Controlling Piezoelectric Response inSemiconductor Quantum Dots via ImpulsiveCharge LocalizationPooja Tyagi, Ryan R. Cooney, Samuel L. Sewall, D. M. Sagar, Jonathan I. Saari, andPatanjali Kambhampati*

Department of Chemistry, McGill University, Montreal, Quebec H3A 2K6, Canada

ABSTRACT By direct observation of coherent acoustic phonons, we demonstrate a novel extrinsic piezoelectric response in colloidalCdSe semiconductor quantum dots. This response is driven by the migration of charges to the surface of the quantum dot on avibrationally impulsive time scale. Surface- and fluence-dependent studies reveal that the observed carrier capture based piezo responseis controllable and is at least an order of magnitude larger than the intrinsic piezo response of wurtzite CdSe.

KEYWORDS Coherent phonons, quantum dots, piezoelectricity, femtosecond spectroscopy

Quantum dots are nanoscale semiconductor crystal-lites which can interpolate between the quantizedlimit of molecules and the continuum limit of bulk

solids.1 The lower excitonic states of the dot are quantized,asdenotedbyatomic-liketermsymbolshavinglowdegeneracy.1-3

These lower states yield phenomena such as quantizedAuger recombination3 and Coulomb blockade.4 In contrast,the higher states converge to a continuum, enabling creationof dozens of excitons per dot5,6 thereby creating an excitonicplasma7 with Fe-h ∼ 1020 cm-3.

Excited electron-hole pairs can launch acoustic vibra-tions of the lattice. The interaction of excited charge carrierswith these vibrations, called phonons, can determine manyaspects of a system’s response, from the Stokes shift inquantized systems8-10 to carrier thermalization11,12 andpiezoelectricity13,14 in continuum systems. A clear under-standing of these interactions in semiconductor nanocrystalsis of particular importance given the increasing demand fornanoelectronic devices.15,16 For example, the saturationvelocity of charge carriers in a system is limited by phononemission, thereby, limiting the operating speed of electronicdevices. Also, the switching time of a laser and otheroptoelectronic devices is limited by the electron thermaliza-tiontimewhichdependsonelectron-phonon(e-ph)interactions.

While the excitonic states can be described at variouslevels of theory,1,2,17,18 colloidal quantum dots are conve-niently described in terms of an effective mass approxima-tion model.1,2 Regardless of theoretical approach, the lowerenergy transitions yield resolvable absorption features (Fig-ure 1a) which reflect quantized excitonic states of lowdegeneracy (Figure 1b). Therefore, one can describe theelectron-phonon interactions in terms of a displaced har-

monic oscillator picture (Figure 1c). In this situation, the e-phcoupling is parametrized in terms of the relative displace-ment of the harmonic potentials for each quantized state.8

In these systems, one can probe e-ph couplings directly inthe time domain by observing coherent phonons via fem-tosecond spectroscopy.

Whether in molecules,19 carbon nanotubes,9 or the ex-citonic states of quantum dots,8,20,21 the coherent phonondynamics can be related to e-ph coupling via direct optical

*Authortowhomcorrespondenceshouldbeaddressed,pat.kambhampati@mcgill.ca.Received for review: 05/05/2010Published on Web: 07/01/2010

FIGURE 1. Absorption spectrum of CdSe colloidal quantum dots (a).Schematic level structure (b). Direct optical excitation of coherentphonons in the displaced harmonic oscillator model for quantizedstates (c). Low fluence excitation into the 1S quantized state yieldsboth coherent optical and acoustic phonons (d). Indirect generationof coherent phonons via impulsive lattice heating for continuumstates (e). High fluence excitation into the continuum yields largeamplitude coherent acoustic phonons (f).

pubs.acs.org/NanoLett

© 2010 American Chemical Society 3062 DOI: 10.1021/nl101605r | Nano Lett. 2010, 10, 3062–3067

excitation of coherent phonons via a femtosecond laserpulse. Figure 1d shows coherent optical and acoustic phononsin CdSe colloidal quantum dots upon excitation directly intothe band edge (1S) exciton with 40 fs laser pulses.

