Organization of this Dissertation - 自組式砷化銦量子點/環及其紅外線偵測器之研究

Chapter 1 Introduction

1.3 Organization of this Dissertation

In this dissertation, we try to get a comprehensive understanding of the growth control and electronic properties of self-assembled InAs 0-D quantum structures, and then put it into practice to realize LWIR QSIPs with improved performances and thus higher operating temperature. The experimental techniques used are described in chapter 2, including MBE growth, material characterization methods, device processing procedures, and QSIPs measurement techniques.

In chapter 3, the growth control on the self-assembled InAs quantum dots (QDs) and their geometric tailoring into quantum rings (QRs) are studied. Then, the influence of the geometric change during dot-to-ring evolution on electronic wavefunctions and state energies is discussed. Finally, the growth condition of high density and high quality QRs which are good for device applications is presented.

In chapter 4, the energy spectra of self-assembled InAs QDs are investigated using the

selective excitation photoluminescence technique. Three distinct regions with different mechanisms in carrier excitation and relaxation are identified in the emission spectra excited with different photon energies. The special joint density of state tail of the QD that extends from the wetting layer bandedge facilitates carrier relaxation and is posited to explain these spectral results.

In chapter 5, InAs QDs with different confinement barrier schemes are used in quantum-dot infrared photodetectors (QDIPs) for the detection wavelength and polarization absorption tuning. A novel confinement-enhanced dots-in-a-well (CE-DWELL) structure is shown to possess both the LWIR detection and the high normal-incident absorption capabilities.

Most importantly, the device quantum efficiency is greatly enhanced with this new design.

In chapter 6, we perform detailed studies on the InAs/AlGaAs/In0.15Ga0.85As CE-DWELL QDIPs. It is found the thickness and Al content of the AlGaAs insertion layers influence much not only the absorption property but also the transport property of the device.

With appropriate device parameters of CE-DWELL, we present LWIR QDIPs with operation temperatures over 200K.

In chapter 7, the infrared photodetectors using InAs QRs as the absorption media are investigated. Compared with QDIPs, quantum ring infrared photodetectors (QRIPs) show wider photocurrent spectra, more stable responsivity with temperature change, and lower dark current activation energy due to the shallower confined states of electrons. With an Al0.27Ga0.73As current blocking layer, the operating temperature of QRIPs is greatly elevated.

Finally, conclusions and a plan for future work are given in chapter 8.

Chapter 2 Experimental Techniques

The samples investigated in this dissertation were all grown by the molecular beam epitaxy (MBE) technology. After growth, the material characteristics of the samples were examined routinely by the atomic force microscopy (AFM) as well as the photoluminescence (PL) spectroscopy. Besides, the transmission electron microscopy (TEM) analysis was also performed for certain samples. For the samples aimed for infrared detectors, standard processing techniques including lithography, etching and metalization were used for the device fabrication. Measurements were then done to characterize the device performance, including I-V characteristics, responsivity spectra, and noise level. In this chapter, all of the above experimental techniques are briefly described.

2.1 Molecular Beam Epitaxy Growth

The epitaxy of nanometer scaled heterostructures is of increasing importance for both the physical studies and the quantum device associated applications. To fulfill this task, precise control over the epi-layer thickness as well as its composition including doping profile is essential. The molecular beam epitaxy (MBE) technique, which allows the epitaxial growth with atomic dimensional precision down to a few angstroms, emerges as the most promising approach. The MBE is a physical deposition process which takes place in an ultra-high vacuum (UHV) environment. The epitaxial growth proceeds via the reaction of molecular beams of the constituent elements with a crystalline substrate surface held at a suitable substrate temperature under UHV conditions. The high vacuum environment is the key leading to the advantages of the MBE growth. In the UHV environment, the mean free path of the molecules evaporated from each effusion cell is longer than the distance between the sources and the substrate, so the molecules strike directly on the substrate without any scattering event. For that reason, with the

fast switching action of the shutter in front of each cell, the amount of respective source molecules sticking onto the substrate can be precisely controlled by time and the cell temperature. Precise layer structures with abrupt interfaces within the atomic-layer thickness can therefore be achieved. The UHV environment also limits the amount of impurities inside the chamber so that the quality of the epitaxy films can be preserved. Moreover, under such high vacuum conditions, many surface analysis techniques can be performed in situ for the sample right after or even during the growth process [2.1].

