Single Quantum Dot Optical Spectroscopy

Single Quantum Dot Optical Spectroscopy

Submitted as a Course Project for ECE 580

By

Amrita Urdhwareshe

Rohini Vidya Shankar

1. Introduction

Quantum dots (QDs) signify a new class of materials, with size-dependent electronic, optical and magnetic properties. These structures are predicted to have discrete atomic-like energy levels (Figure 1), and a spectrum of ultra-narrow transitions that is tunable with the size of the quantum dot. These are interesting because of the possible promising applications making use of these discrete energy levels, and the possibility to study fundamental physical aspects in these man-made “artificial atoms”.

Figure 1 Density of states in a semiconductor as a function of dimension.
Ref [1]

The optical properties of semiconductors can arise either from the recombination of electrons and holes or from the creation and annihilation/excitation of excitons. An exciton is a bound state of an electron and a hole in a semiconductor, and can be modeled on the lines of a hydrogen atom. It therefore, has a characteristic “Bohr radius” and discrete energy levels.

According to the density of states and the energy levels in a quantum dot, optical transitions of single excitons occupying the quantum dot ground states would be expected to exhibit spectrally narrow emission characteristics. However, the conventional spectroscopic measurements of these QDs revealed significantly broadened emission spectra. This is attributed to the fact that these conventional studies measure an ensemble of dots with size and shape fluctuations, which results in inhomogeneously broadened spectral features (Figure 2). Hence, spectroscopy of single QDs becomes essential if one is to understand and evaluate their intrinsic physical properties precisely, without ending up getting ensemble averages.

Figure 2: Inhomogeneous broadening of spectra Ref [2]

In this report, we discuss some of the experimental techniques used for single quantum dot optical spectroscopy, and summarize the key results obtained.

2. Experimental techniques

2.1 Sample preparation

1) Colloidal nanocrystals

This technique is usually used for preparing quantum dots of II-VI compounds (CdSe, CdS, PbS etc.). The quantum dots are prepared via a chemical synthesis, and then deposited onto a substrate.

The synthesis begins with the rapid injection of organometallic reagents into a hot coordinating solvent to produce a temporally discrete homogeneous nucleation. Slow growth and annealing in the coordinating solvent results in uniform surface derivatization and regularity in core structure. Size selective precipitation provides powders of nearly monodisperse nanocrystallites which can be dispersed in a variety of solvents. The crystallites are slightly prolate with an aspect ratio of 1.1 to 1.3. The average crystallite size, defined by its major axis, is tunable from ~12 to ~115 Ao, with a size distribution of about 5 %. For example, a typical preparation of CdSe nanocrystallites would consist of mixing a Me2Cd (dimethyl cadmium) solution and a TOPSe (tri-n-octylphosphine selenide) solution in TOP (tri-n-octylphosphine) in suitable concentrations under suitable reaction conditions depending upon the size of the particles to be synthesized.

Both the average size and the size distribution of crystallites in a sample are dependent on the growth temperature, the growth temperature necessary to maintain steady growth increasing with increasing average crystallite size.

Some core-shell quantum dots (eg. ZnS/CdSe or CdS/CdSe core-shell dots) have also been used for spectroscopic studies. These dots are prepared by drop wise adding of the shell material onto a solution of the core nanoparticles under appropriate reaction conditions.

Single nanocrystal samples are then prepared by spin coating an extremely dilute solution of nanocrystals, usually in poly (methyl methacrylate)/toluene onto a crystalline quartz substrate. The concentration of nanocrystals in the initial solution (~nmol/L) is chosen to produce an areal density of less than one nanocrystal per square micrometer in order to allow spatial resolution of individual nanocrystals using far-field optics.

2) Epitaxially grown quantum dots

Molecular beam epitaxy is a commonly used technique for the fabrication of quantum dots, especially for the III-V compounds (e.g. GaAs, InAs, etc.). The commonly investigated samples under this category are InGaAs /AlGaAs self-assembled QDs grown on (100) GaAs substrate fabricated by gas source molecular beam epitaxy. With the Stranski-Krastanow growth mode process, a thin quantum well structure with thickness 4 multilayer (ML), so called wetting layer (WL), is formed under the QDs. The density of the QDs is about 3X10-8/cm2 and the typical lateral size and height of the dots are 40nm and 10 nm, respectively, measured by an atomic force microscope. The commonly used technique for studying these samples is near-field spectroscopy.

