OPTICAL STUDY OF COUPLING MECHANISMS IN QUANTUM DOT - QUANTUM WELL HYBRID NANOSTRUCTURE

7/29/2019 OPTICAL STUDY OF COUPLING MECHANISMS IN QUANTUM DOT - QUANTUM WELL HYBRID NANOSTRUCTURE

1/159

OPTICAL STUDY OF COUPLING MECHANISMS IN QUANTUM DOT -Q U A N T U M W E L L H Y BRI D N A N O S T RU CT U RE


2/159


3/159

OPTICAL STUDY OF COUPLING MECHANISMS IN QUANTUM DOT -QUANTUM WELL HYBRID NANOSTRUCTURE

A dissertation subm itted in partial fulfillmentof the requirements for the degree ofDoctor of Philosophy in M icroelectronics-Photonics

By

Vitaliy DoroganUzhgorod National UniversityBachelor of Science in Physics, 1999Uzhgorod National UniversityMaster of Science in Physics, 2000

August 2011University of Arkansas


4/159

UMI Number: 3476090

All rights reservedINFORMATION TO ALL USERSThe quality of this reproduction is dependent upon the quality of the copy submitted.

In the unlikely event that the author did not send a com plete man uscriptand there are missing pages, these will be noted. Also, if material had to be removed,a no te will indicate the deletion.

UMIDissertation Publishing

UMI 3476090Copyright 2011 by ProQu est LLC.All rights rese rved. This edition of the work is protected a gainstunauthorized copying under Title 17, United States Code.

ProQuest LLC789 East Eisenhower ParkwayP.O. Box 1346Ann Arbor, M l 48106-1346


5/159

ABSTRACT

The interaction between nanostructures of different dimensionality, two-dimensional quantum well (QW) and zero-dimensional quantum dots (QDs), has beenstudied by photoluminescence (PL) spectroscopy method s in the InAs/GaA s QDs -InGaAs/GaAs QW hybrid nanostructure. A strong dependence of the PL spectra on theseparation barrier thickness and height was o bserved. In sam ples with thick (high) barrierbetween the QW and QDs essentially no carrier transfer took place and thenanostructures behaved independently. If the separation between the QW and QDs wasonly several nanometers, the effective carrier transfer from the QW ground state to the 3 rdQD excited state occurred. It was shown that the QD state filling at elevated excitationintensities had a strong effect on the tunneling efficiency. The PL data were successfullydescribed by the rate equation model which included state filling effects. The tunnelingtimes extracted from the continuous-wave PL data predicted a subpicosecond tunnelingdynamics for the QD-QW samples with spacers < 2 nm, which suggested a microscopiccoupling mechanism in this system to be the resonant electron tunneling.

Anom alous dependence of the QD PL excitation (PLE) intensity at resonance Q Wexcitation on the separation barrier thickness was observed and explained in terms ofintermediate coherent tunnel coupling based on the optical Bloch equation model. It wasshown that the resonant coherent tunneling could be more effective between the QWexcited state and the QD excited state due to the larger overlap of the electronicwavefunctions. T his interaction was observed in the PLE spectra as a peak splitting of thesecond QW subband resonance, indicating the hybridization and derealization of theelectronic states between the QW and the QDs. The possibility of tuning the QW energy


6/159

levels by varying the QW composition was demonstrated on a set of samples. When theQW lowest state was tuned in resonance with the WL state, the anti-crossing of the QWand WL resonances was observed in the PLE spectra which indicated a strong couplingbetween these two states with the tunneling dynamics on a few hundred femtosecondscale.


7/159

This dissertation is approved for recommendationto the Graduate Council.Dissertation Director:

Dr. Gregory Salamo

Dissertation Committee:

Dr. Huaxiang Fu

Dr. Min Zou

Dr. Surendra Singh

Prof. KenVickers (ex officio)

The following signatories attest that all software used in this dissertation was legallylicensed for use by Mr. Vitaliy D orogan for research purposes and publication.

Mr. Vitaliy Dorogan, Student Dr. Gregory Salamo, Dissertation Director

This dissertation was submitted tohttp://www.turnitin.com for plagiarism review by theTurnltln company's software. The signatories have examined the report on thisdissertation that was returned by Turnltln and attest that, in their opinion, the itemshighlighted by the software are incidental to common usage and are not plagiarizedmaterial.

Prof. KenVickers, Program D irector Dr. Gregory Salamo, Dissertation Director
http://www.turnitin.com/http://www.turnitin.com/http://www.turnitin.com/


8/159

2011 by Vitaliy DoroganAll Rights Reserved


9/159

DISSERTATION DUPLICATION RELEASEI hereby authorize the University of Arkansas Libraries to duplicate thisdissertation when needed for research and/or scholarship.

Agreed Vitally Dorog an

Refused Vitaliy Dorogan


10/159

ACKNOWLEDGEMENTS

I would like to express my sincere gratitude to my academic advisor,

Distinguished Professor Gregory J. Salamo, who introduced me to amazing world ofnanoscience and inspired me in many ways. Dr. Salamo, who has a great knowledge ofboth experimental and theoretical physics, set an example of ideal scientist to me. It wasa great pleasure for me to work with Dr. Yuriy Mazur, Dr. Vasyl Kunets,Dr. Dorel Guzun, D r. Georgiy Tarasov, and Dr. Morgan W are. These peop le gave mecontinuous support, encouragement, and help with conducting the experiments anddiscussing the results. I appreciate a cooperation with Prof. Christoph Lienau who made abig contribution to our understanding of the coherent phenomena. I am grateful toDr. Euclydes Marega Jr. for long hours spent and great passion shown during the MBEgrowth of many high quality samples used in this work. I thank Dr. Mourad Benamarafor taking num erous TE M im ages of the samples used in this research. I am also grateful

to m y d issertation comm ittee mem bers Dr. Huaxiang Fu, D r. Min Zou, D r. SurendraSingh, and Professor Ken Vickers for critical reviewing of my dissertation and giving mesuggestions that helped me improve it. Finally, I wish to thank all of the postdocs andstudents in Dr. Salam o's group and M icroEP com munity for being inspiring andsupportive all this time on my long way to achieving this important goal in my life.

The research presented in this dissertation was financially supported by theNational Science Foundation under Grants # DMR-0520550 and DMR-1008107. Anyopinions, findings, and conclusions or recommendations expressed in this material arethose of the author and do not necessarily reflect the views of the National ScienceFoundation.

vii


11/159

D E D I CA T I O N

I dedicate this dissertation to my parents, Gennadiy and Ida Dorogan. I appreciate thatyou have never interfered with my decision to become a scientist and always supportedme in every possible way.

viii


12/159

TABLE OF CONTENTSI. INTRODU CTION 1II . COUPLING IN DOUBLE LAYER NANOSTR UCTURES 3

2.1 Doub le quantum well system 42.2 Doub le quantum dot system 112.3 Quantum d o t - quantum well system 192.4 Strain-induced QDs in a QW 262.5 Applications of QD -QW system 282.6 Summary 30

III. SAMPLES AND EXPERIMENTAL METHODS 313.1 QD-Q W sample design and growth 313.2 Structural characterization of QD-QW samples 353.3 Continuous-wave PL and PL excitation measurem ents 393.4 Time-resolved PL measurem ents 413.5 Summary 42

IV. EFFECT OF QD STATE FILLING ON TUNNELING PROCESSES IN QD-QW NANOSTRUCTURES 43

4.1 Effect of GaAs spacer thickness 444.2 Effect of AlAs and AlGaA s barrier height 474.3 PL decay measurem ents in QD-QW samples 514.4 Rate equation model 564.5 Fitting experimental data and determination of tunneling time 614.6 Effect of QW compo sition variation on tunneling processes 704.7 Strain-induced QDs in dot-in-w ell structure 784.8 Summ ary 86

IX


13/159

V. SIGNATURES OF COHE RENT TUNN ELING IN QD-QWNANOSTRUCTURES 88

5.1 Anomalous dependence of QD PLE signal on barrier thickness under resonantQW excitation 89

5.2 Optical Bloch equation model 945.3 Resonant tunneling time measurement 1005.4 Hybridization of the QD and QW excited states 1025.5 Anti-crossing between WL and QW states tuned in resonance by QW comp osition

variation 1095.6 Summ ary 114

VI. CONCLUSIONS AND OUTLOOK 115References 119Appendix A: Derivation of Equations (4.13) and (4.14) 126Appendix B: Description of Research for Popular Publication 130Appendix C: Executive Summ ary of Newly Created Intellectual Property 134Append ix D: Potential Patent and Com mercialization A spects of listed IntellectualProperty Items 135Append ix E: Broader Impact of Research 137Append ix F: Microsoft Project for PhD MicroEP Degree Plan 139Appendix G: All Publications Published, Submitted and Planned 143

x


14/159

I. INTRODUCTION

Nanotechnology, an amazing field of research, technology, and applications, hasgrown beyond our bravest expectations in the past two decades. Nanostructures,structures made of nanometer-sized pieces of matter with specific shapes, are theelements or building blocks of nanotechnology. By controlling size and shape ofnanoscale material systems, it is possible to engineer materials with unique propertieswhich bring innovation in all fields that one can imagine. Products of nanotechnology canbe found in wide variety of important applications: electronics, optoelectronics,telecommunications, power generation, military and space applications, all kinds ofcoatings, paints, lubricants, biomedical ap plications, etc.

