Stark Spectroscopy and Radiative Lifetimes in Single Self-Assembled CdTe Quantum Dots Ł. Kłopotowski,1 V. Voliotis,2 A. Kudelski,3 A. I. Tartakovskii,4 P. Wojnar,1 K. Fronc,1 R. Grousson,2 O. Krebs,3 M. S. Skolnick,4 G. Karczewski,1 and T. Wojtowicz1 1Institute of Physics, Polish Academy of Sciences, 02-668 Warsaw, Poland 2Université Pierre et Marie Curie, CNRS, 4 Place Jussieu, 75252 Paris, France 3CNRS-Laboratoire de Photonique et Nanostructures, Route de Nozay, 91460 Marcoussis, France 4Department of Physics and Astronomy, University of Sheffield, S3 7RH, United Kingdom 1 (Dated: January 4, 2011) 1 0 We present studies on Coulomb interactions in single self-assembled CdTe quantum dots. We 2 use a field effect structure to tune the charge state of the dot and investigate the impact of the charge state on carrier wave functions. The analysis of the quantum confined Stark shifts of four n excitonic complexes allows us to conclude that the hole wave function is softer than electron wave a function, i. e. it is subject to stronger modifications upon changing of the dot charge state. These J conclusionsarecorroborated bytime-resolvedphotoluminescencestudiesofrecombination lifetimes 3 of different excitonic complexes. We find that the lifetimes are notably shorter than expected for strong confinement and result from a relatively shallow potential in the valence band. This weak ] l confinement facilitates strong hole wave function redistributions. We analyze spectroscopic shifts l a of the observed excitonic complexes and find the same sequence of transitions for all studied dots. h Weconclude that theuniversality of spectroscopic shifts is dueto therole of Coulomb correlations - stemming from strong configuration mixingin thevalence band. s e m PACSnumbers: 78.67.Hc,32.60.+i,71.35.Pq,71.55.Gs . t a I. INTRODUCTION tumconfinedStarteffect(QCSE)15–19. Theshiftscanbe m exploitedas a resonantopticalspectrum analyzerwith a - voltage-tunable sensitivity20, which allowed to measure d In orderto fully exploitthe potential ofquantum dots the lifetime-limited absorption transition linewidth21. n (QDs) in numerous proposed optoelectronic and quan- From the point of view of fundamental studies, QCSE o tum computing applications, a detailed knowledge of c Coulomb interactionbetween carriersoccupying a dot is is sensitive to the charge distribution in the dot and re- [ flectsdotmorphology. Indeed,analysisoftheStarkshifts essential. The descriptionofthese interactions is usually allowedto determine the orientationof the electron-hole 1 done in the language of Hartree-Fock (HF) approxima- dipole in an InAs dot with GaAs barriers22 and inter- v tion,inwhichthecarrierwavefunctionsaregivenbysin- 6 pret this effect as a result of an In/Ga intermixing dur- gle orbitals, subsequently corrected by correlation terms 0 inggrowth18,23. Redistributionsofcarrierwavefunctions originatingfrommixingofdifferentorbitalconfigurations 5 upon chargingwere investigatedby time-resolvedphoto- (configurationinteraction—CI).EmbeddingQDsinver- 0 luminescence as the exciton lifetimes directly reflect the 1. ttircicaallficeoldnterffoelcotfstthreucQtDurecshaprrgoevisdteastet1heanpdosaslilboiwlistytoofsteuledcy- electron-hole overlap4,24. It was found that upon charg- 0 ingholesundergoastrongermodificationthanelectrons, many fundamental properties related to Coulomb inter- 1 owing to a weaker confinement in the valence band – actions. In particular, relative impact and magnitude of 1 a result subsequently confirmed by a theoretical work5. : the directHFandcorrelationtermsonthe spectroscopic v shiftsofQDtransitionsuponchargingwereevaluated2–5. Moreover, it was shown that weaker confinement leads i to larger oscillator strengths (i.e. shorter lifetimes) of X It was also