Development and investigation of II-VI semiconductor microcavity structures / von Alexander Pawlis [original]

Development and investigation

of II-VI semiconductor

microcavity structures

Alexander Pawlis

Amelgatzen, im Dezember 2003

Development and investigation

of II-VI semiconductor

microcavity structures

Von der Fakult¨at f¨ur Naturwissenschaften

der Universit¨at Paderborn

zur Erlangung des akademischen Grades eines

Doktor der Naturwissenschaften

genehmigte

Dissertation

von

Alexander Pawlis

Amelgatzen, im Dezember 2003

Abstract

Due to the strong coupling interaction between photon and exciton in semiconductor

microcavities, two new quantum mechanical eigenstates, the so-called polariton states are

manifested and show extraordinary optical properties. One of these properties is an ”anti-

crossing” behavior of the coupled polariton dispersion with a minimum energy difference,

the Rabi-splitting energy ΩRabi, which is observed in the resonance between photon and

exciton.

In this thesis I report on the observation and investigation of strong coupling between

photonic and excitonic modes in ZnSe/(Zn,Cd)Se multi quantum well microcavities. The

active layer, which is a ZnSe/(Zn,Cd)Se quantum structure, is grown by molecular beam

epitaxy. The results of the optimization of crystallographical and optical properties of

the ZnSe/(Zn,Cd)Se quantum structure as well as the implementation of a cavity length

gradient are discussed in detail. The active layer is covered with polycrystalline dielectric

Bragg-mirrors of ZnS/YF3or ZnSe/Y F3and a high reflectivity R >0.995 in the blue and

green spectral range is achieved.

A large room temperature Rabi-splitting energy of Ω >35 meV has been measured

in microcavities containing four ZnSe/(Zn,Cd)Se quantum wells as active layers. The

”anti-crossing” behavior of the polariton modes has been demonstrated by reflectivity as

well as photoluminescence investigations.

Therefore two different methods of microcavity resonance tuning have been performed.

In reflectivity measurements the photonic mode has been tuned in resonance with the

excitonic mode by varying the spot position on the sample in direction of the micro-

cavity length gradient. In contrast to this, in temperature-dependent photoluminescence

measurements the polariton dispersion is obtained by modifying the resonance condi-

tion between excitonic and photonic mode via the temperature shift of the quantum well

transition energy.

All experimental results are in good agreement with the calculated polariton properties

based on the quantum electrodynamical model of a coupled photon-exciton oscillator.

Contents

1 Introduction 11

2 Exciton-photon interaction in semiconductor microcavity structures 15

2.1 Excitons in strongly coupled microcavities . . . . . . . . . . . . . . . . . . 15

2.1.1 Strong and weak coupling in atomic cavities . . . . . . . . . . . . . 17

2.1.2 Strong coupling in quantum well microcavities . . . . . . . . . . . . 18

2.1.3 Polariton dispersion – effective mass model for a coupled oscillator . 19

2.2 Simulation of the optical response of a microcavity . . . . . . . . . . . . . 21

2.2.1 The Transfer-Matrix model for a single layer . . . . . . . . . . . . . 21

2.2.2 Calculation of the complex refractive index dispersion . . . . . . . . 22

2.3 Resonance tuning of microcavities to obtain the polariton dispersion . . . . 25

2.3.1 Angular dependent microcavity detuning . . . . . . . . . . . . . . . 25

2.3.2 Microcavity detuning with a cavity length gradient . . . . . . . . . 25

2.3.3 Temperature detuning in microcavity structures . . . . . . . . . . . 26

3 Distributed Bragg-mirrors for II-VI microcavities 29

3.1 Optical properties of ZnS/YF3dielectric Bragg-mirrors . . . . . . . . . . . 30

3.1.1 Structural properties of ZnS/YF3DBRs ............... 33

3.2 ZnSe and YF3Bragg-Mirrors ......................... 37

3.2.1 Optical properties of ZnSe/YF3DBRs ................ 37

3.2.2 Structural properties of ZnSe/YF3Bragg-mirrors . . . . . . . . . . 40

Contents

4 ZnSe and (Zn,Cd)Se quantum structures for II-VI microcavities 45

4.1 In-situ growth characterization of ZnSe/(Zn,Cd)Se quantum structures . . 46

4.2 Investigation of (Zn,Cd)Se multi quantum wells . . . . . . . . . . . . . . . 49

4.2.1 Structural investigations of MQWs . . . . . . . . . . . . . . . . . . 50

4.2.2 Optical properties of ZnSe/(Zn,Cd)Se MQWs . . . . . . . . . . . . 57

4.3 ZnSe/(Zn,Cd)Se superlattice structures . . . . . . . . . . . . . . . . . . . . 59

4.3.1 Structural properties of superlattice structures . . . . . . . . . . . . 61

4.3.2 Photoluminescence investigation of SL structures . . . . . . . . . . 68

5 II-VI semiconductor microcavity structures 71

5.1 The combination of Bragg-mirrors and active layer to complete microcavities 71

5.1.1 Experimental growth procedure of microcavity structures . . . . . . 72

5.2 Investigation of a SL microcavity structure with ZnSe/YF3DBRs . . . . . 75

5.2.1 Measurement of the microcavity length gradient off-resonance . . . 76

5.2.2 Reflectivity measurements in resonance condition . . . . . . . . . . 76

5.3 Room temperature Rabi-splitting in a MQW microcavity with ZnS/YF3

DBRs ...................................... 81

5.3.1 Photoluminescence investigations of the MQW microcavity in res-

onancecondition ............................ 84

5.3.2 Reflectivity measurements of the MQW-microcavity in resonance

condition ................................ 87

5.3.3 Calculation of the QW oscillator strength per unit area . . . . . . . 91

6 Conclusions 93

7 Appendix 109

List of Abbreviations

AFM Atomic Force Microscopy

BEP Beam Equivalent Pressure

DBR Distributed Bragg-Reflector

FWHM Full Width at Half Maximum

LD Laser Diode

LED Light Emitting Device

MBE Molecular Beam Epitaxy

ML Monolayer

MQW MultiQuantum Well

MSEO Modified Single Effective Oscillator

PL Photoluminescence

QD Quantum Dot

QW Quantum Well

RHEED Reflection High Energy Electron Diffraction

RMS Root Mean Square

SL Superlattice

TE Transversal-Electric

TEM Transmission Electron Microscopy

TM Transversal-Magnetic

UHV Ultra High Vacuum

VCSEL Vertical Cavity Surface Emitting Laser

Contents

1 Introduction

Semiconductor QWs as part of microcavities allow to study the interconversion between

excitons and photons, which in the strong coupling regime is manifested as a Rabi-splitting

into two resonance features of the optical spectrum. The coupling between excitons and

photons generates new quasi-particles known as polaritons. These particles have some

extraordinary properties, which open up the possibility to develop new types of efficient

light emitters or quantum processors [Ge1998, Ba2002].

After the first observation of polariton splitting in a Fabry-Perot microcavity (Weisbuch

et. al. 1992) [We1992, We2000], the strong coupling regime of QW excitons was studied by

various spectroscopy methods in III-V semiconductor microcavities with epitaxial grown

DBRs [Kh1999, Bl1998] based on GaAs, AlAs and their alloys. Because the exciton

binding energy as well as the oscillator strength of GaAs and its related materials is very

small, most of the studies according to polariton effects are limited to the low temperature

region up to 100 K. Some efforts have been performed to achieve strong coupling in III-

V microcavities at room temperature in complicated microcavities containing up to 30

GaAs/(Al,Ga)As QWs, yielding a Rabi-splitting about 8 meV [Di2001].

For room temperature applications the Rabi-splitting must exceed the thermal energy

of about 25 meV. Since the Rabi-splitting depends on the oscillator strength and the

exciton binding energy, this condition is excellently satisfied with II-VI semiconductor

based quantum structures.

Andr´e et al. [An2000] investigated CdTe MQW structures enclosed in (Cd,Mg)Te/(Cd,Mn)Te

semiconductor Bragg-mirrors and achieved Rabi-splitting energies between 12 meV and

1 Introduction

30 meV in microcavities containing up to 16 CdTe QWs. Kelkar et al. [Ke1995] re-

ported a Rabi-splitting of 17.5 meV at 70 K and 10 meV at 175 K in a (Zn,Mg)(S,Se)

microcavity with three (Zn,Cd)Se QWs as the resonant medium and dielectric SiO2/TiO2

Bragg-mirrors.

The discussion about polariton based applications in future devices such as a polariton

laser, has been opened up recently by Saba et al. [Sa2002, Ci2002]. They have measured

parametric polariton amplification up to 220 K in a CdTe based MQW microcavity. The

experimental observations demonstrate that the polariton amplification cut-off tempera-

ture is directly related to the exciton binding energy. ZnSe and CdSe combine a large

Rabi-splitting energy and a high exciton binding energy up to 40 meV. Therefore ZnSe

and (Zn,Cd)Se compounds are particularly suited for microcavity applications such as

polariton amplification devices.

This work is focused on the development of semiconductor microcavity structures with

ZnSe/(Zn,Cd)Se QWs as the active layer, which are enclosed in distributed Bragg-reflectors

of ZnS/YF3or ZnSe/YF3, respectively. Furthermore such microcavities are investigated

with various spectroscopic methods to demonstrate the existence of the strong coupling

between photon and exciton at room temperature.

In the first chapter a short introduction is given about the theoretical background of the

exciton-photon coupling in atomic as well as semiconductor microcavity structures.

Chapter 2 describes the development of distributed Bragg-reflectors of ZnS/YF3and

ZnSe/YF3, respectively, which are grown by thermal evaporation. Because the maximum

reflectivity in these Bragg-mirrors depends mainly on interface quality, surface roughness

and refractive index difference, detailed investigations are performed by reflectivity as well

as AFM measurements.

The main issue in Chapter 3 is the investigation of different MBE grown ZnSe/(Zn,Cd)Se

MQW- and SL structures. I describe the efforts to maximize the number of QWs, which

might be enclosed in a small resonator length (about 200 nm) before a significant degen-

eration of structural and optical properties of the active layer is observed.

In the last chapter of this thesis, both technologies, the Bragg-mirrors and the quantum

structures are combined to complete microcavity structures, to investigate strong coupling

effects at room temperature. The anticrossing behavior of the polariton dispersion is

measured with different spectroscopic methods of microcavity detuning and a large room

temperature Rabi-splitting energy of 40 meV is obtained in a four-fold ZnSe/(Zn,Cd)Se

MQW microcavity.

1 Introduction

2 Exciton-photon interaction in

semiconductor microcavity structures

2.1 Excitons in strongly coupled microcavities

One of the most important and popular set of applications for semiconductors such as

light emitting devices (LEDs and LDs) or photon detectors are based on the interaction

of light and matter. When light is incident on a bulk direct band-gap semiconductor, it

will be absorbed, if the photon energy is larger then the bandgap energy. Due to this

absorption process free electron-hole pairs are generated.

Electron-hole pairs, which are spatially close together, the so-called excitons, are bound

by the Coulomb force. The energy and momentum of these quasi-particles in the semi-

conductor material is conserved. Excitons can be scattered by collision with photons,

crystal defects, free carriers or other excitons. The scattering enhances the recombination

of electron-hole pairs, which leads to spontaneous photon emission or nonradiative recom-

bination. In this case the excitons interact with a continuum of electromagnetic modes

and can emit light in any direction and frequency within the emission linewidth, which

results in an exponential decay of the generated excitons [Sa1974, Al1975].

The modification of the electromagnetic environment of a semiconductor exciton increases

the possibility of resonant photon-exciton coupling. This can be realized, if the semicon-

ductor is placed into a microcavity as illustrated in Fig. (2.1). In Fig. (2.1 a.) photon

and exciton are uncoupled. By spontaneous exciton recombination a photon is emitted

2 Exciton-photon interaction in semiconductor microcavity structures

Figure 2.1: Cross-section drawing of a semiconductor microcavity structure. a.) The cavity photon with

momentum kCand linewidth γCas well as the exciton with kXand γXrespectively, are uncoupled. By

spontaneous exciton recombination, a photon is generated and emitted without a preferred direction. b.)

The electromagnetic environment of an exciton is modified by a resonantly incident photon. The electric

field of the photon manipulates the exciton dipole momentum qXand strong coupling occurs. The energy

is periodically transferred between photon and exciton. This is manifested in the characteristic Rabi-

oscillation frequency. The linear combinations of the photonic and excitonic eigenstates are the polariton

modes.

without any preferred direction. The homogenous linewidth γXof the exciton transition

results from the decay time of the generated radiation field. An uncoupled photon is con-

fined in the microcavity structure and its lifetime is related to the homogenous linewidth

γCof the resonator.

If the incident light is resonant with the exciton transition, the electric field of the photon

interacts with the exciton dipole momentum qXand creates an exciton polarization, which

has the same ωand kCas the incident light [No1999]. Hence it follows that the coupled

photon-exciton with a given ωand konly can interact with a single mode of the radiation

2.1 Excitons in strongly coupled microcavities

field and its radiation dynamics is different from light emission of non resonant excitons.

This effect is illustrated in Fig. (2.1 b.). During the interaction time of light field and

exciton the energy oscillates between the given excitonic and photonic eigenstate: A

photon is absorbed by the semiconductor and creates an exciton, which re-emits a photon

and so on. The energy exchange of the coupled photon-exciton leads to two stationary

states with the frequencies ωup and ωlp, the so-called normal modes, which are the linear

combinations of electromagnetic and exciton polarization modes (i.e. the polariton modes)

[Ho1958].

2.1.1 Strong and weak coupling in atomic cavities

In a microcavity with a single atom inside the occurrence of strong or weak coupling is

directly related to the strength of the dipole interaction energy gas well as the linewidth

of photon γCand exciton γ0, respectively. For the case that the dipole interaction

g < (γC, γ0), the dipole interaction time to manifest the strong coupling is larger

than the decay time of the exciton as well as the lifetime of the photon in the resonator.

Hence it follows that the microcavity operates in the weak coupling regime. Due to the

modification of the mode density by the cavity confinement, the spontaneous emission is

enhanced, if photonic mode and excitonic mode are in resonance and inhibited in the non

resonant case [Be1994, Yo1995].

If g > (γC, γ0), the dipole interaction time is smaller than the lifetime of the photon

and the decay time of the exciton. Then the microcavity operates in the strong coupling

regime featuring the two polariton states with the average linewidth γ±=1

2(γC+γ0)

[Ca1989]. If γC< γ0, the polariton linewidth γ±is smaller than the free space linewidth

of the atomic transition and the spontaneous emission is inhibited. On the other hand

γC> γ0leads to an enhancement of the spontaneous emission.

The Rabi-oscillation frequency ΩRabi, which is proportional to the energy exchange fre-

quency during the strong coupling between the photonic and excitonic quantum state, is

directly related to the coupling strength ~gand the frequency splitting between the upper

2 Exciton-photon interaction in semiconductor microcavity structures

and lower polariton state ωup and ωlp,

ΩRabi =ωup −ωlp.(2.1)

The probability of re-absorption of the photon by the atom in the strong coupling regime

depends on the number of atoms in the ground state (i.e. the oscillator density). For

Natoms in the microcavity the oscillator strength of the exciton transition is enhanced

by a factor of N. Hence the Rabi-splitting energy ΩRabi will increase by a factor of√N,

which means that the strong coupling regime is achieved more easier with a large oscillator

density in the microcavity [No1999]. Due to their increased oscillator density, QW exci-

tons are particularly suited for the investigation of the strong coupling in semiconductor

microcavities [We1992].

2.1.2 Strong coupling in quantum well microcavities

Due to the large density of oscillators in microcavities with QWs, one photon does not

interact any more with one and the same exciton. Therefore the coupling strength is char-

acterized by the QW oscillator strength per unit area f

S, instead of the dipole interaction

g. Hence it follows that the Rabi-splitting energy ~ΩRabi is given by

~ΩRabi =~s8cNQW γX

nCLeff

,(2.2)

where Leff =Lc+LDBRs is the effective microcavity length including the penetration

depth of the electromagnetic wavefunction into the Bragg-mirrors and NQW is the number

of QWs [Ya2002]. The term

γX=e2

4ε0M∗

XcnC

S(2.3)

describes the decay rate of excitons in the QW ground state (i.e. lifetime broadening,

similar to γ0in atomic cavities) and is proportional to the oscillator strength per unit

area [Sk1998].

2.1 Excitons in strongly coupled microcavities

2.1.3 Polariton dispersion – effective mass model for a coupled os-

cillator

In a microcavity containing a single QW, the photon-exciton coupling is described by

the 2 ×2 matrix Hamiltonian Eq. (2.4), where V=1

2~ΩRabi is the coupling potential

and EX(k) and EC(k) are the dispersion relations (based on the effective mass model) of

uncoupled exciton and photon, respectively.