Coherent phonons are also observed in continuum sys-tems such as bulk solids, metal nanoparticles,11,12 andsemiconductor quantum wells.13,14 In the case of metallicsystems, coherent acoustic phonons are launched indirectlyvia impulsive lattice heating11 rather than directly via thefemtosecond laser pulse. In this mechanism, an ultrafastpulse creates a hot electron distribution which rapidly ther-malizes and impulsively heats the lattice (Figure 1e). Thetwo-temperature model parametrizes this energy exchangein terms of an electron-phonon coupling constant, ge-ph.11

In the case of strain engineered piezoelectric multiple quan-tum wells,13,14 the femtosecond pump pulse creates anexcitonic plasma7 which impulsively screens the intrinsicpiezoelectric field of the system thereby launching piezo-electric coupled acoustic phonons. In these situations, theelectron-phonon coupling and its manifestation via coher-ent phonons is viewed in terms of a continuum response.Since the higher energy states of a quantum dot convergetoward an excitonic continuum, one might anticipate asimilar e-ph response to intense excitation into these non-quantized states.5,6,22

In contrast to such expectations, we show here a com-pletely new manifestation of e-ph interactions due to spatialmigration of charges in the quantum dot: extrinsic piezo-electric coupling. Since CdSe has a wurtzite lattice and isintrinsically piezoelectric, we expect that there would be apiezo response due to the screening of the intrinsic field bythe excited electron-hole plasma, as in the case of quantumwells. On the contrary, we find that unlike quantum wells,the intrinsic response in quantum dots is negligible, and themajority of the piezoelectric response is extrinsic and pre-dominantly arises due to the migration of charges to thesurface of the QD.

By direct observation of e-ph coupling via coherentphonons, we show that the spatial separation of charges ina quantum dot yields large amplitude coherent acousticphonons. Since the charge separation (due to localization ofthe hole at the surface) is on a subpicosecond time scale, itis vibrationally impulsive for acoustic phonons. The extrinsicpiezoelectric response is found to be at least an order ofmagnitude larger than the intrinsic screening response andis furthermore tunable at the single exciton level. Surface-dependent studies reveal that the large amplitude piezoresponse can be controlled by altering the surface passiva-tion of the QD. In addition to revealing a new mechanismof coherent acoustic phonon generation, these results offera way of modulating the optical response of quantum dotsand suggest design principles for quantum dot based opto-electronic devices.

These results were obtained using a state-resolved fem-tosecond spectroscopic technique which has been previously

described.8,23-26 Experiments were performed on colloidalCdSe quantum dots (R ) 1.6 nm) with three surface clas-sifications denoted as (i) phototreated, (ii) untreated, and (iii)capped. The “phototreated dots” were prepared by severalhours of intense illumination without flowing the sample(details in the Supporting Information). These dots, preparedby the illumination procedure, undergo efficient surfacetrapping which competes with intraband relaxation (Figure3b inset). The “untreated dots” are measured in a flow cellat short exposure times, and the “capped dots” have apassivating ZnS shell. The transient absorption measure-ments yield ∆OD ()ODpump-on - ODpump-off), where OD is theoptical density of the sample. The signal obtained (∆OD) hasboth exciton (nonoscillatory) and phonon (oscillatory) con-tributions. To obtain the phonon contribution, the transientswere fit to a multiexponential model function and theoscillations were extracted by a standard subtraction pro-cedure.8 The fast Fourier transform (FFT) of the oscillationsreveals a phonon mode at a frequency of 26 cm-1 (Figure3c) which corresponds to the well-known longitudinal acous-tic (LA) phonon mode.8,27

In our prior work8 the femtosecond pump pulse wasresonant with the lower quantized states with fluence inten-tionally adjusted to maintain low mean occupancies, (⟨N⟩ <0.5). Here, we excite with fluences which create excitondensities of up to ⟨N⟩ ∼ 3 and each exciton having 0.5-1.0eV of excess electronic energy to dissipate. Owing to thelarge excess electronic energy per dot and the fast subpico-second hot exciton cooling times,24,25,28 one might antici-pate that the dissipation of excess electronic energy impul-sively heats the quantum dot lattice, thereby, launchingcoherent acoustic phonons. Alternatively, the excitonicplasma created by the intense pump pulse7 (Fe-h ∼ 1019-1021

cm-3) could screen the intrinsic electric field of wurtzite CdSequantum dots29 and coherent phonons would be launchedas in the case of piezoelectric quantum well structures.

