physrevb.85.125311

Upload: graciely-santos

Post on 14-Apr-2018

222 views

Category:

Documents


0 download

TRANSCRIPT

  • 7/29/2019 PhysRevB.85.125311

    1/7

    PHYSICAL REVIEW B 85, 125311 (2012)

    Influence of electronic coupling on the radiative lifetime in the (In,Ga)As/GaAs quantumdotquantum well system

    M. Syperek,1,* J. Andrzejewski,1 W. Rudno-Rudzinski,1 G. Sek,1 J. Misiewicz,1 E. M. Pavelescu,2, C. Gilfert,2 and

    J. P. Reithmaier2

    1Institute of Physics, Wrocaw University of Technology, Wybrzeze Wyspianskiego 27, 50-370 Wrocaw, Poland2

    Institute of Nanostructure Technologies and Analytics, Technische Physik, Universitaet Kassel, Heinrich-Plett-Strasse 40,34132 Kassel, Germany

    (Received 16 November 2011; revised manuscript received 12 March 2012; published 22 March 2012)

    In this paper, we present time-resolved photoluminescence of self-assembled (In,Ga)As quantum dots (QDs)

    coupled to anInxGa1xAs quantum well (QW) through a thin GaAs barrier.The coupling strength is controlled by

    the well width and material composition and has a direct influence on the QDQW system ground-state radiative

    lifetime, which increases from 1.1 up to 1.82 ns as the coupling strength increases. A thorough explanation is

    given for the reduction of the fundamental transition oscillator strength in the interacting system induced by

    the transformation of related eigenstates from a noncoupled to a coupled regime. The experimental observations

    have been fully confirmed by the results of three-dimensional band structure calculations of a coupled QDQW

    system within the framework of eight-band kp theory.

    DOI: 10.1103/PhysRevB.85.125311 PACS number(s): 78.66.Fd, 78.67.De, 68.65.Hb, 78.47.D

    I. INTRODUCTION

    Recent rapid advancement in the fabrication techniques of

    semiconductor quantum dot (QD) structures has been followed

    by a remarkable amount of research devoted to the exploration

    of their unique optical properties. A number of spectroscopic

    results on a single semiconductor QD have been obtained,

    which confirmed its quasiatomic character being a conse-

    quence of the three-dimensional (3D) carrier confinement.1,2

    However, the QD is very rarely a standalone object. Its

    coupling to the environment is of primary importance for

    current application and future technological implementations,

    including quantum data processing based on spin or charge

    state.3,4 A QD can interact with other dots, defect states,surface states, etc., which in turn leads to the modification of

    its intrinsic physical properties. In the following, we consider

    the coupling of a quasi-0D QD ground state to quantum well

    (QW) 2D extended states.

    The concept of a coupled QDQW system has been

    proposed for the realization of the tunnel injection of carriers

    from their large reservoir (QW) to QDs. It has been suc-

    cessfully implemented for single-electron transistors,5,6 tunnel

    injection lasers,79 and QD-based memories.10 Surprisingly,

    some basic physical phenomena in such coupled systems

    remain unexplored or at least their nature is unclear. A few

    interesting questions are still open related to the full band

    structure picture, the intraband and interband carrier scatteringprocesses, charge- and spin-transfer mechanisms between

    the QW and QD in either coherent or incoherent regimes,

    and the carrier dynamics on the ground state of an entire

    QDQW system in the presence of an electronic coupling.

    The issue of a coherent charge transfer between resonantly

    coupled high-energy states of the QW and QD has been very

    recently addressed.11 In this work, we elucidate the role of a

    coupling regime between the QD and QW and its strength on

    the coupled-system ground-state carrier dynamics beyond the

    coherent regime of interaction.

    In this respect, we have performed continuous-wave

    (CW) photoluminescence (PL) and photoreflectance (PR)

    experiments, which let us identify the dipole-allowed ground-

    state and excited-state interband transitions in a set of

    (In,Ga)As/GaAs coupled QDQW structures. The variable

    indium content in the well and the well-width changes allowed

    us to control the coupling strength between the QD and QW,

    defined as the influence of the well potential on the states

    confined in QDs. For the sake of comparison, a reference

    (In,Ga)As/GaAs QD structure has also been considered.

    Information concerning a possible coupling regime is acquired

    based on the data analysis of a time-resolved photolumines-

    cence (TRPL) experiment, additionally supported by a 3D

    band structure calculation of the entire system within the

    eight-band kp approach.

    II. EXPERIMENTAL DETAILS

    The investigated structures, called CS1, CS2, CS3, and

    QDref, were grown by molecular-beam epitaxy on a semi-

    insulating (001) GaAs substrate.12 The CS1 consists of a

    15-nm-wide In0.2Ga0.8As QW separated by a 2-nm-thick

    GaAs barrier from In0.6Ga0.4As QDs obtained by depositing

    nominally 1.8 monolayers of the (In,Ga)As material. The well

    and dots are surrounded by two 15-nm-thick layers of GaAs,

    and, finally, the entire QDQW system is sandwiched between

    AlGaAs/GaAs superlattices. The CS2 and CS3 have a similar

    layer structure and differ from the CS1 in the well width,

    which is 7 nm for both, and in the In composition in thewell, which is nominally 23% and 27%. These parameters

    are summarized in Table I. Additionally, the (In,Ga)As QDs

    reference structure (QDref) has been produced under the same

    growth conditions without the QW. The surface density of

    dots in each structure is estimated to be on the order of

    1010 cm2. The layer sequence of samples is schematically

    depicted in Fig. 1. For the CW PL and TRPL experiments,

    samples were held in a variable-temperature cryostat. In the

    case of the CW experiment, the photoexcitation was provided

    by the 532 nm line from the yttrium aluminum garnet laser.

