2074
J. Phys. Chem. B 2006, 110, 2074-2079
Recombination Dynamics of CdTe/CdS Core-Shell Nanocrystals O. Scho1 ps,† N. Le Thomas,† U. Woggon,*,† and M. V. Artemyev‡ Fachbereich Physik, UniVersita¨t Dortmund, Otto-Hahn-Str. 4, 44227 Dortmund, Germany, and Institute for Physico-Chemical Problems, Belorussian State UniVersity, Minsk 220080, Belarus ReceiVed: October 6, 2005; In Final Form: December 6, 2005
The recombination dynamics of zinc-blende-type, deep-red emitting CdTe/CdS core-shell nanocrystals is studied over a wide temperature range. Two characteristic decay regimes are found: a temperature-dependent decay component of a few nanoseconds and a long-living temperature-independent component of ∼315 ns. The average decay time of the exciton states changes from 20 to 5ns when the temperature is increased from 15 to 295 K. At low temperatures, the observed decay behavior is assigned to thermally induced population and decay of the allowed exchange-split exciton states. At temperatures above T > 100 K, nonradiative decay channels involving phonons start to contribute to the exciton recombination. The observed broad distribution in decay times, monitored by stretched exponential fitting functions, we explain by variations in the electron-hole overlap caused by a partly incomplete CdTe/CdS core-shell structure and the nearly energydegenerated bright and dark state superposition.
1. Introduction Semiconductor nanocrystals (NCs), or colloidal quantum dots (QDs), are currently of great interest, and their applications in light-emitting devices, cavity quantum electrodynamics and biolabeling demand a narrow size distribution, photostability, designed surface properties, and high luminescence quantum yields. Numerous studies are known focusing on surface chemistry, capping strategies and enhancement of photoluminescence (PL) efficiencies. In particular for CdSe NCs, the optical properties of excitons confined in such nanocrystals have been thoroughly examined, and the important effect of exchange interaction for ground-state symmetries, fine structure, and PL dynamics has been explored.1-4 Unlike nanocrystals of CdSe, cadmium telluride (CdTe) nanocrystals are much less studied, in particular data about the temperature dependence of the radiative recombination are missing. Although several groups have performed the synthesis of CdTe nanocrystals in various environments,5-10 studies of their luminescence properties were mainly carried out on small nanocrystal sizes R < 2.5 nm.11-13 For certain applications, CdTe nanocrystals of larger sizes (R > 3 nm) whose emission is in the deep-red or near-infrared (NIR) spectral range are of particular interest, e.g., for solar cells, in vivo biomedical detections, or as donors/acceptors in systems designed for efficient energy transfer.14 Large sizes of CdTe NCs were obtained by growth in coordinating5 or noncoordinating solvents,9 by making use of S-Te exchange reaction by capping CdTe by CdS shells14 or by preparing nanocrystalline CdTe films by electrodeposition.15 Several strategies of improvement of the PL efficiency of capped CdTe nanocrystals have been developed.4,5,16-18 The PL dynamics shall then be predominantly governed by the zinc-blende-type crystal symmetry and let us expect a behavior distinctly different from what is found, e.g., in II-VI nanocrystals with wurtzite crystalline structure. * To whom correspondence
[email protected]. † Universita ¨ t Dortmund. ‡ Belorussian State University.
should
be
addressed.
E-mail:
Figure 1. Room-temperature absorption and PL spectra of the studied CdTe/CdS core-shell nanocrystals in solution.
In this paper, we study CdTe/CdS core-shell nanocrystals with radii R > 3 nm, which emit in the deep-red spectral range. We investigate the temperature dependence of the PL decay time and discuss the influence of the ground state exciton fine structure on the radiative decay times for a nanocrystal material with cubic crystal structure. We will show that the exciton decay is intrinsically a stretched exponential decay caused by the energy degeneracy of an individual nanocrystal and fluctuations in the confinement potential within the nanocrystal ensemble. The preparation of the CdTe/CdS core-shell NCs and the experiments are described in section 2.1. A study of both the steady-state and time-resolved PL at different temperatures is presented in section 2.2. In section 3, the results are discussed and the influence of the exchange splitting and of the CdTe/ CdS interface on the PL dynamics are analyzed. 2. Experiment 2.1. Sample Preparation and Characterization. The CdTe core nanocrystals have been synthesized by the standard high-
10.1021/jp0557013 CCC: $33.50 © 2006 American Chemical Society Published on Web 01/13/2006
Dynamics of CdTe/CdS Core-Shell Nanocrystals
Figure 2. Exemplary fit of a decay curve obtained from a spectral cut at half of the maximum on the low energy side (λ ) 775 nm) of the emission spectrum (see inset, note logarithmic PL intensity scale). The decay is fitted by a function A1 exp(-t/τ1)β1 + A2 exp(-t/τ2)β2 giving for the fast and slow decay component stretched exponentials with coefficients β1 ) 0.585 and β2 ) 1.