In principle, the large amplitude coherent acousticphonons observed under high excitation density into thecontinuum may arise from any of the three well establishedmechanisms of electron-phonon coupling: (i) direct optical,(ii) impulsive heating, and/or (iii) screening of intrinsicpiezoelectricity. In order to identify the mechanism drivingthese large amplitude coherent acoustic phonons, we per-form these femtosecond pump/probe experiments on dotsof the same size with different surface passivation. Thesurface-dependent studies show that none of the abovemechanisms account for the observed results.

Figure 1f shows coherent acoustic phonons in CdSequantum dots upon excitation with intense femtosecondpulses at 3.1 eV. Under these excitation conditions theoptical phonon is no longer coupled as we have previouslydescribed,8 in contrast the coherent acoustic phonon am-plitude increases with excitation intensity. Figure 2b showsthe pump/probe transients upon excitation with ⟨N⟩ ∼ 2.Since the excitation of the acoustic phonon mode produces

© 2010 American Chemical Society 3063 DOI: 10.1021/nl101605r | Nano Lett. 2010, 10, 3062-–3067

frequency modulation of the absorption spectrum,8 theprobe is tuned to the red of the band edge (1S) exciton tomaximize the oscillation amplitude. The pump is tuned to3.1 eV to access the continuum states. Adopting the notationof Klimov and co-workers, we have labeled the transientspectral feature in the probed region (Figure 2c) as A1.1,23,28

We have shown that this feature can have a positive ornegative sign based upon the excitonic state which isprobed.23,26,30 Indeed, the A1 transients in Figure 2b arepositive or negative based upon surface passivation and insome cases can change sign with time. This sign change willbe important in subsequent analysis.

The transients for the three surface conditions are shownin Figure 2b. It is clear that the oscillations are much morepronounced in the phototreated dots, being visible withoutrequiring extracting residual oscillations. In contrast, theoscillations are much weaker for the untreated and ZnS-capped dots. We note that the sensitivity to observing theoscillations is nearly identical ((10%) in all cases, as theprobe is tuned to the peak in the derivative of the absorptionspectrum.8 Attenuated oscillations in ZnS-capped CdSe nano-rods were similarly observed by Lanzani and co-workers.31

The assignment of surface characterization is based uponthe transient absorption (TA) data in parts c and d of Figure2. Our prior work has shown that the sign and amplitude ofthe pump/probe signal in the probed spectral region (A1)

reflects the excitonic state of the system via biexciton-induced level shiftings.23,26,30 The pump pulse produces anexciton which undergoes intraband relaxation24,25 and ul-timately undergoes surface trapping.26 Our earlier resultsshowed that it is specifically the hole that gets trapped atthe surface in CdSe quantum dots.26 When the exciton isnear the band edge (t ) 0.1-5 ps), the A1 signal is negativeand a surface trapped exciton (t ) 50-500 ps) yields apositive A1 signal for this size of dot. We refer the reader toour prior works for further details of the pump/probesignals.23-26,30

In the case of untreated CdSe dots, the A1 signal isnegative at t ) 0.1, 1, 3 ps. At long time (t > 100 ps) the A1signal goes positive due to a surface-trapped hole.26 Incontrast, the phototreated dots have a positive A1 signal atall times. This implies that the phototreated dots reach thesurface trapped state within 1 ps, whereas the untreated dotstake much longer (t > 100 ps) to reach this state. Therefore,in the case of phototreated dots, the surface trapping processis impulsive with respect to the acoustic phonon period.

Due to the efficiency of this competing trapping pro-cess,26 some fraction of the hot excitons (specifically holes)moves to the surface without cooling to the 1S band edgeexciton as can be understood from simple competitionkinetics (Figure 3b). We expect that the illumination proce-dure adjusts the equilibrium of the weakly bound ligandwhich passivate the surface of the dot.32 In order to confirmthis assignment, we performed transient absorption experi-ments on ZnS-capped CdSe quantum dots. In these well-passivated ZnS-capped dots, we were never able to observepositive A1 signals even under prolonged intense illumina-tion. This indicates that these dots do not reach the surface-trapped state, a result completely consistent with the ex-pected passivating function of the capping layers.33

The surface-dependent studies are essential toward de-termining the mechanism of coherent acoustic phonongeneration. In all three cases, the size of the dot and theexcitation and observation conditions are identical. Yet, onlythe phototreated dot couples to the acoustic phonons withsufficient efficiency to be observed directly in the A1 pump/probe transients. In contrast, no coherent phonons areobserved in the ZnS-capped dots even at high excitationdensities with subtraction of the fits.