shown that the dots can be fed with individ- excitons confined in monolayer fluctuation QDs.25,26 r ual chargecarriersone by one6 as eachof them raises an a electrostatic barrier, which needs to be overcome to add Charge tunability is obtained routinely in III-V QDs thenextcarrier. ThisCoulombblockaderesultsinapos- andmostofthe reportscitedaboverelatetotheInGaAs sibilityofobtainingacharge-tunabledevice,inwhichthe system. On the other hand, the physics of Coulomb chargingofasingledotispreciselycontrolledbyexternal blockade in II-VI systems is relatively unknown. Charge voltageandmanifestedbydistinctchargingstepsinpho- tunabilitywasdemonstratedinCdSeQDs27 andthenon toluminescence (PL) spectra7–10. In turn, the feasibility a CdTe QD with a single Mn ion14. QCSE was reported ofpreparingadotinagivenchargestateleadstodiscov- onlyforanelectricfieldappliedintheQDplane17. These eries of effects particular to a given charge configuration achievements notwithstanding, a lot of vital information such as creation of dynamical nuclear polarization11,12, on Coulomb interactions in II-VI systems is missing. two-qubit conditional quantum-logic operations13, and In this report, we present studies of Coulomb interac- electrical control of a spin state of a single Mn atom14. tions in CdTe QDs by two methods. In Section III, we Moreover, application of a voltage results in an electric show PL studies of a QD embedded in a Schottky diode field, which shifts QD transition energies due to a quan- structure, where the charge state can be tuned from −e 2 to +eby anapplicationofaverticalbias. Itallowsus to laser,yielding 2 ps pulses at 532 nm. The PL signalwas access the charge distributions and their modifications time-resolved with a streak camera providing an overall under external electric field for various few-body com- temporal resolution of 10 ps. All measurements were plexes. Stark effectspectroscopyprovidesadirectaccess performed at 10 K. to electron-holepolarizabilityandpermanentdipole mo- ment — both intrinsic parameters reflecting the form of electronic wave-functions. We address the redistribution III. CHARGE TUNABILITY AND STARK of carrier wave functions upon charging and find that EFFECT the electron is stiffer than the hole. We attribute this effect to stronger correlations in the valence band. This We start with discussing bias-dependent PL of a QD conclusion is further supported in Section IV, where we embeddedinaSchottkydiodestructure. InFigure1,we study PL lifetimes of different excitonic complexes. We present spectra from the low energy tail of the ensemble find that the lifetime is more affected by an addition of PLbandunderappliedbias: fromreverse(−2V,bottom anextraholethanbyanextraelectron. This correlation spectrum)toforward(+5V,topspectrum). Weidentify driven wave function redistribution results in a decrease the observed transitions as recombinations of four exci- of the electron-hole overlapand as a consequence, an in- toniccomplexesconfinedto the samesingleQD.Highest creaseinthetransitionlifetime. InSectionV,wediscuss energy transition is the neutral exciton (X0) recombina- the results in the language of configuration mixing and tion. At zero bias, positively (X+) and negatively (X−) address the magnitudes of spectroscopic shifts recorded chargedexcitonsandabiexciton(XX)arered-shiftedby for the observed excitonic complexes. 