temperature reaction as reported for CdSe NCs.5 Briefly, an equimolar amount (0.1-0.5 mmol) of elemental Te and dimethylcadmium was dissolved in 2 mL of trioctylphosphine TOP and quickly injected into the mixture of trioctylphosphine oxide TOPO (1.5 g), hexadecylamine HDA (2.5 g), and octadecene (5 mL) at 200 °C and under intensive stirring. Then, CdTe NCs were grown at 180 °C to the desired size, which was periodically controlled by the spectral position of the NCs PL band. The isolated CdTe core NCs in toluene solution were injected into a fresh portion of a TOPO/HDA mixture to which the precursors for the CdS shell (1 mmol of Cd ethylhexanoate and 1 mmol of thiourea in TOP) were added dropwise at 180 °C. The CdTe/CdS core-shell NCs were isolated, purified, and redissolved in chloroform. From transmission electron microscopy (TEM) experiments, we derive an average NC size of diameter d ≈ 7 nm. The samples of the CdTe/CdS core-shell NCs for the optical investigations were prepared by depositing the NCs on a quartz substrate. Great care was taken in preparing an ensemble of diluted, noninteracting nanocrystals. The observed identity of the PL spectra of the low-concentrated nanocrystals in solution and after drying on the substrate indicates that no aggregates or close-packed layers of coupled nanocrystals are formed, which would result in spectral red shifts.19-21 Figure 1 shows the room temperature absorption and emission spectrum of the prepared CdTe/CdS core-shell NCs.22 By a simple effective mass approximations (EMA)-based modeling of the size-dependent absorption maximum (see also Figure 6), we estimate a radius of R ) 3.5 nm for the CdTe core. This assignment is in good agreement with absorption and structural data from literature and the measured TEM data.4,5,9,15 2.2. Time-Resolved PL. To investigate the PL dynamics, the CdTe/CdS nanocrystals were excited by a Nd:YVO4 laserpumped Ti:Sa laser system providing 150-fs pulses at 864 nm with a repetition rate of 75.4 MHz. After frequency-doubling using a BBO crystal and passing a pulse picker to reduce the repetition rate to 1.2 MHz, the incident pulses are focused in a spotsize of about 20 µm providing thus a pump power density of ∼1 W/cm2. The luminescence is spectrally dispersed by a Chromex 500 spectrometer (spectral resolution 0.7 nm). For the luminescence detection a single-photon-counting system (time resolution 500 ps) has been used. The used excitation density results in the excitation of less than one electron-hole pair per NC, i.e., we are in the low-density limit and can neglect any excitation of multiexcitons or the presence of Auger processes. The PL decay signal IPL(t) is recorded with a signal dynamics covering at least 3 orders of magnitude (see Figure 2). The logarithmic plot of the PL survey spectrum in the inset of Figure
J. Phys. Chem. B, Vol. 110, No. 5, 2006 2075 2 shows that the main radiative recombination channel is at a wavelength around 750 nm. No significant trap emission is present which might open an additional recombination path at lower energies. The weak emission at 600 nm arises from a precursor of colloidal CdTe in the equilibrium of Ostwald ripening.18 To extract the radiative lifetime from the PL decay time usually a suited kinetic model has to be applied providing a reasonable fitting function. The current studies in the literature refer to luminescence decays that are highly nonexponential10,23-25. For CdTe NCs, as will be discussed below, the microscopic origin of this multiexponential decay behavior has still to be clarified. The CdTe nanocrystals studied here are well isolated and cannot interact with each other which excludes microscopic hopping and migration processes as the origin for stretched exponential decays.26 In the following, we analyze the measured PL-decays with the bath temperature as the external parameter and derive average decay times, which will then be discussed within a qualitative model. The decay curves show best-fitting parameters by using stretched exponential fitting functions. The well-known biexponential fitting functions usually applied for CdSe nanocrystals and caused by a thermally induced brightstate filling from dark states could not be used to describe the CdTe decay dynamics. An example for the applied fitting routine is given in Figure 2. Different fit routines are tested both biand multiexponential fit functions as well as a single-exponent stretched exponential decay. The best fit is obtained with a two-component Kohlrausch-Williams-Watts function A1 exp(-t/τ1)β1 + A2 exp(-t/τ2)β2 using two stretched exponentials with coefficients β1 and β2. Although the applied fit model starts with these two stretched exponential functions, we will show later that the best fits resulted in β2 ) 1, which means the second, long decay component is essentially a monoexponential decay. The parameter A1 and A2 characterize the relative weighting of the fast and slow decay component. The value of β has implicitly the information about the spread in the decay time distribution, i.e., as smaller β as larger the variation in τ. From the stretched exponential fit of the decay curve we can derive an average lifetime ) (τ/β)‚Γ(1/β) as introduced, e.g., in ref 27. 