Since all three systems under investigation have bandedge (1S) at 2.29 eV, they will have equal excess energy todissipate to the lattice under identical pumping conditions(Eexcess ) Epump - E1S). This implies that if lattice heating isthe driving force for the LA phonons, all three systemsshould have the same phonon amplitude, which is incon-sistent with our observations. Under different pumpingconditions, there would be more lattice heating for high-frequency pumps than the band edge pump. This predictsan increase in the phonon amplitude with increasing pumpfrequency. In contrast, our previous state-dependent studiesshow a decrease in the amplitude of LA phonons with

FIGURE 2. Schematic illustration of phototreated, untreated and ZnS-capped dots at t ∼ 1-3 ps (a). Femtosecond transients of CdSe dotsunder high excitation density, for 3.1 eV pump to access thecontinuum states. Coherent acoustic phonons are readily visible inthe A1 transients for phototreated dots (b). Femtosecond transientabsorption spectra of untreated and phototreated dots. The un-treated dots reach a surface trapped state in t ) 100 ps (as shownby the positive A1 signal in the probed region), whereas thephototreated dots reach such a state in t < 3 ps (c) and (d).

© 2010 American Chemical Society 3064 DOI: 10.1021/nl101605r | Nano Lett. 2010, 10, 3062-–3067

increasing pump frequency.8 Hence, we can rule out latticeheating as being the dominant mechanism for generationof coherent acoustic phonons in CdSe QDs.

CdSe lattice has a nonzero intrinsic dipole moment.Photoexcitation of charge carriers will screen the intrinsicpiezo field, thereby launching coherent acoustic phonons.The contribution of the field screening to the generationof coherent phonons would be the same in capped,untreated, and phototreated dots since they have thesame lattice structure. In contrast, we find that theresponse of the quantum dots due to the screening of theintrinsic electric field is an order of magnitude smallerthan that due to coupling to the extrinsic electric field. Thisis revealed by comparing the FFT phonon amplitude foruntreated and ZnS-capped dots, which have no surface

trapping and therefore no external electric field, to pho-totreated dots (Figure 4a).

In particular, perfectly passivated ZnS-capped dots musthave contribution from lattice heating and intrinsic piezofield screening. In our experiments, we observe negligibleelectron-phonon coupling in capped dots and a largecoupling in phototreated dots. This implies that the ampli-tude of the driving force for coherent acoustic phonons islarge in phototreated dots, small in untreated dots, andnearly zero in capped dots. On the basis of these observa-tions, we must necessarily rule out the previously describedmechanisms and invoke a new mechanism that is intimatelyrelated to surface properties of the QD.

The data suggest that the process of surface trappingcreates the impulse which launches coherent acousticphonons. When the exciton gets trapped at the surface, itcreates an electric field which turns on at the time scale ofthe trapping process. The electric field arises from themigration of charges to the exterior of the quantum dot. Dueto the piezoelectric nature of the wurtzite CdSe quantumdots,29 this impulsive electric field launches coherent acous-tic vibrations of the lattice. A necessary feature of thisprocess is that it proceeds at the single charge level owingto the excitation and trapping of discrete charges.

The lower limit of the surface piezo response is from theZnS-capped dots. In the limit of perfect passivation, theimpulse would purely arise from impulsive lattice heatingand from screening of the intrinsic piezo response of wurtz-ite CdSe.29 In these capped dots we see no observablecoupling to acoustic phonons suggesting that the contribu-tion of the above listed mechanisms is at least an order ofmagnitude smaller than the extrinsic piezo response ob-served here (Figure 4a).