H=



EX(k)V

V EC(k)

(2.4)

The eigenvalues of the coupled oscillators are obtained by diagonalization of the Hamil-

tonian

EP±=EX(k) + EC(k)

2±1

2p(~ΩRabi)2+ (EX(k)−EC(k))2.(2.5)

The in-plane exciton dispersion EX(k) is described by the effective mass model

EX(k) = E0+

~2k2

2M∗

,(2.6)

where M∗

Xis the reduced effective mass of the exciton and E0the QW ground state

transition energy. The kz=2πN

LCof the photonic modes are confined by the resonator

length LCand Nis the mode number. Therefore the photon dispersion in the cavity is

given by

EC(k||) = ~c

nCs2πN

LC2

+k2

||.(2.7)

The confined photon may also be associated to an effective mass MC=hnC

cLC, which is

about five orders of magnitude smaller than the exciton effective mass M∗

X[An2000]. The

dispersions of pure photon, pure exciton and the upper and lower polariton are shown

in Fig. (2.2). They are calculated for a typical λ-sized (i.e. nC∗LC=λ) microcavity

structure with one QW in its center and an emission wavelength of λ= 517 nm (i.e.

E0= 2.4 eV).

2 Exciton-photon interaction in semiconductor microcavity structures

Figure 2.2: Calculated energy dispersion of photon (dotted grey curve), exciton (dashed black curve)

and coupled polaritons (full curves in black and grey) as a function of the in-plane cavity wavevector

k||. The pure exciton dispersion is almost flat compared to the pure photon dispersion. Due to the

strong coupling between photon and exciton, the degeneracy of both states is lifted near the resonance

at k|| = 0. This results in a the splitting between the two polariton branches. The minimum splitting,

which is the Rabi-splitting energy ~ΩRabi, is obtained in the resonance.

The exciton dispersion is nearly flat on the scale of the photon dispersion. The resonance

condition of the photonic and excitonic mode is fulfilled at k|| = 0. For large k|| the upper

polariton is almost ”excitonlike” and the lower polariton ”photonlike”, respectively. Near

the resonance, the degeneracy of uncoupled photon and exciton is lifted due to the strong

coupling interaction potential Vand the anticrossing behavior of the polariton dispersion

relation is observed. In the resonance condition the splitting between the polariton modes

is minimal, yielding the Rabi-splitting energy ~ΩRabi.

2.2 Simulation of the optical response of a microcavity

2.2.1 The Transfer-Matrix model for a single layer

The optical response of a semiconductor microcavity structure such as reflectivity, absorp-

tion and transmission can be calculated with the Transfer-Matrix model, which is based

on the boundary conditions of the Maxwell equations. The behavior of an incident plane

electromagnetic wave on the boundary between two different materials is illustrated in

Fig. (2.3a.) [He1979].

Figure 2.3: a.) An incident plane electromagnetic wave with the wavevector kin is partially reflected and

transmitted at the boundaries of a semiconductor material and its environment with the characteristic

refractive indices n1and n0, respectively. During the propagation of the transmitted wave kt1through

the material with the length d1, the electric field gets a phase shift δ, which depends on the transmission

angle a1and the refractive index n1.b.) For variation of α, the angle of incidence, the electric field of

the TM-mode is modified, but the electric field of the TE-mode does not change with α.

The plane wave with the wavevector kin is partially reflected and transmitted at the

boundaries of a semiconductor with the complex refractive index dispersion n1and its

surrounding medium n0. The transversal component of the electric field must be steady

on both sides of the boundaries. During the propagation of the transmitted part of

the incident wave kt1through the medium, the electric field is phase shifted by δ=

kt1n1d1cos(α1). This shift depends on the length d1and the refractive index n1of the

2 Exciton-photon interaction in semiconductor microcavity structures

material as well as the transmission angle α1. Since the transmission angle is related to

the angle of incidence by the refractive law, it is possible to modify the phase shift in the

material by variation of the angle of incidence. As shown in Fig. (2.3b.) the TE- and

TM-polarization must be treated separately, due to their different symmetry with respect

to the incidence angle. The electric field of the TM-mode is angular dependent, while the

electric field of the TE-mode does not change with α.

The electromagnetic field of the transmitted and reflected wave is directly linked to the

electromagnetic field of the incident wave by the 2 ×2 Transfer-Matrix M[Pa2000]

Ein

Hin=M×Et2

Ht2.(2.8)

For the single layer in Fig. (2.3) the characteristic matrix Mfor TE and TM-polarization

are defined as

MTE =



cos(δ)−ı

γ1

sin(δ)

cos(α1)

−ıγ1sin(δ)cos(α1)cos(δ)

,(2.9)

MTM =



cos(δ)−ı

γ1sin(δ)cos(α1)

−ıγ1sin(δ)

cos(α1)cos(δ)

,(2.10)

where γ1=qε0

µ0n1describes the dielectric properties of the layer material.

For a typical microcavity structure including Bragg-mirrors and the active layers, each

layer is assigned to a characteristic matrix Miand the optical response of the whole

system is obtained by multiplying the single layer matrices according to their structural

sequence.

2.2.2 Calculation of the complex refractive index dispersion

The accuracy of the Transfer-Matrix model to describe multilayer structures depends

appreciable on the precise knowledge of the complex refractive index dispersion of the

layer materials. All microcavity structures studied in this thesis are based on the materials

ZnSe, ZnS and YF3for the dielectric mirrors as well as ZnSe/(Zn,Cd)Se QWs in the active

2.2 Simulation of the optical response of a microcavity

layer and GaAs as the substrate material, respectively. For the II-VI and III-V compounds

the modified single effective oscillator (MSEO) method [Af1974] is appropriate for the

calculation of the refractive index dispersion below the semiconductor bandgap.

The absorption near the bandgap is not negligible any more, since the typical emission

energy of the ZnSe/(Zn,Cd)Se microcavities is between 2.2 eV and 2.6 eV (i.e. 564 nm

and 477 nm). Therefore the MSEO method with excitons, which considers an additional

broadening effect γX, is more appropriate [Ta1997]. In this formulation, the imaginary

parts of the complex dielectric constant of the materials are approximated by a delta

function with strength 1

2πEdat the energy E0[We1971].

Figure 2.4: Calculated complex dielectric constant of the materials ZnS (grey curves) and ZnSe (black

curves) separated in real (left side) and imaginary (right side) parts, respectively. Left side: Near the

bandgap of the materials the exciton generation is taken into account by a Lorentzian lineshape with the

exciton broadening γX.Right side: The extinction coefficient is modelled by a Delta function. Below

the bandgap the absorption of both materials is negligible. The parameters of the complex dielectric

constant of the materials are summarized in Tab. (2.1).

Fig. (2.4) shows the calculated refractive index dispersion of ZnS and ZnSe separated

into real (left side) and imaginary part (right side). Near the bandgap of the materials

the exciton contribution is considered by a Lorentzian lineshape in the real part of the

2 Exciton-photon interaction in semiconductor microcavity structures

dielectric function. The absorption also increases near the bandgap. Below the bandgap

the extinction coefficient of both materials is smaller than one percent.

material Eg(eV) E0(eV) Ed(eV) γX(meV)

ZnSe 2.67(1) 5.54(2) 27.0(2) 14(1)

ZnS 3.54(1) 6.36(2) 26.1(2) 70(1)

GaAs 1.424(1) 3.55(2) 33.5(2) 35(2)

Table 2.1: Parameters for the calculation of the complex refractive index dispersions of ZnSe, ZnS and

GaAs with the modified MSEO method. References: (1)experimental fits, this work and (2)[We1971].

The parameters for the calculation of the dielectric constant of all the materials, which

are used in this thesis, are summarized in Tab. (2.1). The material YF3has a large

bandgap energy and therefore no strong dispersion. The refractive index dispersion is

approximated by

n(Y F3) = 1.417 ·exp 17

λ−206(2.11)

and the extinction coefficient is neglected [Pa2000].

For the calculation of the dielectric constant of the (Zn,Cd)Se QWs, the confined exciton

contribution is included with a standard Lorentzian oscillator form

(ω) = B+4πfoszω2

ω2

0−ω2−ıγXω,(2.12)

where fosz is the oscillator strength and γXis the inhomogeneous broadening of the QW

transition [Tr1996]. ω0is the transition frequency and the background refractive index of

the QW material is included by B. Due to the fact that the QW dispersion is integrated

into the Transfer-Matrix model, the light matter interaction between photon and exciton

is considered. Hence it is possible to calculate reflectivity, transmission and absorption of

a microcavity structure with the Transfer-Matrix model.

2.3 Resonance tuning of microcavities to obtain the polariton dispersion

2.3 Resonance tuning of microcavities to obtain the po-

lariton dispersion

For measurement and calculation of the anticrossing of the polariton dispersion relation in

microcavities it is necessary to modify the resonance condition between the photonic and

excitonic mode. Actually three methods may be used: Angular detuning, cavity length

gradient detuning and temperature detuning, respectively.

2.3.1 Angular dependent microcavity detuning

In angular resolved reflectivity measurements the angle of incidence αin and the angle of

detection (αd=−αin) is varied from normal incidence. Due to the fact that the polariton

dispersion Eq. (2.5-2.7) is directly related to k||, a standing wave with a determined in-

plane wavevector is excited in the microcavity. With increasing αin the cavity mode is

shifted to higher energy, due to the additional k|| and can be tuned through the resonance

with the QW transition. In the case of the angular detuning a polariton dispersion as

shown in Fig. (2.2) is measured.

2.3.2 Microcavity detuning with a cavity length gradient

The energy of the photonic mode in the microcavity is tunable as a function of the spot

position on the sample by including a thickness gradient during growth of the active

layer. With the method of MBE growth such a wedged active layer is obtained with a

special geometric construction of the vacuum chamber. The most important point is the

alignment of the evaporation cells with respect to the sample position.

Fig. (2.5) depicts the geometric position of the zinc and selenium evaporation cell as well

as the sample position in our MBE system. The cells are aligned radialsymmetric right

opposite to the sample. Due to the fact that the sample is centered in the chamber,

the molecular beams of each material are tilted by the same angle αwith respect to the

substrate normal y.

2 Exciton-photon interaction in semiconductor microcavity structures

Figure 2.5: Alignment of the evaporation cells in our MBE chamber to implement growth gradient in the

active layer. Due to the angular tilt αof the zinc and selenium cell openings with respect to the substrate

normal y, the molecular beam path length of both materials varies by ∆z=y·sin(α). Therefore a

molecular flux gradient along the y-axis of the sample is obtained.

This geometry results in a variation of the molecular beam path length ∆z=y∗sin(α).

The zinc and selenium cell are parallel to the y-axis of the sample. Therefore a maximum

concentration gradient of zinc and selenium atoms on the sample surface is obtained and

yields a growth gradient in the active layer in direction of the y-axis [Ar2003, Ba2003].

2.3.3 Temperature detuning in microcavity structures

The third possibility to modify the resonance conditions in microcavities is the tempera-

ture detuning as illustrated schematically in Fig. (2.6). The QW transition energy (dotted

line) is shifted to lower energy with increasing temperature, due to the temperature de-

pendence of the bandgap of barrier and QW material, respectively. The spot position on

the sample and therefore the cavity mode energy (dashed line) is constant.

2.3 Resonance tuning of microcavities to obtain the polariton dispersion

Figure 2.6: Principle of temperature detuning of a microcavity. The cavity mode energy (dashed line) is

constant, while the QW transition energy (dotted line) shifts to lower energy with increasing temperature.

From the PL spectra at certain temperatures, indicated by the grey curves, the temperature dependent

polariton dispersion (black curves) are obtained.

The upper and lower polariton dispersion curves, Fig. (2.6), black curves, are obtained

by measuring reflectivity or luminescence of the microcavity structure in a small temper-

ature range near the resonance between photonic and excitonic mode. The grey curves,

indicated in Fig. (2.6), schematically depict various PL spectra. The temperature de-

pendence of the cavity mode is neglected and the QW transition energy shift is linear

approximated. These assumptions are sufficiently fulfilled in ZnSe/(Zn,Cd)Se QWs in a

temperature range between 250 K and 350 K.

2 Exciton-photon interaction in semiconductor microcavity structures

3 Distributed Bragg-mirrors for II-VI

microcavities

An important issue for the design of microcavity structures is the optimization of high

reflectivity Bragg-mirrors for the photon confinement. In typical III-V structures based

on GaAs, MBE grown AlAs/GaAs DBRs [Ba1995] are particular suited on account of

lattice matched growth on GaAs as well as a large refractive index difference of both

materials in the near infrared spectral range.

For ZnSe based microcavity structures operating in the green spectral range, several II-

VI materials and their alloys such as (Zn,Mg)(S,Se)/Zn(S,Se) [Uu1995] or ZnSe/MgS

[Ta2000] are used as semiconductor DBRs. Despite the high structural quality of these

mirrors, the refractive index difference is very small and leads to large stack numbers

for the achievement of high reflectivity. Alternatively a large family of polycrystalline

dielectric DBRs such as SiO2/TiO2(∆n≈1.0) [Ke1995] have been recently used.

In this work dielectric DBRs of ZnS/YF3(∆n≈1.0) and ZnSe/YF3(∆n≈1.3) are

optimized for the blue and green spectral range. Additionally to the large refractive index

difference of these materials, the II-VI compounds ZnS and ZnSe are closely related to

the cavity material with respect to the structural properties (i.e. thermal extension etc.).

Furthermore YF3is a chemical stable anorganic compound with a small optical dispersion

and a negligible absorption in the visible spectral range.

3 Distributed Bragg-mirrors for II-VI microcavities

3.1 Optical properties of ZnS/YF3dielectric Bragg-mirrors

Several ZnS/YF3DBRs with four up to ten stack periods are grown on (001)-GaAs

substrate by thermal evaporation. The reflectivity spectra are measured with a standard

reflection measurement setup.

Figure 3.1: Reflectivity of the eight-fold stack of ZnS/YF3Bragg-mirror DBR-048 (dots) as a function of

the photon energy. The theoretical reflectivity of the mirror (full curve) is calculated with the Transfer-

Matrix model. The Bragg-wavelength λBand the refractive indices of both materials are used as simu-

lation parameters. The experimental maximum reflectivity is about Rmax. = 0.95 at EB= 2.638 eV

(i.e. λB= 470 nm).

Fig. (3.1) shows the reflectivity spectrum (dots) of an eight-fold stack of ZnS/YF3DBRs

on GaAs (sample DBR-048). The stop-band of the Bragg-mirror is clearly resolved be-

tween 2.4 eV and 2.9 eV, yielding a maximal reflectivity about 0.950±0.005 in its center.

The experimental data are fitted with the Transfer-Matrix model (full curves) to obtain

3.1 Optical properties of ZnS/YF3dielectric Bragg-mirrors

the refractive index difference ∆nbetween ZnS and YF3and the stop-band centrum (i.e.

the Bragg-wavelength λB).

In the stop-band region the simulated curve is in good agreement with the experimental

data and the energetic position of the interference fringes are also matched. However, in

the low energy region between 1.6 eV and 2.2 eV, the measured spectrum has a signifi-

cantly larger reflectivity than it is calculated in the simulation. The reflectivity depends

especially in that region mainly on the refractive index and absorption coefficient of the

GaAs substrate, but does not influence the important mirror parameters.

Figure 3.2: Reflectivity of the ten-fold stack Bragg-mirror DBR-045 (dots) and the calculated data

(full curve) as a function of the photon energy. The experimental maximum reflectivity is about

Rmax. = 0.924 at EB= 2.725 eV (i.e. λB= 470 nm).

In Fig. (3.2) the reflectivity spectrum of the sample DBR-045 (dots), a ten-fold stack

of DBRs, is illustrated. The full curve denotes the calculated data obtained by the

3 Distributed Bragg-mirrors for II-VI microcavities

Transfer-Matrix simulation. The stop-band is shifted to higher energies compared to

that of the spectrum of the eight-fold stack, but has a similar size and shape. The

experimental maximum reflectivity of about 0.924±0.005 is significantly smaller compared

to the reflectivity of the eight-fold stack. This fact reveals that the reflectivity of ZnS/YF3

DBRs is limited by additional structural parameters and not only by the number of

periods. A more detailed discussion of this observation is given in Sec. (3.1.1).

The fitting parameters, which are used for the Transfer-Matrix simulation, are summarized

in Tab. (3.1). The six-, eight-, and ten-fold stack are deposited by thermal evaporation.

In case of these mirrors the refractive index difference ∆n(λB) between ZnS and YF3is

significantly smaller than the theoretical value.

The five-fold stack of ZnS/YF3DBRs (see also Fig. (3.3)) is grown by electron beam

assisted evaporation. The maximum reflectivity as well as the refractive index difference

is in excellent agreement with the theoretical data. The lower refractive index difference in

periods Rmax. λB(nm) ∆n(λB)Rtheo. ∆ntheo.

