DOI: 10.1002/cphc.201402826
Articles
Localization and Dynamics of Long-Lived Excitations in Colloidal Semiconductor Nanocrystals with Dual Quantum Confinement Su Liu,[a, b] Nicholas J. Borys,*[a, c] Sameer Sapra,[d] Alexander Eychmìller,[e] and John M. Lupton[a, f] Semiconductor nanocrystals consisting of a quantum dot (QD) core and a quantum well (QW) shell, where the QD and QW are separated by a tunneling barrier, offer a unique opportunity to engineer the photophysical properties of individual nanostructures. Using the thicknesses of the corresponding layers, the excitons of the first and second excited states can be separated spatially, localizing one state to the QD and the other to the QW. Thus the wave function overlap of the two states can be minimized, suppressing non-radiative thermalization between the two wells, which in turn leads to radiative relaxation from both states. The molecular analogy to such dual emission would be the inhibition of internal conversion, a special case that violates Kasha’s rule. Using nanosecond time-resolved spectroscopy of QDQW CdSe/ZnS onion-like nanocrystals, an intermediate regime of exciton separation and suppressed thermalization is identified where the non-radiative relaxation of the higher-energy state is slowed, but not completely inhib-
ited. In this intermediate thermalization regime, the temporal evolution of the delayed emission spectra resulting from trapped carriers mimic the dynamics of such states in nanocrystals that consist of only a QD core. In stark contrast, when a higher-energy metastable state exists in the QW shell due to strongly suppressed interwell thermalization, the spectral dynamics of the long-lived excitations in the QD and QW, which are spectrally distinct, are amplified and differ from each other as well as from those in the core-only nanocrystals. This difference in spectral dynamics demonstrates the utility of exploiting well-defined exciton localization to study the nature and spatial dependence of the intriguing photophysics of colloidal semiconductor nanocrystals, and illustrates the power of nanosecond gated luminescence spectroscopy in illuminating complex relaxation dynamics which are entirely masked in steadystate or ultrafast spectroscopy.
1. Introduction Colloidal semiconductor heterostructure nanocrystals unify the ease of wet-chemical-based synthesis with the quantum me[a] Dr. S. Liu, Dr. N. J. Borys, Prof. J. M. Lupton Department of Physics and Astronomy The University of Utah Salt Lake City, UT 84112 (USA) E-mail: [email protected] [b] Dr. S. Liu Department of Applied Physics Chalmers University of Technology Kemivagen 9 41296 Gothenburg (Sweden) [c] Dr. N. J. Borys Molecular Foundry Lawrence Berkeley National Laboratory Berkeley, CA 94720 (USA) [d] Prof. S. Sapra Department of Chemistry Indian Institute of Technology Delhi New Delhi, 110016 (India) [e] Prof. A. Eychmìller Physical Chemistry TU Dresden, 01062 Dresden (Germany) [f] Prof. J. M. Lupton Institut fìr Experimentelle und Angewandte Physik Universitt Regensburg, 93053 Regensburg (Germany)
ChemPhysChem 2015, 16, 1663 – 1669
chanical wave-function-engineering capabilities of traditional semiconductor heterostructures.[1–3] These single-particle architectures are commonly limited to variants of the core–shell geometry, a structural configuration where a spherical core of one material is surrounded by a shell of a second material. Through the selection of the constituent materials, significant control of the spatial localization of the wave functions of the electron and hole can be achieved, thereby enabling singleparticle systems that either maximize or minimize the carrier wave function overlap and hence the oscillator strength of the lowest-energy exciton state.[2, 3] Similarly, nanocrystal heterostructures that consist of three or more components and are more reminiscent of semiconductor superlattices[4–6] enable engineering of exciton localization by supporting the formation of two or more well-defined, spatially separated quantum wells (QWs) on individual nanocrystals.[7–11] The structural parameters include the size of the wells and barrier layer thickness that separates the wells. With these parameters, the quantum-confined energies of the wells and their electronic coupling can be tuned in a remarkably systematic manner. As such, these systems have attracted attention as low-cost platforms for broadband lighting,[12] optical gain media,[13] multicolor fluorescent probes,[14] and coupled qubit architectures.[11]
1663
Ó 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Articles Although more complex multi-well colloidal nanosystems are continually emerging,[15–17] the traditional workhorse configuration is an onion-like sphere combining quantum dot (QD) and quantum well (QW) structures on the same particle. The first reported colloidal QDQW system was comprised of a larger-bandgap QD material with an embedded QW of a different, smaller-bandgap material.[18, 19] In this structure, the coupling between the QD and QW of different materials alters the energetics, dynamics, and spatial distribution of the first excited state in the nanocrystal.[20–22] Alternatively, inverse QDQW structures can be engineered where the QD and QW are comprised of the same material: the QD core is encapsulated by a larger-bandgap material, which acts as a tunneling barrier, followed by a second layer of the QD material that forms a QW.[12, 14, 23] In this configuration, the radius of the QD and radial thickness of the QW control their respective quantumconfinement energies, whereas the thickness of the tunneling barrier controls the electronic coupling between the two potential wells.[8, 10, 13, 14] The coupling between the QD and QW critically depends on the overlap of the wave functions of the two states and includes mechanisms such as Dexter-type energy transfer. Therefore the structural parameters of such an “inverse” QDQW architecture can be tailored to produce nanocrystals with different non-radiative relaxation rates between the QD and QW.[10, 13] For example, a tunneling barrier with a thickness of ~ 1 monolayer permits efficient coupling between the QD and QW, enabling thermalization from the higher-energy exciton state of the QW (or QD) to the lowerenergy exciton state of the QD (or QW).[14] On the other hand, a thicker tunneling barrier (~ 3 monolayers) has been shown to electronically isolate the QD and QW, effectively blocking the thermalization pathway between the two potential wells.[14] Efficient radiative relaxation then occurs from the lowest-energy excitons of both the QD and QW, giving rise to dual-color emission from single nanoparticles (provided the lowest energy excitons in the QD and QW have different energies).[10, 12–14] Because the energy of the emission can be used to identify the region of exciton recombination (i.e., in the QD or QW), these dual-color QDQW architectures and related multi-well systems provide a unique opportunity to explore the effects of exciton localization (i.e., the spatial distribution of the exciton wave function) on the rich photophysics of colloidal nanocrystals such as carrier trapping and relaxation dynamics,[24] photon antibunching[16, 25] as well as single-particle blinking and spectral diffusion phenomena.[26] Here, we use ensemble-level gated fluorescence spectroscopy on onion-like CdSe/ZnS inverse QDQW nanocrystals to investigate intraparticle, interwell thermalization processes and the relaxation dynamics of long-lived excitations. Whereas the presence of two emission bands in time-integrated spectroscopy suggests that thermalization between the QD and QW is indeed suppressed, time-resolved spectroscopy reveals an intermediate regime of inhibited relaxation between the two potential wells. The degree of interwell thermalization suppression appears to have a dramatic effect on the energetic dynamics of long-lived excitations (i.e., excited states with lifetimes that exceed that of conventional band-edge excitons) as ChemPhysChem 2015, 16, 1663 – 1669
www.chemphyschem.org
observed through the evolution of their emission spectra with increasing delay after the initial excitation. When thermalization between the two wells is permitted, we find that the spectral dynamics of the luminescence are indistinguishable from a nanoparticle architecture that is comprised of only a surfacepassivated core. However, in the case when the QD and QW are isolated, we identify distinctly different spectral progressions of the two emission bands, revealing the potential to utilize exciton localization phenomena in QDQW systems to investigate the underlying physics of long-lived (trap-mediated) excitations in colloidal semiconductor nanocrystals.