    For the TRPL experiment, the sample was excited by a

    mode-locked Ti:sapphire laser, which generates pulses with

    125311-11098-0121/2012/85(12)/125311(7) 2012 American Physical Society

    http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://dx.doi.org/10.1103/PhysRevB.85.125311http://dx.doi.org/10.1103/PhysRevB.85.125311http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-
  • 7/29/2019 PhysRevB.85.125311

    2/7

    M. SYPEREK et al. PHYSICAL REVIEW B 85, 125311 (2012)

    TABLE I. The parameters of the coupled QDQW structures.

    Structure QW width (nm) QW [In] content (%)

    CS1 15 17

    CS2 7 23

    CS3 7 27

    a duration of 140 fs at a repetition rate of 76 MHz. In the

    case of the TRPL experiment, the PL signal was dispersed

    by a 0.3-m-focal-length monochromator and detected by a

    streak camera system equipped with the S1 photocathode. The

    overall time resolution was on the level of20ps.The CWPL

    emission was measured using the same setup with the InGaAs

    CCD linear array detector. The PR experiment has been

    performed at room temperature on a separate setup described

    in Ref. 13 (low-temperature PR has not been used due to strong

    PL background). In order to compare the low-temperature CW

    PL spectra with those obtained in the PR experiment, the latter

    were shifted according to the temperature-induced band-gap

    shift.

    III. EXPERIMENTAL RESULTS

    The high-excitation CW PL and PR spectra for investigated

    structures are shown in Fig. 2. The QDref PL spectrum reveals

    two emission bands centered at 1.25 and 1.31 eV. They are

    FIG. 1. (Color online) The layer sequence of the coupled-system

    (CS) structure (top), together with a simple sketch of the valence

    band (VB) and conduction band (CB) energy profiles of the region

    of interest (bottom). The QW ground state (GS) and QD GS denote

    optical transitions between states primarily localized in the QW and

    QD, respectively.

    FIG. 2. (Color online) High-excitation, low-temperature CW

    photoluminescence (red line) and room-temperature photoreflectance

    spectra (blackline, shifted to higher energy accordingto the band-gap

    temperature change) of a reference QD structure (QDref, top panel)

    and coupled quantum dotquantum well systems (CS1CS3, bottom

    panels).

    attributed, respectively, to the ground-state and excited-state

    transitions in the ensemble of dots. In the case of coupled

    systems (CS1CS3), the interpretation of PL spectra is not so

    evident. However, the lowest PL emission band centered at

    1.22 eV can be assigned to the dipole-allowed ground-statetransition of the entire QDcoupledQW system. To identify

    the origin of the higher-energy optical transitions, we per-

    formed the PR experiment. The PR spectra presented in Fig. 2

    for the coupled structures reveal a weak feature at 1.22 eV,

    which corresponds well to the ground-state QD transition

    observed in the emission spectrum. At the higher-energy side,

    there appears a considerably more intensive PR resonance,

    which changes itsposition from 1.348eV for theCS1 structure,

    to 1.315 eV for CS2, down to 1.278 eV for CS3. This optical

    transition involves confined states in thecoupled system, which

    have acquired a significant amplitude of the eigenfunction on

    the QW side (see sketch in Fig. 1, bottom). The transition

    matrix elements between states primarily localized in the

    QW (QW-like states) will exceed those of the QD-like or

    mixed-type character, which are indirect in the real space.

    Consequently, this should lead to the experimentally observed

    increase in the PR resonance intensity contributed by a large

    oscillator strength of such an optical transition. The spectral

    redshift of the QW-like transition in the coupled system is

    strongly related to the In content in the well and its width.

    These two parameters define the QW potential depth and

    thus can control the coupling regime of the coupled-system

    fundamental state. One would expect that if the QW-like

    transition is energetically closer to the ground-state transition

    of the entire system, then the coupling strength increases. Such

    a tentative hypothesis holds for the structures under study,

    125311-2

  • 7/29/2019 PhysRevB.85.125311

    3/7

    INFLUENCE OF ELECTRONIC COUPLING ON THE . . . PHYSICAL REVIEW B 85, 125311 (2012)

    FIG. 3. (Color online) Schematic picture of the square of wave

    function moduli in the x-y plane (top panels) and in the cross section

    along the z (growth) axis (bottom panels) for three coupling regimes.

    as we will prove later. The energy distance E between the

    QW-like andthe ground-state transition is equal to 128, 95, and

    48 meV, respectively, for the CS1, CS2, and CS3 structures,

    as illustrated in Fig. 2. The three possible coupling regimes

    are schematically depicted in Fig. 3. The noncoupled regimeis characterized by the carrier probability density |(r)|2

    restricted essentially to the QD area, while in the weakly

    coupled QDQW system, the probability density slightly leaks

    out from the QD confining potential toward the QW. Finally,

    in the strongly coupled system, the carrier |(r)|2 is smearedover both the QD and QW confining potentials.

    Let us turn to the QDref and CS structures ground-state

    emission dynamics. The electron-hole recombination process

    can be described within the scattering theory, which is usually

    summarized in Fermis golden rule, which connects the

    electron-hole pair radiative decay time () and the transition

    oscillator strength (fosc) in the following expression: 1

    fosc, while fosc |e,h(r)|2

    . It leads to the conclusion thatmeasuringthe radiative decay timemaps out the spatial overlap

    between the electron and hole wave functions and thus gives

    insight into the coupling regime of the investigated system.

    Figure 4(a) presents TRPL traces taken at the maximum

    of the ground-state emission band for (In,Ga)As/GaAs QDs

    (QDref) and (In,Ga)As/GaAs coupled QDQW structures

    (CS1CS3) with different In content in the QW, and thus

    different E. The structures were excited nonresonantly with

    photons of 1.6 eV energy and the average pump power density

    ofP = 0.42 W cm2. These excitation conditions correspond

    to less than a single electron-hole pair generated per dot.