6 0.930 487 0.58 0.990 0.97

8 0.950 470 0.64 0.996 1.03

10 0.924 455 0.55 0.998 1.05

5 0.979 508 0.96 0.987 1.01

Table 3.1: Experimental (left side) and theoretical (right side) parameters of different ZnS/YF3DBRs.

The six-, eight- and ten-fold stacks are deposited by thermal evaporation of ZnS and YF3, the five-fold

stack is grown by electron beam assisted evaporation of the materials.

the six- eight- and ten-fold stack of Bragg-mirrors indicates an interdiffusion between ZnS

and YF3during the thermal evaporation process. This effect is amplified with increasing

stack periods and growth time. Furthermore a dissociation of the binary ZnS and a

change in the composition during evaporation has to be considered, due to the high vapor

pressure of sulphur.

The measured and the simulated reflectivity spectrum of the five-fold stack are illustrated

in Fig. (3.3). The stop-band is nearly rectangular and significantly larger compared to

3.1 Optical properties of ZnS/YF3dielectric Bragg-mirrors

Figure 3.3: Reflectivity of a five-fold stack of ZnS/YF3Bragg-mirrors (dots) grown by electron beam

assisted evaporation. The experimental maximum reflectivity is about Rmax. = 0.979 at EB= 2.441 eV

(i.e. λB= 508 nm). The stop-band size is larger than that of the eight- and ten-fold stack, revealing

better interface quality and reduced interdiffusion between ZnS and YF3.

that of the eight- and ten-fold stack. Furthermore the five-fold stack has an increased

maximum reflectivity of about Rmax. = 0.979 ±0.005. This result reveals that no

significant interdiffusion at the interfaces between ZnS and YF3or composition changes

in the materials are observed, if the Bragg-mirrors are grown by electron beam assisted

evaporation.

3.1.1 Structural properties of ZnS/YF3DBRs

The interface roughness and the surface structure of the DBRs is an important factor,

which determines the Bragg-mirror quality. If the surface roughness increases with the

3 Distributed Bragg-mirrors for II-VI microcavities

number of periods, the reflectivity is limited, due to scattering loss at the interfaces.

Figure 3.4: AFM picture of a 20 µm×20 µm scan of the ZnS surface of sample DBR-034, a six-fold

stack of ZnS/YF3DBRs. The figure illustrates a clear polycrystalline surface. The RMS roughness in

the black frame is about 16±1 nm.

Fig. (3.4) shows the surface morphology of the top ZnS layer of a six-fold stack of ZnS/YF3

Bragg-mirrors. The surface is measured in a 20 µm×20 µm AFM scan directly after

mirror deposition, to avoid a significant surface modification, due to oxidation in atmo-

spheric environment. The picture illustrates a clear polycrystalline surface with cubic ZnS

microcrystals of a size in the order of 400 nm. The RMS surface roughness, calculated in

the small rectangular area, is about 16±1 nm.

Fig. (3.5) shows the RMS surface roughness (dots) of different DBRs with four up to ten

pairs of ZnS/YF3mirrors. The surface roughness increases with the number of periods

and yields that the mirror interface quality is reduced in larger stacks. The full curve

describes a linear dependence between surface roughness and number of periods.

Hence the mode leakage (i.e. increased scattering loss of photons) limits the maximum

3.1 Optical properties of ZnS/YF3dielectric Bragg-mirrors

Figure 3.5: RMS surface roughness of several ZnS/YF3DBRs (dots) as a function of the number of

stacks. The line graph depicts a linear dependence between surface roughness and number of periods.

reflectivity of ZnS/YF3Bragg-mirrors with a large number of periods. This fact is in

good agreement with the reflectivity measurements in Sec. (3.1) and explains that the

ten-fold stack DBR has a lower reflectivity than the eight-fold stack. Fig. (3.6) shows the

experimental maximum reflectivity Rexp.(N) (dots) as well as the calculated maximum of

reflectivity Rtheo.(N) (grey curve) as a function of the number of Bragg-mirror periods N

and assuming perfect planar interfaces.

The deviation between experimental and calculated data increases with the number of

periods, due to the scattering loss at imperfect interfaces. The black curve depicts the

calculated maximum reflectivity considering a scattering factor

σ=Rexp.

Rtheor.

= 1 −0.007 ∗N, (3.1)

3 Distributed Bragg-mirrors for II-VI microcavities

Figure 3.6: Maximum reflectivity of several ZnS/YF3DBRs as a function of the number of periods (dots).

The full curves depict the calculated maximum reflectivity for planar interfaces without any scattering

loss (grey curve) and for imperfect interfaces (black curve) including a scattering factor σ, which is

proportional to the surface roughness, respectively.

which is proportional to the number of periods. This assumption is in good agreement

with the experimental data and confirms the model that the maximum reflectivity of

ZnS/YF3DBRs, grown by thermal evaporation of the binary materials, is limited by

the interface roughness. The black dotted lines in Fig. (3.6) show that six-, seven- and

eight-fold stacks of ZnS/YF3Bragg-mirrors have an adequately reflectivity for microcavity

applications.

3.2 ZnSe and YF3Bragg-Mirrors

Due to the fact that the reflectivity of ZnS/YF3DBRs is limited by structural properties

such as interface roughness and sulphur interdiffusion, alternatively Bragg-mirrors of the

materials ZnSe and YF3are characterized in this section.

As shown in Fig. (2.4), the most important advantage of ZnSe/YF3DBRs is the larger

refractive index difference compared to that of ZnS/YF3mirrors. Hence it follows that

a larger reflectivity can be achieved with a smaller number of periods in the ZnSe/YF3

mirrors. Furthermore the material sulphur is exchanged by selenium. This material has a

vapor pressure, which is some orders of magnitude smaller compared to that of sulphur.

Therefore the composition of the ZnSe vapor pressure is stabilized. Additionally to this

interdiffusion between ZnSe and YF3during the mirror deposition is reduced.

The disadvantage of ZnSe/YF3mirrors is the increased absorption of light in ZnSe above

its bandgap energy of about 2.72 eV. This can limit the maximum reflectivity in the blue

spectral region.

3.2.1 Optical properties of ZnSe/YF3DBRs

For further improvement of the microcavity structures, several ZnSe/YF3DBRs are grown

by thermal evaporation and also with processing parameters, which are comparable to

that of the ZnS/YF3Bragg-mirrors. Fig. (3.7) depicts the measured and the theoretical

reflectivity spectrum of the ten-fold stack of ZnSe/YF3Bragg-mirrors (DBR-219) grown

on GaAs substrate. The experimental data show clear interference fringes and a well

determined and rectangular shaped stop-band between 2.15 eV and 3.15 eV, yielding a

good interface quality in the mirrors.

A clear reduction of the reflectivity which stems from the ZnSe bandgap absorption is

measured above 2.65 eV. The maximum reflectivity of the ZnSe/YF3ten-fold stack is

about Rmax. = 0.999±0.005 at EB= 2.610 eV (i.e. λB= 475 nm). The reflectivity is

significantly larger compared to that of the ZnS/YF3ten-fold stack mirror in Fig. (3.2).

3 Distributed Bragg-mirrors for II-VI microcavities

Figure 3.7: Reflectivity spectrum of the ten-fold stack of ZnSe/YF3Bragg-mirrors DBR-219 (dots) as well

as the approved Transfer-Matrix calculation (full curve). The refractive indices of both materials and the

Bragg-wavelength are used as simulation parameters. The maximum reflectivity is about Rmax. = 0.999

at EB= 2.610 eV (i.e. λB= 475 nm). At energies larger than 2.65 eV the reflectivity is significantly

reduced due to the ZnSe bandgap absorption.

The calculated reflectivity curve in Fig. (3.7) is in good agreement with the experimental

data between 1.6 eV and 3.2 eV.

Despite that, the experimental measured reflectivity is above 3.2 eV larger than the cal-

culated values. For the Transfer-Matrix simulation of the measured spectrum in Fig. (3.7)

the refractive index dispersion of monocrystalline ZnSe has been used, but in polycrys-

talline ZnSe the absorption might be reduced.

The reflectivity spectrum of the five-fold stack of ZnSe/YF3DBRs (sample DBR-252) is

illustrated in Fig. (3.8). The experimental measured reflectivity about Rmax. = 0.993 at

3.2 ZnSe and YF3Bragg-Mirrors

Figure 3.8: Reflectivity spectrum of the five-fold stack of ZnSe/YF3Bragg-mirrors DBR-252 (dots) and

the calculated data (full curve). The experimental maximum of reflectivity is about Rmax. = 0.993 at

EB= 2.755 eV (i.e. λB= 450 nm). The ZnSe bandgap absorption has a significant influence on

the reflectivity at energies larger than 2.85 eV. This value is shifted about ten percent to larger energies

compared to that of the ten-fold stack Bragg-mirror.

EB= 2.755 eV (i.e. λB= 450 nm) is in good agreement with the simulation parameters

and just slightly smaller compared to the reflectivity of the ten-fold stack DBR-219.

The fact that size and shape of the stop-band in both spectra is quite similar indicates that

interdiffusion of the materials at the mirror interfaces is not increasing with the number of

periods as it has been observed in the ZnS/YF3DBRs. The influence of ZnSe absorption

on the reflectivity is measured in both mirrors, but in the five-fold stack the bandgap

energy is shifted about ten percent to higher energies compared to the data of DBR-219.

Since the bandgap energy is related to the crystal structure, the energy difference indicates

3 Distributed Bragg-mirrors for II-VI microcavities

a fluctuation of the structural quality of the polycrystalline ZnSe layers.

periods Rmax. λB(nm) ∆n(λB)Rtheo. ∆ntheo. Eg(ZnSe) (eV)

5 0.993 450 1.43 0.992 1.30 2.99

6 0.990 488 1.13 0.996 1.30 2.60

8 0.996 500 1.06 0.998 1.23 2.60

10 0.999 475 1.28 0.995 1.29 2.67

Table 3.2: Experimental (left side) and theoretical (right side) parameters of different ZnS/YF3DBRs.

The mirrors are deposited by thermal evaporation of ZnSe and YF3. The bandgap energy of the poly-

crystalline ZnSe layers is considered in the Transfer-Matrix simulation and responsible for the reduction

of the reflectivity in the stop-band.

The parameters used for the Transfer-Matrix simulation of the reflectivity of different

ZnSe/YF3Bragg-mirrors are summarized in Tab. (3.2). The experimental maximum

reflectivity is in every case larger than 0.99 and also in good agreement with the theoretical

data. For five up to ten Bragg-mirror periods, no reduction of the reflectivity is observed,

unlike it has been shown for ZnS/YF3DBRs in the previous section.

The refractive index difference between ZnSe and YF3differs slightly from the theoretical

values, but has no clear decreasing trend. Therefore material interdiffusion between ZnSe

and YF3does not depend on the number of mirror periods and is negligible. The bandgap

energy, which has been extracted from the ZnSe refractive index dispersion, varies between

2.6 eV and 3.0 eV. This reveals that the structural quality of polycrystalline ZnSe is very

sensitive to fluctuations of the mirror growth parameters.

Despite the increased absorption in ZnSe/YF3Bragg-mirrors near the ZnSe bandgap,

the optical properties of thermal evaporated ZnSe/YF3DBRs are significantly improved

compared to that of ZnS/YF3Bragg-mirrors.

3.2.2 Structural properties of ZnSe/YF3Bragg-mirrors

Hence the reflectivity of ZnSe/YF3DBRs is not decreasing in larger mirror stacks, the in-

terface roughness, which is related to the surface roughness, is expected to be significantly

3.2 ZnSe and YF3Bragg-Mirrors

Figure 3.9: AFM picture of a 20 µm×20 µm scan of the ZnSe surface of the ten-fold stack of ZnSe/YF3

DBRs (sample DBR-215). The RMS surface in the black square region is about 5±1 nm, which is

significantly smaller compared to the typical surface roughness of ZnS/YF3DBRs (about 10-20 nm).

smaller than the roughness of ZnS/YF3Bragg-mirrors.

Fig. (3.9) illustrates the AFM picture of a 20 µm×20 µm scan on the ZnSe surface of the

ten-fold stack of Bragg-mirrors DBR-215. The image is taken directly after the mirror

deposition to avoid a significant surface modification due to oxidation in atmospheric

environment. The surface is relatively plane and the RMS roughness is about 5±1 nm.

This value is significantly smaller compared to the RMS roughness of ZnS/YF3Bragg-

mirrors.

The RMS roughness and the maximum reflectivity of the ZnSe/YF3DBRs are summarized

in Fig. (3.10). The black dots represent the experimental maximum of reflectivity of the

different ZnSe/YF3DBRs. The full curve describes the reflectivity at λB= 480 nm as

a function of the number of periods and is calculated with the Transfer-Matrix model

assuming perfect planer interfaces.

3 Distributed Bragg-mirrors for II-VI microcavities

Figure 3.10: Reflectivity (left scale) and surface roughness (right scale) of different ZnSe/YF3DBRs as

a function of the number of periods. The black curve describes the reflectivity of the mirrors calculated

with a Bragg-wavelength of λB= 480 nm and assuming perfect planar interfaces. The experimental

results (black dots) are in excellent agreement with the theoretical data. The grey squares depict the

RMS surface roughness of the DBRs measured by AFM. The dashed grey line denotes a linear increase

of the RMS roughness with the number of stacks.

Experimental and theoretical data are in excellent agreement yielding that the inter-

faces between the materials are well determined. The reflectivity, calculated for a Bragg-

wavelength of λB= 480 nm and as a function of the number of periods, is significantly

limited to Rmax. = 0.9966, due to the ZnSe bandgap absorption.

The grey squares and the grey dashed line in Fig. (3.10) represent the RMS surface

roughness of the ZnSe/YF3DBRs as a function of the number of periods. The surface

roughness is about three times smaller compared to that of ZnS/YF3Bragg-mirrors.

Despite that, it also shows a linear increase for larger stack periods, but no decrease of

3.2 ZnSe and YF3Bragg-Mirrors

the reflectivity is observed within the accuracy of the measurement system.

In conclusion, both Bragg-mirror systems are particularly suited for deposition on ZnSe

based microcavity structures. In the green spectral region the ZnSe/YF3DBRs seem to

be superior, due to their better structural quality and larger refractive index difference.

In the blue spectral range the ZnS/YF3DBRs have the advantage of reduced absorption

in the DBRs.

3 Distributed Bragg-mirrors for II-VI microcavities

4 ZnSe and (Zn,Cd)Se quantum

structures for II-VI microcavities

This chapter is focused on the development and optimization of the active layer in the

microcavity structures. The II-VI semiconductor materials, especially ZnSe and CdSe,

have been investigated intensively to develop optoelectronic devices in the blue and green

spectral range [Ea1995]. Due to the fact that (001)-GaAs substrates are nearly lattice

matched on ZnSe, dramatic advance of the crystallographic properties of ZnSe layers has

been achieved by MBE growth.

Furthermore the exciton binding energy (about 30 meV [Gu1982]) and the oscillator

strength of bulk ZnSe is significantly larger compared to that of other II-VI based semi-

conductors such as ZnTe or CdTe. Therefore the ZnSe and CdSe based materials are

particularly suited for room-temperature applications of optoelectronic devices. First

success to develop a room temperature VCSEL structure with (Zn,Cd)Se QWs enclosed

in Zn(S,Se) barriers has been achieved in 1995 by Jeon et. al. [Je1995].

One of the most important points in this work is the adjustment of the (Zn,Cd)Se QW

emission (i.e. the excitonic mode) to the cavity mode energy (i.e. photonic mode). The

cavity mode depends on the resonator length and is manifested, after the active layer has

been enclosed in the DBRs of ZnS/YF3or ZnSe/YF3, respectively.

Due to the fact that the strong coupling between excitonic and photonic mode is es-

tablished only when the strong coupling conditions in Sec. (2.1.1) are fulfilled, another

important step is the optimization of the QW luminescence with respect to the FWHM

4 ZnSe and (Zn,Cd)Se quantum structures for II-VI microcavities

of the transition energy. Last but not least the strain status of the (Zn,Cd)Se QWs has

to be considered.

4.1 In-situ growth characterization of ZnSe/(Zn,Cd)Se

quantum structures

The most common in-situ characterization of the crystallographic properties during MBE

growth are RHEED-measurements. Due to the formation of two dimensional islands on

a plane semiconductor surface during growth, the surface roughness changes periodically.

The reflectivity is maximal, when the surface is fully covered with one monolayer of atoms

and minimal for half coverage, respectively.

From the intensity oscillations of the specular spot (i.e. the reflected primary electron

beam), the growth rate (i.e. the time, which is needed to grow one monolayer of atoms)

is obtained and opens up the possibility to determine the thickness of the layer.