2.1-1 MBE Apparatuses and Pre-growth Procedure

The MBE system in our lab is the Varian (or Veeco) Gen II (solid-source based) system, one of the most commonly used experimental commercial systems. It includes two individual epitaxy units (named LM and RN respectively) that are linked by an extension chamber. The two epitaxy units have basically the same configuration. Fig. 2.1 sketches the configuration for the LM system. The whole system comprises three chambers: the entry/exit chamber, the buffer chamber, and the growth chamber. The pumping machines are basically oil free. A turbo-molecular pump cascaded by a scroll pump (by a diaphragm pump for the RN system) is used for rough pumping. In the growth chamber, a cryopump and an ion pump are used to maintain high vacuum environment and a titanium-sublimation pump (TSP) is also equipped and used when the vacuum level is not good. In addition, the liquid-nitrogen-cooled cryopanel is installed within the growth chamber, surrounding the substrate manipulator and each cell, to improve the vacuum level during sample growth and also eliminate the cells from thermal cross-talk. For the buffer chamber, an ion pump and a TSP are used to keep the vacuum level.

In the entry/exit chamber, a smaller cryopump is used.

Eight effusion cells are installed in the LM chamber. Two gallium cells, one aluminum cell and one indium cell are used for group III sources. One arsenic Kunsden cell and one arsenic valved cracker cell are used for group V sources. The other two cells are charged with

Fig. 2.1 Sketch of the Varian Gen II MBE system in our lab.

silicon and beryllium as n-type and p-type dopant sources. Thus, the LM chamber supplies the growth of arsenide based III-V materials. For example, AlGaAs, InGaAs on GaAs substrates;

InAlAs, InGaAs, and InGaAlAs on InP substrates. For the RN chamber, there exist ten ports for source cell installation. Thus, besides the eight cells stated above, one antimony valved-cracker cell is also used for another group V source. So the RN chamber provides the opportunity for growing antimony related III-V materials on the substrates with larger lattice constant, such as InAs, GaSb, etc.

Also equipped in the growth chamber are two analysis instruments: the residue gas analyser (RGA) and the reflection high-energy electron diffraction (RHEED) monitor. The RGA is used to analyze the residue gas in the chamber and thereby help us to understand the cleanness in the chamber. By setting the detection range to only for the atomic mass unit (AMU) of 4, i.e. only monitoring the helium element, the RGA also serves as a very sensitive leak

detector. The RHEED is a powerful surface analysis instrument and is commonly used in the MBE system. More description about it will be given later.

To maintain the cleanness of the growth chamber and ensure the quality of grown samples, a standard treatment procedure for wafers before growth is necessary. Epi-ready GaAs wafers are mounted on molybdenum holders with springs and then loaded on a wafer trolley.

After loading the trolley into the entry/exit chamber, the whole chamber is baked at 200 ^oC to remove the residual water molecules on the wafers. When the pressure is lower than 1x10^-8torr, the trolley is transferred into the buffer chamber, in which the standby pressure is typically 1-5x10^-10torr. In this chamber, each wafer is put on a heated station and then heated up to 400

oC for further removing other contamination such as organic species. After about three hours, when the pressure in the buffer chamber is definitely lower than that in the growth chamber (stand by at around 3x10^-9torr), the wafer is ready for epitaxy growth and can be transferred into the growth chamber. It should be noted that, before the cells are warmed up for sample growth, the cryopanel is fed with continuous-flowed liquid nitrogen to further lower the pressure of the growth chamber, typically, from ~3x10^-9torr to less than 2x10^-10torr. Just before growth, the thin native oxide on the wafer surface is decomposed and desorbed by heating under enough arsenic flux. The desorption temperature depends on wafer materials and is about 620 ^oC for GaAs wafers. The temperature of the wafer is read by a pyrometer, which faces the heated viewport in front of the growth chamber (Fig. 2.1) and detects the thermal radiation from the wafer.

2.1-2 In Situ Surface Diagnosis and In Situ Growth Rate Calibration by Reflection High Energy Electron Diffraction