Semiconductor nanostructures hold promise to drastically improve thefunctionality and increase the variety of devices based on existing semiconductormicroelectronic technology. Moreover, semiconductor nanostructures are to bring thequantum phenom ena to a practical use , e. g. quantum comp utation. Amo ng thesemiconductor nanostructures are quantum wells (QWs), quantum wires (QWRs),quantum dots (QDs), and all possible combinations of the three, as well as multilayeredstructures. Unique properties of the QDs allow for realization of single QD devices [1]such as single-electron transistors [2,3], single-photon emitters [4,5,6], devices forstorage of a single electron [7], electron spin memory devices [8,9], and a concept ofquantum information processing [10].

The next step needed for successful optical quantum computing is a creation ofeven more complex quantum systems that will contain various nanostructures interactingwith each other [11,12]. Such systems that are made of different nanoscale components

1


15/159

with quite different properties are called hybrid nanostructures. For example, combiningmetal nanostructures with semiconductor nanostructures is of great interest because thiscombination gives an opportunity to utilize a unique ability of metals to confine the lightwithin nanometer-sized structures along with the ability of semiconductors to detect thislight and convert it into an electrical signal [13]. These metal-semiconductor hybridnanostructures are predecessors of future nanophotonic integrated circuits. Of course,hybrid nanostructures could be made of solely semiconductor components that exhibitvery different properties. The fundamental questions about the details of mechanisms ofinteraction, often referred to as coupling, between different nanostructures are still to beanswered.

One can consider that coupling between two nanostructures exists if exchange ofcharge carriers and/or energy takes place in such system, but how does one confirmexperimentally that two nanostructures are coupled? If one of the two couplednanostructures is being disturbed externally, for exam ple by light of a certain wav elength,it will change its energy state. The other nanostructure should respond to that change bychanging its own state, which can be observed. How does this transfer occur? How manydifferent mechanisms do exist in a particular system? What is the contribution of eachmechanism? How does the coupling change if one changes the distance or thesurrounding material that separates two nanostructures? An attempt to answer these andmany other questions about coupled nanostructures has been made in this dissertationresearch by exploring optical properties of the quantum dot - quantum well (QD-QW)hybrid nanostructure.

2


16/159

II. COUPLING IN DOUBLE LAYER NANOSTRUCTURES

To build a quantum well nanostructure one must use at least two kinds ofsemiconductor m aterials, one of which has a greater bandgap than the other. The size of asmaller bandgap material in at least one dimension should be only several nanometersthick, close to the de Broglie wavelength of the carrier. The smaller bandgap materialshould be surrounded by the larger bandgap material. This provides a potential barrier atthe interface of the two materials that confines charge carriers within the nanostructure.The schematic way of presenting such nanostructures is the one-dimensional energydiagram along one of the directions of quantum confinement as shown in Figure 2. 1.

GaAs

CDCLU

GaAs r f * I

ConductionBand

-^- ValenceBand

Direction of confinementFigure 2.1 Schematic representation of the energy band diagram of asemiconductor nanostructure using InGaAs QW as an example.

3


17/159

In the nanostructure, both electrons and h oles are confined within a potential well and areallowed to possess only certain discrete energies, quantum states. The values and thenumber of the discrete energy levels depend on the height of the potential barriers and onthe width of the potential well.

The most suitable way to study coupling between two semiconductornanostructures is to make a structure that consists of two layers (each layer contains thesame or different type of nanostructures) separated by a barrier material of knownthickness, so called double layer nanostructures. Coupling of two nanostructures mayinvolve many different mechanisms that can play more or less significant roles: coherenttunneling [14,15], phonon- and Auger-assisted (incoherent) tunneling [16,17], dipole-dipole or Forster interaction [18,19], long-range coupling through the radiation field[20,21], long-range polariton interaction [22]. The main coupling mechanism that takesplace at low temperature (~10 K) in the semiconductor nanostructures discussed in thisstudy is quantum (coherent or incoherent) tunneling. The strength of tunnel couplingdepends on the distance that separates two adjacent nanostructures, on the barriermaterial, and on whether the energy levels in the two nanostructures are aligned.

2.1 Double quantum well systemQuantum wells, thin layers of semiconductor material sandwiched between a

different bulk-sized semiconductor with high quality abrupt interfaces (as in Figure 2.1),were among the first semiconductor nanostructures that crystal growth techniques wereable to produce about two decades ago. Hence, a double QW structure was a logical

4


18/159

opportunity for exploration of coupling mechanisms between semiconductornanostructures.

There are two types of double QW structures: symmetric double QWs (both wellshave the same width and composition) and asymmetric double QWs (the structureconsists of a narrow QW and a wide QW , both wells are of the same com position). In thesymmetric double QW system, the energy levels of electrons and holes in both wells areperfectly aligned. As one brings two identical QWs together to a distance of severalnanometers the wave functions overlap and the energy levels start to interact and pushapart from each other. This is a direct analogy to the situation when two separatehydrogen atoms are brought together and form the H2 molecule, their atomic states splitinto doublets. On the other hand, if the distance between the two QWs is rather large(> 20 nm ), each QW can be co nsidered as an identical standalone QW with no anyinteraction with the o ther QW . It is shown in Figure 2.2 how the lowest conductionenergy states splitting in the coupled QW s depends on the barrier thickness in the systemswith (a) 40 A thick QWs and (b) 80 A thick QWs [23]. In superlattices, heterostructureswith many identical QWs divided by thin barriers, the splitting of the interacting levelsresults in a formation of minibands, which is analogous to a formation of bands instead ofdiscrete atomic states when individual atoms are bound together forming a solid matter.

An asymmetric double QW system has an advantage over a symmetric double

QW system, because the optical signal (for example, PL or absorption) from the twoQW s is spectrally separated (see Figure 2.3). Another advantage of an asymm etric doubleQW system is that by applying an external electric field along the growth direction it ispossible to tune the energy levels in and out of the resonance (see Figure 2 .4).

5


19/159

160

0 20 40 60 80 100B ( A )

Figure 2.2 Conduction energy levels splitting as a function of barrierthickness B of (a) 40 A-B-40 A and (b) 80 A-B-80 A sym metric do ubleQW s systems. Inset is a schematic cond uction band energy diagram [23].

CB

(a)

VB

WQW NQW

c

E 2 > E 1

(b)

E2 EnergyFigure 2.3 Asymmetric double QWs system: (a) energy band diagram and(b) corresponding PL spectrum of wide QW (WQW) and narrow QW(NQW).

6


20/159

- 40 A), the electrons tunnel faster than the holes(the tunneling rate of the electrons is about one order faster than that of the holes) due tothe large difference of the effective masses. This results in a charge buildup at highexcitation intensities and, hence, to a slight band bending due to the internal electric field,which in turn shifts the electron and hole energy levels [33,34]. In a nonresonant regime,if the energy level mismatch is larger than the energy of a longitudinal optical (LO)phonon then the electron tunneling transfer via thick barrier turns out to be LO-phonon-assisted process [34]. If this energy m ismatch is smaller than the value of the LO phono n,the nonresonant electron tunneling occurs via scattering on defects and interfaceimperfections of the double QW heterostructure [35]. The electron tunneling ratedecreases exponentially with an increase of the barrier thickness [23 ].

10


24/159

2.2 Double quantum dot systemA quantum dot is often embedded in the matrix made of a material with a larger

bandgap which creates a quantum confinement for charge carriers in all three dim ensions,and has attracted attention of researchers in the past 15-20 years. The QDs are oftencalled artificial atoms because of their discrete, atomic-like, energy states for theelectrons and holes. The properties of the QDs (optical and electrical) depend on therelative energy po sition of these levels and can be easily en gineered by changing the sizeand chemical composition of the QDs.

There are different methods of the QD growth. Spherical QDs with a precisecontrol of the QD size and hom ogeneity can be obtained by chemical colloidal synthesis.Usually, these nanoparticles are made of II-VI sem iconductors and are dispersed in somekind of a chemical solution. Additional procedures are required to attach colloidal QDs tothe semiconductor surface if a combination with a planar structure is needed. In epitaxialgrowth, for example by molecular beam epitaxy (MBE) or metal-organic chemical vapordeposition (MOCVD), the QDs are grown on a semiconductor substrate with an atomic-layer-thickness control. During the growth, a thin layer of semiconductor material withslightly different lattice constant from that of the substrate (for example, between GaAsand InAs lattice constant difference is ~7%) is deposited on the substrate. The QDs formspontaneously due to the strain relaxation in a lattice mismatched structure (Stranski-Krastanov grow th mode) [36]. Varying the growth conditions and choosing differenttypes of substrates to grow on, one can control QD shape, size, density, and even sitelocation [37,38,39,40]. Using a droplet epitaxy MBE growth, the QD clusters of variousconfigurations can be created [41,42]. All above mentioned techniques are so-called

11


25/159

"bottom-up" approaches to the QD formation. In a "top-down" approach, the QDs arefabricated using patterning of the QW (or any other heterostructures that con tains a two-dimensional electron gas) by means of various techniques such as lithography. Onlysamples with Stranski-Krastanov based QDs grown by MBE were used in this work.

If two QDs are placed close to each other so that the wave functions of thecarriers in both dots overlap, coupling may occur between these two QDs. This situationis very similar to a formation of a molecule out of the two atom s. Following this analogy,the system of the two (or more) interacting QDs is often called a "QD mo lecule".

