Chapter 2: Experimental techniques
2.2 Material characterization
2.2.1 Photoluminescence
Photoluminescence (PL) is a nondestructive characterization technique to identify the optical quality of semiconductors. It is a measurement defined as the creation of electron-hole (e-h) pairs in the semiconductor by optical radiation and subsequent radiative recombination photon emission. Briefly, there are three basic physical processes involved in the PL: e-h excitation, carriers’ thermalization and diffusion, and e-h recombination.
The PL system in our laboratory (as shown in Fig. 2.4) consists of an argon-ion pumping laser, an optical chopper, a lock-in amplifier, a closed-cycle cryostat, a 0.85m double grating monochromator, two photodetectors (Si-PMT and TE-cooled InGaAs)
and a set of focused/collected lens. The samples to be measured are kept in the cryostat, which provides for the varied-temperature measurements (about 20K ~ room temperature). The pumping laser is focused to excite the samples, and then the resulting luminescence is collected into the input slit of the monochromator. The grating used in the system is 600 /mm with blazing wavelength at 1000nm. The dispersed light is imaged on the output slit of monochromator and detected by the photodetector. The whole PL system is controlled by a personal computer.
2.2.2 Atomic force microscopy
• Introduction
The invention of the scanning tunneling microscope (STM) has revolutionized the field of microscopy: scanning probe microscopy (SPM). It relies on a feedback loop to control a fine tip only a few nanometers away from the sample surface while the tip is raster scanned in X and Y to record an image. Since a tunneling current is employed in STM, the application is limited to conductive surfaces. In order to analyze the features of insulting surface, a new kind of SPM, which is called atomic force microscope (AFM), was developed immediately. At first, the contact mode AFM was developed, which relied on the repulsive forces experienced by the tip measured by recording the cantilever deflection. Afterwards, a new modulation technique in AFM (tapping mode) was invented to overcome the limitations of contact mode. For consideration of sample damage and fast wear-out of tip operated in contact mode, we characterize the samples in the tapping mode. Moreover, we operate the AFM in the constant-force (or constant-interaction) mode. In the constant-force mode, in which the feedback mechanism is activated, one detects the variations of the local z-height of the tip with respect to the sample surface at the fixed force strength.
Our system (as shown in Fig.2.5) is a Digital Instruments MultiMode SPM, which basically consists of the optical head and the base. The scanner is installed within the base, and the measured sample is mounted on the top of the scanner. The optical head consists of sample space, a laser diode, mirrors, and a four-quadrant positional photodetector. The tip on the cantilever is vibrated by bimorph at its resonance. As it approaching to sample surface, the tip-sample interaction causes a change in the amplitude, the phase, and the resonance frequency of the vibrating cantilever. Therefore, during operating in tapping mode, the feedback loop keeps the cantilever to vibrate at constant amplitude (constant-force mode) by extending or retracting the scanner as it is simultaneously raster-scans in X and Y directions. Finally, the “history” of scanner movement in Z across the sample surface is converted into a 3D image of the height data.
Finally, it is necessary to note that we should check the tips used in the measurement carefully because the results of AFM images are strongly dependent on the shape and radius of the tip. Figure 2.6 shows that the surface morphology of QDs in the same sample with different tips. The measured result is approximately the sum of the nanostructures’ real size and the tip’s diameter. Therefore, a ‘very uniform’ size distribution of the nanostructures would be obtained and is nearly equal to the tip’s diameter when the tip’ size is much lager than nanostructures’.
• Brief description of operation principle
The basic principle of operation in the tapping mode AFM can be simply and phenomenally described with a forced oscillation with damping. The equation of motion and its solution have the form
m t
where F, ω, ωo , and β are driving force, driving frequency, cantilever’s resonant frequency and damping constant respectively. The solution consists of homogeneous (which is omitted above) and particular parts. The homogeneous solution comprises an exponent term derived from the damping, and would decay rapidly. The particular term is a steady-state solution.
From the formula above, the amplitude and phase of the oscillation changes with damping β. The relations between the amplitude (phase) and oscillation frequency of the cantilever are shown in Fig.2.7. Figure 2.8 gives the shift in amplitude (phase) at the fixed frequency due to the tip-sample interaction. There are at least three types of data recorded in the tapping mode AFM measurement: height, amplitude, and phase image.
As mentioned above, the height image records the ‘traveling’ of scanner needed to keep the amplitude of the cantilever constant. On the other hand, the amplitude image is the change in amplitude of cantilever due to tip-sample interaction, and phase imaging is the mapping of the phase lag between the periodic signal that drives the cantilever and the oscillations of the cantilever.