3. Results and Discussion The current literature presents only a few reports about PL decay studies on CdTe NCs,8,10,11,28,29 mainly focusing on thioland TOP/DDA-capped nanocrystals8,10,11 or CdTe NCs grown in noncoordinating solvents.29 The reported nonexponential decay times cover an extremely broad range in lifetimes, between a few hundreds of picoseconds to a few hundreds of nanoseconds, for which no clear explanation exists until now. Surface-related traps as a result of incomplete capping were discussed. The exciton could possibly be localized on a bond between the crystallite core and a capping thiol group. Surfacerelated electron and hole traps have been likewise introduced in refs 29 and 30 to explain recombination data obtained by PL-upconversion techniques at uncapped CdTe nanocrystals. To clarify the recombination process of CdTe/CdS NCs further and to discriminate between effects of excitonic fine structure and surface-related trap processes, we discuss in the following the temperature-dependence of the PL decay and intensity. 3.1. Temperature Dependence. Figure 3 and Figure 4 show the results for both steady-state and time-resolved PL of deepred emitting CdTe/CdS core-shell NCs measured in a wide temperature range. With increasing temperature the emission maximum shifts toward lower energies and the PL band gets broader.
2076 J. Phys. Chem. B, Vol. 110, No. 5, 2006
Figure 3. Steady-state PL spectra of CdTe/CdS core-shell NCs (dcore ∼ 7 nm, 2 ML CdS shell) at different temperatures. The inset shows the temperature-dependent peak shift fitted with the linear equation 1.716 eV - 0.22 meV/K.
Scho¨ps et al.
Figure 5. Emission intensity (integrated area) vs temperature. The solid line is a fit with two exponential functions with activation/deactivation energies of 0.8 and 4.0 meV, respectively (for details see text).
TABLE 1: Fit Parameter Derived for Different Temperatures Using the Fitting Function A1 exp(-t/τ1)β1 + A2 exp(-t/τ2)β2
Figure 4. Decay curves for the spectra of Figure 3 measured at different temperatures. The inset shows the temperature dependence of 〈τ1〉, the averaged initial decay time, showing the transition between two almost constant values of 20 and 5 ns at low and high temperatures, respectively.
The PL bandwidth (full width at half-maximum, fwhm) changes from 38 nm at low temperatures to 54 nm at room temperature according to the linear equation of fwhm(T) ) 35.8 nm + 0.06 nm/K. The shift of the PL-peak maximum with temperature (inset of Figure 3) can be fitted with dE/dT ) 2.2 × 10-4 eV/K, which is very similar to the known temperaturedependent band gap shift of bulk CdTe of dE/dT ) 3 × 10-4 eV/K31 indicating that the phonon-mediated gap-shift is similar for nanocrystals and bulk CdTe. Figure 4 shows the decay curves for the spectra of Figure 3 measured at different temperatures. The energy shift dE/dT derived from Figure 3 is used to calibrate the detection energy for the PL-decay measurements. The decay curves plotted in Figure 4 have been obtained from narrow spectral cuts at the corresponding half-maximum position at the low-energy side of the PL-spectrum. The average decay times 〈τ1〉 plotted in the inset are derived according to the procedure described in section 2.2 (see ref 27). By application of a stretched exponential fit function according to A1 exp(-t/τ1)β1 + A2 exp(-t/τ2)β2, the PL dynamics can well be described over more than 3 orders of magnitude in the signal dynamics and covering a wide temperature range from 15 to 295 K. The corresponding fit parameter A1, τ1, β1, A2, τ2, and β2 are listed in Table 1. The temperature
T (K)
A1
τ1
β1
A2
τ2
β2
15 25 50 100 150 200 250 295
0.996 0.99422 0.99452 0.99762 0.9985 0.99945 0.9995 0.9995
13.195 13.725 12.08 7.74982 4.54885 2.8 2.17367 2.25
0.585 0.58 0.5865 0.53526 0.47831 0.462 0.48014 0.5
0.004 0.0057 0.0055 0.00238 0.0015 0.00055 0.0005 0.0005
350 314 315.6 328 333 333 300 260
1 1 1 1 1 1 1 1