8.2,11.8and14.9meV,respectively. Withincreasingex- citation power, we observe a roughly quadratic increase of the XX line and linear increase of the X0 and charged II. SAMPLES AND EXPERIMENT exciton transitions. Thesamplesweregrownbymolecularbeamepitaxyon a (100)-oriented GaAs substrate. Schottky diode struc- p-ZnTe Schottky tures contained a ZnTe buffer layer, ∼ 4µm thick, p- s)s) QDs ZnMgTegate dpoerpecdmw3,itwhhnicihtraocgteendaatsaalbevaeclkocfoanbtaocutt.1I0t1w8aNssaecpcaerpattoerds unitunit XX X- X+ ZnTe ZnTe X0 byan80nmwideintrinsicZnTespacerfromasinglelayer b.b. +5 V rr of CdTe QDs. Dot formation was induced by changing aa (( thesurfaceenergyofastrainedCdTelayerbydeposition y y of amorphous tellurium28. The above procedure yields sitsit nn approximatelylens-shapedots,withbasediameterinthe ee tt range between 20 and 40 nm and heights between 2 and nn II 8 nm. QD layer was capped by a 100 nm layer of ZnTe L L PP 0 V and another 100 nm layer of Zn0.9Mg0.1Te. The latter servedas a blockingbarrierto preventthe escapeof car- riers to the surface. On top, a semitransparent, 15 nm -2 V thick Ti/Au Schottky gate was deposited. 1980 1985 1990 1995 The cw PL signal was excited slightly below the ZnTe Energy (meV) barrier with a 532 nm laser beam focused onto a ∼2µm spot. The estimated dot density is on the order of 1010 FIG. 1. PL spectra of a single CdTe QD embedded in a cm−2, so we excited roughly 500 dots. Individual dots field effect structure under bias: from −2 V on the bottom were accessed by tuning the detection energy either to to +5 V on top. Thick line marks the spectrum at 0 V. high-orlow-energytailoftheinhomogenouslybroadened Identified transitions, from low to high energies, correspond (FWHM ∼100 meV) PL band. The signal was detected to biexciton, negatively and positively charged and neutral by a nitrogen-cooled CCD camera coupled to a double exciton recombinations. The inset shows the band structure monochromator. underreversebias. Samples used for the time-resolved PL studies con- sistedofanundopedZnTebufferontopofwhichasingle The identification of the transition lines is based on layerofCdTeQDscappedwithanotherZnTebarrierwas the analysis of the intensity dependence on applied bias. deposited. To facilitate the accessto single dots,200nm Application of negative (reverse) bias (see inset in Fig. shadow mask apertures were processed by spin casting 1) enhances the tunneling of photocreated carriers out polybeads before metallization of a 100 nm gold layer. from the dots21,29. Tunneling of holes is faster owing to For measurements of the PL lifetime, as an excitation a weak confinement in the valence band, which results source we used a frequency doubled output of an optical from a vanishing valence band offset at the CdTe/ZnTe − parametric oscillator pumped with a Ti:sapphire pulsed heterointerface. Therefore,atnegativebiastheX dom- 3 inates the spectrum. Applying an increasingly forward ourspectralresolutionof70µeVandandmuchlessthan bias voltage restores the barrier between the back con- the spectral linewidth of ∼ 220µeV. Therefore, we con- tactandtheQD,andshiftstheFermilevelabovetheQD cludethattheeffects relatedtothe build-upofthe space states, thus injecting holes into the dots. As a result, at charge are negligible. positive (forward) bias X+ dominates the spectrum (see Fig. 1). In a charge tunable device, in order to inject a 1995.2 next carrier, the Fermi level needs to be lifted by a cer- tain addition energy to suppress Coulomb blockade6,30. 1995.0 1981.0 As a result, distinct charging steps are usually observed inthebias-dependentPLspectra7,8,10,11. AsseeninFig. 