References
[2.1] E. H. C. Parker, The technology and physics of molecular beam epitaxy, London, England (1985).
[2.2] M. A. Herman and H. Sitter, Molecular beam epitaxy Fundamentals and current status, Springer (1996).
[2.3] V. Swaminathan and A.T. Macrander, Materials aspects of GaAs and InP based structures, Prentice Hall, Inc. (1991)
[2.4] S. N. Magonov and M.H. Whangbo, Surface analysis with STM and AFM (1996).
Cryo-pump
Fig.2.1 A schematic of our Varian Gen II MBE system.
Fig.2.2 One of measured RHEED oscillations in our system.
In820 Growth Rate
Fig.2.3 One of calibrated plots for growth rate and beam flux equivalent pressure.
Argon Laser
Lock-in Amp. PC
Si-PMT
PD
Monochromator (SPEX-1404) M1
L2 L1 Cryo-Stat M2
Sample Chopper
Fig.2.4 A schematic of the PL measurement system.
Fig.2.5 MultiMode SPM (upper image) and optical head (lower image).
Figure 2.6 AFM images of QDs in the same sample taken with different tips
Fig. 2.7 The relations between the amplitude (phase) and oscillation frequency of the cantilever
Fig. 2.8 The amplitude (phase) shift at the fixed frequency due to the tip-sample
Chapter 3
InAs/GaAs quantum dots growth and characterization
In this chapter, the studies of self-assemble InAs quantum dots (QDs) growth on (100) GaAs substrate are presented. The investigation of QDs growth and characterization were carried out, which were based on the previous results in our laboratory. First of all, the fabrication methods of QDs and the mechanism of the self-assembled growth (Stranski-Krastanow mode) are introduced. Then, the previous results of growth conditions dependence of the QDs are reviewed briefly. At last, the studies of growth parameters dependence of the QDs are demonstrated.
3.1 Introduction to QDs fabrication
3.1.1 Fabrication methods of QDs
There has been a great amount of effort expended in the QD fabrication methods during the last two decades. The most conventional method is to lateral pattern the quantum well structure by electron beam lithography and wet or dry etching. However, lithography- and etching-based technologies will damage the structure and lead to defect problem. In order to overcome the surface state problem, two new techniques were introduced in early 1990s. In 1993, a novel fabrication technique, which was so-called thermal etching, was developed to construct high-quality and surface-state-free QDs [3.1]. This technique utilizes the property of different thermal desorption rates between different materials to result in nonuniform thermal evaporation in a thin epilay to create QDs. Figure 3.1 illustrates the growth procedure
of thermal etching. Besides, Stranski-Krastanow (S-K) growth mode is also a promising approach to produce QDs [3.2]. The pyramidal-shaped QDs are formed by S-K transition at the early stage of the strained growth and are a consequence of the elastic relaxation of the strain caused by the lattice mismatch between the epilay and the substrate. In the InAs/GaAs system, the lattice mismatch is about 7%. Figure 3.2 illustrates the typical growth procedure of InAs QDs on GaAs. The deposition of InAs on GaAs substrate maintains 2-D layer by layer growth in the first few monolayers (MLs). However, with the increasing strain energy induced by the lattice mismatch, the S-K transition takes place at certain deposition, so-called critical thickness. At the same time, the RHEED pattern transforms from a (2x4) streaky pattern to a spotty pattern, indicating the presence of 3D islands on the sample surface. From previous results, the typical critical thickness of InAs on GaAs system is about 1.4~1.7 MLs.
3.1.2 Growth mechanism of self-assembled QDs: Stranski-Krastanow mode There are three modes of crystal growth on the surfaces, which was classified by Ernst Bauer in 1958 [3.3-3.6]. They are Frank-van der Merwe (F-M), Volmer-Weber (V-W), and Stranski-Krastanow (S-K) modes, as illustrated in Fig.3.3. In the epitaxy of hetero-structures, the growth modes are determined with the overall free energy, such as surface, volume and interface free energies. If we deposit the material “A” on substrate “S” under definite conditions, and the overall free energy of the system decreases faster over the first monolayer (or two), 2-D layer by layer growth mode will proceed, that is called F-M mode (left one in Fig.3.3). On the other hand, it favors the island or the V-W mode (middle one in Fig.3.3) as the overall energy increases faster over the first monolayer (or two). In an intermediate case, the layer-plus-island or S-K mode turns out (right one in Fig.3.3).