The optimized growth regime is obtained under slightly selenium rich conditions [Wo2000],

hence then the metal flux will determine the growth rate. Furthermore the re-evaporation

of selenium atoms from the surface is avoided, which results in an appreciable smooth

surface.

Fig. (4.1) depicts the typical intensity oscillations of the specular spot after growth start of

ZnSe on a atomically flat GaAs surface. The intensity oscillations of the specular spot are

clearly visible and show a well defined periodicity. The damping stems from simultaneous

growth of several monolayers at the same time. The average growth rate (GR), which is

obtained from the oscillations in Fig. (4.1), is GR(ZnSe) = 0.29 ±0.01 ML·s−1.

The calculation of the ZnSe growth rate from the growth oscillations is the most important

aspect to determine the thickness of the ZnSe/(Zn,Cd)Se quantum structure and to fix the

energetic position of the cavity mode. Analyzed in the concept of growth rate calculation

from RHEED-oscillation measurements, the typical resonator length is about 200±7 nm.

Therefore it is necessary to implement an additional thickness gradient in the structure to

4.1 In-situ growth characterization of ZnSe/(Zn,Cd)Se quantum structures

Figure 4.1: Intensity oscillations of the specular spot during deposition of ZnSe on a two dimensional plane

GaAs surface. From the time difference between two minima or maxima of the intensity, respectively,

the growth rate is determined. In this case the average growth rate is GR(ZnSe) = 0.29 ±0.01 ML·s−1.

obtain a tunable cavity mode, which can be brought into resonance with the QW emission

[We1992].

The transition energy of the ZnSe/(Zn,Cd)Se QWs is determined by the cadmium mole

fraction as well as the thickness of the (Zn,Cd)Se layers. Therefore it is necessary to

monitor the atomic flux of the zinc and the cadmium atoms, which are incorporated into

the structure.

The RHEED intensity oscillations during the (Zn,Cd)Se QW growth are shown in Fig. (4.2).

Due to the lattice mismatch of (Zn,Cd)Se on GaAs (about 2.5 % for a cadmium mole frac-

tion of about 0.3), the oscillations are not as pronounced as in ZnSe and are also strongly

4 ZnSe and (Zn,Cd)Se quantum structures for II-VI microcavities

Figure 4.2: Intensity oscillations of the specular spot during deposition of a (Zn,Cd)Se QW. The oscil-

lations are strongly damped and hardly resolved. The average growth rate is GR((Zn,Cd)Se) = 0.37 ±

0.05 ML·s−1.

damped. The average (Zn,Cd)Se growth rate is GR(ZnCdSe) = 0.37 ±0.05 ML·s−1.

Therefore the thickness of the QW is obtained with an accuracy of ±0.5 nm.

Hence the growth regime of ZnSe, CdSe and their alloys is connected to selenium rich

conditions, the growth rate of the materials is limited by the amount of metal, which

is incorporated into the growing surface. Therefore the Cd mole fraction in the QWs is

obtained from the relative difference of the ZnSe and (Zn,Cd)Se growth rates

x(Cd) = GR(ZnCdSe)−GR(ZnSe)

GR(ZnCdSe),(4.1)

respectively. The dependence of the cadmium mole fraction from the Zn:Cd flux ratio

4.2 Investigation of (Zn,Cd)Se multi quantum wells

χ=BEP (Zn)

BEP (Cd)has been investigated in [Pa2000], yielding

x(Cd) = (1 + 1.15 ∗χ)−1.(4.2)

Analyzed via the concept in Eq. (4.1) and Eq. (4.2), it is possible to calculate the growth

time of the QWs, to achieve a well defined thickness and cadmium mole fraction, respec-

tively.

4.2 Investigation of (Zn,Cd)Se multi quantum wells

For the optimal achievement of exciton-photon coupling in microcavities, it is necessary

to include a maximum number of QWs into a resonator as short as possible. In this thesis

the typical optical resonator length (ZnSe barriers plus QWs) is about nC∗LC= 1 λ,

where nCand LCare the refractive index and length of the cavity, respectively, and λis

the emission wavelength of the structure.

The active medium of the resonator is represented by the (Zn,Cd)Se QWs. These are

placed near the antinodes of the standing wave in the microcavity, to achieve optimized

interaction of photons and excitons.

Fig. (4.3) depicts a cross-section drawing of the two types of MQW structures, which

are investigated in this section. The cavity length of about 192 nm (i.e. λ-resonator) is

adapted to the QW emission in the green spectral range. The active layers are represented

by (Zn,Cd)Se QWs enclosed in ZnSe barriers. The MQW structure on the left side in

Fig. (4.3) contains two QWs in the center and also two QWs near the resonator edge.

The double QW configuration on the right side in Fig. (4.3) contains only the two QWs

in the center. If both structures are placed between Bragg-mirrors, the antinodes of the

cavity mode are formed near the QWs and an optimal interaction of the photonic mode

with the QW excitons is obtained. Due to the symmetry of both structures, the electric

field distribution in each of the QWs is exactly the same.

4 ZnSe and (Zn,Cd)Se quantum structures for II-VI microcavities

Figure 4.3: Cross-section drawing of a MQW structure with four (left side) and two (right side) (Zn,Cd)Se

QWs, respectively. The resonator length of about 192 nm is adapted to the QW emission in the green

spectral range. If both structures are enclosed between Bragg-mirrors, the electric field distribution is

exactly the same in each of the QWs.

4.2.1 Structural investigations of MQWs

Besides the in-situ RHEED-measurements described in Sec. (4.1), the MQW structures

are characterized ex-situ by high resolution x-ray diffraction, to get detailed information

about the structural parameters.

For detailed characterization of the strain state of a MQW structure similar to that in

Fig. (4.3), left side, a reciprocal space map around the asymmetric GaAs-(224) Bragg-

reflex has been measured. From such a measurement the reciprocal lattice parameters qx

and qz, which are directly related to the lattice vectors in surface and growth direction,

respectively, are extracted.

The reciprocal space map of the MQW structure McS-803 is shown in Fig. (4.4). The

ZnSe Bragg-reflex is shifted to a lower qxand qzrelatively to the GaAs reflex. This

observation results from a tilt or a partial relaxation of the ZnSe. From the measurement

4.2 Investigation of (Zn,Cd)Se multi quantum wells

Figure 4.4: Reciprocal space map around the GaAs-(224) reflex of the MQW structure McS-803. The

ZnSe reflex is observed on the dark grey line, which is the relaxation line for ZnSe. From the measured

ZnSe peak position, the relaxation grade has been estimated of about 38 %. Furthermore the black line

indicates that the (Zn,Cd)Se QWs are lattice matched on ZnSe.

of a symmetric reciprocal space map of the structure a tilt has been excluded, which

denotes that the ZnSe layers are partial relaxed.

The dark grey line is the relaxation line for ZnSe. The points of intersection of the

relaxation line with the dashed as well as the dotted grey line depict the peak position

for fully strained and fully relaxed ZnSe on GaAs, respectively. From this geometric

construction the relaxation of the ZnSe barriers has been estimated of about 38 %. The

black line shows that the (Zn,Cd)Se QWs are lattice matched on ZnSe.

Hence now the strain state of the MQW structure McS-803 is well determined, the im-

4 ZnSe and (Zn,Cd)Se quantum structures for II-VI microcavities

portant structural parameters (i.e. thickness and mole fraction of the QWs as well as the

cavity length) are obtained from the measurement of a symmetrical ω−2θ−scan. Such a

scan yields information about the lattice parameters in growth direction (i.e. perpendic-

ular to the surface). With the full dynamical x-ray simulation theory [Kh2003, Pa2001]

the measured spectrum is fitted by varying the lattice parameters of the materials as well

as the thickness of (Zn,Cd)Se QWs and ZnSe barrier layers, respectively.

Figure 4.5: Symmetric ω−2θ−scan of sample McS-803 (dots) measured relative to the GaAs-(004)-reflex.

The red curve denotes the full dynamic x-ray diffraction simulation of the MQW structure. The GaAs

and ZnSe peaks are clearly resolved, but the ZnSe peak is not superimposed by Pendell¨osung fringes as

shown in the calculated data. The structural parameters of McS-803 in the small inset are obtained from

the x-ray simulation (red curve).

Fig. (4.5) depicts the ω−2θ−scan spectrum of the MQW structure McS-803 (dots)

measured relative to the GaAs-(004) Bragg-reflex as well as the simulated data (red line).

4.2 Investigation of (Zn,Cd)Se multi quantum wells

The ZnSe reflex is clearly resolved at the relative angle ∆2θ=−0.335 ◦, which is

significantly smaller than that, calculated for a fully strained ZnSe layer on GaAs (i.e.

∆2θ=−0.405 ◦). This fact is in good agreement with the results in Fig. (4.4)

and confirms the partial relaxation of the ZnSe. Furthermore no Pendell¨osung fringes

are observed in the ZnSe reflex region of the measured spectrum, yielding a reduced

crystallographic quality of the ZnSe barrier layers (i.e. dislocations at the GaAs/ZnSe

heterointerface).

From a fit of the x-ray simulation to the experimental data. the structural properties of

McS-803 are obtained. The results are illustrated in the small inset of Fig. (4.5). The

(Zn,Cd)Se reflex has been estimated at ∆2θ=−3.618 ◦, resulting in a cadmium mole

fraction of x(Cd) = 0.31±0.02. Furthermore the (Zn,Cd)Se reflex is superimposed by

Pendell¨osung fringes, from which a periodicity of 22±1 nm is obtained. The experimen-

tal results are in good agreement with the simulation data, yielding a QW thickness of

d(QW) = 7.0±0.5 nm and a ZnSe spacer layer thickness of 14.0±0.5 nm between the

central QWs. For the two ZnSe barrier layers a thickness of 68±5 nm and 73±5 nm has

been assumed in the calculation.

The crystallographic quality of the MQW structure is investigated by measuring a sym-

metric ω-scan (the so-called ”rocking-curve”) of the (004)-reflex of ZnSe. The FWHM

value of the peak in such a spectrum is directly related to the distribution of the in-plane

lattice parameter. Due to the relaxation of an epitaxial layer, the dislocation density in

the structure increases and results in a larger FWHM value of the Bragg-reflex in the

ω-scan spectrum.

Fig. (4.6) illustrates the ω-scan spectrum of the MQW structure McS-1072, which has

similar structural parameters as McS-803. The peak intensity is measured as a function

of the relative difference ∆ωfrom the (004)-ZnSe Bragg-angle. The FWHM value of the

peak is about ∆ω= 11.1±0.1 arcmin, which is significantly larger than the FWHM

value of fully strained ZnSe structures on GaAs (about 1 - 3 arcmin) [Wo1995].

The observations in Fig. (4.4) and Fig. (4.6) corroborate that the relaxation is induced

4 ZnSe and (Zn,Cd)Se quantum structures for II-VI microcavities

Figure 4.6: Symmetric ω-scan of the (004)-reflex of ZnSe in the MQW structure McS-1072 (dots). The

intensity distribution is measured as a function of the relative difference ∆ωfrom the ZnSe-(004) Bragg-

angle. The FWHM value of ∆ω= 11.1±0.1 arcmin is obtained by a Lorentzian fit (full curve) of the

measured data.

during growth of the first ZnSe barrier layer and the dislocation density in the structure

is increased. The strain energy near the GaAs/ZnSe heterointerface is unequivocally

increased, due to the placement of the first (Zn,Cd)Se QW and accelerates the relaxation

in the subsequent ZnSe barrier layer.

The influence of strain relaxation in the ZnSe barrier layers is reduced in a double QW

structure. The design of such a structure is illustrated in Fig. (4.3), right side. In contrast

to the MQW structure, the two (Zn,Cd)Se QWs near the GaAs/ZnSe interface and the

ZnSe/air boundary are skipped.

The symmetrical ω−2θ−scan of the double QW structure McS-775 (dots) measured

around the GaAs-(004)-reflex is illustrated in Fig. (4.7). The relative angular difference

4.2 Investigation of (Zn,Cd)Se multi quantum wells

Figure 4.7: Symmetric ω−2θ−scan of sample McS-775 (dots) measured relative to the GaAs-(004)-reflex.

The red curve denotes the full dynamic x-ray diffraction simulation of the double QW structure. The

GaAs and ZnSe peaks are clearly resolved and the ZnSe as well as the (Zn,Cd)Se peaks are superimposed

by Pendell¨osung fringes. The structural parameters of McS-775 in the small inset are obtained from the

x-ray simulation.

of the ZnSe- and GaAs-reflex in the x-ray spectrum is about ∆2θ=−0.411 ◦. A

comparison of this value with the data in Fig. (4.5) shows that the ZnSe layers in both

structures have a different strain state. This observation has been analyzed via the concept

of asymmetric reciprocal space maps in various similar double QW structures, yielding

that the ZnSe layers as well as (Zn,Cd)Se QWs are fully strained on the GaAs substrate.

The angular position of the (Zn,Cd)Se reflex is calculated at 2θ=−3.325 ◦from

the simulation parameters. Both, the ZnSe and the (Zn,Cd)Se peaks are superimposed

by significant Pendell¨osung fringes, yielding the improved interface quality between the

4 ZnSe and (Zn,Cd)Se quantum structures for II-VI microcavities

materials in the fully strained double QW structure. The results, which are summarized

in the small inset in Fig. (4.7), are obtained by simulation (red curve) of the measured

x-ray spectrum (dots). The thickness of the QWs is about 5.8±0.5 nm with a cadmium

mole fraction of x(Cd) = 0.32±0.02. The thickness of the ZnSe barriers is about 81±2 nm

with a spacer layer of 12.5±0.5 nm between the central QWs.

Figure 4.8: Symmetric ω-scan of the ZnSe-(004)-reflex of the double QW structure McS-775 (dots). The

intensity distribution is measured as a function of the relative difference ∆ωfrom the ZnSe-(004) Bragg-

angle. The FWHM value of ∆ω= 2.9±0.1 arcmin is obtained by a Lorentzian fit (full curve) of the

measured data.

The ω-scan of the double QW structure is illustrated in Fig. (4.8). In this case also

the intensity distribution of the (004)-ZnSe reflex has been measured as a function of

the relative angular difference ∆ω(dots), to get information about the crystallographic

quality. From the Lorentzian fit of the measured spectrum (full curve) a FWHM value

of ∆ω= 2.9±0.1 arcmin is obtained. This value is significantly smaller compared to

4.2 Investigation of (Zn,Cd)Se multi quantum wells

the FWHM value of the ω-scan spectrum in Fig. (4.6), which has been measured on the

MQW structure.

This result shows that the crystallographic quality of the fully strained double QW struc-

ture is at least four times better than that of the MQW configuration with partial relaxed

ZnSe barriers. Despite this structural advantage, the polariton splitting is reduced by

a factor of two in a double QW microcavity, hence the number of QWs determines the

Rabi-splitting energy.

4.2.2 Optical properties of ZnSe/(Zn,Cd)Se MQWs

Due to the fact that the transition energy and the FWHM of the QW emission mainly

determines the conditions for the observation of the strong coupling, the optical properties

of the (Zn,Cd)Se QWs are as important as the structural parameters.

Figure 4.9: PL spectra of the MQW structure McS-1072 at room temperature (left side) and 4.3 K (right

side), respectively. Left side: The measured spectrum (dots) is fitted with two Gaussian functions,

where the full grey curve depicts the QW transition at E0= 2.282 eV with an emission FWHM of about

33 meV. The dotted grey curve considers the thermal luminescence broadening. Right side: At 4.3 K

a QW transition at E0= 2.328 eV with a FWHM of about 27 meV is observed.

4 ZnSe and (Zn,Cd)Se quantum structures for II-VI microcavities

The PL spectra of the MQW structure McS-1072 are summarized in Fig. (4.9) at room

temperature (left side) and liquid helium temperature (right side), respectively. The room

temperature spectrum (dots) is fitted with two Gaussian functions, where the full grey

curve depicts the homogenous QW transition at E0= 2.282 eV with an emission FWHM of

about 33 meV. The dotted grey curve considers luminescence broadening due to thermal

activation of transitions between higher quantum states and has a FWHM value of about

58 meV.

In the low temperature PL spectrum on the right side in Fig. (4.9) a QW transition at

E0= 2.328 eV with a FWHM of about 27 meV is observed. This value reflects the

inhomogeneous broadening in the structure due to fluctuations of strain, thickness and

cadmium mole fraction of the QWs. The typical low temperature luminescence FWHM

of high quality ZnSe/(Zn,Cd)Se QWs is about 8-12 meV [Se2000]. The luminescence is

shifted about 46 meV to higher energy compared to the room temperature emission.

Fig. (4.10) depicts the PL spectra of the double QW structure McS-772 as described

in the small inlay in Fig. (4.7) at room temperature (left side) as well as at 5 K (right

side), respectively. The room temperature spectrum (dots) is fitted with two Gaussian

functions (grey curves). The FWHM of the QW transition, which is about 23 meV, is

clearly smaller than it has been measured on the MQW structure. The FWHM of the

second peak (dotted grey curve), which reflects the thermal broadening, is about 66 meV.