1994.8 ) 1980.8 1, in our results these steps are blurred and at a given V X XX e 1994.6 bias voltagetransitions relatedto various excitonic com- m plexes are observed simultaneously. This coexistence of y ( 1980.6 various charge states may result from a relatively weak g1986.60 + tunnel coupling between the back contactandthe dots9. er X 1983.2 n Moreover, single carriers can be captured from e.g. the E wetting layer. Although we observe no emission from 1986.55 1983.0 - a wetting layer in our samples, the dots are reportedly X formed on top of two dimensional platelets that provide a sequence of spatially extended excited states31. In the 1986.50 1982.8 -100 -50 0 50 -100 -50 0 50 experiment,weexcitethedotabovethesestatesenabling Electric Field (kV/cm) astochasticcaptureofseparateelectronsandholes. The capture competes with bias-controlledchargingandcon- FIG. 2. Transition energies as a function of the electric field sequentlychargingstepsarestronglymaskedandvarious forthefourexcitoniccomplexes. Linesarefittedsecondorder charge states coexist. polynomials. The bias applied between the top Schottky gate and the p-type back contact generates an electric field along Measured transition energies as a function of the ap- the growth axis. As a result, the transition lines are plied electric field together with fitted curves are pre- shifted due to the QCSE. For moderate electric fields, sented in Fig. 2. Positive and negative F values cor- these shifts can be approximated by first two orders of respond to electric fields applied parallel and antiparal- theperturbationexpansion: E(F)=E(0)−p·F+β·F2, lel to the growth axis, respectively. Clearly, the Stark where E(0) is the transition energy at zero electric field, shiftsarecorrectlyreproducedbyasecondorderpolyno- andpandβ arebuilt-indipolemomentandelectron-hole mial. From the fits in Fig. 2 we gain access to the built polarizability, respectively18. In order to quantitatively in dipole moment p and the electron-hole polarizability analyze the Stark shifts, electric field magnitude F has β. The former corresponds to the zero field distance be- tobedetermined. Inprinciple,F isgivenbyasimpleca- tween the centers of gravity of electron and hole wave pacitor formula: F = (U −U )/d, where U and U are functions. In the case ofX0, we obtainp/e=0.89±0.03 bi bi theappliedandbuilt-involtage,respectively,anddisthe Åand β =5.7±0.89nm2/V.Positivesignofp indicates widthoftheintrinsicregion. However,atasmallforward that inabsence of electric field, the hole is located above bias the flow of charge screens out the external electric the electron. This inverted carrier alignment points out field. To analyze the shifts under such conditions, we fit atranslationalasymmetryalongthegrowthaxis22 possi- the X0 PL energy dependence on electric field E (F) bly related to a Zn/Cd intermixing. As negative electric X0 for the negative bias, and assume that X0 transition at field is increased, the centers of electron and hole wave- forward bias follows this dependence. In this way the functions are brought together towards the dot center. voltage-to-electric field conversion is performed for the Eventually,acancelationofthedipolemomentoccursat positive bias range. Another uncertainty in the deter- an electric field F0 ≈−78kV·cm−1, where the transition mination of F is related to a build-up of a space charge energy dependence on F reachesa maximum. The value upon optical excitation. Photoexcited electrons can be- ofβ is ofthe same orderofmagnitude asin CdSe QDs17 cometrappedatthe interfacebetweenthe ZnTe capping and roughly an order of magnitude smaller than in InAs layerandZn0.9Mg0.1Teblockingbarrier(seeinsetinFig. QDs18. This latter results displays the influence of a 1), creating an electric field opposite in polarity to the stronger electron-hole attraction in II-VI nanostructures appliedbias–aneffectwhichwasshowntoaffectthePL with respect to their less polar III-V counterparts. transitionenergies down to lowestexcitationdensities21. Values of p/e retrieved from the quadratic fits for all In orderto investigateCoulombinteractionsin our dots, fourinvestigatedexcitoniccomplexesarecollectedinTa- we chose to work at an excitation density, at which the ble I together with the reduction ∆p relative to the X0. biexciton transition is