This assignment of extrinsic piezoelectric coupling viasurface charge trapping is consistent with the standarddamped driven oscillator model commonly used to modelcoherent phonons in nanoparticles11,12 (Figure 3c). Wemodel the system as a damped driven harmonic oscillator,the damping time and frequency of which are determinedfrom the experimentally observed oscillations. The mainissue is to derive the driving impulse due to surface chargetrapping. This is accomplished by solving the kinetic rateequations for the system shown in the inset of Figure 3b.We fix t1 and t2 and change the ratio R to control thecompetition kinetics in order to study the role of surfacepassivation. We define R ) t2S/t1 ) t3S/t2, so that R ) 0.1 (R) 10) corresponds to the case where trapping is 10 timesfaster (slower) than the intraband relaxation. Solving the rateequations, we obtain the population of the surface state asa function of time for R ) 0.1/1/10. Since the impulse, I(t),due to the charge buildup is proportional to the number ofcharges on the surface ⟨Nsurface(t)⟩, we can write

FIGURE 3. Timeline of exciton relaxation and the competing processof hole trapping (a). Representative pump/probe signals (S) in theprobed spectral region for band edge (1S) and 3.10 eV pumps. Thesignal monitors the development of surface charges. St becomesmore positive with increase in the rate of surface trapping. The insetshows a schematic of the intraband relaxation and the competingsurface trapping process (b). FFT spectra of the data (symbols) andthe results of the model (lines) for the three surface passivationsrevealing the acoustic phonon amplitude at an incident fluence of2.2 µJ. The inset shows the driving impulse due to charge trappingin capped (blue), untreated (green), and phototreated (red) dots (c).

I(t) ) C⟨Nsurface(t)⟩

© 2010 American Chemical Society 3065 DOI: 10.1021/nl101605r | Nano Lett. 2010, 10, 3062-–3067

where C is a constant. The inset of Figure 3c shows theimpulse obtained for different rates of surface trapping (R) 0.1/1/10). Using I(t) as the driving impulse, we solve thedamped driven harmonic oscillator for the three cases. Theoscillations thus obtained are Fourier transformed and theresults are plotted in Figure 3c for comparison with the data.The symbols are the FFT of the experimentally observedoscillations and the lines are the model results. The resultsfor R ) 0.1 (red), R ) 1(green), and R ) 10 (blue) fit well tothe phototreated, untreated, and capped dots data, respec-tively, and are consistent with the proposed mechanism.

The inset in Figure 3b shows a schematic of the competi-tion kinetics between intraband relaxation and excited statesurface trapping. The main panel of Figure 3b shows howthe A1 pump/probe signals are related to the excited statesurface trapping process. In simple terms, the A1 signal willbe positive for a hot exciton, will relax toward the negativefor a cold (relaxed) exciton, and ultimately converge upon apositive sign for a surface trapped exciton.26

We simulate the A1 transients from the solution of thekinetic rate equations using

where ni(t) is the population of state “i” and Ai is theamplitude for ∆OD associated with state “i”. These calcula-tions illustrate the dependence of A1 transients on thesurface trapping rate and initial pumping conditions (Figure3b). The results of this calculation are in good agreementwith our previous experimental work.23-26 Regardless of thevalue of R and the initial pumping conditions, all A1 signalsmeet at late times, implying that in all cases, the systemeventually ends up in the same surface trapped state.However, the early time A1 signal depends on the rate ofsurface trapping. As trapping becomes more efficient thanthe intraband relaxation, some population directly goes tothe surface and the A1 signal becomes more positive at earlytimes. It is this population which forms the vibrationalimpulse.

The number of surface-trapped holes can be extractedfrom the A1 signal. A hole that relaxes to the band edge hasa specific value of the early time A1 signal (St) and anydeviation from St (δSt) is due to surface trapping. Noting thatthe surface-trapped hole (t ) 50-100 ps) has a specific valueof the signal (Ssurface) (Figure 3b), one can calculate thefraction (F) of the excitons which have a surface trapped holeby taking the ratio δSt/Ssurface (Supporting Information).

We estimate that the fraction F ∼ 0.0/0.02/0.6 for capped/untreated/phototreated dots, respectively. Essentially, alter-ing the surface passivation provides a control of the com-petition kinetics which determines the dominant couplingmechanism. Since exciton relaxation completes in ∼1 ps,24,25

we estimate that the polarizing process via excited statesurface trapping26 proceeds on a distribution of time scalesfrom 0.1 to 1 ps, precisely as needed to be in the impulsiveregime.