dependence of the average, fast initial decay time 〈τ1〉 is plotted in the inset and shows a continuous decrease with increasing temperature. The most striking results of that temperaturedependent decay time analysis are: (i) both the average time constant 〈τ1〉 of the fast initial decay (see inset of Figure 4) as well as τ1 itself (see Table 1) show a continuous decrease in the decay time with increasing temperature, (ii) the coefficient β1 is approximately 0.5 with a slight tendency to decrease with increasing temperature, (iii) a temperature-independent, monoexponential (i.e., β2 ) 1) long-time behavior exists (represented by the term A2 exp(-t/τ2)β2), which can be fit to a single decay time of approximately τ2 ) 315 ns. In the following subsection we discuss a proposal for possible origins of the observed features i-iii. We conclude this section with an analysis of the change in PL intensity when the temperature is increased from 15 to 295 K. In Figure 5 the emission intensity derived from the integrated area below the emission band is plotted vs temperature. Effects of photodegradation during data acquisition are taken into account by measuring the intensity before and after data accumulation and correcting the PL intensity by this degradation factor. In the temperature range between 15 < T < 25 K, a slight increase by a factor of 1.2 is observed followed by a steady decrease for temperatures above T > 50 K. The data can be fit to a superposition of two Boltzmann distribution functions with activation/deactivation energies of 0.8 and 4.0 meV, respectively. As will be discussed in more detail in section 3.2., we assign the slight increase in efficiency at low temperatures to a thermally induced population change among the allowed exchange-split exciton states having different oscillator strength. At higher temperatures 50 > T > 295 K, the increasing phonon population (both confined acoustic phonons and LO phonons) supports nonradiative recombination which results in a decrease of fluorescence intensity by about a factor of 8, a similar factor as measured for the fastening of the decay time τ1 in the same temperature range.
Dynamics of CdTe/CdS Core-Shell Nanocrystals
Figure 6. Radius-dependent exchange splitting energies calculated for zinc-blende-type CdTe nanocrystals. The parameter used are ωST3D ) 0.07 meV, aex ) 6.5 nm, γ1 ) 4.8, γ2 ) 1.5. The inset shows the size-dependent first absorption peak position at T ) 295 K derived from a simple EMA model. The bulk CdTe band gap is Eg ) 1.606 eV at T ) 4 K.
3.2. Influence of the Exchange Splitting and Exciton Fine Structure on the PL Dynamics. In zinc blende CdTe, the conduction and valence bands are made from s orbitals of Cd and p orbitals of Te, respectively, in first approximation. The orbital angular momentum is mixed with the spin angular momentum, and the valence band is split into the topmost J ) 3/ band and the split-off J ) 1/ band. Since CdTe has the largest 2 2 spin-orbit splitting of 0.927 eV among CdS, CdSe, and CdTe, the split-off band is expected to mix only weakly with the topmost valence band. The energies of the lowest electron-hole pair transitions have been computed for CdTe NCs using different methods based on EMA or tight-binding calculations.32-38 In experiments, the quantized electronic levels are observed to shift monotonically with decreasing size without any crossing or anticrossing, reflecting the rather simple valenceband structure of CdTe.7,33 We therefore use likewise a simple EMA model to estimate the theoretical exciton band edge structure of the CdTe NCs neglecting the thin CdS shell and assume complete strain relaxation. The inset of Figure 6 shows the size dependence of the first absorption peak position at T ) 295 K. On the basis of the size-dependent absorption maximum and the data from TEM, we use the size of R ) 3.5 nm as a base for the discussion of effects of electron-hole exchange interaction in our CdTe NCs in the following. For the CdTe nanocrystals studied here, we first model the electronhole exchange interaction following the theory of invariants of ref 2. The 8-fold degenerate exciton ground state seS3/2 is then split into a group of lower levels of dark 0L, (1L, and (2 exciton states and a group of upper levels of bright states with total angular momentum projection 0U and (1U. To calculate the splitting between the dark and bright states we take for the ratio of the light to heavy hole effective masses the value 0.23, for the bulk exciton Bohr radius 6.5 nm and for the bulk exciton splitting ωST3D ) 0.07 meV.39 In Figure 6 we plot the calculated energies of the exciton band edge structure vs the NC radius R relative to the seS3/2 state with no exchange interaction. As can be seen (see also ref 2), in zinc-blende-type CdTe NCs, the ground-state exciton is split by exchange interaction; however, the fine structure is different from that known for wurtzite-type CdSe NCs with five states. In CdTe NCs we obtain two groups of energetically degenerate exciton states and both have contributions from allowed optical transitions. Their energy splitting is smaller and amounts to only a few millielectron volts, in particular in the size range of R > 3 nm, allowing a simultaneous, thermal population of all states, even at cryogenic temperatures. Important consequences of that excitonic fine structure on the emission dynamics of excitons in CdTe are:
J. Phys. Chem. B, Vol. 110, No. 5, 2006 2077 (i) the lack of a temperature-dependent change in population as usually observed in CdSe nanocrystals,40 i.e., the population of a single bright (1U exciton state thermally filled from a single, energetically nondegenerated (2 dark exciton state; (ii) the energy degeneracy of allowed exciton states with different oscillator strength can result in a decay with different radiative lifetimes, i.e., a stretched exponential decay is possible already for a single CdTe NC; (iii) the specific exciton fine structure of zinc-blende-type CdTe nanocrystals let us expect an increase in quantum efficiency due to contributions of the upper (1U and 0U bright states, in particular in the low temperature range between 5 and 50 K. This is indeed observed until a certain temperature where the effect is then overcompensated by the increase in phonon population (see Figure 5). The activation energy of ∼1 meV coincides with the energy difference between the lower (1L, 0L, and (2 exciton states and the upper (1U and 0U bright states in a cubic CdTe nanocrystal (Figure 6). For the explanation of features i and ii derived from our experimental data in Figure 4 and Table 1, we can conclude that for CdTe NCs the total decay is determined by a superposition of several decay times which is one of the possible origins for the strong nonexponential decay behavior. On the basis of the results presented in Figure 6, we can likewise derive an explanation for the variation in τ1 and in the parameter β1 from 0.58 to 0.5 for temperatures above T ) 50 K. According to ref 2, the optically allowed exciton states have different oscillator strength, which varies, relative to the 0U state, between the three states (1U, 0U, and (1L with factors 2, 1, and 0 for the ideal spherical NC and with factors 1.6 (1.4), 1, and 0.4 (0.6) for oblate (prolate) NCs. The PL decay is then expected to exhibit a multiexponential dynamics due to the thermally induced population of states with different oscillator strength with a tendency of a fastening of the radiative recombination with higher temperatures. This could explain why in the Figure 4 the first part of the decay curve (that is to say between 0 and 200 ns) has a higher stretched exponential character for T > 100 K. The above-discussed effect of the exchange interaction on the PL dynamics, however, cannot explain why in some reports on room temperature PL decay measurements monoexponential decay curves of several tens of nanoseconds are observed.11,12,29 Likewise it cannot explain the observation of a temperature-independent decay component of ∼315 ns observed in this study. Also the low value of the stretched exponential coefficient β of 0.5 indicates a much larger variation in the radiative lifetime τ1, over more than 1 order of magnitude, as can be explained by the variation in oscillator strength of the different contributing bright states. To understand these facts, we have to look closer into the surface-related contributions to the emission dynamics, which will be done in the following subsection. 3.3. Influence of the CdTe/CdS Surface Boundary on the PL Dynamics. The discussion in subsection 3.2 is valid for ideal exciton states, i.e., both electron and hole are localized in the CdTe core surrounded by vacuum. Uncapped CdTe core NCs prepared via a high-temperature reaction in organic solvents, however, have a very low quantum efficiency caused by surface-related nonradiative recombination centers. To increase the quantum yield, the CdTe core can be capped by thin shells of various materials, such as CdS, CdSe, or HgTe.14,18,41-44 Core/shell CdTe/CdSe nanocrystals are grown in ref 43 and introduced as structures with spatially indirect band gap, so-called type-II quantum dots. In literature14,17,18,44 it has been proposed that thiol-capped CdTe nanocrystals represent a kind of naturally sulfur-capped surface (CdS shell)
2078 J. Phys. Chem. B, Vol. 110, No. 5, 2006
Scho¨ps et al. 23 and 25, which resulted in a multiexponential decay behavior already for a single CdSe/ZnS nanocrystal. The low β-value, i.e., the broad distribution in radiative decay times can also be discussed in terms of differently localized electron and hole states for each NC caused by the individual configuration of the CdTe/CdS boundary.49 This effect adds to the spread in radiative lifetimes which arises already from the exchange interaction lowering the β value in the stretched exponential function further. A single monoexponential decay might still be observable for the limit of localization of the electron in the shell and the hole in the NC center (without electron escape and nonradiative loss). The observation of such monoexponential decays, in particular at higher temperatures, indicates the deviation of the exciton from an ideal exciton with identical electron and hole overlap toward an indirect-type exciton. 4. Summary
Figure 7. Schematic plot of the electron and hole wave functions using the sketched band alignment for an indirect-type CdTe/CdS core-shell structure.