resolved. However, we checked We note that although the accuracy of the fit for the thatdecreasingthepowerbyafactorof6shiftsthetran- X+ is lower than for the remaining complexes, it is clear sitionenergiesonlybyabout100µeV,slightlymorethan that the X+ energy shift in electric field is much smaller 4 p/e (Å) ∆p/p(X0) modifications upon changing of the dot occupation, one X0 0.89 ± 0.03 — expects the XX lifetime to be exactly 0.5 ·τ(X0) since X+ 0.08 ± 0.02 −91 % the biexciton has two decay channels, while X0 has only X− 0.45 ± 0.02 −49 % one24. Instead, we only observea decreaseby a factor of XX 0.48 ± 0.01 −46 % 0.73, which directly reflects the decrease of the electron- holeoverlap. Wesupposethatthedecreasestemsmostly TABLEI.Valuesofbuilt-indipolemomentobtainedforeach from the presence of a second hole. oftheexcitoniccomplexesfrom fittingofthetransition ener- Thelifetimedataallowsustoaddressalsotheconfine- giesdependenceonelectricfieldandtheitsreductionrelative to theX0. ment conditions in our QDs. In the strong confinement limit, carrier wave functions are determined by the QD than for the other complexes. Therefore, from Table I potentialandCoulombinteractionsareonlyasmallper- we infer that an addition of a second hole nearly cancels turbation. Insuchacase,thewavefunctionsarestiffand the built in dipole, while addition of a second electron changing the dot occupancy modifies them only slightly. reduces it only by roughly 50%. This result implies that As a consequence,the transitionlifetimes do not depend upon charging the redistribution of the hole wave func- onthe dot chargestate andwithin a two levelmodel are tion is more pronounced than the modification of the given by33: electron wave function. IV. TIME-RESOLVED PHOTOLUMINESCENCE τ = 3λ2PLǫ0cm0 (1) 2πne2f SPECTROSCOPY whereλ isthePLwavelength,nistherefractiveindex PL In order to gain more insight into the modifications ofthemediumsurroundingtheQDandf –theoscillator of carrier wave functions upon charging, we performed strength proportional to the overlapintegral hφ |φ i:34 e h measurements of recombinations lifetimes τ. In Fig. 3a, we present PL transients from a single QD, collected for the X0, X+, X−, XX, and another transition, labeled as 100 a) X0 t = 212 aXnXd∗,hawshaichsimapilpaeraprsowreedrsdheifpteenddwenitche.reMspoescttptroobtahbelyXXit nits) XX+- ttrr == 320110 u r is a recombination of a negatively charged biexciton32. b. 10 XX tr = 199 Alltransientsexhibitsmallslowcomponent,treatedasa ar XX* tr = 184 ( background, related to carrier recapture or dark exciton y recombination. A single exponential decay was fitted to sit n e extract the lifetimes and the numerical results are pre- nt 1 sented in the legend. In Fig. 3b, we collect the lifetimes L I obtained from six different dots. To address a relative P modification of carrier wave function upon charging, we present the lifetimes in units of the X0 lifetime τ(X0) , 0 250 500 750 1000 given in the legend. We find that upon charging with Time (ps) an extra hole, the recombination lifetime is notably in- creased–onaverageτ(X+)/τ(X0)=1.18±0.06. Onthe 1.4 QD1 tr = 225 ps QD2 t = 232 ps other hand, charging with an extra electron affects the QD3 tr = 166 ps r lifetime only slightly – averaging over our data we find 1.2 QD4 tr = 334 ps τ(X−)/τ(X0)=1.05±0.04. ThebiexcitonandXX∗ life- X0 QD5 tr = 212 ps atinmdeτs(aXreXsh∗)o/rτte(nXe0d)–=τ0(X.72X±)/0τ.