The surface treatments show that the impulse whichlaunches the coherent acoustic phonons does not arise fromintrinsic mechanisms of direct optical coupling, impulsiveheating, or impulsive screening of the intrinsic piezoelectricfield of wurtzite CdSe quantum dots. Instead, the couplingto acoustic phonons is primarily dominated by this newmechanism of extrinsic piezoelectric field due to surfacetrapped holes, within the detection limits.

Figure 4a shows the fluence dependence of the intensityof the FFT of the acoustic phonon. A similar fluence depen-dence was obtained by Alivisatos and co-workers.22 Inreality, the untreated dots most likely have some polarizationaccumulated during the course of the experiment. Werelated the peak FFT intensity to the number of surface-trapped holes, Figure 4b. In the case of untreated dots, nocorrelation is found between the phonon amplitude and thenumber of surface trapped charges (⟨Nsurface⟩) whereas inphototreated dots, the amplitude increases with increasing⟨Nsurface⟩ (Figure 4b).

We find that excitation into the continuum in colloidalquantum dots creates a completely new coupling mecha-nism which reflects a large amplitude, extrinsic piezoelectricresponse. In addition to identifying the mechanism of

FIGURE 4. The fluence dependence of acoustic phonon FFT amplitude for the three surface passivations (a). Relating the phonon FFT amplitudeto the mean occupancy of surface trapped charges. No correlation is observed in the case of untreated dots whereas there is a positive correlationin phototreated dots (b). Schematic illustrating the generation of coherent acoustic phonons via surface charge trapping (c).

∆OD(t) ) ∑i

Aini(t)

© 2010 American Chemical Society 3066 DOI: 10.1021/nl101605r | Nano Lett. 2010, 10, 3062-–3067

phonon generation in quantum dots, this method alsoreveals a way of modulating the optical response of thesesystems. The piezo response can be enhanced or suppressedbased upon surface treatment as shown here. The tunabilityand large amplitude of this extrinsic response suggest itsimportance for quantum dot based optoelectronic devices.

Acknowledgment. Financial support from the CFI, NSERC,FQRNT, and McGill University is acknowledged. P.T. ac-knowledges support from the Max Binz Fellowship. Wethank Jonathan Mooney, Michael M. Krause, and especiallyEva A. Dias for careful reading of the manuscript and usefulcomments. We also thank the McGill University Center forSelf-Assembled Chemical Structures for use of their facilities.

Supporting Information Available. Additional experi-mental details, characterization of phototreated dots, calcu-lation of the number of surface trapped charges, evidenceof hole trapping, and mechanisms for generation of coherentacoustic phonons. This material is available free of chargevia the Internet at http://pubs.acs.org.

REFERENCES AND NOTES(1) Klimov, V. I. Annu. Rev. Phys. Chem. 2007, 58, 635–673.(2) Efros, A. L.; Rosen, M. Annu. Rev. Mater. Sci. 2000, 30, 475–521.(3) Klimov, V. I.; Mikhailovsky, A. A.; McBranch, D. W.; Leatherdale,

C. A.; Bawendi, M. G. Science 2000, 287 (5455), 1011–1013.(4) Livermore, C.; Crouch, C. H.; Westervelt, R. M.; Campman, K. L.;

Gossard, A. C. Science 1996, 274 (5291), 1332–1335.(5) Son, D. H.; Wittenberg, J. S.; Alivisatos, A. P. Phys. Rev. Lett. 2004,

92 (12), 127406/1–127406/4.(6) Burda, C.; Link, S.; Green, T. C.; El-Sayed, M. A. J. Phys. Chem. B

1999, 103 (49), 10775–10780.(7) Hendry, E.; Koeberg, M.; Bonn, M. Phys. Rev. B 2007, 76 (4),

045214–6.(8) Sagar, D. M.; Cooney, R. R.; Sewall, S. L.; Dias, E. A.; Barsan,

M. M.; Butler, I. S.; Kambhampati, P. Phys. Rev. B 2008, 77 (23),235321–14.

(9) Gambetta, A.; Manzoni, C.; Menna, E.; Meneghetti, M.; Cerullo,G.; Lanzani, G.; Tretiak, S.; Piryatinski, A.; Saxena, A.; Martin,R. L.; Bishop, A. R. Nat. Phys. 2006, 2 (8), 515–520.

(10) Norris, D. J.; Efros, A. L.; Rosen, M.; Bawendi, M. G. Phys. Rev. B1996, 53 (24), 16347–16354.