This value is in good agreement with the PL data of the MQW structure in Fig. (4.9).

Considering time dependent fluctuations of the metal fluxes during MBE processing, the

cadmium mole fraction of the outer QWs in the MQW structure might deviate from the

mole fraction of the central QWs, which are grown in a small time interval of about five

minutes. This fact results in an increased emission linewidth compared to the double QW

structure, in which only the two central QWs contribute to the luminescence.

The low temperature PL spectrum of the double QW structure is illustrated on the right

side of Fig. (4.10) as well as the Gaussian fit function. The QW transition at E0= 2.483 eV

has a linewidth with a FWHM of about 9 meV, which is significantly smaller compared

4.3 ZnSe/(Zn,Cd)Se superlattice structures

Figure 4.10: PL spectra of the double QW structure McS-772 at room temperature (left side) and 4.3 K

(right side), respectively. Left side: The measured spectrum (dots) is fitted with two Gaussian functions.

The full grey curve represents the QW transition at E0= 2.403 eV with an emission FWHM of about

23 meV. The second peak (dotted grey curve) considers the thermal broadening effect. Right side: At

4.3 K a homogenous transition peak at E0= 2.483 eV with a FWHM of about 9 meV is observed.

to the transition FWHM value in the low temperature spectrum of the MQW structure.

This observation reveals that the inhomogeneous broadening of the emission is signifi-

cantly reduced in case of the double QW structure. Considering the structural properties

of the MQW structure, as described in Sec. (4.2.1), the additional inhomogeneous broad-

ening of the PL emission results from the partial relaxation of the active layer.

4.3 ZnSe/(Zn,Cd)Se superlattice structures

In Sec. (4.2.1) and Sec. (4.2.2) it has been shown that MQW structures with four QWs

arranged as indicated in Fig. (4.3), left side, are influenced by the partial relaxation of

the ZnSe barriers. This aspect results in an increased inhomogeneous broadening of the

luminescence. This effect is reduced in a fully strained double QW structure, but yields

4 ZnSe and (Zn,Cd)Se quantum structures for II-VI microcavities

the disadvantage that the number of QWs in the microcavity is reduced by a factor of

two.

Figure 4.11: Cross-section drawing of typical SL structures with four (left side) and six (right side)

(Zn,Cd)Se QWs, respectively, as a superlattice. The resonator length of about 180 nm is adapted to the

QW emission in the blue-green spectral range. The thickness of (Zn,Cd)Se QWs and ZnSe barriers is

reduced to diminish the strain energy in the structure.

Therefore the design of the active layers is modified as it is illustrated in Fig. (4.11).

Firstly the number of QWs in the center of the structure is increased and secondly the

thickness of the (Zn,Cd)Se QWs as well as the ZnSe spacer layers is reduced. In this case

a better crystal quality might be achieved, due to the absence of a large strain field near

the sensitive GaAs/ZnSe heterointerface and the ZnSe barrier relaxation is inhibited.

Furthermore the compact arrangement of the QWs promise a better reproducibility of

their structural parameters (i.e. thickness and cadmium mole fraction) and the strain is

distributed homogenously along all QWs.

Nevertheless, an increase of the PL transition linewidth, due to QW coupling in the SL

structure has to be considered. Additionally to this, the exciton distribution in the QWs

4.3 ZnSe/(Zn,Cd)Se superlattice structures

with respect to the photonic field distribution is different. In the SL microcavities the

outer QWs have a weaker coupling to the photonic mode, due to their increased distance

from the antinodes of the standing wave.

4.3.1 Structural properties of superlattice structures

Fig. (4.12) shows an asymmetric reciprocal space map around the GaAs-(224)-reflex of

the four-fold SL structure McS-851, similar to that, which is described in Fig. (4.11), left

side. The ZnSe Bragg-reflex as well as the superlattice peaks SL 0 and SL-1 are well

Figure 4.12: Reciprocal space map around the GaAs-(224)-reflex of the SL structure McS-851. The ZnSe

reflex is observed on the grey line, which indicates the qxlattice vector in reciprocal space for a fully

strained lattice on GaAs. The superlattice peaks SL-1 and SL 0 are also clearly resolved at the same qx

as the ZnSe reflex.

4 ZnSe and (Zn,Cd)Se quantum structures for II-VI microcavities

resolved in the reciprocal space map. The grey dashed line indicates the qxlattice vector

in reciprocal space, yielding that the ZnSe barriers as well as the QWs and the spacer

layers are fully strained on the GaAs substrate.

Figure 4.13: Symmetric ω−2θ−scan of the SL structure McS-851 (dots) with four QWs. The scan

is measured relative to the GaAs-(004)-reflex. The ZnSe peak is resolved at ∆2θ=−0.406◦and

also three superlattice peaks (SL+1, SL 0 and SL-1) are observed in the spectrum. From their relative

angular distance, a superlattice period of 8.0 nm is estimated. The important structural parameters are

extracted from the full dynamical x-ray diffraction simulation (red curve) of the measured spectrum and

summarized in the cross-section drawing in the small inset.

Hence the ZnSe layers as well as the (Zn,Cd)Se QWs are fully strained on GaAs, the

important structural parameters are obtained from a symmetric ω−2θ−scan. Fig. (4.13)

shows the ω−2θ−scan of the SL structure McS-851 (dots) with four QWs measured

relative to the GaAs-(004)-reflex. The relative angular difference between the ZnSe- and

the GaAs-(004)-reflex is about ∆2θ=−0.406◦. This value is in good agreement with

4.3 ZnSe/(Zn,Cd)Se superlattice structures

the relative angular position of fully strained ZnSe on GaAs (i.e. ∆2θ=−0.405◦) and

confirms the results of Fig. (4.12).

From the Pendell¨osung fringes, which superimpose the ZnSe peak in the spectrum, a ZnSe

barrier layer size of 76±2 nm is estimated. The three additional Bragg-reflexes SL-1, SL 0

and SL+1 stem from the periodicity of the (Zn,Cd)Se superlattice. The relative angular

position of SL 0 is directly related to the composition of the (Zn,Cd)Se layers as well as

the thickness ratio between the ZnSe spacer and the (Zn,Cd)Se QWs [Kh2003]. From the

relative angular distance between the superlattice peaks, the periodicity of the structure

(i.e. the sum of the thickness of ZnSe spacer and (Zn,Cd)Se QW is obtained.

The x-ray simulation spectrum (red curve) in Fig. (4.13) yields the best fit to the experi-

mental data, An accurate characterization of the structural properties of sample McS-851

is performed. The structural properties are summarized in the small inlay in Fig. (4.13).

The thickness of the (Zn,Cd)Se QWs is about 3.9±0.2 nm with a cadmium mole fraction

of 0.20±0.01, and the thickness of the ZnSe spacer layers between the QWs is about

4.1±0.2 nm.

One of the important features of the x-ray spectrum in Fig. (4.13) are the Pendell¨osung

fringes in the ZnSe reflex. These fringes indicate a good interface quality between ZnSe

barriers and QWs and are as well resolved as in the spectrum of the double QW structure

in Fig. (4.7).

Fig. (4.14) illustrates the ω-scan spectrum (dots) of McS-851. The intensity is measured

as a function of the relative angular difference from the ZnSe-(004)-reflex Bragg-angle.

The spectrum is fitted with a Lorentzian function and a FWHM value of ∆ω= 1.9±

0.1 arcmin is obtained.

The ω-scan FWHM value of McS-851 is about half an order of magnitude smaller than that

of the partial relaxed MQW structure McS-803. This result corroborates the improved

crystallographic quality of the SL structure, which is fully strained on the GaAs substrate.

Due to the fact, that thickness and cadmium mole fraction of the QWs are reduced in

McS-851, the ω-scan FWHM value is in fact significantly smaller than that of the fully

4 ZnSe and (Zn,Cd)Se quantum structures for II-VI microcavities

Figure 4.14: ZnSe-(004)-reflex ω-scan of the SL structure McS-851 (dots). The intensity distribution is

measured as a function of the relative difference ∆ωfrom the ZnSe Bragg-angle. The FWHM value of

∆ω= 1.9±0.1 arcmin is obtained by a Lorentzian fit (full curve) of the measured data.

strained double QW structure McS-775.

The most important parameter regarding the SL configuration is the number of QWs,

which can be integrated in the active layer, to increase the oscillator density and therefore

the Rabi-splitting in a SL microcavity. Analyzed via this concept, a significant reduction

of the crystallographic quality, which results from a relaxation of the QWs might be

observed. In case of the MQW structure the number of QWs is limited by four, as shown

in the previous section.

The ω−2θ−scan spectrum of the six-fold SL structure McS-1014 (dots) measured relative

to the GaAs-(004)-reflex is illustrated in Fig. (4.15). The relative angular difference

between the ZnSe- and the GaAs-(004) Bragg-reflex is about ∆2θ=−0.403◦. The

comparison of this value with the data of McS-851 in Fig. (4.13) and Fig. (4.12) shows

4.3 ZnSe/(Zn,Cd)Se superlattice structures

Figure 4.15: ZnSe-(004)-reflex ω−2θ−scan of the six-fold SL structure McS-1014. The ZnSe peak is

measured at ∆2θ=−0.403◦, yielding fully strained barrier layers. Furthermore four superlattice

peaks (SL+2, SL+1, SL 0 and SL-1) are resolved in the spectrum. A superlattice period of 7.1 nm and

a cadmium mole fraction of about 0.33 is estimated from the relative angular distance of the SL peaks.

The structural parameters of McS-1014, which are summarized in the inset, are performed by the x-ray

diffraction simulation (red curve).

that McS-1014 is also fully strained on the GaAs substrate. Despite that, the typical

Pendell¨osung fringes, which superimpose the ZnSe peak in the four-fold SL structure, are

scarcely resolved in Fig. (4.15).

The structural parameters of McS-1014 are extracted from the x-ray simulation (red curve)

of the ω−2θ−scan spectrum (dots) and summarized in the small inset in Fig. (4.15). Four

superlattice peaks are clearly resolved in the measured data and a superlattice period of

7.1±0.1 nm is obtained. The cadmium mole fraction of 0.33±0.01 is calculated from the

relative angular position of the SL 0 reflex.

4 ZnSe and (Zn,Cd)Se quantum structures for II-VI microcavities

Figure 4.16: ZnSe-(004)-reflex ω-scan spectrum of the six-fold SL structure McS-1014 (dots). The inten-

sity distribution is measured as a function of the relative difference ∆ωfrom the ZnSe Bragg-angle. The

FWHM value of ∆ω= 4.4±0.1 arcmin is obtained by a Lorentzian fit (full curve) of the measured

spectrum.

Fig. (4.16) depicts the ZnSe-(004)-reflex ω-scan spectrum of the six-fold SL structure McS-

1014. The spectrum is fitted by a Lorentzian function and a FWHM value of ∆ω= 4.4±

0.1 arcmin is obtained, which is about a factor of two larger than that of the four-fold

SL structure in Fig. (4.14). This result is consistent with the degradation of the interface

quality (reduction of the intensity of the Pendell¨osung fringes), which is observed in the

ω−2θ−scan spectrum in Fig. (4.15). The reduced interface quality of McS-1014 stems

from the substantial increased cadmium mole fraction (x(Cd) = 0.33) compared to that

of McS-851 (x(Cd) = 0.20). The increased cadmium mole fraction in McS-1014 ensures

a better exciton confinement at room temperature.

The ω−scan spectrum FWHM value of the six-fold SL sample is about 50 % larger

4.3 ZnSe/(Zn,Cd)Se superlattice structures

in comparison to the data of the double QW structure McS-775 in Fig. (4.8), despite

similar structural properties according to cadmium mole fraction and the entire thickness

of all (Zn,Cd)Se QWs. Nevertheless, the investigation of several SL structures clearly

shows an improvement of the structural quality compared to that of the MQW samples.

Additionally to this, it is possible to increase the number of QWs up to six in the SL

structures.

Figure 4.17: Cross section TEM-picture of a six-fold SL structure McS-1035. Left side: Overview of the

whole structure. The six (Zn,Cd)Se QWs (dark color) are sandwiched between 76 nm ZnSe barriers. The

strain contrast reveals excellent homogeneity of the QWs and spacer layers. Right side: Detailed view

of the QW superlattice region in atomic resolution. The superlattice period is about 8.3 nm divided up

into 2.9 nm QWs and 5.4 nm ZnSe spacers. The strain contrast reveals slightly fluctuations of thickness

and cadmium mole fraction of the QWs.

Fig. (4.17) depicts a cross-section TEM-picture of the six-fold SL structure McS-1035.

4 ZnSe and (Zn,Cd)Se quantum structures for II-VI microcavities

On the left side the strain contrast of the whole structure is resolved. The dark regions

are the six (Zn,Cd)Se QWs enclosed in 76 nm ZnSe barriers (bright regions). The picture

demonstrates the excellent uniformity of the crystallographic structure. On the right side

of Fig. (4.17) the SL region is illustrated in atomic resolution. Slightly fluctuations of

the strain contrast reveal small variations in thickness and cadmium mole fraction of the

(Zn,Cd)Se QWs on atomic scale.

The superlattice period is about 8.3 nm divided up into 2.9 nm (Zn,Cd)Se QWs and

5.4 nm ZnSe spacers in between. From the MBE growth parameters of the structure

a barrier thickness of 73 nm, a QW thickness of 2.5 nm and a ZnSe spacer thickness

of 5.0 nm is obtained. These values are in excellent agreement with the real structural

parameters measured on the TEM investigations.

4.3.2 Photoluminescence investigation of SL structures

Hence a coupling of the QWs has to be considered in the SL structures, the optical

properties are different compared to that of MQW structures. For the estimation of

the (Zn,Cd)Se QW transition energy, the finite barrier model for fully strained QWs is

used, yielding an excellent agreement between PL data and structural properties [Pa2001].

The penetration depth of the electron and hole wavefunction in the ZnSe barrier layers

is estimated of about 2.5-4.0 nm for typical QW configurations with a cadmium mole

fraction of 0.20-0.38 as well as a QW thickness of 3-5 nm.

The PL spectra (dots) of the four-fold SL structure McS-851 at room (left side) as well

as at low temperature (right side) are illustrated in Fig. (4.18). The room temperature

spectrum is fitted by two Gauss functions to consider the QW transition (full grey curve)

at E0= 2.526 eV and additionally thermal activation of transitions between higher

quantum states (dotted grey curve). The FWHM value of about ∆E= 23 meV of

the room temperature luminescence is nearly equal to that of the uncoupled double QW

structure in Fig. (4.10).

In contrast to this, the low temperature PL transition in Fig. (4.18), right side, is observed

4.3 ZnSe/(Zn,Cd)Se superlattice structures

Figure 4.18: PL spectra (dots) of the four-fold SL structure McS-851 at room temperature (left side) and

4.3 K (right side), respectively. Left side: The measured spectrum is fitted with two Gauss functions,

to consider the QW transition (full grey curve) as well as thermal luminescence broadening (dotted

grey curve). The best fit yields a transition energy of E0= 2.526 eV with a FWHM value of about

∆E= 23 meV. Right side: At 4.3 K the PL spectrum is fitted by one Gaussian function. The QW

transition is observed at E0= 2.610 eV with a FWHM value of about ∆E= 20 meV.

at E0= 2.610 eV with a FWHM value of about ∆E= 20 meV. This FWHM value is

about two times larger compared to that of the double QW structure. This fact confirms

an increase of the luminescence broadening, due to the QW coupling in the SL structure.

Fig. (4.19) depicts the PL spectra measured at the six-fold SL structure McS-1014. The

PL measurements show results similar to that of the four-fold SL structure McS-851,

although the cadmium mole fraction and the number of QWs is increased in McS-1014.

At room temperature the luminescence transition energy is E0= 2.470 eV with a

slightly larger FWHM value of about ∆E= 24 meV. Both effects, a lower luminescence

energy and an increased FWHM value, respectively, stem from the increased cadmium

mole fraction of the QWs in the six-fold SL sample.

The low temperature PL spectrum Fig. (4.19), right side, is fitted with a Gauss function,

4 ZnSe and (Zn,Cd)Se quantum structures for II-VI microcavities

Figure 4.19: PL spectra (dots) of the six-fold SL structure McS-1014 at room temperature (left side) and

4.3 K (right side), respectively. Left side: The measured spectrum is fitted with two Gauss functions

(grey curve), the best fit yields a transition energy of E0= 2.470 eV with a FWHM value of about

∆E= 24 meV. Right side: At 4.3 K the PL spectrum is fitted by one Gauss function, yielding a

transition energy of E0= 2.543 eV with a FWHM value of ∆E= 22 meV.

yielding a transition energy of E0= 2.543 eV with a FWHM value of ∆E= 21.8 meV.