    Note that the CW PL measurements presented above were

    performed at a much higher excitation density than in the

    TRPL experiment in order to obtain the emission from the

    excited states of the system. Each of the experimental TRPL

    traces can be described by a single exponential decay process:I (t) = A e(t/PL), where A is the preexponential factor andPL is the emission decay time constant. The fitting function

    is given as a straight line in Fig. 4(a) in a semilogarithmic

    scale. Since the experiment was performed at low temperature

    (T = 4.5 K) and in weak excitation conditions below the level

    corresponding to a single electron-hole pair created per dot,

    we believe that the obtained PL is not considerably modified

    by nonradiative recombination processes and/or multicharging

    effects. Thus, PL is very close to the intrinsic electron-hole

    radiative recombination time; see, e.g., Refs. 1416. However,

    arb.units

    FIG. 4. (Color online) (a) Low-temperature TRPL traces (open

    circles) taken at maximum of PL intensity for (In,Ga)As/GaAs

    QDs (QDref) and (In,Ga)As/GaAs QDQW coupled systems (CS1

    CS3). The solid line indicates the fit to the experimental data.

    (b) Experimental (full circles) and theoretical (open circles) PL

    decay time constants vs detuning energy (E). Error bars are for

    the theoretical values based on the uncertainty of the value of the

    refractive index (5%). Dashed line indicates the PL decay timeconstant for the reference QD structure. Dotted line is a guide to

    the eye. Laser photon energy is 1.6 eV, Ppump = 0.42 W cm2.

    recent papers by the group of P. Lodahl17,18 claim that the

    nonradiative component to the measured decay time of QD

    emission cannot be neglected, thus the radiative decay time

    is, in fact, probably longer than the measured one (the

    difference being on theorder of 1020%, accordingto Ref. 18).

    Even if this is true also for the structures investigated here

    (which is not evident, since the PL decays we registered are

    monoexponential, contrary to the ones in Refs. 17 and 18),

    the potential changes in the nonradiative processes, which

    decrease the total decay time, explain neither the directionof the changes in the lifetimes measured for coupled systems

    nor their magnitudes.

    The PL vsE in the coupled system is plotted in Fig. 4(b).

    The QDref ground-state recombination time is shown by a

    dashed line at the level of PL = 1.1 ns resulting from the

    intrinsic properties of a set of similar QD emitters of a given

    size, shape, and chemical content. If we assume that an in-

    plane coupling between QDs is negligible, these QDs can

    be treated as a good example of a quantum system being in

    the noncoupled regime. The PL is defined in this case by

    the spatial coherence of an electron-hole pair confined in the

    QD.19,20 A very similar PL decay time of 1.1 ns has been

    obtained for the CS1, where the QD structure was modified

    by introducing the 15-nm-wide (In,Ga)As QW with 17% of

    indium, and separated from the QD layer by a thin 2-nm-wide

    GaAs barrier. An almost identical decay time constant for

    the CS1 and QD structures suggests that the emitter has the

    same or very similar parameters in both cases.It brings us to the

    conclusion that the ground-state emission in the CS1 system

    originates from the radiative recombination of electron-hole

    pairs localized primarily in the quasi-0D potential of the dot.

    Moreover, the dot itself remains in the noncoupled regime,

    where the CS1 ground state is not significantly perturbed by

    the presence of the QW [see Fig. 3(a)].

    A considerable increase in the PL lifetime is observed for

    the CS2 and CS3 structures compared to the QDref one. In

    125311-3

  • 7/29/2019 PhysRevB.85.125311

    4/7

    M. SYPEREK et al. PHYSICAL REVIEW B 85, 125311 (2012)

    FIG. 5. Excitation-power dependence of low-temperature PL

    decay times.

    order to explain the increase in PL for CS2 and CS3, we

    exclude a number of effects which could have an impact on

    the elongation of the decay time constant, and conclude that

    it can only be affected by the increasing electronic couplingstrength in these structures, leading to the reduction of the

    electron-hole wave-function overlap. We discard the following

    alternative explanations:

    (1) the ground-state filling effect similar to the one observed

    in QDs;

    (2) modification of the QD surface density, leading to the in-

    plane coupling between them, thus increasing the probability

    of charge-transfer processes or smearing the carrier probability

    density function between adjacent dots;

    (3) change of the dot size and/or shape, thus affecting the

    coherence volume.

    The ground-state filling effect is commonly observed at

    high optical pumping regimes. In order to investigate thispossibility, we have measured the excitation-power depen-

    dence of PL decay times for all samples. Figure 5 shows the

    determined PL lifetimes for theQDref and CS1CS3structures

    as a function of excitation-power density. The PL, for the

    QDref and CS1, does not exhibit significant changes over the

    three decades of the average excitation-power densities. On

    the other hand, for CS2 and CS3, an increase in the excitation

    density from 0.42 to 30 Wcm2 causes a noticeable increase

    in PL from 1.33 up to 1.7 ns, and from 1.82 up to

    3.5 ns, respectively. This can be explained by the state filling

    due to the constant refilling of the ground state from the upper

    lying states of the system. However, for both structures, there

    is a range of excitation-power densities where the lifetimes

    are almost constant andstill therelation between therespectivePL is conserved, meaning that the state filling effect alone

    does not explain the observed behavior.

    The QD ground-state radiative decay time can be modified

    by the electronic coupling within the QD layer, which is

    essentially controlled by the QD surface density or by changes

    in the QD size. However, the growth process parameters for

    the QDref and CS structures were the same. Although some

    change in the QD size parameters could occur due to the

    fact that the dots in the CS structures are grown on top of

    the QW, we do not expect a considerable increase in the QD

    surface density because the modification of strain is negligible,

    as shown by our calculations and structural data. For the

    same reason, the QD size will be affected only very slightly

    and hence the coherence volume can also be assumed to be

    constant. Considering the interdot coupling, for the dot surface

    density of 1010 cm2, the average distance between two

    adjacent dots in the layer exceeds 70 nm, which makes the

    direct electronic coupling between them negligible.