(0X8τ0()X=0)0..73±0.05τ(X0) tetime / 1.0 QD6 tr = 285 ps These results againpoint outthe relative stiffness and Lif 0.8 softness of the electron and hole wave functions, respec- tively. Indeed, the increase of the X+ lifetime together with almost complete cancelation of the built-in dipole 0.6 b) (seeTableI)indicatethatuponadditionofanextrahole, its wave function expands laterally decreasing both the X+ X- XX XX* electron-holeoverlapandthe dipole. Onthe otherhand, Excitonic Complex additionofanextraelectrondoesnotresultinanimpor- − tantchangeoftheelectronwavefunctionasboththeX FIG. 3. a) PL time traces of various excitonic complexes built-in dipole and the lifetime remain on average only from a single quantum dot. Lines are single exponential de- weakly affected. The redistribution of carrierwave func- cays with lifetimes of each complex given in the legend. b) tionsuponchargingaremostclearlydemonstratedinthe Lifetimesoftheobservedexcitoniccomplexescollectedonsix behavior of the XX decay. Neglecting the wave function different dots, normalized toneutral exciton lifetime. 5 offinding the electronandthe holeatthe samelocation, i.e. to s(0) and is expected to increase with increasing E f =hφ |φ i P (2) QD radius25,26,36. The weak confinement is also mani- e h 2EPL fested by a clear dependence of the lifetime on the dot charge state – an effect which would be absent, if the where E is the PL photon energy and E is the Kane PL P wave functions were frozen and given by single orbitals energy,17.9eVforCdTe. Inthestrongconfinementlimit determined solely by the confinement. hφ |φ i=1andthe aboveformulayieldsforaCdTeQD e h The dominant role of correlations is further demon- inaZnTematrix(n=3.0)emittingin2.0eValifetimeof stratedin the shape of the emissionspectrum of a single 1.3ns. Thisismuchlongerthanexperimentallyobserved dot. We find the same sequence of the transitionlines in lifetimes on the order of 300 ps. Moreover, as evidenced all the dots studied and found in literature. In Figure 4, in Fig. 3, the lifetimes clearly depend on the QD charge we present the spectroscopic shifts of the observed exci- state. We can therefore conclude that our system is far tonic complexes taken as the difference between a given from the strong confinement limit. Indeed, bulk exciton complex and a neutral exciton X0. We collect the data Bohr radius for CdTe is 3 nm, about ten times smaller from the samples used in the present study – the field thanthedotlateralsize. Moreover,aspointedoutabove, effect sample discussedin Sec. III, and the samples used the confinement in the valence band is particularly weak in the time-resolvedPL studies in Sec. IV. We note that owing to a vanishing valence band offset and it results in all the studied dots, the charged excitons and biexci- mainly from strain. On the other hand, confinement in tons appearredshifted with respectto the X0. Moreover, the conduction band is stronger, since all the band mis- forallthestudieddots,werecordthesametransitionse- match between CdTe and ZnTe gives rise to a potential well for electrons. quence: EX0 > EX+ > EX− > EXX > EXX∗. We also note that the same sequence was observed in all other CdTeQDsreportedpreviously,wherethechargestateof the emitting complex was identified14,32,37,38. V. DISCUSSION AND CONCLUSIONS -6 Theanalysisofthereductionofthebuilt-indipolemo- ) V ment(seeTableI)andthelifetimedependenceonthedot e -8 m occupation lead us to a conclusion that the hole wave ( functionis soft, while the electronwavefunctionis rigid. ft -10 The former undergoes a redistribution upon charging a hi S dot, while the latter remains almostunaffected. The rel- c -12 ative softness and stiffness of the hole and electron wave pi QD1 o QD2 functions, respectively point to a stronger correlation sc -14 QQDD34 amongholesthanelectrons. Thestrongholecorrelations o QD5 abraendre,lawthedichtoprreolvaitdiveeslyclwoseealkycsopnaficendemsheneltlsi.n tIhnetvhaelelnance- pectr -16 QFRRieeDgff.6.. 