When considering the MBE-grown double layer QDs, one has to deal with twoensembles of QDs. This means that due to the size distribution of the QDs the sum ofslightly different discrete energy levels ends up as a wide band in the optical spectrum.Also, it is impo rtant to distinguish b etween lateral and vertical cou pling. Figu re 2.8illustrates an ideal case where the QDs in the layer 1 are located directly on top of theidentical Q Ds in the layer 2. In the Figure 2.8, the arrows indicate interdot distanceswithin the layer and between the layers. Because of the flattened hem ispherical shape ofthe QDs, the lateral coupling can be achieved only in dense QD arrays (with QD density

QD Lateral

QD Vertical QD

Layer 1

Layer 2Figure 2.8 Schematic cross-section view of a double layer QD system.

12


26/159

greater than 1010 cm 2) [43,44]. In the vertical QD arrangement, the interdot distance canbe significantly reduced compared to the lateral QDs and, therefore, much strongercoupling can be achieved. Moreover, it is more difficult to control the lateral interdotdistance as compared to the vertical distance between the QD layers, which is importantfor the study of the tunneling time as a function of the barrier thickness [45].

By analogy with the double QW system, the double layer QDs can be symmetricand asymmetric, and the carrier tunneling can be resonant and nonresonant. It isreasonable to perform an explicit study of resonant carrier transfer only on a single pairof QDs, where the energy states are sharply and clearly defined in the optical spectra. Onthe other hand, if the system consists of the two QD ensembles with the same averagesize, due to the QD size variation, som e portion of the QDs from b oth layers is always in-resonance and another portion of the QDs is always off-resonance. Therefore, to studynonresonant carrier transfer between the two QD ensembles it is reasonable to have twolayers with different sizes of the QDs with uniform distributions within each layer. In thiscase, the optical signal from the ground exciton transition of each QD layer has differentenergy in the spectrum [46,47].

The PL spectra of the asymmetric double QD samples with various spacerthicknesses are shown in Figure 2.9(a) [47]. The QDs in the first (seed) layer formed by1.8 monolayers (ML) of InAs deposition were smaller in size (small QDs - SQD) andwere kept the same for the whole set of samples. For the second layer, QDs wereintentionally made larger (large QDs - LQD) by means of 2.4 ML InAs deposition. Mostof the QDs in the second layer, correlated by a strain field from the seed layer, grewdirectly on top of the first layer QD s. Den sity and size of the QD s in the second layer

13


27/159

1.0 1.2 1.4Photon energy (eV)Figure 2.9 (a) PL spectra of the asymmetric double QD structures withdifferent spacer thicknesses taken at T= 10 K. (b) Schematic energy bandstructure along the growth direction of the double QD system with theprocesses of carrier generation, recombination, and interdot tunnelingrepresented by arrows [47].

depended on the spacer thickness [47,48]. The QD layers were divided by a GaAs spacerwith thicknesses of 30, 40, 50, and 60 ML (which corresponded to 85, 113, 142, and170 A, respe ctively). As one can see from the Figure 2.9(a), for the smallest spacer (30ML) only the emission peak from the LQDs in the second layer was present in the PLspectrum. The PL peak from the SQ Ds in the seed layer started to appear gradually as thespacer thickness increased from 30 to 60 ML. This behavior of the PL spectra indicatedthat tunneling probability w as high with 30 M L spac er: most of the carriers tunneled fromthe SQD to the LQD and recombined there, giving rise to a single PL peak. Withincreasing spacer thickness tunnel coupling decreased due to reduced carrier wavefunction overlap, which eventually resulted in an independent recombination of excitonsin both QD layers. The carrier dynamics in such a system (shown by arrows in

14


28/159

Figure 2.9(b)) was probed by the time-resolved PL measurem ents using pulsed laserexcitation. The PL transients (curves that show the PL intensity decay after the laserpulse as a function of time) recorded at the emission energies of SQDs and LQDs of the30 ML sample (Figure 2.10(a)) clearly indicated that the L QD s continued to receivecarriers for some time after the laser pulse (rise part of the LQD PL transient) and showedlonger PL decay tim e, while the PL signal from the SQD s started to rapidly decay

T3N

10

10"1 - < 8" < ' % 3 0 M L ^ * j1 1 i

(b)^ 6 0 MLW 5 0 M L^ ^ 4 0 MLZr iir

0 500 1000Time (ps) 1500

Figure 2.10 (a) Normalized PL intensities detected at the emission peaksof the SQDs and LQDs of the structure with 30 ML spacer thickness as afunction of the time delay after the laser pulse, (b) Normalized PLtransients detected at the emission peak of the SQDs for the samples withdifferent spacer thicknesses [47].15


29/159

immediately after the laser pulse. The comparison of PL transients from the SQDs forsamp les with different spacer thicknesse s is depicted in Figure 2.10(b ). The PL decaytime TPL can be extracted from the slope of the exponential decay of the PL transients.The PL decay time (measured experimentally) is related to the transfer (or tunneling)time r tr from the SQD s to the LQD s by the following equation:

r P L = ^ - , (2-2)Ts+ Ttrwhere rs is the radiative lifetime in the SQDs. The rs can be obtained from the PL

transients measured on the reference sample containing solely SQDs grown under thesame conditions as the SQDs in the double layer QD samples. The tunneling timebetween two layers of the QDs depends exponentially on the spacer thickness [45,47].This dependence is described by the Wentzel-Kramers-Brillouin (WKB ) approxim ation:

r lr cc ex p 2*W ^ ) (2.3)Here, d is the spacer thickness, rn is the effective mass in the spacer material, Fis theconduction band discontinuity, and E s is the energy of the lowest quantum state in theSQDs. Several studies [45,47] have shown that the tunneling times in the asymmetricdouble QD system were longer than those in the asymmetric double QW system. Thiscan be explained by the carrier localization in the QDs and the possibility of tunnelingonly between the two neighboring dots, whereas in the double QW system, carriers cantunnel at any location in the QW and find available empty states in the second QW.

Using near-field optical microscopy [49] or micro-PL techniques together with ashadow mask that has an aperture with diameter of < 1 um [50,51,52], or with processing

16


30/159

of the low density QD samples using an electron beam lithography and dry etching [53],mesas are left with supposedly only one QD pair, allowing the study of coupling in asingle QD molecule. Exploring single QDs biased by an external electric field gives anopportunity to trace resonant tunneling of single excitons, electrons, and holes. Excitonenergy dependences on the electric field applied along the growth direction calculated forthe system of asym metric pair of large InAs Q D (4 nm h eight) and small InAs QD(2.5 nm height) with 4 nm GaAs spacer are shown in Figure 2.11 [51 ]. The excitons are

100

5 50 -E> 0a >cCDcooa 100.>JS 50

0

X s|_ / (b)

a n1 0 Y 01 0 *

/fix" X B100 -50 0 50 100Electric Field (kV/cm)

T o p Q Dsiiii

4 nmy^WA

Bottom Q D

T o p Q D

4 nm

Bottom Q D

Figure 2.11 (a) and (b) represent the calculations of exciton energies as afunction of electric field in the two types of single asymmetric QDmolecule schematically shown on a right-hand side. "B" and "T" denotehorizontal lines of a direct exciton transitions in a bottom and top QDs,respectively. Yellow circles indicate the anti-crossings observed in PLexperiment [51].

17


31/159

labeled ashBlTXQ, where eg and er (hB and hi) indicate the number of electrons (holes)located in the bottom ("B") and top ("T") QD s, Q is the total charge of the exciton. Directexcitons, both electron and hole are in the same dot, ( ^ J 0 and QIX) only slightlydepend on the electric bias, whereas indirect excitons, electron and hole are in differentdots, (l^X0 and ^X ) exhibit strong linear electric field dependence (Stark shift). Whentwo levels (either electron or hole) in two QD s align the resonance o ccurs, which appearsas anti-crossing in the exciton energy diagrams in Figure 2.11. If the larger QD sits on topof the small QD (Figure 2.11(a)), the electron resonance occurs at positive bias and thehole resonance occurs at negative bias. The situation is reversed for the opposite structure(Figure 2.11(b)). The mag nitude of the anti-crossing splitting, a measure of the couplingstrength, is related to the resonant tunneling time and depends on the spacer thickness andthe effective mass of the particles that tunnel. In case of electron level resonance, theanti-crossing splitting is much larger (one order for the particular case shown inFigure 2.11) than that of the holes [50,51], which reflects the effective mass differencebetween electrons and holes.

Study of nonresonant tunneling times in a single pair of asymmetric double QDsas a function of spacer thickness revealed that at large spacer (-2 0 nm) there w aspractically no tunneling, at intermediate spacer (-10 nm) mostly electron tunneling tookplace, and at small spacer (-2.5 nm) both electrons and holes tunneled from small tolarge QDs [53]. Transfer times showed an exponential dependence on the spacerthickness [53], which agreed with the WKB approximation. In temperature rangebetween 4 K and 25 K, thermally induced (phonon-assisted) transfer of holes betweentwo neighboring QDs separated by 5 nm spacer has been reported [52].

18


32/159

Thus, double layer QDs have very similar tunneling properties to the double QWstructure. However, these two systems possess quite different two-dimensional (2D) andzero-dimensional (OD) densities of states. For the QDs, two scenarios, ensemble andsingle QDs, should be considered as well as the possibility of lateral and verticalcoupling. Another big QD research field, spin interaction and spintronics applications incoupled QDs (see for example [54,55]), was not touched upon in this review since thisdissertation research does not include any direct information about the interaction of theelectron and hole spins in the QD-Q W system.

2.3 Quantum dot - quantum well systemNow that the coupling in the double QWs and double QDs has been reviewed,

consider a hybrid structure that combines two different nanostructures - QD-Q W system,as shown in Figure 2.12. This system offers a unique opportunity to explore and utilizethe coupling between OD and 2D quantum states in the QD s and QW , respectively.