The luminescence FWHM value is in good agreement with the results of the low temper-

ature luminescence of the four-fold SL structure.

Despite the reduced crystallographic quality of the six-fold SL structure (see also Fig. (4.14)

and Fig. (4.16)), the luminescence properties of both SL structures are equivalent. This

observation yields that at least up to six QWs are successfully included into SL structures

without any significant degradation of the structural and the optical properties.

5 II-VI semiconductor microcavity

structures

5.1 The combination of Bragg-mirrors and active layer

to complete microcavities

In the previous two chapters the optimization of ZnS/YF3and ZnSe/YF3-Bragg mirrors

as well as the design of the active layer containing up to six ZnSe/(Zn,Cd)Se QWs has

been reported. The main focus of this section is the combination of Bragg-mirrors and

active layer to a complete microcavity, which opens up the possibility to investigate the

photon-exciton interaction. Main efforts have been performed to measure the strong

coupling and especially the polariton anticrossing behavior in such microcavities at room

temperature.

Fig. (5.1) shows a cross-line drawing of a ZnSe based microcavity structure with four

(Zn,Cd)Se QWs in the center as the active layer, a six-fold stack of ZnSe/YF3Bragg-

mirrors on the front side and an eight-fold stack on the rear side, respectively. The

microresonator is optimized to emission at λ= 517 nm, which results in a resonator

length of about LC=λ

nC= 190 nm. The thickness gradient of the cavity, which is

exaggeratedly drawn in Fig. (5.1), is about ∆ LC= 1 % mm−1.

The reflectivity of this structure can be calculated with the Transfer-Matrix model de-

scribed in Sec. (2.2) as a function of the cavity length. The spectra are illustrated in

5 II-VI semiconductor microcavity structures

Figure 5.1: Cross-section drawing of a typical ZnSe/(Zn,Cd)Se SL structure with four QWs in the center

as well as an eight-fold stack of ZnSe/YF3DBRs rear mirror and a six-fold stack of DBRs front mirror,

respectively. ∆LCis the cavity thickness gradient, which is implemented during growth of the active

layer with the nominal resonator length LC.

Fig. (5.2) (full curves). With decreasing cavity length, the photonic mode (indicated by

the grey line) shifts to higher energy, while the QW transition energy (black dashed line)

is constant. Due to the strong coupling, the absorption peaks (full curves) show an an-

ticrossing behavior. The anticrossing of the polariton peaks is indicated in Fig. (5.2) by

the dashed lines with black (upper polariton) and white (lower polariton) circles. When

the cavity mode and the QW transition energy are in resonance, the energy difference

between upper and lower polariton is minimal and equal to the Rabi-splitting energy,

which is about ΩRabi = 64 meV in this calculation.

5.1.1 Experimental growth procedure of microcavity structures

The fabrication of the microcavity structure is divided into five steps. First the active

layer of ZnSe/(Zn,Cd)Se QWs is grown by MBE and afterwards characterized by x-ray

5.1 The combination of Bragg-mirrors and active layer to complete microcavities

Figure 5.2: Reflectivity spectra (full curves) of the microcavity structure as a function of LC. Due to the

thickness gradient, the cavity mode shifts to higher energy (grey line), while the QW transition energy is

constant (black dashed line). The strong coupling is manifested in the anticrossing behavior of the two

absorption peaks (dotted lines with white and black circles). The minimum splitting energy between the

lower and the upper polariton peak is about ΩRabi = 64 meV.

5 II-VI semiconductor microcavity structures

diffraction and PL, to obtain the important structural parameters.

After that the sample is transferred into a evaporation chamber to deposit the first Bragg-

mirror, which is a stack of eight pairs of ZnSe/YF3or ZnS/YF3, respectively. In a third

step the ”flip-chip“-technology is used to fix the sample with the mirrored surface on glass

substrate by using epoxy glue.

After this procedure, the most complicated preparation step, the selective wet etching

of the GaAs substrate, is performed. An appropriate solution of 82 ml NaOH (1M) and

18 ml H2O2(35 %) is used. The typical etch rate in GaAs is about 100-150 µm h−1.

Experimental results have shown, that the epitaxial ZnSe layer might be exposed up to

10 min to the etching solution. During this time no significant reduction of the layer

thickness but a slight increase of the surface roughness (about 8 nm after 10 min) is ob-

served [Ar2003]. After the substrate removal process has been performed, the microcavity

is completed by deposition of a six-fold stack of ZnSe/YF3or ZnS/YF3DBRs.

The critical point of the substrate removal process is the relaxation of the epitaxial ZnSe

layer. This has to be considered, because the active layer is strained on the GaAs sub-

strate. Fig. (5.3) depicts a microscope picture of a four-fold SL structure after the selective

wet etching process. The GaAs (brown color) is completely removed in the center of the

sample and the active layer (orange) is uncovered.

The spontaneous relaxation of the active layer leads to the formation of horizontal and

vertical microcracks in the preferred (110) and (-110) crystallographic direction. Then the

etch solution penetrates through this cracks into the interface between the active layer and

the first Bragg-mirror. The green region in the center of Fig. (5.3) shows the Bragg-mirror

surface. In this region the active layer was washed away by the etch solution.

The damage in the active layer is significantly reduced in case of partial or fully relaxed

structures [Ar2003]. Nevertheless, in the previous chapter it has been shown that fully

strained SL structures have significantly better optical and structural properties than the

partial relaxed MQW structures. However, further optimization of the substrate removal

process has to be performed.

5.2 Investigation of a SL microcavity structure with ZnSe/YF3DBRs

Figure 5.3: Microscope picture of a SL structure after selective wet etching. The GaAs substrate (brown)

is removed in the center of the sample. The orange colored pieces near the border to the GaAs are parts

of the active layer. The green region in the center of the picture shows the Bragg-mirror surface. The

whole GaAs-free region is strewn with microcracks, which are the result of spontaneous relaxation of the

active layer.

5.2 Investigation of a SL microcavity structure with ZnSe/YF3

DBRs

Although the substrate removal process is critical in case of fully strained SL structures,

the four-fold SL sample McS-851 has been successfully completed in a microcavity struc-

ture (sample PE-031) with an eight-fold stack of ZnSe/YF3DBRs on the rear side and a

six-fold stack of ZnSe/YF3DBRs on the front side, respectively.

The structural properties of this sample are obtained from the x-ray measurements shown

in Fig. (4.13). The nominal ZnSe barrier thickness is about 76 nm for each barrier,

the thickness of the (Zn,Cd)Se QWs is about 3.9 nm and the QWs are separated by

approximately 4.1 nm ZnSe spacers. From these data a resonator length of about 180 nm

5 II-VI semiconductor microcavity structures

is obtained.

The PL spectrum of this sample measured before mirror deposition and substrate removal,

is illustrated in Fig. (4.18). A QW transition energy of 2.526 eV with a FWHM linewidth

of about 23 meV is observed at room temperature. Using the structural parameters,

which are obtained by x-ray diffraction, the transition energy can be calculated with the

finite barrier model of biaxial strained QWs [Pa2000]. The theoretical value of 2.524 eV

is in excellent agreement with the PL data and the x-ray measurements.

5.2.1 Measurement of the microcavity length gradient off-resonance

To gain information about the microresonator length gradient, the reflectivity of the

microcavity is measured at room temperature on different positions on the sample. The

reflectivity spot size is about 300 µm and the spot position is moved in 250 µm steps

across the sample surface in direction of the resonator length gradient.

Three of these reflectivity spectra are illustrated in Fig. (5.4). The cavity mode is clearly

resolved in each spectrum and has a FWHM value of about 48 meV. Furthermore a clear

shift to lower energy with increasing sample thickness is observed.

The absorption peak energy of the cavity mode as a function of the position son the

sample up to 2000 µm from the origin (black dots, left scale) is summarized in Fig. (5.5).

From the linear approximation an energy shift of about ∆EC= 13 meV mm−1on

the sample is obtained. Considering the refractive index of ZnSe, the resonator length

is calculated from the absorption peak energies (grey dots, right scale). This results in

a cavity length of LC= 172.8 + 1.19 ∗10−3∗s, which is in excellent agreement with

the nominal cavity length of 180 nm measured by x-ray diffraction before the mirror

deposition.

5.2.2 Reflectivity measurements in resonance condition

The position dependent cavity mode energy of the sample is obtained from the linear

approximation EC= 2.579 −1.3∗10−5∗s(eV). The transition energy of the (Zn,Cd)Se

5.2 Investigation of a SL microcavity structure with ZnSe/YF3DBRs

Figure 5.4: Reflectivity spectra of the SL microcavity PE-031 measured at various positions on the sample

(0 µm, 500 µm and 1000 µm). A linear shift of the cavity mode to lower energy is observed (dashed line).

The FWHM value of the cavity mode is about 48 meV.

QWs has been measured by PL at 2.526 eV. Therefore the resonance position between

photonic and excitonic mode at room temperature is calculated of about 3500 µm from

the origin.

The spectra in Fig. (5.6) are measured near the calculated resonance position as a function

of the position on the sample (dots). The spectra measured at 2250 µm, 3100 µm and

3200 µm are each fitted by a Lorentzian function, where the spectra at 2500 µm, 2750 µm

and 3000 µm are each fitted with two Lorentzian peaks, respectively (full curves). The

energetic peak positions, indicated by the arrows are obtained from the best fit of each

measured spectrum.

5 II-VI semiconductor microcavity structures

Figure 5.5: Absorption peak energy ECof the cavity mode as a function of the position son the sample

PE-031 (black dots, left scale). From the linear approximation (black line, left scale) a cavity mode energy

shift of about 13 meV mm−1on the sample is obtained. The grey dots and the grey line, respectively,

describe the resonator length gradient LC(right scale).

In the spectrum measured at 3000 µm two absorption peaks with an energy difference of

about Ω = 35 meV and a FWHM value of 15 meV and 47 meV are clearly resolved

(black and grey curve). The average linewidth (i.e. the splitting to linewidth ratio)

of both absorption peaks is about 32 meV. The strong coupling condition in Sec. (2.1.1)

denotes that the average linewidth of both polariton peaks should not exceed the splitting

energy. This condition is barely fulfilled in the spectra in Fig. (5.6) and explains, that

the polariton splitting is not resolved more pronounced.

In the spectra measured at 2250 µm, 3100 µm and 3200 µm, the splitting is not resolved

any more, due to the fact that the cavity mode linewidth exceeds the splitting to linewidth

ratio. In the spectra measured at 2500 µm and 2750 µm still a slight shoulder in the

5.2 Investigation of a SL microcavity structure with ZnSe/YF3DBRs

Figure 5.6: Room temperature reflectivity spectra of the SL microcavity PE-031 near the resonance

position between photonic and excitonic mode. In the spectrum at 3000 µm two absorption peaks with

an energy difference of Ω = 35 meV and a FWHM value of 15 meV and 47 meV are resolved from a

fit with two Lorentzian peaks (black and grey curve). In the spectra measured at 2500 µm and 2750 µm

also a slight shoulder in the absorption peak is observed, while the other spectra show the cavity mode

only. The fitted peak energies of each spectrum are indicated by the black arrows.

5 II-VI semiconductor microcavity structures

absorption peaks is observed, yielding that the coupling is at least stable in the range

between 2500 µm and 3000 µm. However, there is a slight variation between the calculated

and experimental resonance position (at 3500 µm and 3000 µm, respectively). This fact

yields that the strain relaxation in the sample after the substrate removal process has a

perceptible influence on QW emission energy.

The absorption peaks energies of each spectrum in Fig. (5.6) is plotted as a function

of the resonator detuning energy ~(ω−ω0) (dots) in Fig. (5.7). The detuning energy

(bottom scale) is calculated from the difference between QW transition energy and cavity

mode energy, respectively. The full curves in Fig. (5.7) depict the theoretical polariton

Figure 5.7: Absorption peak energy (dots) of each spectrum in Fig. (5.6) as a function of the resonator

detuning energy ~(ω−ω0) (bottom scale). The top scale denotes the position on the sample. The full

curves depict the calculated polariton dispersion curves for the microcavity. However, a good agreement

between experimental and theoretical data is achieved, yielding the typical anticrossing behavior of the

polariton dispersion curves in the strong coupling regime. The calculated Rabi-splitting energy is about

ΩRabi = 33 meV at zero detuning.

5.3 Room temperature Rabi-splitting in a MQW microcavity with ZnS/YF3DBRs

dispersion for the microcavity structure. For the calculation Eq.( 2.5) is used, where the

photon dispersion ECis given by the resonator length gradient as shown in Fig. (5.5). The

exciton dispersion EX, which is position independent, corresponds to the QW transition

energy. The Rabi-splitting energy as well as the QW transition energy are used as fitting

parameters, to achieve an agreement with the experimental data in Fig. (5.7).

From the calculation a Rabi-splitting energy of ΩRabi = 33 meV and a QW transition

energy of EX= 2.538 eV is obtained. Both results are in good agreement with the

experimental observations, yielding the typical anticrossing behavior of the polariton dis-

persion curves in Fig. (5.7). This calculation supposes the existence of the strong coupling

in the SL microcavity PE-031 at room temperature.

5.3 Room temperature Rabi-splitting in a MQW micro-

cavity with ZnS/YF3DBRs

In the previous section the exciton-photon coupling in a fully strained SL microcavity with

four (Zn,Cd)Se QWs has been discussed. The relaxation process of the active layer after

the GaAs substrate removal induces the formation of microcracks in the sample, which

leads to surface destruction. This results in an increased FWHM value of the cavity mode

and decreases the splitting to linewidth ratio.

Due to the partial strain relaxation of MQW structures, the damage of the active layer

is significantly reduced although the optical and structural properties of MQW samples

are not as good as in SL structures. The reduction of microcracks and surface roughness

decreases the FWHM value of the photonic mode in the MQW structure and the Rabi-

splitting condition is sufficiently fulfilled.

The cross-section drawing of the as-grown MQW microcavity PE-013 is illustrated in

Fig. (5.8). The active layer contains four (Zn,Cd)Se QWs with a cadmium mole fraction

of 0.31 and a QW thickness of 7 nm. The QWs are enclosed in ZnSe barriers of 68 nm

and 72 nm, respectively. The nominal resonator length is about 197 nm. The structural

5 II-VI semiconductor microcavity structures

Figure 5.8: Cross-section drawing of the as-grown MQW microcavity PE-013 with six periods of ZnS/YF3

DBRs on the front side and eight periods on the rear side, respectively. The (Zn,Cd)Se QWs are aligned in

MQW configuration and enclosed in ZnSe barriers. The nominal resonator length is about LC= 197 nm

plus the cavity length gradient ∆LC.

data are obtained from the ω−2θ−scan in Fig. (4.5). The microcavity is completed by

a six-fold stack of ZnS/YF3DBRs on the front side and an eight-fold stack of ZnS/YF3

DBRs on the rear side.

For detailed information about the microcavity length gradient ∆LC, the reflectivity of

the structure is measured off-resonance as a function of the position on the sample. Some

of the spectra are summarized in Fig. (5.9) (dots). The cavity mode shifts to higher

energy with increasing position on the sample and the dashed line reveals a nonlinear

dependence. The cavity mode energy shift is sufficiently approximated by

EC= 2.255 + 4 ×10−6∗s+ 8 ×10−8∗s2(eV ),(5.1)

where ECis the cavity mode energy in (eV) and sthe position on the sample in (µm).

Each spectrum in Fig. (5.9) is fitted with a Lorentzian function, revealing cavity mode

FWHM values between 10 meV and 16 meV, which is significantly smaller than it has

5.3 Room temperature Rabi-splitting in a MQW microcavity with ZnS/YF3DBRs

Figure 5.9: Room temperature reflectivity spectra of the MQW microcavity measured off-resonance on

various positions on the sample (dots). The spectra denote a clear shift of the cavity mode ECto higher

energy with increasing sample position. This shift is indicated by the dashed line and shows a nonlinear

behavior. Each spectrum is fitted with a Lorentzian function and the quality factor Qis calculated.

been obtained for the SL microcavity PE-031 (about 48 meV). The cavity quality factor

is given by Q=EC/∆EC, where ECis the energetic position of the photonic mode.

Fig. (5.10) shows the room temperature PL spectrum of the MQW microcavity structure

PE-031 measured off-resonance. The grey curve depicts the Gauss fit of the spectrum,

yielding a QW transition energy of 2.313 eV with a FWHM value of about 26 meV.

This value is significantly smaller than that of the similar MQW-structure McS-1072

5 II-VI semiconductor microcavity structures

Figure 5.10: Room temperature PL spectrum of the MQW microcavity PE-013 measured off-resonance

(dots). The spectrum is fitted with a Gaussian function (grey curve) and from that a QW transition

energy of 2.313 eV with an FWHM of about 26 meV is obtained.

in Fig. (4.9), left side (about 33 meV). This observation reveals that the spontaneous

emission rate of the active layer is enhanced due to the inclusion in the Bragg-mirrors.