2. Results and Discussion The QDQW and core-only CdSe/ZnS nanoparticles were synthesized following the successive ionic layer adsorption and reaction (SILAR)based process described by Peng et al.[14] and Sapra et al.[12] For the optical spectroscopy measurements, the nanocrystal samples were dispersed in hexane and measured under ambient conditions. The absorption spectra were recorded on a standard UV/Vis spectrophotometer. For the timeintegrated and time-resolved photoluminescence measurements, the samples were excited with a pulsed solid-state laser operating at 355 nm with ~ 60 mJ per pulse and a tunable repetition rate from 1–200 Hz. The emission was collected and dispersed with a spectrometer and recorded with a gated intensified charge-coupled device (ICCD) camera (Andor iStar) that was synchronized with the excitation pulses. Peak positions in the emission spectra were extracted by fitting either one (for spectra with a single peak) or the summation of two (for spectra with double peaks) Gaussian functions. Figure 1 illustrates the band alignment scheme of CdSe/ZnS QDQW nanocrystals, the three different nanoparticle architectures that were explored in this study and their corresponding steady-state absorption and time-integrated luminescence spectra. Due to the substantial offsets in the conduction and valance bands between the two materials (Figure 1 a), the ZnS serves as a tunneling barrier for both the excited state electrons and holes, thereby creating quantum wells in the CdSe regions of the nanocrystal.[14] Each nanocrystal system consists of a 3.5 nm CdSe core that is encapsulated by different configurations of subsequent layers of ZnS and CdSe. The core-only particles (Figure 1 b) are augmented only by a surface-passivating layer of ZnS that is ~ 2 monolayers thick. The QDQW structures in Figure 1 c and Figure 1 d are referenced by their structural configuration. In Figure 1 c, for example, the core (C) is progressively encapsulated by ~ 3 monolayers of ZnS, ~ 4 monolayers of CdSe, and finally ~ 1 monolayer of ZnS, and is therefore referred to as a “C341” architecture. Likewise, the QDQW architecture in Figure 1 d, “C422,” contains a CdSe core that is encapsulated with ~ 4 monolayers of ZnS, ~ 2 monolayers of CdSe and finally ~ 2 monolayers of ZnS. Although the precise nature of the interfaces and layers produced by the SILARbased method is still actively investigated,[27] previous optical spectroscopy and structural characterization with transmission electron microscopy (TEM) imaging of these materials[12, 14] pro-
1664
Ó 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Articles
Figure 1. Steady-state absorption and temporally integrated photoluminescence spectra (under pulsed excitation at 355 nm) of core-only and QDQW nanocrystal architectures comprised of CdSe potential wells and ZnS tunneling barriers. a) The energetic alignments of the conduction band (CB) and valence band (VB) of bulk CdSe and ZnS, illustrating the formation of potential wells for electrons and holes in the CdSe along the radial direction of the nanocrystal. b), c) and d) Time-integrated luminescence (red curves) and steady-state absorption (black curves) spectra for the core-only, C341 and C422 nanocrystal architectures, respectively. A transmission electron microscopy (TEM) micrograph of the core-only particles is inset in (b), and high-resolution TEM images of the corresponding QDQW nanocrystals are inset in panels (c) and (d). Schematics of the structural cross-section, conduction band alignment and anticipated exciton wave functions (following the effective mass approximation calculations of Nizamoglu et al.[10]) along the radial direction for the first two excited states n = 1 and n = 2 are shown for each nanocrystal architecture.