Double QD systemr*\ rr\ ST\2LXDouble QW system

S QD-QW system

Figure 2.12 Double layer nanostructures and a hybrid QD-QWnanostructure with arrows defining separation distance between the layers.19


33/159

Resonant and nonresonant tunneling from a QW to a QD have been consideredtheoretically [16,56]. In the resonant case when the QW ground state is closely alignedwith one of the QD excited states, the Bardeen transition-probability approach was usedto calculate the tunneling times of electrons and h oles depending on the barrier thickness[56]. It is seen from Figure 2.13 that according to this model electrons tunnel from a QWto the QDs much faster than holes (~ 4 orders of magnitude difference at 10 A spacer).

^ V

C

"53cc3h-

1 0 - 5ic r6lO" 7

ic r810 " y1 0 - i u10" 1110" 1 21 0 " u10" 1 410" 1 50 10 20 30 40 50

Barrier width (A )Figure 2.13 Calculated resonant tunneling time for electrons and holesthat tunnel from the QW ground state to the QD 1 st (long dash line) and2 nd (dotted line) excited states (solid line is a total tunneling time)depending on the separation barrier thickness [56].

In the nonresonant case when energy levels in the QW and QD are not aligned, LO-phonon-assisted and Auger-assisted tunneling of electrons and holes can take place [16].For the former mechanism, that is most effective at low excitation powers, the energydifference between the Q W and QD ground states must be equal or larger than the energyof one LO phon on w hich carrier emits while tunneling from th e QW to the QD . The lattermechanism does not have such a limitation on the energy difference between the Q W and

20

Hole tunneling^ ^ ^ ^ ^ ^ ^ ^ ^ T ^ m - p T ^ T - p - T

Electron tunneling


34/159

QD levels and dominates the tunneling at high excitation powers. In Auger-assistedtunneling, two carriers in the QW take part in the process: one carrier transfers part of itsenergy via Coulomb repulsion to another carrier exciting it to a higher state in the QWand tunnels to the QD state. Tunneling time for both phonon- and Auger-assistedmechanisms can vary between a few and few hundreds of picoseconds depending on aspacer thickness [16].

Several QD-QW structures slightly different by design and material compositionhave been studied experimentally using steady-state and time-resolved PL. The PL andPL excitation (PLE) spectra of the QD-QW structure are shown in Figure 2.14(b) along

QD QW T = 8 Ki lex = 0.1W/cm2

1.2 1.3 1.4 1.5Photon energy (eV)Figure 2.14 (a) PL spectra of InAs QDs and Ino 3Gao 7As QW referencesamples, (b) PL and PLE spectra of the QD-QW sample. PLE signal wasdetected at the QD and QW exciton ground state energies (shown byvertical down-arrows). Iex = 0.1 W/cm 2; T = 8 K [57].

21


35/159

with the PL spectra of InAs QDs and 7.5 nm thick Ino.3Gao.7As QW reference samples(Figure 2.14(a)) grown under the same conditions as the QD-QW sample [57]. The QDsand QW separated by a 4.5 nm GaAs spacer were designed in such a way that the lowestQW electronic state was located above the QD ground state by more than one LO -phononenergy. The PLE spectrum measured at the QD PL maximum clearly showed QW to QDcoupling by a spike and abrupt absorption edge at energy position that coincided with theQW PL emission (Figure 2.14(b)). Another evidence to the favor of the carrier transferfrom the QW to the QDs w as the excitation intensity evolution of the PL spectra dep ictedin Figure 2.15. At the lowest excitation intensity, only the QD PL peak was present in the

1.15 1.20 1.25 1.30 1.35Photon energy (eV)Figure 2.15 PL spectra (normalized with respect to the QD PL maximum)of QD-QW sample at different excitation intensities. The spectra areshifted vertically for clarity. Inset shows the integrated PL intensity ratioIQD/IQW as a function of excitation intensity. T= 10 K [57].

22
http://ino.3gao.7as/http://ino.3gao.7as/


36/159

spectrum. As excitation intensity increased the QW PL peak started to gradually grow inintensity and eventually dominated the spectrum at the highest excitation intensities. Atlow excitation intensity the electrons and holes initially trapped by the QW wereeffectively transferred to the QDs with emission of LO-phonons in the process. Withincreased excitation power more QDs became populated with excitons and, due to thePauli exclusion principle, could not accept additional carriers from the QW which led tothe appearance of the QW PL peak (Figure 2.15). Based on the experimental results, itwas suggested that the tunneling in this particular system occured in several stages. First,an electron tunneled to the QD while a heavy hole remained for some time in the QW,temporarily forming an indirect exciton with an electron in the QD. In the next stage, aheavy hole tunneled to the QD restoring a direct exciton with an electron in the QD [5 7].Time-resolved PL results for the same QD-QW sample fitted by the rate equation modelthat included QD state filling gave the tunneling time of 1.12 ns [58]. Another time-resolved PL study was done o n a QD -QW system that consisted of 7 nm thickIn03Gao7As QW and Ino6Ga04As QD s separated by a 2 nm GaAs spacer [59]. This studyalso indicated the significance of a state filling effect in tunneling process at highexcitation powe rs and the nonresonant tunneling time w as determined to be 200 ps .

Systematic investigation of the QD-QW tunnel injection structure as a function ofa separation spacer thickness was performed by Talalaev et al. [60]. A set of the QD-QW

structures with different spacer thicknesses was grown in an opposite order: first, InAsQDs were formed, the n the QDs were cov ered by a GaA s spacer, and 11 nm thickIno lsGao 85AS QW was grown on top followed by a GaAs cap layer (see Figure 2.16(a)).

23


37/159

Figure 2.16 (a) Schematic illustration of the inverse QD-QW structurestudied in [60]. (b) Formation of nanobridges at spacers thinner than 5 nm.From high resolution transmission electron microscopy measurements it was determined

that in samples with a small separation barrier the QDs were touching the QW formingnanobridges defined as areas that contained more than 15% of Indium (Figure 2.16(b)).The coupling between the QW and QDs was confirmed by the PLE measurements.Tunneling times obtained from the time-resolved PL experiments agreed with the WKBapproximation w ithin the barrier thickness range from 10 to 5 nm. The deviation from theW KB approximation w as observed below 5 nm w hich was attributed to the effect ofnanobridges that connected Q Ds to the QW [60].

Nonradiative Forster-type energy transfer, which involves Coulomb interactionsas oppose to the wave function overlap, from the 3 nm thick InGaN QW to the C dSe/ZnScore/shell nanocrystals through the 3 nm G aN barrier in a hybrid Q D-QW structure w asexplored experimentally [61] and theoretically [62]. Electron-hole pairs can be c reated inthe QW either optically or electrically (as in Figure 2.17(a)). The PL emission of the Q Wis in a strong absorption range of the CdSe nanocrystals which favors a dipole-dipolecoupling between the QW and QDs. The study showed that the energy transfer was fastenough to compete w ith recom bination processes in the QW resulting in a more than 50%

24


38/159

Figure 2.17 (a) Proposed hybrid light emitting QD-QW device thatconsists of epitaxial InGaN QW sandwiched between GaN barrier layersand a monolayer of densely packed CdSe/ZnS QDs coated with organicmolecules placed on top of GaN barrier. Metal contacts on the top and thebottom are used to electrically inject carriers in the QW. ET denotesenergy transfer, (b) Schematic illustration of the energy transfer, carrierrelaxation and recom bination in the hybrid QD -QW structure [62].

of energy transferred to the QDs. Resonantly created electron-hole pairs in the QDsquickly relaxed to the ground state and recombined radiatively (see Figure 2.17(b)). Theeffectiveness of the energy transfer depended on the temperature and the carrierconcentration in the QW. The emission wavelength of the nanocrystals was determinedby their geometrical sizes, which can be easily tuned in a wide range.

25


39/159

2.4 Strain-induced QDs in a QWSelf-assembled epitaxial QDs grown by Stranski-Krastanov mechan ism, i. e.

formed due to the strain relaxation of the lattice mismatched materials, create a strainfield that spreads around a dot in the surrounding matrix. The larger the lattice mismatchand the size of the Q D, the greater the strain and the further it extends. No w, if the QW isplaced in a close proximity to the QD (Figure 2.18(a)), the strain deformation potentialcaused by the dot will induce a lateral 2D confinement in the QW (Figure 2.18(b)). TheQW itself provides confinement in a third (vertical) dimension, hence, resulting in a 3Dcarrier confinement which is nominally required for a formation of a QD. Such QD iscalled strain-induc ed QD (SIQD) [63]. The strain field creates almost ideally parab olic

(a) SIQDInAs IslandInP barrierInGaAsP QWInP bufferInP s ubs.

(c)QD1

- i 0.70 0.75 0.80PL energy (eV) 0 60 120Radial distance from island center (nm)Figure 2.18 (a) Cross-section of the structure with SIQD; (b) lateralconfinement potential created by the strain field in the QW plane; (c)typical PL spectrum of the SIQD; (d) potential profile as a function of astressor dot - QW separation distance d [64].