The situation of desorption can be easily traced with the RHEED instrument in the growth chamber. As mentioned previously, the RHEED is a powerful surface analysis

technique. It allows us to in situ monitor the surface construction of wafer or epi-layers. By using high-energy (about 10 keV) incident electron beam with a very small glancing angle (about 1-2^o), the structure of the outermost layers of atoms of the sample can be analyzed with the diffraction patterns. When the RHEED pattern changes from an obscure pattern to a bright streaky pattern, it is suggested that the wafer surface becomes clean enough and desorption process is completed. After desorption, the epitaxy growth can be started. In principle, higher temperature can improve the quality of epi-layers, but the sticking coefficient of molecules decreases, especially for the species with smaller chemical bonding energy. In our system, the growth temperature for AlGaAs and GaAs is roughly at 600 ^oC, and for the pseudomorphic growth of InGaAs the temperature is at about 510^oC. While for the strain-induced three-dimensional (3-D) island growth of InAs, the so-called self-assembled quantum dot (QD) growth, the temperature is generally in the range between 480^oC and 520^oC, depending on the desired QD condition. Such 3-D island growth can be also in situ identified by use of the RHEED technique. Specifically, if the epitaxy growth goes in a 2-D layer-by-layer mode, a streaky RHEED pattern will be obtained in the meantime (see Fig. 2.2(a)). Once the surface construction transition from 2-D to 3-D occurs, the growth will proceed subsequently in a 3-D mode with the RHEED pattern also transiting from the streaky pattern (Fig. 2.2(a)) to the spotty pattern (Fig. 2.2(b)). Such a spotty RHEED pattern is commonly regarded as the indication

Fig. 2.2 The RHEED patterns that generated by (a) a 2-D surface construction and (b) a 3-D surface construction.

that the deposition amount of InAs is sufficient for QD formation. More details about the growth of the self-assembled InAs QDs will be described in the next chapter.

Besides the capability of surface construction analysis, the oscillatory behavior of the RHEED specular beam intensity also provides us a convenient method to in situ calibrate the growth rate. An explanatory diagram is shown in Fig. 2.3, in which θ is the fractional layer coverage. The equilibrium surface existing before growth is smooth, corresponding to high reflectivity of the specular beam. As growth starts, nucleation islands will form at random sites on the surface, leading to a decrease in reflectivity (due to enhanced random scattering). These islands grow in size until they coalesce into a smooth surface again. It is expected that the minimum in reflectivity would correspond to 50% coverage by the growing layer. Hence, the period of the oscillations corresponds precisely to the elapsing time for the growth of one monolayer. The behavior of layer-by-layer epitaxy growth can be thereby appreciated by observing the oscillation of the RHEED intensity.

Fig. 2.3 Real space representation of the formation of the first complete monolayer of (001) GaAs with respect to RHEED intensity oscillations. θ is the fractional layer coverage [ref. 2.2].

Fig. 2.4 shows an example for the layer-by-layer growth of GaAs. Clear oscillatory behavior is observed for the RHEED intensity in spite of damped amplitude. This damping characteristic suggests that the growth gradually changes from a sequentially layer-by-layer mode to a stable step-flow mode. A growth rate of about one-fourth monolayer per second can

be derived from Fig. 2.4. It should be noted that one monolayer contains both Ga and As atoms and is equivalent to a half of the lattice constant. It is only the beam flux of Ga that determines the growth rate under As rich conditions [2.3]. For GaAs, the V-III flux ratio used in our lab is roughly 10 for As2 and 20 or higher for As4. Excess As2/As4 is lost by re-evaporation and therefore stoichiometric GaAs can be grown. The beam flux ejected from each cell is measured by a beam flux monitor equipped on the substrate manipulator (Fig. 2.1). The orientation of the substrate manipulator is changed by 180^ofor the beam flux monitor to face the effusion cell for beam flux sensing.

Fig. 2.4 One of the measured RHEED oscillations for GaAs in our MBE system.

2.2 Material Characterization

2.2-1 Atomic Force Microscopy

Atomic force microscopy (AFM) is a fast and convenient method for nano-scale surface characterization in semiconductor industry due to its operability in ambient air and minimal sample preparation for investigation. The AFM operation principle is illustrated in Fig. 2.5. The instrument consists of a cantilever with a sharp tip mounted on its end. It is the force between

Fig. 2.5 Schematic illustration of an atomic force microscope.

the apex of the tip and the surface atoms in the sample that produces the AFM operation [2.4].

Such force depends sensitively on the distance between the tip and the sample. Considering the compromise between high image resolution and minimal damage creation for both samples and tips, the tapping mode was used throughout in this dissertation. The cantilever is excited close to its resonant frequency by a driving signal applied to the piezoelectric ceramic to which the cantilever is attached. The excited oscillation of the tip/cantilever is appropriately damped by the force between the tip and the sample due to an appropriate tip-sample distance, and the oscillation amplitude is recorded by a positional detector which detects the laser signal reflected by the end of the cantilever (see Fig. 2.5). With a feedback-control electronics and a 3-D scanning mechanics, when the oscillating tip scans over the surface of the sample, one can keep the tip-sample distance constant by keeping the oscillation amplitude constant. In this manner, the topography on the sample surface can be obtained with the resolution reaching ~1 nm. In our lab, the AFM is performed by a Veeco Digital Instrument D3100 commercial system, primarily for understanding the geometry, sheet density, and uniformity condition of the 3-D quantum objects in the samples.