vides convincing evidence that this process enables tunable growth of multiple quantum wells on individual particles. From previous work based on effective mass approximation (EMA) calculations on these structures by Nizamoglu et al.,[10] the time-integrated luminescence spectra of these particles ChemPhysChem 2015, 16, 1663 – 1669
www.chemphyschem.org
can be rationalized in terms of exciton localization in the corresponding wells (defined by the band-offsets of the bulk materials) and quantified by the radial distribution of the product of the electron and hole wave functions (j fefh j). For the sake of consistency, we maintain the nomenclature conventions established by Nizamoglu et al.[10] In the absence of a second well (Figure 1 b) in the core-only particles, the excitons of both the first (1S exciton; n = 1) and second excited states (2S exciton; n = 2) are localized in the core, permitting rapid thermalization to the lowest-energy exciton state and consequently, a singlepeaked luminescence spectrum. In the C341 QDQW architecture which includes an additional tunneling barrier of 3 monolayers of ZnS (Figure 1 b), the EMA calculations can be extrapolated, predicting that the lowest-energy exciton state will be localized to the core while the next-higher energetic state will be primarily localized to the shell (inset, Figure 1 b).[10] The reduced wave function overlap between the n = 2 and n = 1 states in this configuration inhibits efficient thermalization between the two states,[13] leading to dual-color emitting nanocrystals: one color from the quantum-confined QD and one from the quantum-confined QW. Indeed, a subtle shoulder is observed in the higher-energy region of the main emission peak. This feature is at ~ 2.2 eV, the same energy that has previously been observed for a ~ 4 monolayer thick CdSe QW.[14] Increased quantum confinement and separation of the QW from the QD by ~ 4 monolayers of ZnS (C422 in Figure 1 d) further reduces the wave function overlap and thermalization rate between the two states, leading to significantly more high-energy emission from the QW of the nanocrystal. We note that both multi-layer particle systems (Figure 1 c and 1d) exhibit a slight red-shift of the QD emission peak with respect to the core-only configuration (Figure 1 a) and attribute this to increased delocalization of the electron and hole in the multilayer particles.[14] In Figure 2, time-resolved spectroscopy is used to separate the nanocrystal luminescence into “prompt” and “delayed” emission based on the corresponding delay after excitation. The prompt emission immediately follows laser excitation and includes relaxation processes that occur on shorter timescales than the experimental resolution, whereas the delayed emission includes the slower relaxation processes of long-lived excitations (discussed in more detail below). In contrast to the delayed (4–8 ns) emission, the prompt emission (0–4 ns) exhibits a subtle low-energy shoulder (slightly red-shifting the main exciton peak) and contains a distinct higher energy peak for the core-only nanocrystals in Figure 2 a. Qualitative comparison of the energetic positions of these features in the prompt emission spectrum here to those in the emission spectrum of similar CdSe/ZnS nanocrystals acquired within 1 ps of excitation[28] indicates that these features arise due to the radiative recombination of multiexciton (MX) states. Owing to the short lifetimes of these states (less than 100 ps),[29–31] neither feature is present in the delayed emission, which exhibits only a single peak and is nearly identical to the corresponding temporally integrated spectrum (Figure 1 b). In Figure 2 b, the higherenergy shoulder seen in the temporally integrated emission spectrum of C341 in Figure 1 c is much more pronounced in
1665
Ó 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Articles
Figure 2. Comparison between the prompt and delayed luminescence of core-only and QDQW CdSe/ZnS nanocrystals. a) The prompt emission from 0-4 ns after optical excitation (black curve) compared with the delayed emission from 4–8 ns (red curve) for the core-only particle (MX: core multiexciton). b), c) Similar comparison of prompt emission from 0–25 ns (black curves) to delayed emission from 25–50 ns (red curves) for the C341 and C422 QDQW nanocrystals, respectively. As depicted in the inset schematics, higher-energy excited states typically thermalize non-radiatively (solid black arrows) to lower-lying states as opposed to relaxing radiatively (sinusoidal arrows) to the ground state. This competition between relaxation rates usually results in emission from higher-energy states only appearing in prompt luminescence and necessitates inhibition of non-radiative relaxation processes for it to be observed in the delayed luminescence.
the prompt gated spectrum (0–25 ns, black curve) and completely absent in the corresponding delayed emission (25– 50 ns, orange curve). The relative quantum yield of the higherenergy state with respect to the lowest-energy exciton can be roughly estimated by the ratio of the respective intensities in ChemPhysChem 2015, 16, 1663 – 1669
www.chemphyschem.org
the fast (i.e. prompt) emission spectrum. In the C341 structure, the relative quantum yield of the higher-energy state is at least an order of magnitude larger than that of the MX state in the core-only structure. We exclude the possibility that this higher-energy shoulder in the C341 PL spectrum arises from MX states in either the QD or QW alone because of the following: 1) the relative quantum yield of the state is significantly higher than that of the conventional MX state in the core-only structure; 2) its spectral position agrees well with the exciton energy of a 4 monolayer thick QW;[14] and 3) its energetic position relative to the prominent, single-exciton peak does not coincide with that of the higher-energy multiexciton state that is observed in the core-only particle (Figure 2 a). Rather, we attribute this emission to quasi-isolated exciton states in the QD and QW where fast emission from the QW occurs before thermalization to the lower-lying QD exciton state: in the C341 architecture, inter-well thermalization between the QW and QD is only slowed and not completely suppressed. While the spectroscopic evidence is strong, unambiguous confirmation of this assertion ultimately requires that the intensity of this band is measured as a function of pump fluence to confirm that it scales linearly at low excitation intensities as would be expected for an isolated single exciton state of a quantum well.[28] However, such power-dependent experiments are highly susceptible to material degradation, in contrast to the transient spectroscopy at constant power used here. In stark contrast, the prompt (0–25 ns) and delayed (25–50 ns) emission spectra for the C422 QDQW nanocrystals with increased separation between the potential wells (Figure 2 c) are nearly identical. The long-lived nature of both of these states clearly differentiates them from multiexciton states of a conventional semiconductor nanocrystal that consists only of a single quantum well. Here, the QD and QW in the C422 structure can both be occupied independently by a single exciton, and the high degree of carrier localization in the QD and QW not only slows but completely inhibits the intra-particle, inter-well relaxation between the two potential wells, leading to efficient radiative recombination from both exciton states.[10, 13] On longer timescales that exceed the lifetime of primary quantum-confined direct excitons in semiconductor nanocrystals, the origin of the band-edge emission remains to be unambiguously established and has been attributed to the detrapping of carriers from dark electronic states into the bright excitonic state of the nanocrystal,[32–36] as well as direct radiative relaxation from shallow trap or surface states.[37, 38] Further ambiguity is found in the interpretation of the evolution of the emission spectrum on these timescales, where the observed spectral shifts could arise from a number of effects, including energy-dependent relaxation rates due to a phonon bottleneck,[39, 40] inter-particle energy transfer processes,[41] the quantum-confined Stark effect arising from a build-up of surface charges in the nanocrystals[42–45] or subtle photo-induced changes in the chemical state of the surfaces of the nanocrystals.[34] Although the precise origin of this delayed emission is unclear, these spectral dynamics likely reflect a complex interplay of the nanocrystals with their surrounding environment as well as their own surface and defect states. The influence of
1666
Ó 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Articles wave function localization and excited-state thermalization processes on these long-lived excitations is unknown and could provide further hints regarding the underlying physics of these states and their dynamics. Therefore, the QDQW nanocrystal architectures present an opportunity to explore how the different degrees of localization of the QD and QW excitons as well as the presence of a higher-energy metastable intraparticle state alter these relaxation phenomena. Figure 3 shows the spectral dynamics of the band-edge luminescence from 0 ns to 1300 ns (in 25 ns bins) after laser excitation for the core-only nanocrystals and the C341 QDQW system that exhibits partially inhibited thermalization from the QW to the QD. At delay times longer than ~ 10 ns, the majority of excitons and all of the multiexcitons that formed directly from photoexcitation have relaxed. The remaining emission arises from the long-lived excitations as discussed above. This long-lived emission, which can also be isolated accurately using time-resolved magnetic resonance techniques,[32, 46] is the
Figure 3. Spectral dynamics of the band-edge emission from long-lived excitations in the QD core of the core-only and C341 nanocrystals. a),b) Emission spectra as a function of delay time after optical excitation, plotted on a logarithmic false-color scale. The emission spectrum at each time-slice was fitted with a Gaussian to extract peak position and intensity values. The emission peak position is overlaid on top of the false-color images (white data points) and displayed at higher energy resolution in (c) and (d). e), f) Peak intensity, normalized to the initial value, for the core-only and C341 nanocrystals, respectively.