26


40/159

deformation potential with evenly spaced confined states that give rise to a distinctoptical transitions in PL spectra (Figure 2.18(c)). The depth of a deformation potentialwell in the QW caused by the stressor dot depends on the distance from the stressorisland to the QW (Figure 2.18(d)) [64] as well as the size (height) of the stressor island[65]. There are different material combinations suitable for building a system with largestressor island that causes SIQDs in a near QW . Coherent InP Q Ds with diameter of up to120 nm can be grown on a GaAs surface; the SIQDs are formed in the InGaAs QW [66].Both surface [64,65] and buried [67,68] InA s QDs with lateral sizes from 40 to 110 nmhave been reported as suitable stressor islands for creating SIQDs. It was show n from theoptical and transmission electron microscopy study that when small InAs Q Ds (40-60 nmin diameter) were used as stressors, well detectable SIQDs were formed only in the Q W-above-the-stressor-dot configuration (Figure 2.19(a)), whereas for inverted configurationof the structure, shown in Figure 2.19(b), the formation of SIQDs was possible only forvery small distances between the stressor and the QW [68]. Stressor islands made ofGaSb were also used to create SIQDs in the GaAs/AlGaAs QW [69 ].

(a) (b)

Figure 2.19 (a) Structure with GaAs QW above the stressor InAs QDs and(b) inverted structure with G aAs QW below the stressor InAs QDs [68].

27


41/159

Although SIQDs appear to be an ideal model system for the fundamental studiesof various physical processes in the QDs and Q W s, it may be difficult to find a practicalapplication for these objects. There was one experimental result that may prove apossibility of an exciton storage in an InAs Q D coupled to a SIQD [70].

2.5 Applications of QD-QW systemQuantum dots with their discrete, delta-function-like transitions and emission

wavelength tunable by simply varying the size of the QDs are considered as idealcandidates for building the micro-scale light emitting diodes (LEDs) and semiconductorlasers. Such small lasers could be integrated into the telecommunication schemes andfuture optical computing circuits. The possibility of the QD laser operation has beensuccessfully dem onstrated u sing InG aAs /GaA s Q Ds em itting at 1.3 urn at roomtemp erature [71]. Ho wev er, there were limitations to the efficiency of the QD laserscaused by a rando m Q D size and shape distribution , low density, inefficient carriercollection and redistribution among the dots, room temperature thermal carrier escapeand hot carrier relaxation.

These problem atic issues can be partially resolved if a QW is fabricated in a closeproximity to the Q D layer. In this case, the Q W acts as an efficient carrier collector (dueto its larger area and higher density of states) and reservoir that supplies carriersuniformly to all QDs via tunneling. The carriers that tunnel through the barrier to the do tsare already "cold" and, hence, do not cause the problems related with the hot carrierrelaxation. Also, the strain from the QW affects the growth of the QDs improving theiruniformity and increasing density [72,73]. At high temperature, when thermally activated

28


42/159

carriers escape from the QDs, the QW recaptures some of those carriers and again injectsthem into the QDs increasing the probability of their radiative recombination inside theQDs [74]. It has been reported that QD-QW diode lasers based on single layer InP QDs(Figure 2.20(a)) resonantly coupled to the InGaP QW throu gh a 2 nm tunnel barrier werecapable of room temperature continuous-wave laser operation in visible range (630-680 nm) and had a steeper current-voltage characteristic as com pared to the same InP QDdiode laser without a QW , as illustrated in Figure 2.20(b) [74 ,75].

(a) (b)Figure 2.20 (a) Schematic structure of a QD-Q W diode laser, where CL isa cladding layer, WG is a waveguide, B is a barrier, LB = 20A andLQW = 20A. (b) Comparison of I-V characteristics of the QD and QD+QWdiode lasers [75].

The efficiency of the carrier tunneling strongly depends on the relative position ofthe energy levels in the QW and QDs. Therefore, the QD-QW structure should becarefully engineered so that the tunnel coupled states are in resonance [74,75] or theenergy difference between these levels is equal to the LO-phon on energy [76,77].

The coupled InAs QD - InGaAs QW lasers target the near-infrared operationregion (at 1.3 and 1.55 urn) which is solicited by the fiber-optic telecommunicationindustry [76,77].

29


43/159

The possibility of a white light generation based on the Forster type energytransfer from the epitaxially grown InGaN QWs to the core/shell CdSe/ZnS colloidalnanocrystals has been proposed and experimentally demonstrated [78]. The color of lightgeneration in this type of LED s can be tuned by changing the size of the colloidal QD s.

2.6 SummaryFrom the above discussion, a QD-QW hybrid nanostructure makes a convenient

system to explore coupling between different types of semiconductor nanostructures. Themain coupling mechanism observed in this system is quantum tunneling of electrons,holes, and excitons. Effectiveness of tunneling strongly depends on the relative positionsof energy levels in the QW and QDs as well as the tunneling barrier thickness. The mostefficient tunneling occurs when the energy levels in the QW and QDs are resonantlyaligned or the energy difference between the levels equals to LO-pho non en ergy. Ano ther

coupling mechanism that may play a dominant role in systems that consist of colloidalQDs coupled to an epitaxial QW is nonradiative Forster energy transfer that is governedby the Coulomb interaction of electron-hole pairs. The QD-QW hybrid structure can besuccessfully used in the LED and laser diode fabrication for emission in visible and near-infrared (telecomm unication range) parts of spectrum.

Systematic study of the QD-QW system with different spacer and energy levelconfigurations at various excitation conditions is still needed in order to clarify thecontribution of different coupling mechanisms and carrier dynamics in the coupledsemiconductor nanostructures.

30


44/159

III. SAMPLES AND EXPERIMENTAL METHODS

3.1 QD-QW sample design and growthFor com prehensive study of the QD-QW system, a variety of high quality sam ples

were to be grown. The structure was designed in such a way that the PL peaks from theground states of the QW and QDs were spectrally separated from each other allowingseveral QD excited states to find room between the QW and QDs ground states. Such adesign may prove useful for clarifying the role of QD excited states in tunnelingprocesses of carriers and their further relaxation within the QDs. Furthermore, if theenergy of emission from one of the QD excited states overlaps with the QW excitonicemission, the phenomenon of coherent tunneling can be thoroughly explored in suchsystem. The strength of the tunnel coupling can be effectively varied by changing thethickness and/or height of the potential barrier between the QW and QD s.

A series of QD-QW samples with one layer of InAs QDs and an In xGai_xAs QWseparated by a barrier with variable thickness and height were grown using molecularbeam epitaxy (MBE) Riber 32 system. In order to smooth the surface of semi-insulatingGaAs (001) substrate, a 0.3 um thick GaAs buffer layer was first deposited at temperatureof 580C. After th at the temperature decreased to 530C and a 14 run thick InxGai_xAsQW with x = 0.15 was grown followed by a 30 s growth interruption. The thickness ofthe next layer, GaAs spacer, was varied between 1 and 20 nm in the main set of samples.In a number of samples an additional Al(Ga)As barrier with thickness of 1 to 4 nm wasinserted in the middle of the GaA s spacer (see Figure 3.1(a)). Then, 2 monolayers of InAswere depo sited on top of the GaAs spacer followed by a self-assembled Q D formation.

31


45/159

(a) (b)GaAs cap layer

InAs QDsAlAs orAIGaAs barriers (optional)

W LGaAsspacer

Surface InAs QDs

GaAs buffer layer

GaAs (001) substrate

GaAs buffer layer

GaAs (001) substrate

Figure 3.1 Schematic structure of QD-QW samples used for (a) PLmeasurements and (b) AFM studies.

The entire structure was capped with 50 nm GaAs layer. Sam ples containing only Q Wand only QDs were grown under the same conditions to be used as reference structures.An additional set of QD-QW samples was grown with variation of In composition(x = 0.07, 0.10, 0.15, and 0.18) in the QW , whereas the thickness of GaAs spacer waskept the same, 4 nm. A lso, several QD -QW samples with different GaAs spacers wereleft uncapped for atomic force microscopy (AFM) studies as shown in Figure 3.1(b). Areference QD sample with an additional layer of uncapped QDs on top of the structurewas used in AFM studies as well. The quality of each sample was tested after everygrowth by means of PL measurements and, if necessary, corresponding corrections weremade in the parameters of subsequent growths. The summary of all samples used in thisdissertation is given in the Table 3.1.

32


46/159

Table 3.1 Classification of the QD-QW samples by the barrier type andthickness as well as by In content in the QW. Each shaded cell representsone sample.ReferenceIn As QDs

Referencelno.i5Gao85AsQW

Dot-Wellseparationd s P GaAsspacerAlAs1 nmAlAs2 nmAlAs3 nmAlAs4 nmAIGaAs1 nmAIGaAs2 nm

UncappedQD s

1.5 nm 2 n m 3 nm 5 nm 8 nm 11 nm 15 nm 20 nm

Indiumcontent in QWReferenceQWsQD-QWdSD = 4 nm

7 % 1 0 % 1 5 % 1 8 %

Figure 3.2(a) illustrates the comparison of PL spectra from the referenceInAs/GaAs QD sample (upper part), reference Ino.15Gao.85As/GaAs QW sample (middlepart), and representative Q D-QW sample with 11 nm thick G aAs spacer (bottom pa rt)taken at low excitation intensity and T= 10 K. The PL peaks originated from the groundstate exciton radiative recom bination in the QW at 1.345 eV and in QDs at 1.145 eV.

33
http://ino.15gao.85as/GaAshttp://ino.15gao.85as/GaAs


47/159

AlAs Barr ier. r V.1 -4 nm | |QDsIn As

1.1 1.2 1.3 1.4P hoton Energy (eV)i iL J

Figure 3.2 (a) PL spectra of reference QD sample, reference QW sample,and QD-QW sample with d sp = 11 nm measured at low excitation intensity7exc = 2xlO"57o (/o = 1000 W/cm 2, T= 10 K). High excitation (7eXc = 10/0)PL spectrum of reference QD sample shown by dashed line, (b) Electronenergy band diagram of the QD-QW system based on the growthparameters, TEM and PL results. Recombination and relaxation processesare shown by straight and wavy down-arrows, carrier transfer process isshown by a horizontal wavy arrow [79].