    IV. NUMERICAL CALCULATIONS

    The TRPL experiment cannot identify whether the elec-

    tronic coupling is related to the conduction- or valence-band

    states simply due to the fact that we probe the temporal

    evolution of the product of photogenerated electron and hole

    distribution functions. To shine some light on this issue, we

    have performed three-dimensional band structure calculations

    of reference QD and coupled QDQW systems. We have

    developed a 3D eight-band kp model, whose implementationincludes strain fields,21 piezoelectric effects, and spin-orbit

    interaction.22,23 For simplicity, we neglected the Coulomb

    correlation between electron-hole pairs (excitons). We believe

    that even without taking into account excitons in our model,we can still well explain the observed experimental results

    in a qualitative way. However, we discuss later the possible

    influence of the Coulomb correlation between the electron and

    hole on the radiative lifetimes. Potential approaches to the

    modeling of the exciton are given in Ref. 24. All physical

    equations were numerically solved using the finite-difference

    method.22,23 The calculation of the ground-state energy level

    in the QD and CS structures is performed in a similar approach

    to Ref. 25, with some modifications. One is the introduction

    of the effect of nonlinear piezoelectricity (which, however,

    does not have a considerable influence on the current system

    due to the small sizes of investigated QDs).22,26 Another,

    more important modification is the proper inclusion of theboundary conditions that would correctly take into account the

    existence of a QW-like substructure separated from the QD.

    Typically, in order to calculate QW energy states, the 1D

    picture is usedwith the confinement potential along the z

    direction and a geometric translational symmetry used for

    other directions, with free particlelike picture solutions. On

    the other hand, in the standard QD-like structures, imposing

    the Dirichlet boundary condition is sufficient. However, in our

    case, we have to be able to calculate the energy levels in a

    QW in the same 3D picture as for the QDs. Therefore, we

    impose in our model the periodic boundary conditions so that

    we can calculate the QW states in the 3D model, which agree

    very well with the states calculated in the 1D model, even with

    the strain-related effect switched on. The material parameters

    connected with the piezoelectric effect are taken from Ref. 26.

    Other necessary material parameters are from Ref. 27 without

    any changes.

    As a test of the model, we have determined the energy-

    level structure of the standalone QD. Based on the structural

    data, we assumed a pyramidal QD with a base length of 16

    nm, truncated at 1.7 nm and sitting on a wetting layer, which

    is 3 ML thick for this In content.28 The In0.5Ga0.5As QD is

    immersed in the GaAs matrix. Lattice constants and energy-

    band gaps are taken at T = 4.2 K. The calculated QD ground-

    state energy is 1.25 eV, whereas the excited-state energy is

    1.31 eV. It remains in a very good quantitative agreement with

    125311-4

  • 7/29/2019 PhysRevB.85.125311

    5/7

    INFLUENCE OF ELECTRONIC COUPLING ON THE . . . PHYSICAL REVIEW B 85, 125311 (2012)

    FIG. 6. (Color online) Projection of the 3D probability density

    function for electron and hole ground state onto the x-y-z plane

    calculated for the QD (top panels) and coupled QDQW system

    (bottom panels) with 7-nm-wide QW and In content of 27% in the

    well.

    the PL spectrum in Fig. 2. In this respect, such a QD can be

    treated as a good representative of the real QD family probed

    in the experiment.

    Once the QD band structure was established, we performed

    further numerical calculations of the coupled QDQW system.

    For the sake of the discussion given above, we were mostly

    interested in fundamental energy levels and respective eigen-

    functions for conduction and valence bands. We consistently

    kept the same QD parameters, but changed the indium content

    of the InxGa1xAs QW. In order to obtain a good agreement

    between the numerical and experimental values of the energiesof the ground-state transitions, we tuned the [In] of the CS1

    sample to 17%. In the next step, we performed calculations

    for QW compositions around the nominal values for the two

    QW widths ofd= 7 and 15 nm to determine the dependence

    of PL lifetimes on QW parameters.

    Figure 6 shows an example of the ground-state electron and

    hole probability density ||e,h(r)|2 in the modeled standalone

    QD and for coupled systems. It is clearly visible that in the

    former case both the electron and hole are tightly localized

    within the QD confining potential. However, if we only

    introduce a QW with sufficiently high In content beside

    the QD, then the electron wave-function density previously

    localized in the dot pours out and feels the 2D confining

    potential of the well. Note that the hole still remains strongly

    confined within the QD, showing no difference with respect

    to the standalone QD case. To provide more details, we have

    calculated the fraction of a probability density function for the

    electron and hole within the QD spatial area. We have plotted

    it in Fig. 7 as a function of the In content in the InxGa1xAs

    QW and the QW width.

    First of all, the hole wave function is always strongly

    limited to the QD area independently of the considered In

    content in the well and the well width [see open and closed

    circles in Fig. 7(a)]. Moreover, it is exactly the same as for

    the standalone QD. It brings us to the conclusion that there

    is no electronic coupling between the lowest-lying hole states

    FIG. 7. (Color online) (a) The probability of finding a hole

    (circles) or an electron (squares) on the QD side of the barrier

    as a function of indium content in the QW, for two QW widths.

    (b) Analogical dependence for calculated radiative lifetimes. Dotted

    lines are a guide to the eye.

    confined in the system with the QW. It becomes considerably

    different in the case of electrons. The square of the electron

    wave-function modulus even for the standalone QD is only

    partially localized within the QD area, while the rest leaks

    out to the environment. It does not change for the QDQW

    system as far as the In content is low and is weakly dependent

    on the well width. However, when the In content increases,

    the lowest-lying electron state in the structure with QW is

    transformed from noncoupled first to weakly and next-to-

    strongly coupled regimes. The electron probability density is

    then distributed over the QW and QD confining potential. For

    a relatively high percentage of indium in the (In,Ga)As QW,

    the lowest-lying electron wave function is almost fully out of

    the QD. We can now translate the observed effects into the

    radiative decay time of an electron-hole pair in the coupled

    system being in its ground state.