11385 guage of configurationmixing it implies that the ground S Ref. 40 Ref. 41 -18 state holes admix easily other configurations, while elec- X+ X- XX XX* tronwavefunctionispredominantlybuiltfromthesingle Excitonic Complex particle s-shell orbital since higher lying shells are well separated in energy.35,36 We remark that the same con- FIG. 4. Spectroscopic shifts for X+, X−, XX, and XX∗ re- clusions were drawn for InGaAs QDs despite an entirely solved in the present study and compared to values found different potential depths and dielectric constants4,5. in literature. In all cases, these excitonic complexes appear The regime of the weak confinement and strong redshifted with respect to the X0 and the same sequence of Coulomb correlationsin the valence band are also mani- transitions is found. fested in absolute values of the exciton lifetimes. In the strong confinement limit, we expect the lifetimes about The redshift of all the charged exciton and biexciton four times longer than the observed ones. Under weak transitions points out that the direct Coulomb interac- confinement, the lifetime of anexcitonic complex cannot tionsareweakerthancorrelationsasaresultofrelatively be described by a simple overlap integral as in Eqs. 1 weak confinement. Consequently, the emission spectrum and 2, since the wave function is now a correlated one, qualitatively resembles the one observed in 2D systems, e.g.:35 wherethechargedexcitonbindingenergiesarealsodom- inatedbycorrelationeffects39. Thisremainsinstarkcon- trast with InGaAs system, where charged exciton tran- Ψ(r ,r )=φ (r )φ (r )s(ρ −ρ ) (3) sitions may appear on both sides of X0 depending on e h e e h h e h the localization degree of electrons and holes, which is where ρ are in-plane electron and hole coordinates. controlled by the dot morphology3–5,8,18. e,h Therecombinationrateisproportionaltotheprobability In conclusion, we investigated the impact of Coulomb 6 interactions on carrier wave functions in single CdTe single dots exhibit the same sequence of transitions re- quantum dots by means of Stark spectroscopyandtime- lated to different charge states with the neutral exciton resolved photoluminescence. We found that in absence emitting at highest energy as in 2D systems. of electric field, the hole is located above the electron We would like to thank M. Korkusiński, P. Hawrylak givingrisetoanelectricdipole. Moreover,wediscovered and Ł. Cywiński for fruitful discussions. This research that the hole wave function undergoes strong modifica- was supported by a Polish Ministry of Science and Edu- tions upon changing of the dot charge state, while the cation grant no. 0634/BH03/2007/33and the Polonium electronwavefunctionisnearlyunaffected. Weattribute Programme and by European Union within European thiseffecttoarelativelyweakconfinementinthevalence Regional Development Fund, through Innovative Econ- band, which makes Coulomb correlations dominate over omygrant(POIG.01.01.02-00-008/08)andSANDiENet- the confinement. As a result, the emission spectrum of work of Excellence. 1 H. Drexler, D. Leonard, W. Hansen, J. P. Kotthaus, and Liew, A.G. Cullis, and M.Hopkinson,Phys.Rev.B, 70, P. M. Petroff, Phys.Rev.Lett., 73, 2252 (1994). 201308 (2004). 2 D. V. Regelman, E. Dekel, D. Gershoni, E. Ehrenfreund, 19 K. Kowalik, O. Krebs, P. Senellart, A. Lemaître, B. Eble, A. J. Williamson, J. Shumway, A. Zunger, W. V. Schoen- A. Kudelski, J. A. Gaj, and P. Voisin, phys. stat. sol. c, feld, and P.M. Petroff, Phys.Rev. B, 64, 165301 (2001). 3, 3890 (2006). 3 G.BesterandA.Zunger,Phys.Rev.B,68,073309(2003). 20 A. Zrenner, E. Beham, S. Stufler, F. Findeis, M. Bichler, 4 P. A. Dalgarno, J. M. Smith, J. McFarlane, B. D. Ger- and G. Abstreiter, Nature, 418, 612 (2002). ardot, K. Karrai, A. Badolato, P. M. Petroff, and R. J. 21 S. Seidl, M. Kroner, P. A. Dalgarno, A. Hogele, J. M. Warburton,Phys. Rev.B, 77, 245311 (2008). Smith, M. Ediger, B. D. Gerardot, J. M. Garcia, P. M. 5 J.I.Climente,A.Bertoni, andG.Goldoni,Phys.Rev.B, Petroff, K. Karrai, and R. J. Warburton, Phys. Rev. B, 78, 155316 (2008). 