In practice, the SK growth mode often applies to the strained layer heteroepitaxy, such as QDs growth. The accommodation of misfit strain between the epilayers and substrate changes the balance among the free energies because the strain energy increases with the film thickness. Therefore, a 2D layer growth may be favored initially, and the further deposition of material lead to the appearance of 3D islands within which strain is relaxed and their free energies decrease.
3.1.3 Brief overview of previous results [3.7]
There has been a great mount of effort in studies of growth condition dependence of InAs QDs in our laboratory. The results showed that substrate temperature, As beam flux, and InAs growth rate play important roles on InAs growth and have strong influence on transition energy levels [3.8-3.11]. Roughly speaking, higher substrate temperature, higher As beam flux, and lower InAs growth rate for QDs growth give a lower ground state transition energy due to larger indium diffusion length or less In-Ga intermixing.
There still exist several parameters for QDs growth, such as GaAs capped growth rate, As2/As4 ratio, As beam flux, and so on. In this chapter, we will discuss the effects of these factors on InAs growth based on previous results.
3.2 Self-assembled InAs/GaAs QDs growth by MBE
3.2.1 MBE growth and sample structures
A typical structure is designed for PL and AFM studies of InAs QDs, which is schematically shown in Fig. 3.4. The PL structure comprises a layer of InAs QDs, a pair of GaAs spacers, and a pair of Al0.3Ga0.7As carrier confinement layers. Besides, there is
an uncapped InAs QDs layer grown on top surface of the structure, which is used for AFM studies. The basic growth procedure and condition are given here. However, other special approaches will be described in the respective chapter.
After native oxide desorpted under As flux at 610oC, a 2500Å GaAs buffer layer was deposited at 570oC to recover the substrate surface. 300 Å Al0.3Ga0.7As and 1400 Å GaAs were deposited at a growth rate of 1µm/hr at the same temperature. Then, 100 Å GaAs was grown at a growth rate of 0.3µm/hr. In the meanwhile, the substrate temperature is lowered down (510oC ~530oC) for InAs deposition, and the desired As flux for QDs growth was achieved by adjusting the needle valve of the As cracker cell.
Afterwards, 10nm GaAs capped layer was deposited at the same temperature of QDs growth with a growth rate of 1µm/hr (or 0.3µm/hr.). Then the substrate temperature and As flux were raised to the original values for GaAs and Al0.3Ga0.7As growth. Finally, a layer of uncapped QDs was grown with the same conditions, and then the substrate cooled down under As flux immediately.
3.2.2 Sample characterization
After MBE growth, the samples were characterized by AFM and PL. One of the uncapped InAs QDs AFM images is shown in Fig.3.5. The sheet density of the QDs is about 3x1010cm-2, and the height is about 5nm, respectively. As mentioned in the previous chapter, the tip used for AFM measurement has strong influence on shape and size. Therefore, it is difficult to identify the real shape of QDs from AFM. Besides, it should be noted that the shape and size of QDs buried in the GaAs would change due to strong In-Ga intermixing induced by lattice mismatch between InAs and GaAs, [3.12].
We will study this in the next section and the following chapter.
Figure 3.6 shows the results of the PL under different excitation powers at ~30 K.
We can observe the state filling effect in the figure. We also found that, from the curve with smallest excitation power (1mW) fitting by a multi-peak Gaussian function, the ground state energy, full-width-half-maximum (FWHM), and the energy level difference between ground and first excitation states are about 1.07eV, 29meV, and 65meV, respectively.
3.3 Growth conditions of self-assembled InAs/GaAs QDs
Several growth conditions are studied in this section. They are As2/As4 ratio, GaAs capped growth rate, and As beam flux. In order to study the effects on transition energy and uniformity of QDs for clarity, higher substrate temperatures (510~530oC), and lower InAs growth rates (0.056µm/hr and 0.0373µm/hr) are used in the experiments.
Detailed growth conditions are listed in Table 3.1.