    The oscillator strength fosc is given by the equation

    fosc =2

    m0

    |e |e p|h|2

    Ee Eh, (1)

    FIG. 8. (Color online) The calculated coupled QDQW system

    band structure.

    125311-5

  • 7/29/2019 PhysRevB.85.125311

    6/7

    M. SYPEREK et al. PHYSICAL REVIEW B 85, 125311 (2012)

    where e, h are the 3D electron and hole single-particle

    envelope functions with the energy Ee and Eh for electron

    and hole, respectively, e is the polarization unit vector ofthe electromagnetic field, and p is the electron momentum.The momentum matrix element calculation also takes into

    account the strain,29 and because of time-reversal symmetry

    degeneration, incoherent averaging30 is used. The QD radiative

    lifetime is inversely proportional to the oscillator strengthfosc as described by the following equation:

    31

    =60m0c

    3

    ne21

    20fosc, (2)

    where 0 = (Ee Eh)/h is the optical transition frequency, n

    is the refractive index, m0 is the free electron mass, e is the

    electron charge, and 0 is the vacuum permittivity.

    Figure 7(b) illustrates calculated electron-hole radiative

    lifetimes of the CS ground state vs the In content in the

    QW. The point at In = 0% corresponds to the calculated

    lifetime for the standalone QD, which is 1.0 ns. It fits quite

    well the measured PL lifetime of 1.1 ns. As expected, forthe low [In], the ground state remains in the noncoupled regime

    where the radiative lifetime is governed by the coherence

    volumeoccupiedby the electron-hole pair. However, when[In]

    exceeds a certain value, the radiative lifetime increases rapidly.

    It is driven by the reduction of the electron-hole wave-function

    overlap, which decreases the transition oscillator strength. In

    this case, the system is already transformed to the coupled

    regime. Qualitatively, the same tendency as in Fig. 7(b) is

    observedfor measured PL lifetimes presentedin Fig.4(b), with

    experimental lifetimes shorter than the calculated ones. This

    discrepancy can be related to the lack of Coulomb interactions

    in the model. The effect of Coulomb correlation on oscillator

    strength depends on the confinement degree of carriers in eachof the QDQW coupled systems. Whereas holes are strongly

    confined to QDs for all of the samples, the electron wave

    function is localized in the QD for the reference and CS1

    samples only, and spills out progressively more to the QW

    layer for samples CS2 and CS3. Since the dots under study

    can be considered small (the excitons Bohr radius exceeds the

    dot dimensions), both typesof carriers in the referenceand CS1

    samples are in the strong confinement regime, where quantum

    confinement effects overwhelm the Coulomb interactions and

    excitonic effects can be neglected.3234 The situation becomes

    different for samples CS2 and CS3, where a considerable part

    of the electron wave function is in the QW layer, i.e., separated

    in the real space from the hole wave functions, meaning that

    the confinement has been weakened. In this case, the Coulomb

    attraction might be moreimportant; electronsare pulled toward

    the positions closer to QDs, increasing the oscillator strength

    and thus decreasing the lifetimes. The theoretically predicted

    radiative lifetimes presented in Fig. 4(b) are expected to be

    in better agreement with the measured ones if the Coulomb

    correlation is included. Another reason for the divergence

    between experimental and calculated lifetimes for the samples

    with electron wave functions delocalized from QDs may

    be the increasing nonradiative component to the measured

    values, resulting from the electron scattering on the GaAs

    barrier/interface states, providing nonradiative recombination

    channels.

    V. APPLICATION OUTLOOK

    The experimental results presented above, supported by

    theoretical consideration, can be of immediate relevance to the

    device performance, where the coupled QDQW system plays

    a crucial role. Here we will particularlyreferto tunnelinjection

    (TI) lasers.79,12 Despite the fact that the results presented

    above on the carrier dynamics and electronic structure inInGaAs/GaAs TI structures concern low-temperature data, the

    conclusions can be at least extrapolated to room temperature,

    i.e., normal conditions of the laser device operation. The

    energy-level structure and hence the spatial distribution of

    the probability densities for both types of carriers will not

    change significantly at room temperature. For the structure

    CS3, the measured energy difference (48 meV) between the

    lowest electron and hole energy levels of the confined carrier

    states in both parts of the system, i.e., in the QD-like and

    the injector QW-like lowest levels, seems to not be very well

    matched with the characteristic energies of the LO phonon

    (35 meV) in GaAs. However, the related energy separation

    between the respective QW-like and QD-like transitions is not

    shared equally between the conduction and valence bands.

    Our calculations show that the ratio is about 1:3 in favor of the

    valence band (mainly due to the larger effective mass of the

    holes and hence a much deeper confinement of the lowest hole

    state within the QD potential; see Fig. 8), which gives around

    36 meV for the holes. Moreover, with increasing temperature,

    the carriers become distributed over the states of the QW sub-

    band in the injector, and hence the energy difference between

    the hot carriers in the injector and the lowest state in the dot can

    approach the LO phonon energy also for electrons. This means

    that an efficient carrier transfer mediated by an LO phonon can

    take place in structures like the CS3. This is actually confirmed

    in the experimental data of TI devices (Ref. 12), where a very

    similar active region has been used, and where improvementsin the laser performance have been observed.