5.3.1 Photoluminescence investigations of the MQW microcavity in

resonance condition

Besides the position dependent reflectivity measurements, room temperature PL investi-

gations have been performed at resonance between excitonic and photonic mode. In this

case, the sample position and therefore the cavity mode energy is fixed, while the QW

transition energy is tuned by varying the sample temperature.

Fig. (5.11) shows the luminescence spectra of the MQW microcavity PE-013 between

5.3 Room temperature Rabi-splitting in a MQW microcavity with ZnS/YF3DBRs

Figure 5.11: PL spectra of the microcavity PE-013 measured between 270 K and 330 K (dots). The curves

represent Lorentzian fits. The increase of the temperature leads to a redshift of the QW emission, while

the cavity mode energy is constant at EC= 2.294 eV. The experimental data show a clear anticrossing

behavior and the minimum splitting Ωmin = 44 meV is observed at 300 K.

5 II-VI semiconductor microcavity structures

270 K and 300 K (dots) and the full curves represent Lorentzian fit functions. The energy

of the photonic mode is constant at EC= 2.294 eV. A QW transition energy redshift of

about 1 meV K−1with increasing temperature is estimated. This shift is confirmed by

off-resonance PL measurements between 250 K and 270 K.

The experimental data in Fig. (5.11) show two well separated luminescence peaks in each

spectrum. Furthermore a clear anticrossing behavior between both peaks is observed and

denotes the strong coupling between photon and exciton in the MQW microcavity. The

minimum splitting between the luminescence peaks of about Ωmin = 44 meV is measured

at 300 K. From the FWHM values of both peaks, which are 15.5 meV and 39.7 meV,

respectively, an average linewidth of 27.6 meV is obtained.

This results reveals that the conditions for the strong coupling (i.e. the splitting to

linewidth ratio) are fulfilled. Furthermore in Fig. (5.11) a redistribution of the intensity

from the upper to the lower polariton peak is observed with increasing temperature. This

behavior is also a clear indication in the strong coupling regime.

Fig. (5.12) shows the peak energies of each spectrum in Fig. (5.11) plotted as a function of

the temperature (dots). The full curves depict the best fit of the calculated temperature

dependent polariton dispersion to the experimental data. For the calculation the cavity

mode energy EC= 2.294 eV is derived from the cavity length gradient Eq. (5.1), while

the temperature dependent QW transition energy is given by

EX= 2.627 −1.1×10−3eV

K∗T, (5.2)

where Tis the sample temperature. The QW transition energy redshift of 1.1 meV K−1,

which has been assumed in Eq. (5.2) is in good agreement with the experimental data

measured between 250 K and 270 K. The Rabi-splitting energy ΩRabi = 52 meV is obtained

from the fit of the experimental data.

5.3 Room temperature Rabi-splitting in a MQW microcavity with ZnS/YF3DBRs

Figure 5.12: Temperature dependent polariton dispersion of the MQW microcavity PE-013. The lumines-

cence peak energies, which were fitted with Lorentzian functions in Fig. (5.11), are plotted as a function

of the temperature (dots). The full curves correspond to the calculated polariton dispersion. From this

calculation a Rabi-splitting energy of ΩRabi = 52 meV is determined.

5.3.2 Reflectivity measurements of the MQW-microcavity in reso-

nance condition

The information about the microcavity length gradient in Eq. (5.1) and the QW transition

energy of 2.313 eV (see also Fig. (5.10)) are used to estimate the resonance position

between the photonic and excitonic mode at room temperature. It is found to be about

830 µm from the origin. Near the resonance position the reflectivity of the microcavity is

measured as a function of the position on the sample. The measurements are performed

with a 100 µm spot focus on the sample.

The spectra measured at 200 µm, 500 µm, 700 µm and 800 µm from the origin (A, B, C

5 II-VI semiconductor microcavity structures

Figure 5.13: Overview of the room temperature reflectivity spectra (dots) of sample PE-013 measured

in the range between 200 µm and 800 µm (A, B, C and D). Spectrum A and B show the cavity mode

only, while in C and D a splitting into several absorption peaks is observed. The red full curves depict

the polariton peaks calculated with the Transfer-Matrix model. The spectra B, C and D are in good

agreement with the calculated curves and denote an anticrossing behavior of the polariton modes. From

Lorentzian fits of the experimental data in spectrum C and D with three and two peaks, a splitting of

36 meV and 41 meV is extracted. The blue curve depicts the cavity photoluminescence. However, in the

grey region in spectrum C additional absorption, which might be related to the QWs, is detected.

5.3 Room temperature Rabi-splitting in a MQW microcavity with ZnS/YF3DBRs

and D, respectively) are depicted in Fig. (5.13) (dots). Spectrum A and B show the cavity

mode only, which is shifted to a higher energy with increasing position on the sample.

Near the resonance position at 700 µm (spectrum C) a splitting into three absorption

peaks is observed. At 800 µm two clearly separated peaks are resolved. The minimum

splitting between the polariton peaks is about 41 meV.

The red full curves in spectrum B, C and D depict the polariton absorption peaks cal-

culated with the Transfer-Matrix model. For this calculation the structural parameters

of the MQW microcavity shown as well as the resonator length gradient Eq. (5.1) as the

QW transition energy EX= 2.313 eV (blue curve), are used. A good agreement between

the calculated polariton curves and the experimental spectra B, C and D is obtained.

This result yields that two of the absorption peaks in spectrum C are related to the

polariton splitting. From a Lorentzian fit of the experimental data with three peaks, a

minimum Rabi-splitting energy Ωmin = 36 meV between the polariton absorption peaks

at 2.288 eV and 2.324 eV, respectively, is extracted.

The third peak in spectrum C (grey region) as well as the enhanced FWHM value of the

higher energy peak in spectrum D might be explained by QW absorption, which results

from structural defects such as microcracks or holes in the front side DBR. In this case the

polariton splitting is superimposed by the QW absorption in the range between 2.29 eV

and 2.32 eV, which is in good agreement with the cavity PL spectrum (blue curve).

The average linewidth of the absorption peaks in spectrum C and D is about 13 meV and

17 meV, respectively. This observation is consistent with the calculated average linewidth

∆Epol =1

2(∆EC+ ∆EX) = 18 meV and also significantly smaller than the Rabi-

splitting energy Ωmin = 36 meV. Therefore the conditions of the strong coupling in the

MQW microcavity are fulfilled.

The reflectivity spectra in Fig. (5.13) were fitted with Lorentzian functions and the ab-

sorption peak energies are plotted as a function of the microcavity detuning energy in

Fig. (5.14) (black dots). The detuning energy is due the energy difference between the

uncoupled excitonic and photonic mode. The resonance condition in the microcavity is

5 II-VI semiconductor microcavity structures

Figure 5.14: Absorption peak energies of the reflectivity spectra in Fig. (5.13) as a function of the mi-

crocavity detuning energy (black dots). The red curves denote the polariton dispersion curves calculated

with the Transfer-Matrix model (i.e. the red curves in Fig. (5.13)). The anticrossing of the upper and

lower polariton branch is well resolved. The black full curves are the calculated polariton dispersion

curves using Eq. (2.5). From this calculation a Rabi-splitting energy of ΩRabi = 37 meV is obtained.

achieved at zero detuning energy. The red curves denote the polariton dispersion curves

calculated with the Transfer-Matrix model (see also the red curves in Fig. (5.14)). A good

agreement of the Transfer-Matrix calculation and the experimental data is obtained.

The black full curves in Fig. (5.14) depict the polariton dispersion curves calculated with

Eq. (2.5), where ECis given by Eq. (5.1) and the QW transition energy EXas well as

the Rabi-splitting energy ΩRabi are used as fitting parameters. The best agreement with

the experimental data in Fig. (5.14) as well as the Transfer-Matrix calculation is achieved

for EX= 2.308 eV and ΩRabi = 37 meV. From the PL spectrum in Fig. (5.10) a QW

transition energy of EX= 2.313 eV is obtained. Finally the good agreement between

5.3 Room temperature Rabi-splitting in a MQW microcavity with ZnS/YF3DBRs

the reflectivity measurements in Fig. (5.14) and the calculated polariton dispersion curves

confirms the strong coupling in the MQW microcavity PE-013.

Although the Rabi-splitting energy is larger in the PL measurements of the MQW mi-

crocavity (about 44 meV), calculated and experimental data of both, PL and reflectivity

investigations, are in good agreement with the expected polariton properties.

5.3.3 Calculation of the QW oscillator strength per unit area

In Sec. (2.1) it has been shown that the Rabi-splitting energy ΩRabi depends on the number

of QWs, the effective length of the microcavity Leff. =LC+LDBRs and the oscillator

strength per unit area f

S. Therefore the oscillator strength per unit area can be calculated

from the experimental Rabi-splitting values Ωmin for both, the SL microcavity and the

MQW microcavity, respectively. The penetration depth LDBRs of the photonic mode into

the Bragg-mirrors is obtained from

LDBRs =λ

2nC∗nLnH

nH−nL

(5.3)

where nCis the refractive index of the active layer and nHand nLare the higher and

lower refractive index of the Bragg-mirrors, respectively [Sk1998].

sample x(Cd) ΩRabi (meV) Leff. (nm) Eresonance (eV) f

SL, reflection 0.20 35 176 2.541 0.037±0.005

MQW, reflection 0.35 36 200 2.308 0.044±0.005

MQW, photoluminescence 0.35 44 201 2.298 0.067±0.005

Table 5.1: Calculation of the QW exciton oscillator strength per unit area from the Rabi-splitting energies

measured on the SL and MQW microcavity, respectively. The SL microcavity has a significantly smaller

oscillator strength of 0.037±0.005 compared to the average oscillator strength of the MQW microcavity

of about 0.056±0.005.

The oscillator strength per unit area for both, the SL and MQW microcavity, are sum-

marized in Tab. (5.1). The average oscillator strength from the reflectivity and PL inves-

5 II-VI semiconductor microcavity structures

tigations of the MQW microcavity is about 0.056, which is significantly larger than the

oscillator strength in the SL microcavity (about 0.037).

This observation is consistent with the difference of the cadmium mole fraction in the

(Zn,Cd)Se QWs of both structures. With decreasing cadmium mole fraction, the exciton

confinement in the QWs is reduced, yielding a delocalization of the exciton wavefunction.

This results in a decrease of the electron- and hole-wavefunction overlap integral, which

is related to the oscillator strength. In [Di1993] an oscillator strength per unit area of

about 0.02 is obtained for a Zn0.76Cd0.24Se QW structure enclosed between relaxed ZnSe

barriers. This value is slightly smaller than that obtained with the SL microcavity. This

maybe due to the fact that the ZnSe barriers enclosing the (Zn,Cd)Se QWs in the SL

structure are still partial strained, yielding an additionally confinement of the QWs and

therefore an increased oscillator strength.

6 Conclusions

Since Weisbuch et. al. 1992 has demonstrated the existence of a new quasi-particle, the

polariton, which results from the strong coupling between excitons and photons in semi-

conductor microcavities, this phenomenon has been studied with expanding interest in

common high quality III-V semiconductors such as GaAs, AlAs and their alloys. However,

it has been found out that the polariton properties strongly depend on the exciton binding

energy and the oscillator strength in the semiconductor material, yielding that GaAs and

AlAs are unsuitable for the development of room temperature polariton applications and

-devices.

The focus of this thesis has been the development and investigation of a semiconductor

microcavity based on the II-VI materials ZnSe and CdSe and enclosed in dielectric DBRs

of ZnS/YF3or ZnSe/YF3, respectively.

First of all dielectric Bragg-mirrors of ZnS/YF3were grown on standard GaAs substrate by

thermal evaporation. From optical characterization, high reflectivity coefficients between

0.92 and 0.95 are obtained in the visible spectral range with six up to ten Bragg-mirror

pairs. Simulations of the spectral reflectivity with the Transfer-Matrix model indicate

that sulphur interdiffusion between the mirror interfaces decreases the refractive index

difference between ZnS and YF3and reduces shape and reflectivity of the Bragg-mirror

stop-band. Furthermore structural investigations by AFM yield a significant increase of

the surface roughness, which is proportional to the number of periods and limits the

maximal reflectivity.

In dielectric Bragg-mirrors of ZnSe/YF3, the sulphur was exchanged by selenium. There-

6 Conclusions

fore interdiffusion effects are significantly reduced and the larger refractive index difference

between ZnSe and YF3induces a better photonic confinement in the DBRs. From the re-

flectivity investigations of ZnSe/YF3Bragg-mirrors, high reflectivity coefficients of 0.993

up to 0.999 are achieved with five up to ten mirror periods. However, the increased re-

flectivity compared to that of ZnS/YF3DBRs is only obtained for a wavelength larger

than 470 nm, since strong bandgap absorption in ZnSe occurs in the blue and near UV

light region.

In the second part of this work, MBE grown ZnSe/(Zn,Cd)Se MQWs with up to four,

and SL structures with up to six (Zn,Cd)Se QWs as the active layer in II-VI microcavities

were investigated. The main issue was the in-situ and ex-situ control of the structural

properties as well as the implementation of a cavity length gradient in the ZnSe barrier

layers, which allows the resonance tuning in the microcavity.

With high-resolution x-ray diffraction it is shown that MQW structures are partial re-

laxed, due to the large strain field near the first (Zn,Cd)Se QW closely to the interface

between GaAs and ZnSe. This fact leads to an inhomogeneous strain distribution along

the growth direction of the MQW structure. Additionally to this, a degeneration of the

crystallographic quality is measured with x-ray rocking curves. Due to the inhomogeneous

strain distribution, a large QW transition linewidth up to 33 meV at room temperature

and 27 meV at 4 K is obtained from PL investigations.

In contrast to this, the SL structures with four (Zn,Cd)Se QWs have significantly better

crystallographic quality and are also fully strained on the GaAs substrate. Therefore

a homogenous strain distribution along the QWs is achieved, yielding a smaller QW

transition linewidth of 22 meV at room temperature and of 20 meV at 4.3 K. Furthermore

up to six QWs are successfully included into SL structures without significant degradation

of the structural and optical properties of the active layer.

The main issue in the last part of this work was the combination of the ZnSe/(Zn,Cd)Se

quantum structures with the dielectric Bragg-mirrors to a complete microcavity. There-

fore a SL microcavity with ZnSe/YF3DBRs as well as a MQW microcavity with ZnS/YF3

mirrors were successfully performed into complete microcavities.

The interaction of the photonic and excitonic mode was characterized in both structures by

room temperature reflectivity measurements at various positions on the sample. Therefore

the photonic mode was tuned into resonance with the excitonic mode via the microcavity

length gradient. In both structures the anticrossing dispersion of the two polariton states

was measured. A minimum Rabi-splitting energy of 35 meV and 36 meV is obtained with

the SL and MQW microcavity, respectively. Calculations of the polariton dispersion are

in good agreement with the experimental data of both structures. This fact yields that

the strong coupling regime is successfully demonstrated at room temperature with the SL

as well as the MQW microcavity.

Additionally to this temperature dependent photoluminescence measurements were per-

formed on the MQW microcavity. Both polariton luminescence peaks were resolved

clearly. A minimum Rabi-splitting of 44 meV is measured in the resonance condition

at 300 K and the typical anticrossing behavior is observed. These results are in good

agreement with the experimental results of the position dependent reflectivity measure-

ments.

Furthermore the QW oscillator strength per unit area was calculated from the Rabi-

splitting energy. For the SL microcavity with four Zn0.8Cd0.2Se QWs a value of f

S= 0.037

is obtained. In case of the MQW microcavity with four Zn0.65Cd0.35Se QWs the oscillator

strength per unit area is about f

S= 0.056, averaged from the PL and reflectivity mea-

surements. The comparison between both structures yields that the oscillator strength

increases with an increased cadmium mole fraction in the (Zn,Cd)Se QWs.