ChemPhysChem 2015, 16, 1663 – 1669
www.chemphyschem.org
focal point of these datasets. The top panels (Figure 3 a and Figure 3 b) report the time-resolved emission spectra of each system on a logarithmic false-color scale. The emission spectra for both systems exhibit two distinct features: the band-edge emission peak as well as a broadband lower energy feature that is attributed to direct radiative recombination of deeplevel traps in the nanocrystals.[47, 48] The spectral position of the band-edge emission peaks are overlaid on top of the temporal traces of the emission spectra and displayed in higher resolution in Figure 3 c and Figure 3 d. Evidently, the band-edge emission exhibits remarkably similar behavior in both systems: emission intensity decays with similar dynamics (Figure 3 e and Figure 3 f), and the emission peak shows a subtle shift to lower energies with increased delay. In terms of long-scale emission dynamics, the presence of the QW does not alter the spectral dynamics of the QD core from that of the core-only particle when thermalization is permitted from the higher-energy QW state to the QD. As shown in Figure 4, the emission dynamics of the longlived excitations in the core-only system are distinctly different to those of the QDQW nanocrystals when thermalization between the lowest-energy exciton states of the QD and QW is inhibited. Figure 4 a again displays the band-edge emission spectra as a function of delay time after laser excitation for the uncoupled QDQW nanocrystal system (C422). Two distinct emission peaks can be discerned in the trace of the luminescence spectrum over the entire delay range: the high-energy peak that is attributed to band-edge recombination in the CdSe QW and the lower-energy peak that is attributed to band-edge recombination in the CdSe QD. The spectral positions of these two peaks over the course of the delay time are overlaid on top of the emission spectra and are reported at a higher resolution in Figure 4 b. Unlike the monotonic decrease in emission energy exhibited by the core-only and coupled QDQW nanocrystal architectures in Figure 3, the QD emission here shows a much faster initial drop in emission energy (from 0–250 ns) followed by an increase (250–750 ns), and then seemingly stabilizes with increased delay (750–1300 ns). Surprisingly, the shell emission behaves in a distinctly different manner and continually increases in energy by ~ 20 meV from its initial value at 0 ns. Despite the dramatically different spectral dynamics, the QD and QW both exhibit similar decay dynamics of the emission intensity (Figure 4 c) as seen in the coupled QDQW and core-only nanocrystals. Strikingly, the higherenergy emission from the QW decays at a slower rate than the lower-energy emission of the QD, suggesting a complete absence of energy transfer processes (such as FRET) from the QW to the QD. Again, we note that the dynamics reported here occur on timescales that exceed the lifetimes of multiexciton states in individual QDs or QWs. Additional observations emerge from the spectral dynamics that are presented in Figures 3 and 4. The progression of the band-edge emission of the QD core in the QDQW nanocrystals mimics that of the core-only nanocrystals when a thermalization pathway exists between the QD and QW as in the case of the C341 nanocrystals in Figure 3. However, the scenario is very different when thermalization is inhibited between the
1667
Ó 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Articles only nanocrystals could arise due to several effects. One tempting explanation is that the distinct degrees of exciton localization in the QD and QW of the C422 nanocrystals probe the spatial dependence of the spectral dynamics of the longlived (presumably trapped) excitations, which are otherwise masked by efficient thermalization in the other systems. For example, it is plausible that the shell is more sensitive to surface-related mechanisms, effectively shielding the interior of the nanocrystal and leaving the core susceptible to effects that are internal to the nanocrystal. On the other hand, a Coulombic interaction between the charge carriers of the QD exciton state and the metastable QW exciton state could amplify the spectral dynamics in analogy to enhanced spectral diffusion in multichromophoric conjugated polymers, where the range of random spectral dynamics is increased when multiple chromophores couple to each other.[49] With further studies that perhaps implement an external bias[35] or magnetic resonance techniques,[32, 48] the well-defined exciton localization in these systems may provide a novel route to further elucidate the origins of these intriguing spectral dynamics.