(11) Hartland, G. V. Phys. Chem. Chem. Phys. 2004, 6 (23), 5263–5274.(12) Perner, M.; Gresillon, S.; Marz, J.; von Plessen, G.; Feldmann, J.;

Porstendorfer, J.; Berg, K. J.; Berg, G. Phys. Rev. Lett. 2000, 85(4), 792.

(13) Sun, C.-K.; Liang, J.-C.; Yu, X.-Y. Phys. Rev. Lett. 2000, 84 (1), 179.(14) Ozgur, U.; Lee, C.-W.; Everitt, H. O. Phys. Rev. Lett. 2001, 86 (24),

5604.(15) Mathias Steiner, M. F.; Perebeinos, Vasili; Tsang, James C.; Small,

Joshua P.; Kinoshita, Megumi; Yuan, Dongning; Liu, Jie; Avouris,Phaedon Nat. Nanotechnol. 2009, 4.

(16) Perebeinos, V.; Tersoff, J.; Avouris, P. Phys. Rev. Lett. 2005, 94(8), No. 086802.

(17) Franceschetti, A.; Zunger, A. Phys. Rev. Lett. 1997, 78 (5), 915–918.

(18) Kilina, S. V.; Kilin, D. S.; Prezhdo, O. V. ACS Nano 2009, 3 (1),93–99.

(19) Bardeen, C. J.; Wang, Q.; Shank, C. V. Phys. Rev. Lett. 1995, 75(19), 3410–13.

(20) Salvador, M. R.; Graham, M. W.; Scholes, G. D. J. Chem. Phys.2006, 125 (18), 184709.

(21) Krauss, T. D.; Wise, F. W. Phys. Rev. Lett. 1997, 79 (25), 5102–5105.

(22) Son, D. H.; Wittenberg, J. S.; Banin, U.; Alivisatos, A. P. J. Phys.Chem. B 2006, 110 (40), 19884–19890.

(23) Sewall, S. L.; Cooney, R. R.; Anderson, K. E. H.; Dias, E. A.;Kambhampati, P. Phys. Rev. B 2006, 74 (23), 235328.

(24) Cooney, R. R.; Sewall, S. L.; Anderson, K. E. H.; Dias, E. A.;Kambhampati, P. Phys. Rev. Lett. 2007, 98 (17), 177403–4.

(25) Cooney, R. R.; Sewall, S. L.; Dias, E. A.; Sagar, D. M.; Anderson,K. E. H.; Kambhampati, P. Phys. Rev. B 2007, 75 (24), 245311–14.

(26) Sewall, S. L.; Cooney, R. R.; Anderson, K. E. H.; Dias, E. A.;Sagar, D. M.; Kambhampati, P. J. Chem. Phys. 2008, 129 (8),No. 084701.

(27) Chilla, G.; Kipp, T.; Menke, T.; Heitmann, D.; Nikolic, M.; Froms-dorf, A.; Kornowski, A.; Forster, S.; Weller, H. Phys. Rev. Lett.2008, 100 (5), 057403–4.

(28) Klimov, V. I. J. Phys. Chem. B 2000, 104 (26), 6112–6123.(29) Liu, T.-M.; Yang, M.-J.; Lai, C.-W.; Chou, P.-T.; Chang, M.-H.; Liu,

H.-L.; Sun, C.-K. Phys. Rev. B 2008, 77 (8), 085428–5.(30) Sewall, S. L.; Franceschetti, A.; Cooney, R. R.; Zunger, A.; Kam-

bhampati, P. Phys. Rev. B 2009, 80 (8), No. 081310(R).(31) Creti, A.; Anni, M.; Zavelani-Rossi, M.; Lanzani, G.; Manna, L.;

Lomascolo, M. J. Opt. A: Pure Appl. Opt. 2008, 10 (6), No. 064004.(32) Norris, D. J.; Bawendi, M. G. Phys. Rev. B 1996, 53 (24), 16338–

16346.(33) Hines, M. A.; Guyot-Sionnest, P. J. Phys. Chem. 1996, 100 (2), 468–

71.

© 2010 American Chemical Society 3067 DOI: 10.1021/nl101605r | Nano Lett. 2010, 10, 3062-–3067