2.2 Spectroscopy

1) Near-field optical spectroscopy

Near-field optical probing is a powerful technique to study the optical properties of semiconductor quantum nano-structures with spatial resolutions well beyond the diffraction limit of light. The high-spatial resolution of near-field optical fiber probe tips, which can approach λ/40, where λ is the optical wavelength of the dot luminescence, makes the imaging and spectroscopy of highly specialized nanostructures such as quantum dots possible, even at a spatial density of the order of 100/µm2. Figure 3 shows a schematic representation of a low temperature nano-probing system based on shear-force distance regulation.

The QD sample on the scanning stage is illuminated with He-Ne laser light through the aperture of a high sensitive double-tapered fiber probe under shear force feedback control. Most of the carriers are generated in GaAs and AlGaAs barrier layers. After diffusing in the barrier layers, the carriers are captured in the confined states of the QDs. The resultant PL signal from a single QD is collected by the same aperture to prevent the reduction of spatial resolution due to carrier diffusion. The PL signal is detected by means of a photon-counting avalanche photodiode through a bandpass filter. In the case of collecting the PL spectra, the PL signal is sent into a monochromator and detected by a cooled charged coupled device. The spectral resolution of this measurement is about 1meV.

Figure 3: Schematic representation of low temperature nanoprobing system used to measure the near field luminescence properties of quantum dots Ref [3]

Figure 4: Near-field luminescence image of a single In0.4Ga0.6As/Al0.5Ga0.5As QD (T = 5 K)

Ref [3]

To image a single dot, the luminescence falling inside a single window of Δλ = 2nm centered at 733nm is collected, and the tip is scanned across a 2μmX2μm area of the sample. Figure 4 shows an image of a single quantum dot obtained using this technique.

The strong point of nano-probing is its ability to map the location, polarization state and emission wavelength of every dot even if the areal density is as high as 100/μm2.

2) Far field Microscopy

Far-field epifluorescence confocal microscopy is another technique used to obtain images as well as spectra of single quantum dots. A typical optical setup is shown in Figure 5.

Excitation light is transmitted through a high reflecting mirror (at an angle of 45o) and focused by a long working distance microscope to a ~30 μm spot on the sample surface. The fluorescent image is collected by the same objective lens, reflected off the high reflecting mirror and passed through a wavelength specific filter to remove any excitation light while allowing all of the fluorescence to pass. The image is then focused onto the entrance slit of a spectrometer and detected with a cooled CCD camera (liquid nitrogen or thermoelectrically cooled). Both images and spectra of the sample can be collected on the same detector by switching between the diffraction grating and a mirror at 0o for spectra and images, respectively.

Figure 5: Far-field epifluorescence microscope Ref [2]

In order to take spectra from single nanocrystals, the entrance slit of the spectrometer is used to spatially isolate individual nanocrystals along a vertical region of the image. It is usually possible to align several nanocrystals at different vertical positions within the narrowed slit. The mirror in the spectrometer is then replaced with a diffraction grating and light from each vertical position is dispersed onto the CCD. What results is an image in which single nanocrystals are spectrally dispersed along the x-axis and are vertically separated along the y-axis corresponding to their spatial position within the entrance slit (Figure 6). In this way, it is possible to simultaneously obtain spectra from several different nanocrystals. In Figure 6, four single nanocrystal spectra are collected. However, careful manipulation of the sample concentration can allow the simultaneous detection of up to 100 spectra.

Figure 6: A) Image of single 45 Ao nanocrystals at 10 K; B) Image of the same region as in (A) with narrowed entrance slit; C) Spectrally dispersed image of the entrance slit in (B) Ref [2]

3. Observations

1) Single quantum dot spectra

Much lower linewidths are obtained in the spectra of single quantum dots. The ensemble emission spectrum can be reproduced by a convolution of the average single nanocrystal lineshape observed for individual quantum dots within the sample (Figure 7). The close agreement between the ensemble average distribution and the average of the distribution of single dot spectra confirms the proposition that the broadening of the ensemble spectra is because of the size distribution of the ensemble.

Figure 7: Ensemble vs single nanocrystal spectra for three different sizes Ref [2]

All single nanocrystal spectra show qualitatively similar spectral characteristics, including narrow peaks and a longitudinal optical (LO) phonon progression with peak spacing comparable to the bulk LO phonon frequency. Differences in phonon coupling are observed between different nanocrystals with an average value that is equal to the phonon coupling measured in ensemble samples. The range of phonon couplings measured in these experiments is something that can only be seen on the single nanocrystal level and represents a clear example of ensemble averaging, in which information about the individual members of a distribution is lost when averaged over the ensemble.