3. Conclusions
Figure 4. Spectral dynamics of the band-edge emission from long-lived excitations in the QD and QW in the C422 nanocrystals. a) Gated emission spectra plotted on a logarithmic false-color scale with spectral peak positions (extracted by fitting the summation of two Gaussian peaks to each timeslice) of the QD (black data points) and QW (red data points) overlaid. The peak positions are reported at a higher resolution in (b). c) Decay of emission intensity in the QD and QW channels normalized to the initial intensity.
two wells in the C422 nanocrystals (Figure 4). First, the emission from the long-lived excited states in the QD and QW evolves in clearly distinct manners from that of the core-only nanocrystals. Second, the energetic ranges of the QD and QW dynamics (~ 20 meV in Figure 4 c) are over 2 Õ larger than those of the QD in the core-only and C341 nanocrystals (~ 10 meV in Figure 3 b). This marked contrast indicates that the spectral dynamics of the long-lived excited states of the QD core are significantly altered when a higher-energy metastable state exists in the QW. Moreover, the emission spectra of the QD and QW in C422 (Figure 4 c) do not evolve in a concerted manner. These pronounced differences in the spectral dynamics of the QD and QW states of C422 with respect to both each other as well as to the QD of the C341 and coreChemPhysChem 2015, 16, 1663 – 1669
www.chemphyschem.org
We have compared the spectral dynamics of long-lived excitations in QDQW nanocrystal architectures with and without intraparticle, interwell thermalization to those of a single-well, core-only system. Higher-energy relaxation is only observable in the long-lived emission when the wave function overlap between the excitons of the QD and QW on the individual nanoparticles is minimized, effectively eliminating the thermalization pathway between the two states. Interestingly, when this thermalization pathway persists in the nanocrystal system, the spectral dynamics—the gradual drifting of the emission spectrum—mimic those of the core-only particle. In contrast, when thermalization between the QW and QD is suppressed, the spectral dynamics of QW and QD emission are distinctly different. This effective blocking of excited-state thermalization, analogous to a breach of Kasha’s rule in molecular materials which has been studied widely,[50] is in agreement with prior reports that reversible shuttling of excitation energy between direct and indirect excitonic states by external electric fields can raise the excitonic lifetime by up to five orders of magnitude.[35, 36] These new results therefore demonstrate how engineering of exciton localization in QDQW nanocrystal architectures can be exploited to reveal subtle details about intraparticle interstate interactions, which are often overlooked in conventional steady-state spectroscopies. An intriguing challenge remains in identifying the precise chemical nature of the trap states responsible for delayed population of the direct exciton. Time-resolved optically-detected magnetic resonance techniques have recently been shown to offer, in principle, chemical fingerprinting of trap states[32, 48] based on delayed luminescence. The nanocrystals reported here, characterized by inhibited trap-state relaxation, are likely to be particularly suited to investigations involving this technique.
1668
Ó 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Articles Acknowledgements We thank Dr. Kipp van Schooten and Alex Thiessen for helpful discussions and technical assistance with the preparation of the figures. The authors are indebted to the Volkswagen Foundation for financial support through grant I/86 242. JML is a David & Lucile Packard Foundation Fellow. Keywords: nanostructures · quantum dots · quantum wells · semiconductors · time-resolved spectroscopy [1] D. V. Talapin, J.-S. Lee, M. V. Kovalenko, E. V. Shevchenko, Chem. Rev. 2010, 110, 389. [2] C. de Mello Doneg, Chem. Soc. Rev. 2011, 40, 1512 – 1546. [3] H. Zhu, T. Lian, Energy Environ. Sci. 2012, 5, 9406 – 9418. [4] E. A. Dias, S. L. Sewall, P. Kambhampati, J. Phys. Chem. C 2007, 111, 708 – 713. [5] A. J. Nozik, Annu. Rev. Phys. Chem. 2001, 52, 193 – 231. [6] U. Soni, A. Pal, S. Singh, M. Mittal, S. Yadev, R. Elangovan, S. Sapra, ACS Nano 2014, 8, 113 – 123. [7] E. Yoskovitz, G. Menagen, A. Sitt, E. Lachman, U. Banin, Nano Lett. 2010, 10, 3068 – 3072. [8] P. Tyagi, P. Kambhampati, J. Phys. Chem. C 2012, 116, 8154 – 8160. [9] M. S¸ahin, S. Nizamoglu, O. Yerli, H. V. Demir, J. Appl. Phys. 2012, 111, 023713. [10] S. Nizamoglu, H. V. Demir, Opt. Express 2008, 16, 3515 – 3526. [11] J. Berezovsky, O. Gywat, F. Meier, D. Battaglia, X. Peng, D. D. Awschalom, Nat. Phys. 2006, 2, 831 – 834. [12] S. Sapra, S. Mayilo, T. A. Klar, A. L. Rogach, J. Feldmann, Adv. Mater. 2007, 19, 569 – 572. [13] E. A. Dias, J. I. Saari, P. Tyagi, P. Kambhampati, J. Phys. Chem. C 2012, 116, 5407 – 5413. [14] D. Battaglia, B. Blackman, X. Peng, J. Am. Chem. Soc. 2005, 127, 10889 – 10897. [15] S. Chakrabortty, G. Xing, Y. Xu, S. W. Ngiam, N. Mishra, T. C. Sum, Y. Chan, Small 2011, 7, 2847 – 2852. [16] Z. Deutsch, O. Schwartz, R. Tenne, R. Popovitz-Biro, D. Oron, Nano Lett. 2012, 12, 2948 – 2952. [17] R. Tenne, A. Teitelboim, P. Rukenstein, M. Dyshel, T. Mokari, D. Oron, ACS Nano 2013, 7, 5084 – 5090. [18] A. Eychmìller, A. Mews, H. Weller, Chem. Phys. Lett. 1993, 208, 59 – 62. [19] A. Mews, A. Eychmìller, M. Giersig, D. Schooss, H. Weller, J. Phys. Chem. 1994, 98, 934 – 941. [20] D. Schooss, A. Mews, A. Eychmìller, H. Weller, Phys. Rev. B 1994, 49, 17072 – 17078. [21] A. Eychmìller, T. Voßmeyer, A. Mews, H. Weller, J. of Lumin. 1994, 58, 223 – 226. [22] A. Mews, A. Eychmìller, Ber. Bunsenges. Phys. Chem. 1998, 102, 1343 – 1357. [23] D. Dorfs, A. Eychmìller, Z. Phys. Chem. (Muenchen Ger.) 2006, 220, 1539 – 52.