created by mercapto groups covalently attached to the surface cadmium atoms. Such a structure results not only in the removal of dangling bonds of Te atoms from the surface, but moreover leads to the formation of a core-shell-like structure. However, until now, the discussion of optical properties of CdTe/CdS core-shell structures is done in the context of surface stabilization and suppression of nonradiative processes, and no analysis of the influence of a CdS shell on the electronic states is done. An exact calculation of the spatial probability of the electron and hole wave function for a CdTe NC completely capped by a closed CdS shell needs the knowledge of the band alignment parameters and of the lattice-mismatch-induced strain. Strain might be negligible as reported in refs 14 and 44 and explained by an incomplete CdS shell causing a release of the strain. In concern of the band alignments, the current literature provides different data sets, both from theory and experiments.45-48 A simple estimate for a possible radial distribution of the electron and hole wave functions in a CdTe/CdS core-shell NC (CdS shell thickness 0.6 nm, CdTe core radius 3.5 nm) is illustrated with the sketch in Figure 7. For the shown calculation, strain has been neglected and the band parameter of ref 45 are used. The most important result for even this very simplified picture is that under certain conditions a CdTe/CdS core-shell NC can develop an indirect-type exciton, i.e., the hole is then predominantly confined to the CdTe core, while the electron is predominantly localized in the CdS shell. The idea of either electron or hole localization inside the NC shell of a type-II quantum dot has been formulated already for other material combinations of CdTe/CdSe(Core/Shell) and CdSe/ZnTe(Core/ Shell) NCs.43 The spatial separation of electron and hole can result in a decrease of the wave function overlap and thus longer radiative lifetimes. Although the model of a CdTe nanoparticle which is fully coated by a CdS bulk like monolayer is not exactly valid in our CdTe/CdS sample (the growth of a complete CdS shell around large NCs is still subject of intensive investigations and the shell produced in this study is still an incomplete one), the result from Figure 7 has implications for the full understanding of the PL dynamics. For example, when the electron is localized closer to the CdTe/CdS boundary and gets a higher probability to get trapped by a defect, the stretched exponential behavior of the PL decay could be explained by a fluctuating nonradiative decay rate of each single nanocrystal as demonstrated in refs
In zinc-blende-type, deep-red emitting CdTe/CdS nanocrystals the PL decay time is characterized by stretched exponential decay curves. The stretched exponentials are from a broad distribution in decay times due to a nearly energy-degenerated bright and dark state superposition and variations in the electronhole overlap caused by a (partly incomplete) CdTe/CdS coreshell structure. The average decay time of the bright exciton states changes from 20 to 5 ns when the temperature is increased from 15 to 295 K. At low temperatures T < 25 K, the emission intensity slightly increases by a factor of 1.2 when the temperature is increased from 15 to 25 K which we assign to a thermally induced population change among the allowed exchange-split exciton states having different oscillator strength. The radiative lifetime is in the order of magnitude of nanoseconds, which implies a possible application of NIR emitting CdTe NCs in CQED experiments or for infiltration in photonic crystals. At higher temperatures 50 > T > 295 K, the increasing phonon population supports nonradiative recombination resulting in a decrease of emission intensity. The monoexponential, almost temperature-independent decay of 315 ns observed for later times is either due to the decay of the optically forbidden (2 dark exciton state, by a deep trap decay or caused by indirect excitons formed in some of the CdTe/CdS core-shell nanocrystals. Acknowledgment. Financial support of this work by the DFG (Graduiertenkolleg GRK 726, Wo477/ 18), the EU Project HPRN-CT-2002-00298, and INTAS 01/2100 is gratefully acknowledged. References and Notes (1) Nirmal, M.; Norris, D. J.; Kuno, M.; Bawendi, M. G.; Efros, Al. L.; Rosen, M. Phys. ReV. Lett. 1995, 75, 3728. (2) Efros, Al. L.; Rosen, M.; Kuno, M.; Nirmal, M.; Norris, D. G.; Bawendi, M. G. Phys. ReV. B 1996, 54, 4843. (3) Franceschetti, A.; Fu, H.; Wang, L. W.; Zunger, A. Phys. ReV. B 1999, 60, 1819. (4) Rajh, T.; Micic, O. I.; Nozik, A. J. J. Phys. Chem. 1993, 97, 11999. (5) Murray, C. B.; Norris, D. J.; Bawendi, M. G. J. Am. Chem. Soc. 1993, 115, 8706. (6) Rogach, A. L.; Katsikas, L.; Kornowski, A.; Dangsheng, Su; Eychmu¨ller, A.; Weller, H. Ber. Bunsen-Ges. Phys. Chem. 1996, 100, 1772. (7) Masumoto, Y.; Sonobe, K. Phys. ReV. B 1997, 56, 9734. (8) Talapin, D. V.; Haubold, S.; Rogach, A. L.; Kornowski, A.; Haase, M.; Weller, H. J. Phys. Chem. B 2001, 105, 2260. (9) Yu, W. W., Wang, Y. A.; Peng, X. Chem. Mater. 2003, 15, 4300. (10) Kapitonov, A. M.; Stupak, A. P.; Gaponenko, S. V.; Petrov, E. P.; Rogach, A. L.; Eychmu¨ller, A. A. J. Phys. Chem. B 1999, 103, 10109.
