was placed to block the scattering of pump light while a plane-convex lens was used to collect the diverging probe light and direct it into a high speed Si photodetector. The photodetector recorded the transmission intensity of the probe light and was put near but not at the lens’ focal point so as not to be damaged.
Moreover, to obtain a high signal-to-noise ratio (SNR) for our measurement, a linear polarizer was employed as the last component before the photodetector to eliminate any scattered s-polarized pump light.
Concerning the signal collecting method, a phase-sensitive detection was adopted with joined appliance of an acousto-optic modulator (AOM) and a RF lock-in amplifier. The AOM can modulate laser beams in much higher frequencies compared to conventional mechano-optical choppers, thus a higher SNR could be achieved due to the reduction of the pink noise (i.e. inverse-frequency noise). In our experiment, a 2MHz electrical square wave signal with 1V peak-to-peak value, 0.5V offset, and a 50% duty-cycle was produced by a function generator and sent to both AOM’s controller and the reference port of the lock-in amplifier synchronously through low noise coaxial cables. The phased-locked signals were recorded by a LabVIEW program as a function of time delay between pump pulse and probe pulse.
For resolving phonon pulses with a FWHM ~600fs in our experiment, we set the moving speed of the translation stage and the sampling rate of lock-in amplifier to be
(roundtrip) between each data point was realized, corresponding to a temporal interval
~66fs and an equivalent sampling frequency of 15THz for our femtosecond acoustic measurement. The time constant of the lock-in amplifier was selected to be 10ms, which was the only option closest to the sampling interval (~7.8ms), so as to minimize the sampling errors.
For the calibration of differentiated transmission changes (T/T), we need to measure the transmission (T) of the probe light as well. However, it will be of huge inconvenience to move the AOM back and forth between pump arm and probe arm because it will affect all the remaining system afterwards. Therefore, we used a movable mechanical chopper instead. Same transient transmission difference measurement like the use of AOM was conducted on the pump arm. For the transmission, the magnitude of the modulated probe beam was measured while the pump beam was blocked. Since we only need a ratio number, mechanical chopper had provided enough SNR for the calibration measurement.