ChemPhysChem 2015, 16, 1663 – 1669
www.chemphyschem.org
[24] C. Mauser, E. Da Como, J. Baldauf, A. L. Rogach, J. Huang, D. V. Talapin, J. Feldmann, Phys. Rev. B 2010, 82, 081306. [25] C. Galland, S. Brovelli, W. K. Bae, L. A. Padilha, F. Meinardi, V. I. Klimov, Nano Lett. 2013, 13, 321 – 328. [26] E. A. Dias, A. F. Grimes, D. S. English, P. Kambhampati, J. Phys. Chem. C 2008, 112, 14229 – 14232. [27] R. Tan, D. A. Blom, S. Ma, A. B. Greytak, Chem. Mater. 2013, 25, 3724 – 3736. [28] V. I. Klimov, Annu. Rev. Phys. Chem. 2007, 58, 635 – 673. [29] P. Kambhampati, Acc. Chem. Res. 2011, 44, 1 – 13. [30] P. Kambhampati, J. Phys. Chem. C 2011, 115, 22089 – 22109. [31] P. Kambhampati, J. Phys. Chem. Lett. 2012, 3, 1182 – 1190. [32] K. J. van Schooten, J. Huang, W. J. Baker, D. V. Talapin, C. Boehme, J. M. Lupton, Nano Lett. 2013, 13, 65 – 71. [33] M. Jones, S. S. Lo, G. D. Scholes, J. Phys. Chem. C 2009, 113, 18632 – 18642. [34] M. Jones, J. Nedeljkovic, R. J. Ellingson, A. J. Nozik, G. Rumbles, J. Phys. Chem. B 2003, 107, 11346 – 11352. [35] S. Liu, N. J. Borys, J. Huang, D. V. Talapin, J. M. Lupton, Phys. Rev. B 2012, 86, 045303. [36] R. M. Kraus, P. G. Lagoudakis, A. L. Rogach, D. V. Talapin, H. Weller, J. M. Lupton, J. Feldmann, Phys. Rev. Lett. 2007, 98, 017401. [37] A. Eychmìller, A. Hsselbarth, L. Katsikas, H. Weller, Ber. Bunsenges. Phys. Chem. 1991, 95, 79 – 84. [38] G. Morello, M. Anni, P. D. Cozzoli, L. Manna, R. Cingolani, M. De Giorgi, J. Phys. Chem. C 2007, 111, 10541 – 10545. [39] D. C. Hannah, N. J. Dunn, S. Ithurria, D. V. Talapin, L. X. Chen, M. Pelton, G. C. Schatz, R. D. Schaller, Phys. Rev. Lett. 2011, 107, 177403. [40] G. Rainý, I. Moreels, A. Hassinen, T. Stçferle, Z. Hens, R. F. Mahrt, Nano Lett. 2012, 12, 5224 – 5229. [41] A. Javier, D. Magana, T. Jennings, G. F. Strouse, Appl. Phys. Lett. 2003, 83, 1423 – 1425. [42] S. A. Empedocles, M. G. Bawendi, Science 1997, 278, 2114 – 2117. [43] R. G. Neuhauser, K. T. Shimizu, W. K. Woo, S. A. Empedocles, M. G. Bawendi, Phys. Rev. Lett. 2000, 85, 3301 – 3304. [44] J. Mìller, J. M. Lupton, A. L. Rogach, J. Feldmann, D. V. Talapin, H. Weller, Phys. Rev. Lett. 2004, 93, 167402. [45] J. Mìller, J. M. Lupton, A. L. Rogach, J. Feldmann, D. V. Talapin, H. Weller, Phys. Rev. B 2005, 72, 205339. [46] K. J. van Schooten, C. Boehme, J. M. Lupton, ChemPhysChem 2014, 15, 1737 – 1746. [47] S. Chandramohan, B. D. Ryu, H. K. Kim, C.-H. Hong, E.-K. Suh, Opt. Lett. 2011, 36, 802 – 804. [48] K. J. van Schooten, J. Huang, D. V. Talapin, C. Boehme, J. M. Lupton, Phys. Rev. B 2013, 87, 125412. [49] F. Schindler, J. M. Lupton, J. Feldmann, U. Scherf, Proc. Natl. Acad. Sci. USA 2004, 101, 14695 – 14700. [50] T. Itoh, Chem. Rev. 2012, 112, 4541 – 4568.
Received: February 23, 2015 Revised: January 30, 2015 Published online on March 24, 2015
1669
Ó 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Articles
Localization and Dynamics of Long-Lived Excitations in Colloidal Semiconductor Nanocrystals with Dual Quantum Confinement Su Liu,[a, b] Nicholas J. Borys,*[a, c] Sameer Sapra,[d] Alexander Eychmìller,[e] and John M. Lupton[a, f] Semiconductor nanocrystals consisting of a quantum dot (QD) core and a quantum well (QW) shell, where the QD and QW are separated by a tunneling barrier, offer a unique opportunity to engineer the photophysical properties of individual nanostructures. Using the thicknesses of the corresponding layers, the excitons of the first and second excited states can be separated spatially, localizing one state to the QD and the other to the QW. Thus the wave function overlap of the two states can be minimized, suppressing non-radiative thermalization between the two wells, which in turn leads to radiative relaxation from both states. The molecular analogy to such dual emission would be the inhibition of internal conversion, a special case that violates Kasha’s rule. Using nanosecond time-resolved spectroscopy of QDQW CdSe/ZnS onion-like nanocrystals, an intermediate regime of exciton separation and suppressed thermalization is identified where the non-radiative relaxation of the higher-energy state is slowed, but not completely inhib-
ited. In this intermediate thermalization regime, the temporal evolution of the delayed emission spectra resulting from trapped carriers mimic the dynamics of such states in nanocrystals that consist of only a QD core. In stark contrast, when a higher-energy metastable state exists in the QW shell due to strongly suppressed interwell thermalization, the spectral dynamics of the long-lived excitations in the QD and QW, which are spectrally distinct, are amplified and differ from each other as well as from those in the core-only nanocrystals. This difference in spectral dynamics demonstrates the utility of exploiting well-defined exciton localization to study the nature and spatial dependence of the intriguing photophysics of colloidal semiconductor nanocrystals, and illustrates the power of nanosecond gated luminescence spectroscopy in illuminating complex relaxation dynamics which are entirely masked in steadystate or ultrafast spectroscopy.