    VI. CONCLUSIONS

    In conclusion, we have investigated the role of cou-

    pling strength on the ground-state carrier dynamics in self-

    assembled (In,Ga)As quantum dots coupled to an InxGa1xAs

    quantum wellthrough a 2-nm-wide GaAs barrier. Thecoupling

    strength is controlled through the variations of the In content

    in the well and the well width, and is directly monitored by

    measuring the fundamental state PL lifetime of the systems

    with the QW and the reference QD structure. The PL decay

    time constant of 1.1 ns obtained for reference QDs and CS1

    confirms that the latter system with In content of 17%

    remains in the noncoupled regime in which the electron and

    hole wave functions are not affected by the QW confining

    potential. Increasing the In content leads to the elongation

    of the PL lifetime of several hundreds of ps. This effect

    is directly connected with the transformation of the related

    fundamental eigenstates from a noncoupled to a coupled

    regime, which was fully confirmed by the 3D band structure

    calculations of the coupled systems. Moreover, the numerical

    calculations suggest that in the structures with high indium

    content, an increase in the coupling strength is related to the

    delocalization of the electron wave function from 0D to 2D

    125311-6

  • 7/29/2019 PhysRevB.85.125311

    7/7

    INFLUENCE OF ELECTRONIC COUPLING ON THE . . . PHYSICAL REVIEW B 85, 125311 (2012)

    confining potential of the well, whereas holes remain strongly

    localized in the QD.

    ACKNOWLEDGMENTS

    This work has been supported by the Polish Min-

    istry of Science and Higher Education within Grant No.

    N N515 518338 and by the DeLight Project No. 224366, 7th

    Framework Programme of the European Commission. The

    fellowship support for M.S. is co-financed by the European

    Union within the European Social Fund. Part of this work

    has been realized within the NLTK infrastructure, Project No.

    POIG.02.02.00-00-003/08-00.

    *[email protected] address: National Institute of Research and Development

    for Microtechnology-IMT Bucharest, Erou Iancu Nicolae 126A,

    077190, Bucharest, Romania.1Single Semiconductor Quantum Dot, edited by P. Michler (Springer-

    Verlag, Berlin, 2009).2Semiconductor Quantum Dots, edited by Y. Masumoto and

    T. Takagahara (Springer-Verlag, Berlin, 2002).3Semiconductor Quantum Bits, edited by F. Henneberger and

    O. Benson (World Scientific, Singapore, 2008).4Optical Generation andControl of Quantum Coherence in Semicon-

    ductor Nanostructures, edited by G. Slavcheva and Ph. Roussignol

    (Springer-Verlag, Berlin, 2010).5K. K. Likharev, Proc. IEEE 47, 606 (1999).6D. L. Klein, R. Roth, A. K. L. Lim, A. P. Alivisatos, and P. L.

    McEuen, Nature (London) 389, 699 (1997).7G. Walter, N. Holonyak Jr., J. H. Ryou, and R. D. Dupuis, Appl.

    Phys. Lett. 79, 1956 (2001).8T. Chung, G. Walter, and N. Holonyak Jr., Appl. Phys. Lett. 79,

    4500 (2001).9Z. Mi, S. P. Bhattacharya, and S. Fathpour, Appl. Phys. Lett. 86,

    153109 (2005).10

    A. Marent, T. Nowozin, M. Geller, and D. Bimberg, Semicond. Sci.Technol. 26, 014026 (2011).

    11Yu. I. Mazur,V. G. Dorogan,D. Guzun, E. Marega Jr., G. J. Salamo,

    G. G. Tarasov, A. O. Govorov, P. Vasa, and C. Lienau, Phys. Rev.

    B 82, 155413 (2010).12E-M. Pavelescu, C. Gilfert, J. P. Reithmaier, A. Martin-Minguez,

    and I. Esquivias, IEEE Photonics Technol. Lett. 21, 999 (2009).13J. Misiewicz, P. Sitarek, G. Sek, and R. Kudrawiec, Mater. Sci.

    (Poland) 21, 263 (2003).14J. M. Gerard, O. Cabrol, andB. Sermage, Appl. Phys. Lett. 68, 3123

    (1996).15J. M. Gerard, B. Sermage, B. Gayral, B. Legrand, E. Costard, and

    V. Thierry-Mieg, Phys. Rev. Lett. 81, 1110 (1998).16F. Hatami, M. Grundmann, N. N. Ledentsov, F. Heinrichsdorff,

    R. Heitz, J. Bohrer, D. Bimberg, S. S. Ruvimov, P. Werner, V. M.

    Ustinov, P. S. Kopev, and Zh. I. Alferov, Phys. Rev. B 57, 4635

    (1998).17J. Johansen, S. Stobbe, I. S. Nikolaev, T. Lund-Hansen, P. T.

    Kristensen, J. M. Hvam, W. L. Vos, and P. Lodahl, Phys. Rev.

    B 77, 073303 (2008).18S. Stobbe, J. Johansen, P. T. Kristensen, J. M. Hvam, and P. Lodahl,

    Phys. Rev. B 80, 155307 (2009).

    19D. S. Citrin, Phys. Rev. Lett. 69, 3393 (1992).20U. Bockelmann, Phys. Rev. B 48, 17637 (1993).21C. Pryor, M-E. Pistol, and L. Samuelson, Phys. Rev. B 56, 10404

    (1997).22A. Schliwa, M. Winkelnkemper, and D. Bimberg, Phys. Rev. B 76,

    205324 (2007).23C. Pryor, Phys. Rev. B 57, 7190 (1998).24Concerning the improvements to the numerical model that would

    include the Coulomb interactions [F. Boxberg and J. Tulkki, Rep.

    Prog. Phys. 70, 1425 (2007)], in the first step one would needto apply the self-consistent Hartree-Fock [HF] method based on

    single-particle electron and hole wave functions calculated in

    the eight-band kp model. Since the HF method does not fully

    account for exchange and correlation effects, it may be necessary

    to additionally develop a more complicated method, like the

    configuration interaction [M. Rontani, C. Cavazzoni, D. Bellucci,

    and G. Goldoni, J. Chem. Phys. 124, 124102 (2006); J. D.

    Plumhof, V. Krapek, L. Wang, A. Schliwa, D. Bimberg, A. Rastelli,

    and O. G. Schmidt, Phys. Rev. B 81, 121309(R) (2010);

    A. Franceschetti, H. Fu, L. W. Wang, and A. Zunger, ibid. 60, 1819

    (1999); W. Sheng, S.-J. Cheng, and P. Hawrylak, ibid. 71, 035316

    (2005)]. A combination of these should allow one to include the

    excitonic effects quite realistically, but this is far beyond the scopeof the current paper.