3.3.1 As2/As4 ratio [3.13]
It is well known that the influence of arsenic species (As2 and As4) on MBE growth is quite different. Especially for InAs QDs growth, the effect of As beam flux is very different with that for 2-D growth. In practice, the sticking coefficient of As4 has been shown to be half that of As2. To investigate this effect on QDs growth, various As2/As4 mixtures are used. We can obtain a desired mixture of As2/As4 vapor by altering the cracker zone temperature of the As valved cracker cell. The cracker zone temperatures used in the experiments are 840oC, 730oC, 680oC, and 570oC. For As2
(As4) mode, the temperature of the cracking zone is kept at about 840oC (570oC). On the other hand, the As cracker cell provides a mixture of As2/As4 as the temperature is set at 730oC (680oC). The AFM images of QDs shown in Fig. 3.7 and Fig. 3.8 demonstrate a significant feature in the surface morphology: There is a tendency that
the density and the size uniformity of QDs are improving with increasing the percentage of As4 in the mixture. This result could be explained in the following way:
The QDs form in the early stage (or so-called “seed”) can collect InAs from their neighborhood and then grow gradually. The indium adatoms are more mobile and has a larger diffusion length in As2 atmosphere. Hence, the QDs using As2 flux can gather much more InAs from farther area. Lager QDs’ size and lower sheet density are achieved. In addition, it also exhibits a large fluctuation in QD size. In contrast, The QDs using As4 flux can only collect InAs from a fixed region of their vicinity. This leads to a smaller size dimension and a higher sheet density of QDs. Furthermore; the size uniformity is also improving due to collecting a fixed amount of InAs under As4
beam flux. The AFM images show these facts no matter what As BEP is (on condition that no growth interruption after InAs deposition).
From the PL results in Fig.3.9, it shows little dissimilarity among various As2/As4, if V/III ratio is lower during InAs deposition. The reason is that, for QD growth under low As BEP, the In adatoms migrate on the grown surface easily and then find their proper sites both in As2 and As4 atmosphere. On the contrary, there is an obvious difference, if V/III ratio remains high. The In adatoms tend to stop at improper sites under high As BEP, resulting in a broad distribution. Especially, it shows a multi-fold distribution if the percentage of As2 is increasing in the mixture. The reason is the same as above.
3.3.2 GaAs capped growth rate
In order to obtain high optical activity, the QDs need to be protected via the growth of the capping GaAs layer. As mentioned above, the inter-diffusion or intermixing would take place as the overgrowth of GaAs on InAs QDs due to a large misfit strain
between them. Therefore, the condition of GaAs overgrowth plays an important role in transition energy level of QDs, such as As beam flux and capped growth rate. Previous studies gave a conclusion that a higher As beam flux would reduce the diffusion length of InAs and GaAs on the surface and then depress the inter-diffusion (or intermixing) between GaAs and InAs. In this subsection, the effect of capped growth rates (0.3µm/hr or 1µm/hr) is studied. Figure 3.10 shows that there is no or little difference between higher and lower capped growth rate when substrate temperature is 520oC. The reason is given as follows: Alloying is a strong temperature-dependent process. The rapid In(Ga)As alloying effect occurs around the periphery of the QDs during GaAs overgrowth even though the capped rate is low. The formed alloy inhibits successive Ga adatoms to diffuse inward and intermix with InAs further. For that reason, the capped growth rate has no or little effect on InAs QD growth, as compared to As BEP. However, as GaAs being capped at low temperature, the growth rate would influence the growth of QDs. We will show this in the chapter 6
3.3.3 As beam flux
The diffusion length of adatoms on the surface is strongly dependent on the growth conditions, such as molecular species (In, Ga, or Al / As2 or As4), substrate temperature, arsenic BEP. Adatoms’ diffusion length and intermixing usually determine QDs’ size distribution and their transition energy.
As warming up the As valved cracker cell for growth, the arsenic bulk material needs at least two hours to reach thermal equilibrium due to its large thermal mass.
However, in practice, the arsenic BEP will increase gradually with time. In other words, the vacuum pressure of growth chamber is rising steadily. By exploiting this feature and without adjusting the As needle valve during epitaxial deposition, we can perform
successively three identical experiments to monitor the growth of QDs. The growth condition is given as follows: 2.6 MLs of InAs is deposited at 520oC with a growth rate of ~0.056µm/hr. In Fig.3.11, there are twofold size distributions of PL in all samples.
There always exists a peak value of PL spectrum, whose transition energy is ~1.235eV (see the arrows), under this growth condition. We attribute this to the emission from the QDs that formed immediately after S-K transition. The other peak values are ranged from 1.14~1.20eV. In the first round of the experiment, most of the QDs grew gradually by absorbing In from their vicinity. However, with increase of As BEP, it would be to suppress the migration of In adatom and result in a group of QDs that is smaller than that of the first round. This result manifests the growth and decline of QDs in the S-K mode.
3.3.4 Remark
In order to solve the problem of slow response of arsenic BEP due to large thermal
In order to solve the problem of slow response of arsenic BEP due to large thermal