1. Introduction Colloidal semiconductor heterostructure nanocrystals unify the ease of wet-chemical-based synthesis with the quantum me[a] Dr. S. Liu, Dr. N. J. Borys, Prof. J. M. Lupton Department of Physics and Astronomy The University of Utah Salt Lake City, UT 84112 (USA) E-mail: [email protected] [b] Dr. S. Liu Department of Applied Physics Chalmers University of Technology Kemivagen 9 41296 Gothenburg (Sweden) [c] Dr. N. J. Borys Molecular Foundry Lawrence Berkeley National Laboratory Berkeley, CA 94720 (USA) [d] Prof. S. Sapra Department of Chemistry Indian Institute of Technology Delhi New Delhi, 110016 (India) [e] Prof. A. Eychmìller Physical Chemistry TU Dresden, 01062 Dresden (Germany) [f] Prof. J. M. Lupton Institut fìr Experimentelle und Angewandte Physik Universitt Regensburg, 93053 Regensburg (Germany)
ChemPhysChem 2015, 16, 1663 – 1669
chanical wave-function-engineering capabilities of traditional semiconductor heterostructures.[1–3] These single-particle architectures are commonly limited to variants of the core–shell geometry, a structural configuration where a spherical core of one material is surrounded by a shell of a second material. Through the selection of the constituent materials, significant control of the spatial localization of the wave functions of the electron and hole can be achieved, thereby enabling singleparticle systems that either maximize or minimize the carrier wave function overlap and hence the oscillator strength of the lowest-energy exciton state.[2, 3] Similarly, nanocrystal heterostructures that consist of three or more components and are more reminiscent of semiconductor superlattices[4–6] enable engineering of exciton localization by supporting the formation of two or more well-defined, spatially separated quantum wells (QWs) on individual nanocrystals.[7–11] The structural parameters include the size of the wells and barrier layer thickness that separates the wells. With these parameters, the quantum-confined energies of the wells and their electronic coupling can be tuned in a remarkably systematic manner. As such, these systems have attracted attention as low-cost platforms for broadband lighting,[12] optical gain media,[13] multicolor fluorescent probes,[14] and coupled qubit architectures.[11]
1663
Ó 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Articles Although more complex multi-well colloidal nanosystems are continually emerging,[15–17] the traditional workhorse configuration is an onion-like sphere combining quantum dot (QD) and quantum well (QW) structures on the same particle. The first reported colloidal QDQW system was comprised of a larger-bandgap QD material with an embedded QW of a different, smaller-bandgap material.[18, 19] In this structure, the coupling between the QD and QW of different materials alters the energetics, dynamics, and spatial distribution of the first excited state in the nanocrystal.[20–22] Alternatively, inverse QDQW structures can be engineered where the QD and QW are comprised of the same material: the QD core is encapsulated by a larger-bandgap material, which acts as a tunneling barrier, followed by a second layer of the QD material that forms a QW.[12, 14, 23] In this configuration, the radius of the QD and radial thickness of the QW control their respective quantumconfinement energies, whereas the thickness of the tunneling barrier controls the electronic coupling between the two potential wells.[8, 10, 13, 14] The coupling between the QD and QW critically depends on the overlap of the wave functions of the two states and includes mechanisms such as Dexter-type energy transfer. Therefore the structural parameters of such an “inverse” QDQW architecture can be tailored to produce nanocrystals with different non-radiative relaxation rates between the QD and QW.[10, 13] For example, a tunneling barrier with a thickness of ~ 1 monolayer permits efficient coupling between the QD and QW, enabling thermalization from the higher-energy exciton state of the QW (or QD) to the lowerenergy exciton state of the QD (or QW).[14] On the other hand, a thicker tunneling barrier (~ 3 monolayers) has been shown to electronically isolate the QD and QW, effectively blocking the thermalization pathway between the two potential wells.[14] Efficient radiative relaxation then occurs from the lowest-energy excitons of both the QD and QW, giving rise to dual-color emission from single nanoparticles (provided the lowest energy excitons in the QD and QW have different energies).[10, 12–14] Because the energy of the emission can be used to identify the region of exciton recombination (i.e., in the QD or QW), these dual-color QDQW architectures and related multi-well systems provide a unique opportunity to explore the effects of exciton localization (i.e., the spatial distribution of the exciton wave function) on the rich photophysics of colloidal nanocrystals such as carrier trapping and relaxation dynamics,[24] photon antibunching[16, 25] as well as single-particle blinking and spectral diffusion phenomena.[26] Here, we use ensemble-level gated fluorescence spectroscopy on onion-like CdSe/ZnS inverse QDQW nanocrystals to investigate intraparticle, interwell thermalization processes and the relaxation dynamics of long-lived excitations. Whereas the presence of two emission bands in time-integrated spectroscopy suggests that thermalization between the QD and QW is indeed suppressed, time-resolved spectroscopy reveals an intermediate regime of inhibited relaxation between the two potential wells. The degree of interwell thermalization suppression appears to have a dramatic effect on the energetic dynamics of long-lived excitations (i.e., excited states with lifetimes that exceed that of conventional band-edge excitons) as ChemPhysChem 2015, 16, 1663 – 1669
www.chemphyschem.org
observed through the evolution of their emission spectra with increasing delay after the initial excitation. When thermalization between the two wells is permitted, we find that the spectral dynamics of the luminescence are indistinguishable from a nanoparticle architecture that is comprised of only a surfacepassivated core. However, in the case when the QD and QW are isolated, we identify distinctly different spectral progressions of the two emission bands, revealing the potential to utilize exciton localization phenomena in QDQW systems to investigate the underlying physics of long-lived (trap-mediated) excitations in colloidal semiconductor nanocrystals.