It is easy to see from this comparison that the PL peak positions from the QD-QWstructure coincided within 5% with the PL peaks from both reference samples. The PLspectrum of the reference QDs taken at high excitation power shown by dashed linerevealed additional emission peaks that identified the transition energies of the QDexcited states. It is shown by vertical dashed lines that the QW emission coincided withthe third excited state of the QDs. Thus, in this structure the resonant tunnel couplingshould be observed. According to the growth parameters, TEM, and PL data the energyscheme of the conduction band (CB) and valence band (VB) for such hybrid QD-QW

34


48/159

structure can be suggested as depicted in Figure 3.2(b). The QDs have at least 4 boundstates (E, E1, E2, and E3), whereas the Q W has two states. One state is also present in thewetting layer (WL). The QW ground state is closely aligned with the third excited QDstate allowing the resonan t tunneling of carriers through the GaA s (AlAs) barrier.

3.2 Structural characterization of QD -QW samplesFor successful study of tunneling processes, knowing the exact parameters such as

distance between the nanostructures is crucial. At a nanometer scale, even a smalldeviation from the nominal dimensions can cause a great difference in physicalproperties. It is also very important for comparison of many samples to be valid that allstructural features of each sample be identical except for the one varied intentionally.

The size of the QDs (it is height in case of MBE-grown self-assembled QDs)determines the energy position of the QD PL peak, given the chemical com position of the

QDs remains the same. The size distribution of the QDs is reflected by the width of theQD PL peak. The AFM 1 um x 1 um images and corresponding QD size distributions of(a) uncapped reference QD sample and uncapped QD-QW samples with (b) 3 nm, (c)5 nm, and (d) 8 nm spacers are show n in F igure 3.3. QD height and den sity extractedfrom the AFM images are collected in Table 3.2.

Table 3.2 Structural data obtained from AFM analysis.SampleReference QDsQD-QW, d s p = 3 nmQD-QW, d s p = 5 nmQD-QW, d s p = 8 nm

QD Height (nm)4.5 0.55.0 0.54.5 0.55.0 0.5

QD Density (cm -2)(1 .220.05)x101 u(1 .150.05)x101 u(1 .11 0.05)x10 l u(0 .860.05)x101 u

35


49/159

15 00 nm

0.00 nm

15.00 nm

0 00 nm

15 00 nm

0.00 nm

15.00 nm

0.00 nm

0 1 2 3 4 5 6 7 8 9QD height (nm)

woO 4

J l

(b)

0 1 2 3 4 5 6 7QD height (nm)

oO

I I I II I

(C )

llllll1 2 3 4 5 6 7QD height (nm)

0 1 2 3 4 5 6 7QD height (nm)Figure 3.3 1 um x 1 um AFM scans and QD size (height) distributions of(a) reference QD s and QD -QW samples with (b) 3 nm, (c) 5 nm, and (d)8 nm GaAs spacer between QW and QDs.

36


50/159

Structural analysis to examine the quality of interfaces and verify the thicknessesof the QW and the spacers between the QW and QDs was done by the cross-sectionaltransmission electron microscopy (TEM) measurements using a FEI Titan 80-300 TEMequipped with an image Cs-corrector (CEOS) and an image filter (Gatan Tridiem). Allsamples were prepared using standard procedure which included mechanical polishing,dimpling (using Fischione Dimpling Grinder 200), and low-angle Ar+ ion-milling (usingFischione Ion Mill 1010). Two-beam or multi-beam conditions at one of the (002)reflections were used to carry out diffraction contrast imaging. For high resolution TEM(HRTEM) measurements (magnification up to 380000x), the sample orientation alongthe [110] axis was used. Bright-field cross-sectional TEM images of QD-QW sampleswith 15 nm and 8 nm thick GaA s spacers are depicted in F igure 3.4 (a) and (b),respectively. One can clearly distinguish regions with QDs, WL, QW, and GaAs spacer.The change in contrast seen in the two images taken under the same conditions is relatedto the thickness differences in the different areas of the sample. QD-QW samples withsmaller features w ere examined by the HR TEM in order to check the quality of interfacesand precisely determine the QW and GaAs spacer (or AlAs barrier) thicknesses.Figures 3.4 (c) and (d) illustrate HRTEM images taken from the QD-QW samples with2 nm A lAs barrier embedded in 8 nm GaAs spacer and w ith jus t 2 nm GaAs spacer,respectively. The Q Ds have the shape of truncated con es with average height of 5 nm anddiameter of -20 nm. The thickness of the QW was found to be 14 0.5 nm. All spacerand barrier thicknesses w ere within 0.5 nm error of the nominal dimen sions set up duringthe MBE growth. The TEM image along with the schematic structure of the QD-QWsample with different indium content (x = 0.18) and ^ sp = 4n m is shown in Figure 3.5.

37


51/159

QD (

Figure 3.4 Cross-sectional TEM images of QD-QW samples: (a) and (b)were taken in bright-field mode from samples with 15 nm and 8 nm GaAsspacers, respectively; (c) and (d) are HRTEM images from sample with2 nm AlAs layer symm etrically embedded in 8 nm GaAs spacer and fromsample with 2 nm G aAs spacer, respectively. The QW composition w asIno.15Gao.85As. White dashed line in (d) defines QD contour [79].

In As QDs

Figure 3.5 (a) Multi-beam bright-field TEM image of the QD-QWstructure with the QW indium content of x = 0.18 and GaA s spacerthickness of d sp = 4 nm. (b) S chem atic structure that mirrors the TEMimage.

38
http://ino.15gao.85as/http://ino.15gao.85as/


52/159

Based on all TEM images taken from many QD-QW samples with different spacerthicknesses, an important fact was confirmed. The strain field caused by the InAs QDsdid not penetrate more than 2 nm dow n into the GaAs spacer or further into the Q W.Thus, the possibility of formation of strain-induced QDs inside the QW was excluded inthis case.

Overall, from the AFM and TEM analysis, the following conclusion was made:the MBE-grown QD-QW samples were of high quality and well fitted for the study ofcoupling effects in such systems.

3.3 Continuous-wave PL and PL excitation measurem entsMost of the experimental studies in this dissertation research were done using

various PL methods. Figure 3.6 depicts the experimental setup used for continuous wave(cw) PL and photoluminescence excitation (PLE) measurements. The sample was

mounted inside the closed-cycle helium optical cryostat (Janis) which allowed varyingthe temperature w ithin the range of 10 to 300 K. The cryostat was d esigned in such a waythat the sample was placed in a close proximity (< 1 cm) to the window, so that the shortfocus objective could be used for illumination and collection of the signal. For theexcitation above the GaAs ban dgap, a 532 nm line of the frequency-doubled neodym iumdoped yttrium aluminum garnet (Nd:YAG) laser (Coherent Verdi-VIO) was used. For theexcitation below the GaAs bandgap for example resonant excitation in the WL or QWstates, a tunable Tksapphire laser (Coherent MIRA 900) was used in the cw mode. Theexcitation power was varied by a set of neutral density filters over the range of (10~ 7-102)mW . T he laser beam wa s focused on a sample surface using a near-infrared objective

39


53/159

Figure 3.6 Schematic diagram of the experimental setup for PL and PLEmeasurements. In the figure: M is mirror, FM is flip mirror, BS is beamsplitter, LP Filter is long pass filter, ND Filter is neutral density filter,CCD is InGaAs photodiode detector array.

(Mitutoyo NIR HR) with 50x magnification to a spot with diameter of ~20 urn. The PLsignal was collected w ith the same objective, dispersed by a 0.5 meter imaging triplegrating monochromator (Acton SpectraPro 2500i), and detected by a liquid nitrogencooled InGaAs linear photodiode array (Princeton Instruments OMA V).

In case of PLE mea suremen ts, the excitation w avelength from the Ti: sapphirelaser was tuned from 750 nm to 970 nm with increments of 1 nm while recording the

intensity of the PL signal at a fixed detection wavelength of either QD or QW excitonground state em ission. The output power of the laser was carefully monitored and k ept atthe same level of 500 mW throughout the tuning range.

40


54/159

3.4 Time-resolved PL measurementsThe dynamics of the carrier excitation, distribution, tunneling, relaxation, and

recombination in semiconductor nanostructures can be traced by time-resolved PL(TRPL) measurements where the time evolution of the PL intensity is being recorded.The experimental setup for TRPL is shown in Figure 3.7. The schem e was very sim ilar tothe cw PL m easurements w ith the difference in excitation and detection. The Ti:sapphirelaser was tuned to a wavelength X = 750 nm and switched to a mode-locked regime. Thelaser produced 2 ps pulses at a repetition rate of 76 MH z. The excitation density on asample surface was varied between 5>


55/159

The PL emission collected by the objective was focused to an entering slit of thespectrometer (Bruker) with a synchroscan streak camera (Hamamatsu C5680) attached toit. The streak camera was equipped w ith an infrared enhanced SI cathode, which alloweddetecting of a reliable signal up to the wavelength of -11 50 nm (QD ground stateemission from the Q D-QW samples was around 1100 nm ). The overall time resolution ofthe system was 15 ps.

The result of the measurements by this technique was either the PL transient, acurve that shows the PL intensity decay as a function of time at a given wavelength, or

the PL spectrum at a given time delay. By fitting the PL transients with exponentialfunction it was possible to extract the exciton recombination time in the semiconductornanostructure u nder investigation.