72, 195339 (2005). 6 S.Tarucha,D.G.Austing,T.Honda,R.J.vanderHage, 22 P. W. Fry,I. E. Itskevich,D. J. Mowbray, M. S. Skolnick, andL.P.Kouwenhoven,Phys.Rev.Lett.,77,3613(1996). J. J. Finley, J. A. Barker, E. P. O’Reilly, L. R. Wilson, 7 R.J.Warburton,C.Schäflein,D.Haft,F.Bickel,A.Lorke, I.A.Larkin,P.A.Maksym,M.Hopkinson,M.Al-Khafaji, K.Karrai,J.M.Garcia,W.Schoenfeld, andP.M.Petroff, J.P.R.David,A.G.Cullis,G.Hill, andJ.C.Clark,Phys. Nature, 405, 926 (2000). Rev.Lett., 84, 733 (2000). 8 J. J.Finley,P.W.Fry,A.D.Ashmore,A.Lemaître, A.I. 23 J. A. Barker and E. P. O’Reilly, Phys. Rev. B, 61, 13840 Tartakovskii, R. Oulton, D. J. Mowbray, M. S. Skolnick, (2000). M. Hopkinson, P. D. Buckle, and P. A. Maksym, Phys. 24 M.Feucker,R.Seguin,S.Rodt,A.Hoffman, andD.Bim- Rev.B, 63, 161305 (2001). berg, Appl.Phys. Lett., 92, 63116 (2008). 9 M. Baier, F.Findeis, A. Zrenner,M.Bichler, and G. Ab- 25 J. Bellessa, V. Voliotis, R. Grousson, X. L. Wang, streiter, Phys.Rev.B, 64, 195326 (2001). M. Ogura, and H. Matsuhata, Phys. Rev. B, 58, 9933 10 M. Ediger, G.Bester, A.Badolato, P. M.Petroff, K. Kar- (1998). rai, A. Zunger, and R. J. Warburton, Nature Physics, 3, 26 J.Hours,P.Senellart,E.Peter,A.Cavanna, andJ.Bloch, 774 (2007). Phys.Rev. B, 71, 161306 (2005). 11 B. Eble, O.Krebs, A.Lemaitre, K. Kowalik, A.Kudelski, 27 J.Seufert,M.Rambach,G.Bacher,A.Forchel,T.Passow, P. Voisin, B. Urbaszek, X. Marie, and T. Amand, Phys. and D. Hommel, Appl.Phys. Lett., 82, 3946 (2003). Rev.B, 74, 081306 (2006). 28 F.Tinjod,B.Gilles,S.Moehl,K.Kheng, andH.Mariette, 12 A.I.Tartakovskii,T.Wright,A.Russell,V.I.Fal’ko,A.B. App.Phys.Lett., 82, 4340 (2003). Van’kov, J. Skiba-Szymanska, I. Drouzas, R. S. Kolodka, 29 R.Oulton,J.J.Finley,A.D.Ashmore,I.S.Gregory,D.J. M. S. Skolnick, P. W. Fry, A. Tahraoui, H.-Y. Liu, and Mowbray, M. S. Skolnick, M. J. Steer, S.-L. Liew, M. A. M. Hopkinson,Phys.Rev.Lett., 98, 026806 (2007). Migliorato, and A. J. Cullis, Phys. Rev. B, 66, 045313 13 A.J.Ramsay,S.J.Boyle,R.S.Kolodka,J.B.B.Oliveira, (2002). J.Skiba-Szymanska,H.Y.Liu,M.Hopkinson,A.M.Fox, 30 A. Franceschetti and A. Zunger, Phys. Rev. B, 62, 2614 and M. S.Skolnick,Phys.Rev.Lett.,100, 197401 (2008). (2000). 14 Y. Léger, L. Besombes, J. Fernández-Rossier, L. Main- 31 T. A. Nguyen, S. Mackowski, T. B. Hoang, H. E. Jack- gault, and H. Mariette, Phys. Rev. Lett., 97, 107401 son, L. M. Smith, and G. Karczewski, Phys.Revi.B, 76, (2006). 245320 (2007). 15 S. A.Empedocles and M. G. Bawendi, Science, 278, 2114 32 T. Kazimierczuk, J. Suffczynski, A. Golnik, J. A. Gaj, (1997). P. Kossacki, and P. Wojnar, Phys. Rev. B, 79, 153301 16 S.Raymond,J.P.Reynolds,J.L.Merz,S.Fafard,Y.Feng, (2009). and S.Charbonneau, Phys. Rev.B, 58, R13415 (1998). 33 R.Loudon,TheQuantumTheory ofLight (Oxford,1973). 17 J. Seufert, M. Obert, M. Scheibner, N. A. Gippius, 34 R. S. Knox, Theory of Excitons (Academic Press, New G. Bacher, A. Forchel, T. Passow, K. Leonardi, and York,1963). D. Hommel, App.Phys.Lett., 79, 1033 (2001). 35 G. W. Bryant,Phys. Rev.B, 37, 8763 (1988). 18 J.J.Finley,M.Sabathil,P.Vogl,G.Abstreiter,R.Oulton, 36 S.Corni,M.Brásken,M.Lindberg,J.Olsen, andD.Sund- A. I. Tartakovskii, D. J. Mowbray, M. S. Skolnick, S. L. holm, Phys. Rev.B, 67, 045313 (2003). 7 37 J. Suffczyński, T. Kazimierczuk, M. Goryca, B. Piechal, 38 K. Kowalik, A. Kudelski, A. Golnik, J. Suffczyński, A.Trajnerowicz,K.Kowalik,P.Kossacki,A.Golnik,K.P. O. Krebs, P. Voisin, G. Karczewski, J. Kossut, and J. A. Korona, M. Nawrocki, J. A. Gaj, and G. Karczewski, Gaj, phys.stat. sol. c, 3, 865 (2006). Phys. Rev.B, 74, 085319 (2006). 39 C. Riva, F. M. Peeters, and K. Varga, Phys. Rev. B, 61, 13873 (2000).