    25J. Andrzejewski, G. Sek, E. OReilly, A. Fiore, and J. Misiewicz,

    J. Appl. Phys. 107, 073509 (2010).26G. Bester, X. Wu, D. Vanderbilt, and A. Zunger, Phys. Rev. Lett.

    96, 187602 (2006).27I. Vurgaftman, J. R. Meyer, and L. R. Ram-Mohan, J. Appl. Phys.

    89, 5815 (2001).28G. Sek, P. Poloczek, K. Ryczko, J. Misiewicz, A. Loffler, J. P.

    Reithmaier, and A. Forchel, J. Appl. Phys. 100, 103529 (2006).29F. Szmulowicz, Phys. Rev. B 51, 1613 (1995).30O. Stier, M. Grundmann, and D. Bimberg, Phys. Rev. B 59, 5688

    (1999).31L. C. Andreani, G. Panzarini, and J.-M. Gerard, Phys. Rev. B 60,

    13276 (1999).32R. J. Warburton, C. S. Durr, K. Karrai, J. P. Kotthaus,

    G. Medeiros-Ribeiro, and P. M. Petroff, Phys. Rev. Lett. 79, 5282

    (1997).33R. J. Warburton, B. T. Miller, C. S. Durr, C. Bodefeld, K. Karrai,

    J. P. Kotthaus, G. Medeiros-Ribeiro, P. M. Petroff, and S. Huant,

    Phys. Rev. B 58, 16221 (1998).34W. M. Que, Phys. Rev. B 45, 11036 (1992).