2.2-2 Photoluminescence Spectroscopy

Photoluminescence (PL) spectroscopy is a very practical and efficient material characterization method for most III-V semiconductors, in which the process of the electron-hole (e-h) recombination is mainly radiative. It serves to determine the fundamental bandgap (E0) of the sample, understand the crystalline quality of the sample, and identify the impurity species in the sample [2.5]. When a sample are excited by an optical source, typically a laser with photon energy hν > E0, e-h pairs are generated, subsequently followed by a carrier thermalization process (including diffusion, capture, and relaxation of the carriers), and then the PL signals are obtained from the spontaneous emission of the sample due to the radiative e-h recombination. In our lab, the samples grown by our MBE machines are routinely characterized by a homemade PL system as shown in Fig. 2.6. Such a system has the capability of, besides the conventional PL, also the PL-excitation (PLE), resonant PL and micro-PL measurements.

Fig. 2.6 The instrument setup for PL, PLE and micro-PL measurements in our lab.

For conventional PL measurement, an argon ion laser is used as the excitation source and the samples for investigation are mounted in a helium close-cycled cryostat, where the sample temperature can be maintained at from ~13K to room temperature. While for PLE or resonant PL, a Ti-sapphire laser pumped by the argon laser is instead used to excite the samples. The excitation photon wavelength can be tuned from ~770 nm to ~1030 nm using different lens kits.

The emission signal of the sample was collected by a couple of plano-convex lenses, transferring through air space, dispersed by a 0.55m monochromator, and then detected by a thermal-electric cooled InGaAs photodiode. Using mechanical choppers to modulate the laser beams, the signal-to-noise ratio can be greatly improved by such a lock-in technique. Besides, a 100 times long-working-distance object lens along with an OXFORD designed liquid nitrogen/helium cooled cryostat is used for the micro-PL measurement, where the spot size of the excitation beam can be focused down to only ~1.5 μm. We take advantage of its ultra-high excitation density to obtain PL spectra revealing high-lying excited states of the InAs quantum objects that are studied in this dissertation.

2.2-3 Transmission Electron Microscopy

Transmission electron microscopy (TEM) is a powerful technique for obtaining highly magnified sample images with extremely high resolution, approaching 1-2 Å [2.6]. A very high energy electron beam, typically around 100-400 keV, is deflected and focused on the sample by electro-optic condenser lenses, and passes through the sample, forming a magnified image in the image plane which is then simply projected onto a fluorescent screen. The sample must be sufficiently thin (a few tens to a few hundred nm) to be transparent to electrons. Due to such a thin thickness, few electrons are absorbed in the sample. Thus, image contrast does not depend very much on absorption, as it does in optical transmission spectroscopy, but rather on scattering and diffraction of electrons in the sample.

Images formed using only the transmitted electrons are bright-field images and that using

a specific diffracted beam are dark-field images. On the other hand, if large aperture is used so that one transmitted beam plus many diffracted beam are all collected in the image plane, high-resolution images are formed. The image brightness is determined by the intensity of those electrons transmitted through the sample that pass through the image forming lenses.

Therefore, heavier elements cause stronger scattering, hinder the electrons from reaching the fluorescent screen, and thus create dark image in the TEM picture. In general, for crystalline specimen, the contrast in the TEM images may arise from mass contrast, thickness contrast, diffraction contrast, etc. In this dissertation, TEM pictures are all obtained by a JEOL JEM-2100F field-emission TEM system using high-resolution image mode. Different from the AFM technique, which was routinely performed for every sample, TEM analysis was only performed for certain samples in this dissertation due to the challenging, time-consuming task on sample thinning.

2.3 Device Fabrication Process

The device samples for the studies of quantum structure infrared photodetectors (QSIP) in this dissertation were all grown on semi-insulating GaAs substrates with the type of vertical transport structure, i.e. active region is sandwiched between a top-contact and a bottom-contact n+ GaAs layers. Discrete devices were fabricated by a two-step process using standard

在文檔中自組式砷化銦量子點/環及其紅外線偵測器之研究 (頁 22-0)