6 Conclusions

List of Figures

2.1 Interaction of excitons and photons in a semiconductor microcavity. . . . . 16

2.2 Calculation of the polariton dispersion of a microcavity . . . . . . . . . . . 20

2.3 Reflection and transmission of an electromagnetic wave on a plane layer

surface. ..................................... 21

2.4 Complex refractive index dispersion relations of ZnS and ZnSe. . . . . . . . 23

2.5 Alignment of the evaporation cells in our MBE chamber to implement

growth gradient in the active layer. . . . . . . . . . . . . . . . . . . . . . . 26

2.6 Principle of temperature detuning of a microcavity. . . . . . . . . . . . . . 27

3.1 Reflectivity of an eight-fold stack of ZnS/YF3DBRs. ............ 30

3.2 Reflectivity of a ten-fold stack of ZnS/YF3DBRs............... 31

3.3 Reflectivity of a five-fold stack of ZnS/YF3DBRs............... 33

3.4 AFM picture of a 20 µm×20 µm scan of the ZnS surface of a six-fold

stack of ZnS/YF3DBRs............................. 34

3.5 RMS surface roughness of several ZnS/YF3DBRs............... 35

3.6 Maximum reflectivity of several ZnS/YF3DBRs as a function of the number

ofperiods..................................... 36

3.7 Reflectivity spectrum of a ten-fold stack of ZnSe/YF3DBRs. . . . . . . . . 38

3.8 Reflectivity spectrum of a five-fold stack of ZnSe/YF3DBRs. . . . . . . . . 39

3.9 AFM picture of a 20 µm×20 µm scan of the ZnSe surface of a ten-fold

stack of ZnSe/YF3DBRs. ........................... 41

3.10 Reflectivity and surface roughness of different ZnSe/YF3DBRs. . . . . . . 42

List of Figures

4.1 Intensity oscillations of the specular spot during deposition of ZnSe on

GaAssurface................................... 47

4.2 Intensity oscillations of the specular spot during deposition of a (Zn,Cd)Se

QW. ....................................... 48

4.3 Cross-section drawing of MQW structures with up to four (Zn,Cd)Se QWs. 50

4.4 Reciprocal space map around the GaAs-(224) reflex of the MQW structure. 51

4.5 Symmetric (004)-reflex ω−2θ−scan of a MQW structure. . . . . . . . . . 52

4.6 Symmetric ω-scan of the (004)-reflex of ZnSe in MQW structure McS-1072. 54

4.7 Symmetric ω−2θ−scan of a double QW structure and x-ray simulation

spectrum. .................................... 55

4.8 Symmetric ω-scan of the ZnSe-(004)-reflex of the double QW structure

McS-775...................................... 56

4.9 PL spectra of a MQW structure at room temperature and 4.3 K. . . . . . 57

4.10 PL spectra of a double QW structure at room temperature and 4.3 K. . . . 59

4.11 Cross-section drawing of SL structures with four and six (Zn,Cd)Se QWs. . 60

4.12 Reciprocal space map around the GaAs-(224)-reflex of the SL-QW struc-

tureMcS-851................................... 61

4.13 Symmetric (004)-reflex ω−2θ−scan of a SL structure (dots) with four QWs. 62

4.14 ZnSe-(004)-reflex ω-scan of the four-fold SL structure McS-851. . . . . . . 64

4.15 ZnSe-(004)-reflex ω−2θ−scan of a six-fold SL structure. . . . . . . . . . . 65

4.16 ZnSe-(004)-reflex ω-scan spectrum of the six-fold SL structure McS-1014. . 66

4.17 Cross section TEM-picture of a six-fold SL structure. . . . . . . . . . . . . 67

4.18 PL spectra (dots) of the four-fold SL structure McS-851 at room tempera-

ture(leftside)and4.3K. ........................... 69

4.19 PL spectra (dots) of the six-fold SL structure McS-1014 at room tempera-

ture (left side) and 4.3 K. . . . . . . . . . . . . . . . . . . . . . . . . . . . 70

5.1 Cross-section drawing of a typical ZnSe/(Zn,Cd)Se SL structure with four

QWs........................................ 72

List of Figures

5.2 Reflectivity spectra of the microcavity structure as a function of LC. . . . . 73

5.3 Microscope picture of a SL structure after selective wet etching. . . . . . . 75

5.4 Position dependent reflectivity spectra of the SL microcavity. . . . . . . . . 77

5.5 Absorption peak energy ECof the cavity mode as a function of the position

sonthesample.................................. 78

5.6 Room temperature reflectivity spectra of the SL microcavity near the res-

onanceposition.................................. 79

5.7 Absorption peak energy of each spectrum in Fig. (5.6) as a function of the

resonator detuning energy. . . . . . . . . . . . . . . . . . . . . . . . . . . . 80

5.8 Cross-section drawing of the as-grown MQW microcavity PE-013. . . . . . 82

5.9 Room temperature reflectivity spectra of the MQW microcavity measured

off-resonance. .................................. 83

5.10 Room temperature PL spectrum of the MQW microcavity measured off-

resonance..................................... 84

5.11 PL spectra of the microcavity PE-013 between 270 K and 330 K. . . . . . 85

5.12 Temperature dependent polariton dispersion of the MQW microcavity PE-

013. ....................................... 87

5.13 Overview of the room temperature reflectivity spectra of sample PE-013

neartheresonance................................ 88

5.14 Absorption peak energies of the reflectivity spectra of PE-013 as a function

of the microcavity detuning energy. . . . . . . . . . . . . . . . . . . . . . . 90

List of Figures

100

List of Tables

2.1 Parameters for the calculation of the complex refractive index dispersions

ofZnSe,ZnSandGaAs............................. 24

3.1 Experimental (left side) and theoretical (right side) parameters of different

ZnS/YF3DBRs. ................................ 32

3.2 Experimental and theoretical parameters of different ZnS/YF3DBRs. . . . 40

5.1 Calculation of the QW exciton oscillator strength per unit area from the

Rabi-splittingenergies.............................. 91

101

List of Tables

102

Bibliography

[Af1974] M.A. Afromowitz,

Solid State Comm. 15, 1974, p. 59.

[Al1975] L. Allen and J.H. Eberly,

”Optical resonance and two-level atoms”,

Wiley, New York, 1975.

[An2000] R. Andr´e, F. Boef, R. Romestain, Le Si Dang, E. P´eronne, J.F. Lampin,

D. Hulin and A. Alexandrou

J. Chrystal Growth 214/215, 2000, p. 1002.

[Ar2003] C. Arens,

”Herstellung und Charakterisierung von Quantenstrukturen in Mikrokavit¨aten”,

Diplomarbeit, Fakult¨at f¨ur Naturwissenschaften,

Universit¨at Paderborn, 2003.

[Ba1995] V. Bardinal, R. Legros and C. Fontaine,

Appl. Phys. Lett. 67, 1995, p. 3390.

[Ba2002] J.J. Baumberg,

Physics World, March 2002.

[Ba2003] M. Bartels,

”Self-assembled epitaxial growth of CdSe Quantum Dot Structures”

103

Bibliography

Dissertation, Fakult¨at f¨ur Naturwissenschaften,

Universit¨at Paderborn, submitted 2003.

[Be1994] P.R. Berman,

”Cavity quantum electrodynamics”,

Academic, Boston, 1994.

[Bl1998] J. Bloch, T. Freixanet, J.Y. Marzin, V. Thierry-Mieg and R. Planel,

Appl. Phys. Lett. 73 Nr. 12, 1998, p. 1694.

[Ca1989] H.J. Carmichael, R.J. Brecha, M.G. Raizen, H.J. Kimble and P.R. Rice,

Phys. Rev. A 40, 1989, p. 5516.

[Ci2002] M. Saba, C. Ciuti, J. Bloch, V. Thierry-Mieg, R. Andr´e, Le Si Dang, S. Kun-

dermann, A. Mura, G. Bongiovanni, J.L. Staehli and B. Deveaud,

Nature 414, 2002, p. 731.

[Di1993] J. Ding, M. Hagerott, T. Ishihara, H. Jeon and A.V. Nurmikko,

Phys. Rev. B 47, Nr. 16, 1993, p. 10528.

[Di2001] J.H. Dickerson, E.E. Mendez, A.A Allerman, S. Manotas, F. Agull´o-Rueda and

C. Pecharrom´an

Phys. Rev. B, /bf 64, 2001, p. 155302.

[Ea1995] D.B. Eason, Z. Yu, W.C. Hughes, W.H. Roland, C. Boney, J.W. Cook Jr. and

J.F. Schetzina,

G. Cantwell and W.C. Harsch,

Appl. Phys. Lett. 66 Nr. 2, 1995, p. 115.

[Ge1998] J.M. G´erard, B. Sermage, B. Legrand, E. Costard and V. Thierry-Mieg,

Phys. Rev. Lett. 81, 1998, p 1110.

[Gu1982] H.E. Gumlich, D. Theis and D. Tschierse,

”Physics of II-VI and I-VII compounds, semimagnetic semiconductors”,

104

Bibliography

edited by O. Madelung, Landolt B¨ornstein,

New Series, Group III, Vol. 17b, Springer-Verlag Berlin, 1982.

[He1979] E. Hecht and A. Zajac,

”Optics”,

Addison-Wesley Publishing Company, 1979.

[Ho1958] J. J. Hoplfield,

Phys. Rev. 112 1958, p. 1555.

[Je1995] H. Jeon, V. Kozlov, P. Kelkar and A.V.Nurmikko, C.-C. Chu, D.C. Grillo,

J. Han, G.C.Hua and R.L.Gunshor

Appl. Phys. Lett. 67 Nr. 12, 1995, p. 1668.

[Ke1995] P. Kelkar, V. Kozlov, H. Jeon and A.V. Nurmikko, C.-C. Chu, D.C. Grillo,

J. Han, C.G. Hua and R.L. Gunshor

Phys. Rev. B 52 Nr. 8, 1995, p. R5491.

[Kh1999] G. Khitrova and H.M. Gibbs, F. Jahnke, M. Kira and S.W. Koch,

Rev. Mod. Phys. 71, 1999, p 1591.

[Kh2003] A. Kharchenko,

”A new x-ray diffractometer for the online monitoring of epitaxial processes”,

Dissertation, Fakult¨at f¨ur Naturwissenschaften,

Universit¨at Paderborn, 2003.

[No1999] T.B. Norris,

”Excitons in strongly coupled microcavities”,

ed. by A. Miller and D. Finlayson,

Institute of Physics, 1999, p.121.

[Pa2000] A. Pawlis,

105

Bibliography

”Epitaktisches Wachstum und Charakterisierung von Mikrokavit¨aten”,

Diplomarbeit, Universit¨at Paderborn, 2000.

[Pa2001] A. Pawlis, A. Kharchenko, O. Husberg, D. Schikora and K. Lischka,

phys. stat. sol. (a) 188, 2001, p. 983.

[Sa1974] M. Sargent, M.O. Scully and W.E. Lamb,

”Laser Physics”,

Addison Wesley, Reading, 1974.

[Sa2002] M. Saba, C. Ciuti, S. Kundermann, J.L. Staehli, B. Deveaud, J. Bloch,

V. Thierry-Mieg, R. Andr´e, Le Si Dang, G. Bongiovanni and A. Mura,

phys. stat. sol. (a) 190 Nr. 2, 2002 p. 315.

[Se2000] T. Seedorf, M. Corneli¨sen, K. Leonardi, D. Hommel, H. Selke and P.L. Ryder,

J. Cryst. Growth 214/215, 2000, p. 602.

[Sk1998] M.S. Skolnick, T.A. Fisher and D.M. Whittaker,

Sem. Sci. Tech. 13, 1998, p. 645.

[Ta1997] Ch. Tanguy, P. Lefebvre, H. Mathieu and R.J. Elliot,

J. Appl. Phys. 82 Nr. 2, 1997, p. 798.

[Ta2000] T. Tawara, H. Yoshida, T. Yogo, S. Tanaka and I. Suemune,

J. of Crystal Growth 221 2000, p. 699.

[Tr1996] A. Tredicucci, Y. Chen, V. Pellegrini, M. B¨orger and F. Bassani

Phys. Rev. B 54 Nr. 4, 1996, p. 3493.

[Uu1995] P. Uusimaa, K. Rakenus, A. Salokatve and M. Pessa,

T. Aherne, J.P. Doran, J. O’Gorman and J. Hegarty

Appl. Phys. Lett. 67, 1995, p. 2197.

106

Bibliography

[We1971] S.H. Wemple and M. DiDomenico, Jr.

Phys. Rev. B 3Nr. 4, 1971, p. 1339.

[We1992] C. Weisbuch, M. Nishioka, A. Ishikawa and Y. Arakawa

Phys. Rev. Lett. 69 Nr. 23, 1992, p. 3214.

[We2000] C. Weisbuch, H. Benisty and R. Houdr´e,

J. of Luminescence 85, 2000, p. 271.

[Wo1995] K. Wolf, S. Jilka, A. Rosenauer, G. Sch¨utz, H. Stanzl, T. Reisinger and W. Geb-

hardt,

J. Phys. D 28, 1995, p. A120.

[Wo2000] D. Wolfframm, D.A. Evans, D.I. Westwood and J. Riley

J. Cryst. Growth 216, 2000, p. 119.

[Ya2002] Y.Yamamoto, F. Tassone and H. Cao,

”Semiconductor cavity quantum electrodynamics”,

Springer-Verlag Berlin Heidelberg, 2000.

[Yo1995] H. Yokoyama and K. Ujihara,

”Spontaneous emission and laser oscillation in microcavities”,

CRC Press, Boca Raton, 1995.

107

Bibliography

108

7 Appendix

List of publications

A. Pawlis∗, O. Husberg, A. Kharchenko, K. Lischka and D. Schikora,

”Structural and optical investigations of ZnSe based semiconductor microcavities”,

phys. stat. sol. (a) 188 Nr. 3, 2001, p. 983.

A. Pawlis∗, A. Kharchenko, O. Husberg, D.J. As, K. Lischka and D. Schikora,

”Large room temperature Rabi-splitting in a ZnSe/(Zn,Cd)Se semiconductor microcavity

structure”,

Solid State Comm. 123, 2002, p. 235.

A. Pawlis∗, A. Kharchenko, O. Husberg, D.J. As, K. Lischka and D. Schikora,

”Large room temperature Rabi-splitting in II-VI semiconductor microcavity quantum struc-

tures”,

Microelect. J. 34, 2003, p. 439.

A. Pawlis∗, A. Kharchenko, O. Husberg, K. Lischka and D. Schikora,

”Preparation and properties of ZnSe/(Zn,Cd)Se multi quantum well structures for room

temperature polariton emission”,

submitted to J. of Phys. Cond. Matter, 2003.

109

7 Appendix

T. Frey∗, D.J. As, M. Bartels, A. Pawlis, K. Lischka, A. Tabata, J.R.L. Fernandez,

M.T.O. Silva, J.R. Leite, C. Haug and R. Brenn,

”Structural and vibrational properties of MBE grown cubic (Al,Ga)N/GaN heterostruc-

tures”,

J. of Appl. Phys. 89 Nr. 5, 2001, p. 2631.

Conference contributions

A. Pawlis∗, D. Schikora and K. Lischka,

”High reflectivity ZnS/YF3distributed Bragg-mirrors for microcavites in the green/blue

spectral range”,

DPG conference 2001, Hamburg.

A. Pawlis∗, D. Schikora, O. Husberg, A. Kharchenko and K. Lischka,

”II-VI semiconductor microcavities for light emission in the green spectral range”,

II-VI conference 2001, Bremen.

A. Pawlis∗, M. Bartels, A. Khartchenko, O. Husberg, D. Schikora and K. Lischka,

”ZnSe/(Zn,Cd)Se semiconductor microcavities with dielectric Bragg mirrors of ZnS and

YF3”,

DPG conference 2002, Regensburg.

110

Acknowledgements

I would like to thank Prof. Dr. Klaus Lischka who gave me the opportunity to work

in his group and for the valuable discussions during my PhD thesis.

The same gratitude goes to Dr. Detlef Schikora who supported me with all his knowl-

edge in epitaxy. My thanks also goes to Apl. Prof. Dr. Donat Josef As for the

profoundly discussions and his help regarding optical properties in semiconductors and

semiconductor microcavities.

Furthermore my gratitude goes to Prof. Le Se Dang,Prof. Gerard and Dr. Ballocci

during my stay at the University of Grenoble for their experience regarding the theory of

strong coupling and optical phenomena in microcavities.

I would like to thank the PhD students Christof Arens,Martin Bartels,Alexan-

der Kharchenko,Olaf Husberg,Ulrich K¨ohler,Stefan Potthast and J¨org Sch¨or-

mann and the master degree student Kathrin Friedrich for their cooperation and help-

ful discussions.

My gratitude goes also to Ralf Winterberg for his help regarding the preparation of

microcavities.

I would like to thank the ”Optoelectronic“ staff namely Irmgard Zimmermann and

Bernard Volmer for their help in bureaucratical things and Siegfried Igges for the

enormous and excellent technical engagement and support during the set-up of the new

MBE machine.

Furthermore my gratitude goes to our collaborators Prof. Yamamoto and Dr. Weihs

from Stanford University for their valuable work and contributions.

Moreover I have to thank Andrea Bartels for her love and encouragement she gave me

during the ”hot-phase“ of my PhD work.

Last but not least I would like to express thanks to my parents Monika Pawlis and

Gunter Pawlis for their love, encouragement and financial support as well as to my best

friend Uwe Klingeberg for encouragement and some good advice in bad days of my

work.

111

7 Appendix

This work was supported by the European Network Project ”Photon Mediated Phenom-

ena”, HPRN-CT-2002-00298.

112