3.5 SummarySeveral sets of the QD-QW samples with various spacer thicknesses and barrier

heights as well as different Indium concentrations in the QW were grown by MBE.Reference samples that contained only QW or QDs were grown as well. The crystalquality of the samples and all geometrical parameters such as QD height, diameter anddensity, QW thicknesses, and spacer thicknesses were determined from the o ptical, AFM ,and TEM characterizations. Experimental setups for the cw PL, PLE, and TRPLmeasurements were described in details.

42


56/159

IV. EFFEC T OF Q D STATE FILLING ON TUNNELING PROCESSES INQD-QW NANOSTRUCTURES

The dynamics of carrier transfer from the QW to the QDs, and then of carrierrelaxation within the QDs, define an efficiency of lasing from the QD ground state. It isimportant that the carrier relaxation from the higher QD excited states occurs at a fasterrate than the tunneling from the QW to the QDs [80]. This fast relaxation helps to avoidthe state filling in the QDs that slows down the tunneling process [81].

In this chapter, an attempt to answer the questions of "What coupling mechanismdoes play the major role in the InAs QD - InGaA s QW hybrid nano structure?" and "Ho wdoes the QD state filling influence the carrier transfer time?" was made by means ofcontinuous-wave and time-resolved PL experiments. To determine the mechanism ofcarrier transfer, a variation of the separation barrier thickness and height as well as tuningthe energy of the coupled QW and QD states were realized. Two sets of the QD-QWsamples were grown with changing profile configurations (includes height and thickness)of the Ga(Al)As barrier and changing Indium content in the InGaAs QW. To explore theeffect of the QD state filling on the carrier tunneling, continuous-wave and time-resolvedPL spectra of the QD-QW samples were measured as a function of laser excitationpower. Experimental results were analyzed using the rate equation mo del, and parametersof the carrier dynamics between the QW and QD coupled states and between the excitedand ground states inside the QD were determined.

43


57/159

4.1 Effect of GaA s spacer thicknessIn this study, the QD-QW structure was designed in such a way that the QD 3 rd

excited state emission energy (E gD ~ 1.350 eV) overlapped with the QW ground stateexciton emission (E Q W ~ 1.345 eV), as shown in Figure 3.2(a). For this arrangement ofenergy levels in the QW adjacent to the QDs, fast carrier tunneling of resonant type wasexpected from the QW ground state to the QD 3 rd excited state (Figure 3.2(b)).

The intensity of the QW PL peak in the spectra of the QD-QW structures wasanticipated to be very sensitive to the separation distance between the QW and Q Ds, thusindicating the coupling strength in the hybrid QD-QW system. Indeed, a cleardemonstration of the QW PL peak sensitivity to a variation of the GaAs spacer thicknesscan be seen in Figure 4.1 that shows the PL spectra of six QD-QW samples with GaAsspacer thicknesses, d sp , ranging from 2 to 20 nm tak en und er different excitationintensities 7exc- All spectra were normalized with respect to the maximum of the QDground state PL peak and vertically shifted for convenience. When the GaAs spacer was20 nm thick, the evolution of the PL spectra of both Q W and Q Ds with increasingexcitation intensity were essentially the same as those of the reference QW and QDsamples. The PL peaks from the lowest states of both QW and QDs were detected at thelowest excitation intensity (Figure 4.1(a)), which indicated that the QW and QDs werepractically decoupled at 20 nm GaAs separation and acted as independent nanostructures.The PL peaks from the QD excited states gradually appeared as the excitation intensityincreased. When GaAs spacer was reduced to 15 nm, the QW peak was barely detectableunder the low excitation conditions and became stronger only with increase of excitationpower (Figure 4.1(b)). Further reduction of the spacer thickness resulted in an increase of

44


58/159

1.1 1.2 1.3 1.4P hot on E ne r gy ( e V ) 1.1 1.2 1.3 1.4P hot on E ne r gy ( e V )cfsp = 20 nm d sp = 15 nm

4 x n r v 2 2x105/

1.1 1.2 1.3 1.4P hot on E ne r gy ( e V )d sp = 11 nm

~ L TQW Q D U ~ L TQW QD ~ L TQW Q D U

1.1 1.2 1.3 1.4 1.1 1.2 1.3 1.4P hoton Energy (eV) P hoton Energy (eV)cfsp = 8 nm dSD = 5 nmSP

1.1 1.2 1.3 1.4Photon Energy (eV)dsp = 2 nmu n r i_nr u i rQW QW QWU Q D U Q D U Q D

Figure 4.1 Normalized PL spectra of the QD-QW samples with (a) 20 nm,(b) 15 nm, (c) 11 nm, (d) 8 nm, (e) 5 nm, and (f) 2 nm thicknesses ofGaAs spacer measured at different excitation intensities denoted on eachcurve. The spectra are vertically shifted for clarity. Diagrams of theconduction band profile illustrate the GaAs spacer thicknesses ofcorresponding samples. 7o= 1000 W/cm 2, T= 10 K [79].

45


59/159

the QD-QW coupling strength, which was reflected in the behavior of the QW PL peak.As shown in Figure 4. 1( c) -4 .1 (f) , the QW PL signal emerged at higher excitationintensities Iexc for smaller spacers dsp . This behavior correlated with the carrier tunnelingthat occurs faster for smaller spacers, whereas the recombination rate in the QWremained the same for all spacer thicknesses. The increasing supply of carriers thattunneled through smaller GaAs spacers made the PL peaks from the QD excited statesappear earlier (at lower excitation intensities) as compared to the reference QD sampleand the QD-QW samples with thick GaAs spacers (15 and 20 nm), as can be seen fromFigure 4.1(f). When the excitation power became high enough to populate the 3 rd excitedstate of the QD and block the tunneling channel, the QW PL peak started to show up inthe spectrum. Thus, intensity dependent state filling of the QDs controls the tunnelingchannel and, hence, the appearance of the QW PL signal. The PL results in Figure 4.1qualitatively show the increasing coupling strength tendency with decrease of the GaAsspacer thickness (or separation distance between the QW and the QDs) in the QD-QWhybrid nanostructures.

46


60/159

4.2 Effect of AIAs and AlGaAs barrier heightThe cou pling between the nano structures can be also affected by the height of the

separation barrier. To determine the influence of the barrier height on the tunnelingefficiency in the QD-QW system, the structures were designed with a comp lex barrier dsthat consisted of the GaAs spacer with fixed total thickness d sp which included asymmetrically incorporated barrier (AIAs or Alo.3Gao.7As layer) of variable thicknesses,as schem atically shown in the right part of Figure 4.2. The effective barrier d-& was

4 nm

w+"E3n> _

"rec_ iQ.

d = cL + dA = 8 nmsp GaAs AIAs/ = 0.004 /exc 0

d A I * < n m )

1.1 1.2 1.3P hoton Energy (eV) "U^ 8 nm-Figure 4.2 PL spectra of the QD -QW samples with constant 8 nm thickspacer dsp that contained AIAs effective barrier JAIAS of 1 -4 nm thicknessinserted in the G aAs layer as schem atically shown on the right. All spectrawere taken at excitation intensity Iexc= 0.004x/0, I0 = 1000 W/cm 2,r = 1 0 K [ 7 9 ] .

47
http://alo.3gao.7as/http://alo.3gao.7as/


61/159

defined by the following expression:dB = [spJV(z)-Edz, (4.1)

where E is the starting level energy and V(z) is the profile of the potential barrier. The PLspectra of the QD-QW samples with dsp = 8 nm and em bedded AlAs layers of variablethicknesses from 1 to 4 nm m easured at the same excitation intensity, along with theschematic representation of their conduction band barrier profiles, are depicted inFigure 4.2. There was significant carrier tunneling that suppressed the QW PL peak in theQD-QW sample with pure 8 nm GaAs spacer. Inclusion of just a 1 nm thick AlAs barrierimmediately "turned on" the QW PL signal, while increasing of the AlAs barrierthickness up to 4 nm completely cut off the tunneling channel between the QW and QDspractically making them independent, non-interacting nanostructures. It is clearly seenfrom Figure 4.3 that the integrated PL signal of the QDs dropp ed exponentially (show n

c3.Q>-c

- JQ .Qa

250200150100


62/159

with dashed line) as the AlA s layer thickness increased and approached the reference QDPL signal at C/AIAS = 4 nm. This meant that with 4 nm of AlAs the QDs did not receive anyadditional carriers from the QW.

Another example of how the incorporated barriers affect the tunneling processesare shown in Figure 4.4. Here, QD-QW samples with a fixed 3 nm thick spacer wereused. Figure 4.4(a) demonstrates that the carrier transfer was reduced by increasing thebarrier height. W hen the QW and QDs w ere separated just by 3 nm thick GaA s spacer,the QW signal was not observed in the PL spectrum (curve 1). Embedding of 1 nm thickAlo.3Gao.7As barrier in the middle of the spacer resulted in the increase of the QW PLintensity by one order of magnitud e (curv e 2). Wh en the height of the effective barrierwas increased even more by incorporating a 1 nm thick AlAs layer, it gave rise to twomore orders of the QW PL intensity increase (curve 3). Almost the same result can beachieved by making the Alo.3Gao.7As effective barrier thicker. As shown in Figure 4.4(b)(curve 3), adding 1 nm of thickness to the effective barrier results in about 2.5 ordersincrease of the QW PL peak. Thus, by varying the aluminum content in the embeddedeffective barrier one can effectively control the coupling strength between adjacentnanostructures while keeping the separation distance con stant.

The results of these PL experimen ts with various barrier profiles indicated that inthe QD-QW system the quantum tunneling is the main coupling mechanism and that theelectromagnetic and polariton coupling could be neglected in further consideration [79].