Fore core-shell quantum dots, it has been observed that the quantum yield of the emission is greatly enhanced as compared to the pure core quantum dots. For example, a ZnS coating over CdSe has been reported to increase the fluorescence quantum yield to as high as 50%. The coating also causes a slight red shift of the emission peak. This can be qualitatively understood as follows: in spite of the larger bandgap of the coating, some of the electron (hole) wavefunctions do extend into the shell, causing an increase in the effective size of the quantum dot, hence lowering (raising) the energy of the electron (hole), and resulting in a net red shift of the emission peak.

Some of the single QD spectra obtained using the above techniques have been illustrated below. Figure 8 shows a low temperature (T = 5K) near-field luminescence spectrum of In0.4Ga0.6As/Al0.5Ga0.5As quantum dots. The five sharp peaks correspond to excitonic emission from five different quantum dots lying under the aperture tip. The linewidths of the emissions are limited by the spectrometer spectral resolution of 0.5 nm. Far field photoluminescence taken from the same sample with a much poorer spatial resolution of 5-10 µm does not show any sharp features, only a broad peak centered at 730nm.

Figure 8: Low temperature (T = 5K) near-field luminescence spectrum of In0.4Ga0.6As/Al0.5Ga0.5As quantum dots Ref [3]

The emission wavelength is a sensitive indicator of each dot's structure and local environment. By analyzing the polarization properties of the emission, it can be

Figure 9: (a) PL spectrum of a single QD at room temperature (b) QD spectra for different excitation intensities Ref [4]

concluded that the five emission lines come from different dots. Since the direction of the

axis of the linearly polarized emission is different for each emission line in the set, it indicates that the emission lines are uncoupled from each other. The lack of a preferred polarization direction in the emission also indicates that the shape of even neighboring dots is not preferentially stretched along a given crystal direction.

Figure 9(a) shows the PL spectrum of a single QD obtained in a high excitation condition at room temperature. The PL signals from the WL and GaAs barrier layer were observed at 1.34 and above 1.36 eV, respectively. The PL below 1.28 eV corresponds to that from the confinement states of the QD. Figure 9(b) shows the PL spectra in single QD at various excitation power densities. The peak intensity of each spectrum is normalized by the excitation power density. In the weak excitation condition, a single PL peak is observed at 1.158eV (E1). For excitation power density higher than 1.0kW/cm2, another PL peak appears and grows at the higher energy of 1.205eV (E2).

To confirm the origin of emission lines, the plots of the integrated PL intensity for the emission lines of both E1 and E2 are illustrated in figure 10 (a). The PL intensity of E1, indicated by the closed circles, almost linearly increases up to 0.8 kW/cm2 and gradually saturates. This fact strongly implies that the emission line of E1 originates from the recombination of the ground state. At the power density of around 0.8W/cm2, where the ground state emission begins to saturate, the PL line of E2 indicated by the open circles appears. From this threshold-like PL behavior as a function of excitation power density, the PL line of E2 can be associated with the emission of the first excited state. The energy separation between the ground state and the first excited state is about 45meV which is in agreement with the prediction of the theoretical calculation by considering the composition and size of the QDs.

.

Figure 10: (a) Excitation power dependence of integrated PL intensity at 1.158 eV (closed circles) and 1.205 eV (open circles) (b) Excitation power dependence of the linewidth of the emission at 1.158 eV Ref [4]

In order to estimate the inhomogeneous width of a single QD at room temperature, the PL line width (FWHM) from the ground state as a function of exciton power density is investigated in Fig.10(b). In the weak excitation region, the linewidth remains constant at the value of about 14meV. With an increasing excitation power density, the linewidth also increases above 0.8kW/cm/cm, where the integrated PL intensity starts to saturate. Such a broadening of the linewidth with increase of excitation power density has also been reported at room temperature.

2) Intermittency/ blinking of the fluorescence

It has been observed that on the timescale of a single image (<0.5 s), the fluorescence from individual nanocrystals flickers on and off (fluorescence intermittency) (Figure 11). The pattern of on/off behavior is strongly dependent on the particular nanocrystal as well as the excitation intensity. This binary fluorescent blinking, rather than a stepwise or continuous dimming of the emission from a single point, is strong evidence that the spots imaged and the spectra obtained originate from single nanocrystals. In the case of an ensemble, this intermittency is averaged over all the particles, and hence a distinct on/off pattern is not observed.