    125311-7

    http://dx.doi.org/10.1109/5.752518http://dx.doi.org/10.1109/5.752518http://dx.doi.org/10.1109/5.752518http://dx.doi.org/10.1038/39535http://dx.doi.org/10.1038/39535http://dx.doi.org/10.1038/39535http://dx.doi.org/10.1063/1.1405153http://dx.doi.org/10.1063/1.1405153http://dx.doi.org/10.1063/1.1405153http://dx.doi.org/10.1063/1.1405153http://dx.doi.org/10.1063/1.1430025http://dx.doi.org/10.1063/1.1430025http://dx.doi.org/10.1063/1.1430025http://dx.doi.org/10.1063/1.1430025http://dx.doi.org/10.1063/1.1899230http://dx.doi.org/10.1063/1.1899230http://dx.doi.org/10.1063/1.1899230http://dx.doi.org/10.1063/1.1899230http://dx.doi.org/10.1088/0268-1242/26/1/014026http://dx.doi.org/10.1088/0268-1242/26/1/014026http://dx.doi.org/10.1088/0268-1242/26/1/014026http://dx.doi.org/10.1088/0268-1242/26/1/014026http://dx.doi.org/10.1103/PhysRevB.82.155413http://dx.doi.org/10.1103/PhysRevB.82.155413http://dx.doi.org/10.1103/PhysRevB.82.155413http://dx.doi.org/10.1103/PhysRevB.82.155413http://dx.doi.org/10.1109/LPT.2009.2021074http://dx.doi.org/10.1109/LPT.2009.2021074http://dx.doi.org/10.1109/LPT.2009.2021074http://dx.doi.org/10.1063/1.115798http://dx.doi.org/10.1063/1.115798http://dx.doi.org/10.1063/1.115798http://dx.doi.org/10.1063/1.115798http://dx.doi.org/10.1103/PhysRevLett.81.1110http://dx.doi.org/10.1103/PhysRevLett.81.1110http://dx.doi.org/10.1103/PhysRevLett.81.1110http://dx.doi.org/10.1103/PhysRevB.57.4635http://dx.doi.org/10.1103/PhysRevB.57.4635http://dx.doi.org/10.1103/PhysRevB.57.4635http://dx.doi.org/10.1103/PhysRevB.57.4635http://dx.doi.org/10.1103/PhysRevB.77.073303http://dx.doi.org/10.1103/PhysRevB.77.073303http://dx.doi.org/10.1103/PhysRevB.77.073303http://dx.doi.org/10.1103/PhysRevB.77.073303http://dx.doi.org/10.1103/PhysRevB.80.155307http://dx.doi.org/10.1103/PhysRevB.80.155307http://dx.doi.org/10.1103/PhysRevB.80.155307http://dx.doi.org/10.1103/PhysRevLett.69.3393http://dx.doi.org/10.1103/PhysRevLett.69.3393http://dx.doi.org/10.1103/PhysRevLett.69.3393http://dx.doi.org/10.1103/PhysRevB.48.17637http://dx.doi.org/10.1103/PhysRevB.48.17637http://dx.doi.org/10.1103/PhysRevB.48.17637http://dx.doi.org/10.1103/PhysRevB.56.10404http://dx.doi.org/10.1103/PhysRevB.56.10404http://dx.doi.org/10.1103/PhysRevB.56.10404http://dx.doi.org/10.1103/PhysRevB.56.10404http://dx.doi.org/10.1103/PhysRevB.76.205324http://dx.doi.org/10.1103/PhysRevB.76.205324http://dx.doi.org/10.1103/PhysRevB.76.205324http://dx.doi.org/10.1103/PhysRevB.76.205324http://dx.doi.org/10.1103/PhysRevB.57.7190http://dx.doi.org/10.1103/PhysRevB.57.7190http://dx.doi.org/10.1103/PhysRevB.57.7190http://dx.doi.org/10.1088/0034-4885/70/8/R04http://dx.doi.org/10.1088/0034-4885/70/8/R04http://dx.doi.org/10.1088/0034-4885/70/8/R04http://dx.doi.org/10.1088/0034-4885/70/8/R04http://dx.doi.org/10.1063/1.2179418http://dx.doi.org/10.1063/1.2179418http://dx.doi.org/10.1063/1.2179418http://dx.doi.org/10.1103/PhysRevB.81.121309http://dx.doi.org/10.1103/PhysRevB.81.121309http://dx.doi.org/10.1103/PhysRevB.81.121309http://dx.doi.org/10.1103/PhysRevB.60.1819http://dx.doi.org/10.1103/PhysRevB.60.1819http://dx.doi.org/10.1103/PhysRevB.60.1819http://dx.doi.org/10.1103/PhysRevB.60.1819http://dx.doi.org/10.1103/PhysRevB.71.035316http://dx.doi.org/10.1103/PhysRevB.71.035316http://dx.doi.org/10.1103/PhysRevB.71.035316http://dx.doi.org/10.1103/PhysRevB.71.035316http://dx.doi.org/10.1063/1.3346552http://dx.doi.org/10.1063/1.3346552http://dx.doi.org/10.1063/1.3346552http://dx.doi.org/10.1103/PhysRevLett.96.187602http://dx.doi.org/10.1103/PhysRevLett.96.187602http://dx.doi.org/10.1103/PhysRevLett.96.187602http://dx.doi.org/10.1063/1.1368156http://dx.doi.org/10.1063/1.1368156http://dx.doi.org/10.1063/1.1368156http://dx.doi.org/10.1063/1.2364604http://dx.doi.org/10.1063/1.2364604http://dx.doi.org/10.1063/1.2364604http://dx.doi.org/10.1103/PhysRevB.51.1613http://dx.doi.org/10.1103/PhysRevB.51.1613http://dx.doi.org/10.1103/PhysRevB.51.1613http://dx.doi.org/10.1103/PhysRevB.59.5688http://dx.doi.org/10.1103/PhysRevB.59.5688http://dx.doi.org/10.1103/PhysRevB.59.5688http://dx.doi.org/10.1103/PhysRevB.59.5688http://dx.doi.org/10.1103/PhysRevB.60.13276http://dx.doi.org/10.1103/PhysRevB.60.13276http://dx.doi.org/10.1103/PhysRevB.60.13276http://dx.doi.org/10.1103/PhysRevB.60.13276http://dx.doi.org/10.1103/PhysRevLett.79.5282http://dx.doi.org/10.1103/PhysRevLett.79.5282http://dx.doi.org/10.1103/PhysRevLett.79.5282http://dx.doi.org/10.1103/PhysRevLett.79.5282http://dx.doi.org/10.1103/PhysRevB.58.16221http://dx.doi.org/10.1103/PhysRevB.58.16221http://dx.doi.org/10.1103/PhysRevB.58.16221http://dx.doi.org/10.1103/PhysRevB.45.11036http://dx.doi.org/10.1103/PhysRevB.45.11036http://dx.doi.org/10.1103/PhysRevB.45.11036http://dx.doi.org/10.1103/PhysRevB.45.11036http://dx.doi.org/10.1103/PhysRevB.58.16221http://dx.doi.org/10.1103/PhysRevLett.79.5282http://dx.doi.org/10.1103/PhysRevLett.79.5282http://dx.doi.org/10.1103/PhysRevB.60.13276http://dx.doi.org/10.1103/PhysRevB.60.13276http://dx.doi.org/10.1103/PhysRevB.59.5688http://dx.doi.org/10.1103/PhysRevB.59.5688http://dx.doi.org/10.1103/PhysRevB.51.1613http://dx.doi.org/10.1063/1.2364604http://dx.doi.org/10.1063/1.1368156http://dx.doi.org/10.1063/1.1368156http://dx.doi.org/10.1103/PhysRevLett.96.187602http://dx.doi.org/10.1103/PhysRevLett.96.187602http://dx.doi.org/10.1063/1.3346552http://dx.doi.org/10.1103/PhysRevB.71.035316http://dx.doi.org/10.1103/PhysRevB.71.035316http://dx.doi.org/10.1103/PhysRevB.60.1819http://dx.doi.org/10.1103/PhysRevB.60.1819http://dx.doi.org/10.1103/PhysRevB.81.121309http://dx.doi.org/10.1063/1.2179418http://dx.doi.org/10.1088/0034-4885/70/8/R04http://dx.doi.org/10.1088/0034-4885/70/8/R04http://dx.doi.org/10.1103/PhysRevB.57.7190http://dx.doi.org/10.1103/PhysRevB.76.205324http://dx.doi.org/10.1103/PhysRevB.76.205324http://dx.doi.org/10.1103/PhysRevB.56.10404http://dx.doi.org/10.1103/PhysRevB.56.10404http://dx.doi.org/10.1103/PhysRevB.48.17637http://dx.doi.org/10.1103/PhysRevLett.69.3393http://dx.doi.org/10.1103/PhysRevB.80.155307http://dx.doi.org/10.1103/PhysRevB.77.073303http://dx.doi.org/10.1103/PhysRevB.77.073303http://dx.doi.org/10.1103/PhysRevB.57.4635http://dx.doi.org/10.1103/PhysRevB.57.4635http://dx.doi.org/10.1103/PhysRevLett.81.1110http://dx.doi.org/10.1063/1.115798http://dx.doi.org/10.1063/1.115798http://dx.doi.org/10.1109/LPT.2009.2021074http://dx.doi.org/10.1103/PhysRevB.82.155413http://dx.doi.org/10.1103/PhysRevB.82.155413http://dx.doi.org/10.1088/0268-1242/26/1/014026http://dx.doi.org/10.1088/0268-1242/26/1/014026http://dx.doi.org/10.1063/1.1899230http://dx.doi.org/10.1063/1.1899230http://dx.doi.org/10.1063/1.1430025http://dx.doi.org/10.1063/1.1430025http://dx.doi.org/10.1063/1.1405153http://dx.doi.org/10.1063/1.1405153http://dx.doi.org/10.1038/39535http://dx.doi.org/10.1109/5.752518