49
http://alo.3gao.7as/http://alo.3gao.7as/http://alo.3gao.7as/http://alo.3gao.7as/


63/159

QW QD

1.1 1.2 1.3 1.4Photon Energy (eV)

1.1 1.2 1.3Photon Energy (eV)

"LT*~LT

I r3nm

"Li1 /

QD

1 3 nmFigure 4.4 PL spectra of the Q D-QW samples with constant 3 nm thickspacer d sv that contained AlAs or Alo.3Gao.7As effective barrier t/B- (a) and(b) demonstrate influence of the effective barrier height and width ontunneling, respectively. Corresponding conduction band energy profilesare shown on the right side. The spectra are normalized, vertically shiftedand plotted on a logarithmic scale. I0 = 1000 W/cm 2, T= 10 K.

50
http://alo.3gao.7as/http://alo.3gao.7as/


64/159

4.3 PL decay measurements in QD-QW samplesThe time-resolved PL measurements allowed one to study the dynamics of the

carrier tunneling and relaxation processes in the QD -QW nanostructure. Figure 4.5 showsthe time evolution of the reference QD PL spectrum after the excitation laser pulse. At0 ps delay, the TRPL signal reflected all the features present in a cw PL spectrum at highexcitation intensity (shown by the red solid line in Figure 4.5) with a strong WL peak,whic h quickly decayed in first 100 ps. The Q D grou nd state exhibited the s lowest

800 900 1000 1100Wavelength (nm)

Figure 4.5 Time-resolved PL spectra of the reference QD sample taken atvarious delay times (from 0 to 1000 ps) after the laser pulse. A cw PLspectrum (red solid line) is shown for comparison. TRPL spectra arevertically shifted for clarity. X,exc= 750 nm, T= 10 K.

51


65/159

recombination rate, whereas every higher QD state decayed faster than the previous(lower) state.

The time-resolved PL spectra recorded at a number of delay times of the QD -QWsample with 3 nm GaAs spacer, along with the cw PL spectrum recorded from the samesample at high excitation intensity (red solid line), are shown in Figure 4.6. The strongemission from the QW ground state and the W L dom inated the spectrum at 0 ps delay. Apart of the spectrum that consisted of PL peaks from three QD states remained practicallyunchanged (peak intensity decayed less than 10%) up to the delay time (around 1000 ps)

Figure 4.5 Time-resolved PL spectra of the QD -QW sample with 3 nmGaA s spacer take n at various delay times (from 0 to 1000 ps) after thelaser pulse. A cw PL spectrum (red solid line) is shown for comparison.TRPL spectra are vertically shifted for clarity. A.exc= 750 nm, T= 10 K.52


66/159

when the QW PL peak vanishes. After that, the QD peaks decayed in a cascade fashiondescribed above for the reference QDs. This was a real-time observation of the QWfeeding the QDs via tunneling until it ran out of the carriers excited by the laser pulse.The PL intensities as a function of time, PL transients, were measured at energies of aQD ground state and QD excited states (denoted as E, E 1, E2, and E 3 in Figures 4.5 and4.6) of the reference QD sample and the QD-QW sample with dsp = 3 nm (plotted inFigure 4.7). Nonresonant excitation above the GaAs bandgap (kexc = 750 nm) was usedfor the PL transient measurements with an excitation power of Ip = 1.3 x10 5 W/cm 2. Aconvex shape of the reference QD ground state (E) and 1 st excited state (E1) PLtransients (Figure 4.7(a)) indicates the presence of the state filling effects, which was wellknown for the QDs optically excited with high intensity [82,83]. In the presence of theproximal QW the PL transients exhibited two stages of decay (see Figure 4.7(b)). If in thecase of the reference QDs optically excited carriers relaxed to the dot states only from theWL and GaAs barrier, then in the QD-QW system an additional portion of carrierstunneled to the QDs from the QW affecting the dynamics of relaxation processes in theQDs. In the first stage the QD states received extra carriers from the QW, which resultedin slower decay rates. When the QW was completely exhausted, the PL decay from theQD states became faster without additional supply (second stage). For each lower energyQD state the second stage started later because it still had additional carrier supply fromthe QD states with higher energies. The PL transient from the 3 rd QD excited state (E3)was not possible to measure since its peak position overlapped w ith the QW ground stateexciton emission energy and played a role of an acceptor state for the carriers thattransferred from the QW.

53


67/159

1000 1500 0 500Time (ps) 1000 1500

Figure 4.7 PL decay transients recorded at the energy positions of eachQD state (E, E1, E 2, and E3) for (a) reference QD sample and (b) QD-QWsample with 3 nm G aAs spacer. The pulse laser excitation at 750 nm had apeak intensity of Iv = 1.3x105 W/cm2 [79].

Similarly, at high excitation powers the QW PL transients became affected by thestate filling and exhib ited clear conve x shap es (Figure 4.8(a)). The Q W PL decaytransients of the QD-QW sample with d sp = 5 nm shown in Figure 4.8(a) were fitted by amonoexponential decay (in order to determine the QW excitonic emission time) only atlow excitation powers (Ip = 1.3 xlO 4 W/cm 2). Figure 4.8(b) demonstrates the dependenceof the QW PL decay transients on the GaAs spacer thickness taken at low power. It wasobserved that for thin spacers the QW PL decay occurred at a faster rate than for thickspacers. As GaAs spacer thickness increased in the QD-QW samples, the QW PL decaytime approached the reference QW PL decay time (see the slopes in Figure 4.8(b)).

54


68/159

100 200 300Time (ps)Figure 4.7 (a) QW PL transients of the QD-QW sample with 5 nm spacertaken at different intensities, (b) QW PL transients taken from the QD-QWsamples with different GaAs spacer thicknesses and the reference QWsample at low excitation intensity (7P = 1.3>


69/159

4.4 Rate equation modelThe experimental results presented in the previous sections of this chapter

described mostly qualitative information about the state filling and tunneling processes inthe QD-QW system. To obtain quantitative information from the experimental data, avalid comprehensive model that describes carrier generation, relaxation, recombination,and tunneling processes in the QD-QW system was developed and applied to fit theexperimental data. A dynamic rate equation model based on the QD-QW energy levelscheme shown in Figure 3.2(b) w as build with the following assump tions: (1) undernonresonant optical excitation, where carriers are generated in the GaAs matrix withfollowing relaxation to the Q W states, WL states, and QDs states (directly or through theWL), the coherent effects quickly disappear during the relaxation processes and,therefore, play a minor role in the dynamics of tunneling and intersublevel relaxationwithin the QW and QDs; (2) excitonic tunneling is considered in the QD-QW system(which is a significant simplification as compared to the separate electron and holetunneling and relaxation).

The dynamic picture in the QD-QW system can be described by introducing thenumber of carriers n, that occupy each fth state of the QD (i = 0 for the QD ground stateand / = 1,2,3 for the QD 1 st , 2 n d, and 3 r d excited states, respectively) and the number ofcarriers nw on the QW ground state. These numbers change in time due to the generation,transfer, and relaxation processes schematically illustrated in Figure 4.9. The rates of theQW and QD state population change can be w ritten in the following equation set [79]:

dn w{t) =r w (J x nw(i) nw{t) ^nw^ Vexc) _R _NR Z_i ^dt NR Ttr N, (4.2)

56


70/159

dn l it)dt

N,-n,(t)N, +

W-nXOW)j N ^

, [ t f , -n , (0 ] , (0 n,(0 n,(f) (4.3)J NJ*,J _N R

where Gw (I exc ) and Gf (Iexc) are the carrier generation rates in the QW and f QD state,respectively; rw, xw and r t , r ; are radiative and n onradiative recombination times in

ththe QW state and the i QD state, respectively; x\ r is a carrier transfer time from the QWstate to the z'th QD state; T is a time of intersublevel relaxation between the z'th and7 th QD

states; N, is the fh QD state degeneracy (A ; = 2/ for QDs of semispherical shape); [Nt -n,(t)]/N, is the probability that the z'th QD state is empty which takes into account the state

filling. Thus, if the carrier occupies one of the QD excited states, it has three choices ofrelaxation: radiative recombination, nonradiative recombination, nonradiative relaxationto the lower state.

QW

5oXoC Ore

z w**


71/159

Equations (4.2) and (4.3) are general and include all possible processes and can beused to derive expressions that describe both time-dependent and steady-state cases ofexcitation. There are two excitation limits: high intensity and low intensity. In first case,almost all the states are filled, N, ~n,(t), and the carrier decay is described by theeffective recombination time fR defined as

1 1 1 TRTNR-R TR+ TNR =* T' T?+Tm ^ }

and1 1 1 TRTNR = + - => f R = ww (4 51-R R NR ~^ ltV R , NR y+-J)Tw Tw ~w Tw+ Tw

for the rth QD state and the QW state, respectively. In case of low excitation intensity,most of the states are empty, N, n,(t), and equations (4.2) and (4.3) can be reduced to

^ =^(/)-=#-X^(/). (4-6)

at , Twhere

ir(U=i = (4.8)is a carrier transfer rate term (coupling term) and

=fe ] (4-9)I > Ttr)

is an effective transfer time. Assuming that the QDs are free of defects, the radiativedecay time from each of the Q D states is about the same [84], and, therefore, the effective

58


72/159

recombination time of z'th state rR is replaced wit

OPTICAL STUDY OF COUPLING MECHANISMS IN QUANTUM DOT - QUANTUM WELL HYBRID NANOSTRUCTURE

Documents