Dynamics of CdTe/CdS Core-Shell Nanocrystals (11) Wuister, S. F.; van Driel, F.; Meijerink, A. J. Lumin. 2003, 102/ 103, 327. (12) Wuister, S. F.; Koole, R.; de Mello Donega, C.; Meijerink, A. J. Phys. Chem. B 2005, 109, 5504. (13) Franzl, T.; Koktysh, D. S.; Klar, T. A.; Rogach, A. L.; Feldmann, J.; Gaponik, N. Appl. Phys. Lett. 2004, 84, 2904. (14) Schreder, B.; Schmidt, T.; Ptatschek, V.; Winkler, U.; Materny, A.; Umbach, E.; Lerch, M.; Mu¨ller, G.; Kiefer, W.; Spanhel, L. J. Phys. Chem. B 2000, 104, 1677. (15) Mastai, Y.; Hodes, G. J. Phys. Chem. B 1997, 101, 2685. (16) Gao, M.; Kirstein, S.; Mo¨hwald, H.; Rogach, A. L.; Kornowski, A.; Eychmu¨ller, A.; Weller, H. J. Phys. Chem. B 1998, 102, 8360. (17) Gaponik, N.; Talapin, D. V.; Rogach, A. L.; Hoppe, K.; Shevchenko, E. V.; Kornowski, A.; Eychmu¨ller, A.; Weller, H. J. Phys. Chem. B 2002, 106, 7177. (18) Borchert, H.; Talapin, D. V.; Gaponik, N.; McGinley, C.; Adam, S.; Mo¨ller, T.; Weller, H. J. Phys. Chem. B 2003, 107, 9662. (19) Gindele, F.; Westpha¨ling, R.; Woggon, U.; Spanhel, L.; Ptatscheck, V. Appl. Phys. Lett. 1997, 71, 2181. (20) Artemyev, M. V.; Bibik, A. I.; Gurinovich, L. I.; Gaponenko, S. V.; Woggon, U. Phys. ReV. B 1999, 60, 1504. (21) Artemyev, M. V.; Woggon, U.; Jaschinski, H.; Gurinovich, L. I.; Gaponenko, S. V. J. Phys. Chem. B 2000, 104, 11617. (22) The decay dynamics will be presented here for one selected energy at the low-energy tail of the spectrum. We also checked the behaviour for a few other energies nearby and did not find a qualitatively different behaviour when going to even larger sizes, i.e., longer wavelengths. For very large nanocrystal sizes, the effect of exchange interaction becomes negligible and we did not gain additional insights into the problem under study here. The data for smaller CdTe nanocrystals did not much differ from the published data and were therefore not included in this study here. (23) Schlegel, G.; Bohnenberger, J.; Potapova, I.; Mews, A. Phys. ReV. Lett. 2002, 88, 137401. (24) de Mello Donega, C.; Hickey, S. G.; Wuister, S. F.; Vanmaekelbergh, D.; Meijerink, A. J. Phys. Chem B 2003, 107, 489. (25) Fisher, B. R.; Eisler, H. J.; Stott, N. E.; Bawendi, M. G. J. Phys. Chem. B 2004, 108, 143. (26) Excitons in disordered systems, such as porous silicon, can migrate inside a potential landscape of local potentials. In such systems, stretched exponential decay behaviours can be observed, however, the microscopic origin of the stretched exponential fitting function is there a broad distribution in decay times due the dispersive motion of exciton. This model cannot be applied here since the nanocrystals are isolated and cannot interact with each other. (27) Lindsey, C. P.; Patterson, G. D. J. Chem. Phys. 1980, 73, 3348. (28) van Driel, A. F.; Allan, G.; Delerue, C.; Lodahl,P.; Vos, W. L.; Vanmaekelbergh, D., Phys. ReV. Lett. 2005, 95, 236804. (29) Wang, W.; Yu, W.; Zhang, Y.; Aldana, J.; Peng, X.; Xiao, M. Phys. ReV. B 2003, 68, 125318.
J. Phys. Chem. B, Vol. 110, No. 5, 2006 2079 (30) Rakovich, Y. P.; Filonovich, S. A.; Gomes, M. J. M., Donegan, J. F.; Talapin, D. V.; Rogach, A. L.; Eychmu¨ller, A. Phys. Status Solidi B 2002, 229, 449. (31) Camasset, J.; Auvergne, D.; Mathieu, H.; Triboulet, R.; Varfainy, M. Solid State Commun. 1973, 13, 63. (32) Esch, V.; Fluegel, B.; Khitrova, G.; Gibbs, H. M.; Xu Juajin; Kang, K.; Koch, S. W.; Liu, L. C.; Risbud, S. B.; Peyghambarian, N. Phys. ReV. B 1990, 42, 7450. (33) de Oliveira, C. R. M.; de Paula, A. M.; Filho, F. O.; Neto, J. A. M.; Barbosa, L. C.; Alves, O. L.; Menezes, E. A.; Rios, J. M. M.; Fragnito, H. L.; Cruz, C. H. B.; Cesar, C. L. Appl. Phys. Lett. 1995, 66, 439. (34) Redigolo, M. L.; Arellano, W. A.; Barbosa, L. C.; Brito Cruz, C. H.; Cesar, C. L.; de Paula, A. M. Semicond. Sci. Technol. 1999, 14, 58. (35) Richard, T.; Lefebvre, P.; Mathieu, H.; Allegre, J. Phys. ReV. B 1996, 53, 7287. (36) Lefebvre, P.; Richard, T.; Mathieu, H.; Allegre, J. Solid State Commun. 1996, 98, 303. (37) Perez-Conde, J.; Bhattacharjee, A. K.; Chamarro, M.; Lavallard, P.; Petrikov, V. D.; Lipovskii, A. A. Phys. ReV. B 2001, 64, 113303. (38) Perez-Conde, J.; Bhattacharjee, A. K. Solid State Commun. 1999, 110, 259. (39) Wardzynski, W.; Suffczynski, M. Solid State Commun. 1972, 10, 417. (40) Bawendi, M. G.; et al. J. Chem. Phys. 1992, 96, 946; Nirmal, M.; et al. Phys. ReV. Lett. 1995, 75, 3728; Fan, X.; et al. Phys. ReV. B 2001, 64, 115310; Labeau, O.; et al. Phys. ReV. Lett. 2003, 90, 257404; Crooker, S. A.; et al. Appl. Phys. Lett. 2003, 82, 2793. (41) Harrison, M. T.; Kershaw, S. V.; Burt, M. G.; Eychmu¨ller, A.; Weller, H.; Rogach, A. L. Mater. Sci. Eng. B 2000, 69, 355. (42) Kershaw, S. V.; Burt, M.; Harrison, M.; Rogach, A. L.; Weller, H.; Eychmu¨ller, A. Appl. Phys. Lett. 1999, 75, 1694. (43) Kim, S.; Fisher, B.; Eisler, H. J.; Bawendi, M. G. J. Am. Chem. Soc. 2003, 125, 11466. (44) Rockenberger, J.; Tro¨ger, L.; Rogach, A. L.; Tischer, M.; Grundmann, M.; Eychmu¨ller, A.; Weller, H. J. Chem. Phys. 1998, 108, 7807. (45) Ablyazov, N. N.; Areshkin, A. G.; Melekhin, V. G.; Suslina, L. G.; Fedorov, D. L. Phys. Status Solidi B 1986, 135, 217. (46) Niles, D. W.; Ho¨chst, H. Phys. ReV. B 1990, 41, 12710. (47) Su-Huai Wei, Zhang, S. B. Phys. ReV. B 2000, 62, 6944. (48) Su-Huai Wei, Zhang, S. B.; Zunger, A. J. Appl. Phys. 2000, 87, 1304. (49) The observed temperature-dependent change in decay times from 13 to 2.25 ns cannot be explained by a thermally activated change in electron localization from shell to core states. The thermal activation energy necessary for such a direct-indirect exciton transformation is much higher than the provided thermal energy and the observed change in decay times does not correspond to the decay time difference expected for a transformation from an indirect-type core-shell exciton towards a direct-type core exciton state.