2. Results and Discussion The QDQW and core-only CdSe/ZnS nanoparticles were synthesized following the successive ionic layer adsorption and reaction (SILAR)based process described by Peng et al.[14] and Sapra et al.[12] For the optical spectroscopy measurements, the nanocrystal samples were dispersed in hexane and measured under ambient conditions. The absorption spectra were recorded on a standard UV/Vis spectrophotometer. For the timeintegrated and time-resolved photoluminescence measurements, the samples were excited with a pulsed solid-state laser operating at 355 nm with ~ 60 mJ per pulse and a tunable repetition rate from 1–200 Hz. The emission was collected and dispersed with a spectrometer and recorded with a gated intensified charge-coupled device (ICCD) camera (Andor iStar) that was synchronized with the excitation pulses. Peak positions in the emission spectra were extracted by fitting either one (for spectra with a single peak) or the summation of two (for spectra with double peaks) Gaussian functions. Figure 1 illustrates the band alignment scheme of CdSe/ZnS QDQW nanocrystals, the three different nanoparticle architectures that were explored in this study and their corresponding steady-state absorption and time-integrated luminescence spectra. Due to the substantial offsets in the conduction and valance bands between the two materials (Figure 1 a), the ZnS serves as a tunneling barrier for both the excited state electrons and holes, thereby creating quantum wells in the CdSe regions of the nanocrystal.[14] Each nanocrystal system consists of a 3.5 nm CdSe core that is encapsulated by different configurations of subsequent layers of ZnS and CdSe. The core-only particles (Figure 1 b) are augmented only by a surface-passivating layer of ZnS that is ~ 2 monolayers thick. The QDQW structures in Figure 1 c and Figure 1 d are referenced by their structural configuration. In Figure 1 c, for example, the core (C) is progressively encapsulated by ~ 3 monolayers of ZnS, ~ 4 monolayers of CdSe, and finally ~ 1 monolayer of ZnS, and is therefore referred to as a “C341” architecture. Likewise, the QDQW architecture in Figure 1 d, “C422,” contains a CdSe core that is encapsulated with ~ 4 monolayers of ZnS, ~ 2 monolayers of CdSe and finally ~ 2 monolayers of ZnS. Although the precise nature of the interfaces and layers produced by the SILARbased method is still actively investigated,[27] previous optical spectroscopy and structural characterization with transmission electron microscopy (TEM) imaging of these materials[12, 14] pro-
1664
Ó 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Articles
Figure 1. Steady-state absorption and temporally integrated photoluminescence spectra (under pulsed excitation at 355 nm) of core-only and QDQW nanocrystal architectures comprised of CdSe potential wells and ZnS tunneling barriers. a) The energetic alignments of the conduction band (CB) and valence band (VB) of bulk CdSe and ZnS, illustrating the formation of potential wells for electrons and holes in the CdSe along the radial direction of the nanocrystal. b), c) and d) Time-integrated luminescence (red curves) and steady-state absorption (black curves) spectra for the core-only, C341 and C422 nanocrystal architectures, respectively. A transmission electron microscopy (TEM) micrograph of the core-only particles is inset in (b), and high-resolution TEM images of the corresponding QDQW nanocrystals are inset in panels (c) and (d). Schematics of the structural cross-section, conduction band alignment and anticipated exciton wave functions (following the effective mass approximation calculations of Nizamoglu et al.[10]) along the radial direction for the first two excited states n = 1 and n = 2 are shown for each nanocrystal architecture.
vides convincing evidence that this process enables tunable growth of multiple quantum wells on individual particles. From previous work based on effective mass approximation (EMA) calculations on these structures by Nizamoglu et al.,[10] the time-integrated luminescence spectra of these particles ChemPhysChem 2015, 16, 1663 – 1669
www.chemphyschem.org
can be rationalized in terms of exciton localization in the corresponding wells (defined by the band-offsets of the bulk materials) and quantified by the radial distribution of the product of the electron and hole wave functions (j fefh j). For the sake of consistency, we maintain the nomenclature conventions established by Nizamoglu et al.[10] In the absence of a second well (Figure 1 b) in the core-only particles, the excitons of both the first (1S exciton; n = 1) and second excited states (2S exciton; n = 2) are localized in the core, permitting rapid thermalization to the lowest-energy exciton state and consequently, a singlepeaked luminescence spectrum. In the C341 QDQW architecture which includes an additional tunneling barrier of 3 monolayers of ZnS (Figure 1 b), the EMA calculations can be extrapolated, predicting that the lowest-energy exciton state will be localized to the core while the next-higher energetic state will be primarily localized to the shell (inset, Figure 1 b).[10] The reduced wave function overlap between the n = 2 and n = 1 states in this configuration inhibits efficient thermalization between the two states,[13] leading to dual-color emitting nanocrystals: one color from the quantum-confined QD and one from the quantum-confined QW. Indeed, a subtle shoulder is observed in the higher-energy region of the main emission peak. This feature is at ~ 2.2 eV, the same energy that has previously been observed for a ~ 4 monolayer thick CdSe QW.[14] Increased quantum confinement and separation of the QW from the QD by ~ 4 monolayers of ZnS (C422 in Figure 1 d) further reduces the wave function overlap and thermalization rate between the two states, leading to significantly more high-energy emission from the QW of the nanocrystal. We note that both multi-layer particle systems (Figure 1 c and 1d) exhibit a slight red-shift of the QD emission peak with respect to the core-only configuration (Figure 1 a) and attribute this to increased delocalization of the electron and hole in the multilayer particles.[14] In Figure 2, time-resolved spectroscopy is used to separate the nanocrystal luminescence into “prompt” and “delayed” emission based on the corresponding delay after excitation. The prompt emission immediately follows laser excitation and includes relaxation processes that occur on shorter timescales than the experimental resolution, whereas the delayed emission includes the slower relaxation processes of long-lived excitations (discussed in more detail below). In contrast to the delayed (4–8 ns) emission, the prompt emission (0–4 ns) exhibits a subtle low-energy shoulder (slightly red-shifting the main exciton peak) and contains a distinct higher energy peak for the core-only nanocrystals in Figure 2 a. Qualitative comparison of the energetic positions of these features in the prompt emission spectrum here to those in the emission spectrum of similar CdSe/ZnS nanocrystals acquired within 1 ps of excitation[28] indicates that these features arise due to the radiative recombination of multiexciton (MX) states. Owing to the short lifetimes of these states (less than 100 ps),[29–31] neither feature is present in the delayed emission, which exhibits only a single peak and is nearly identical to the corresponding temporally integrated spectrum (Figure 1 b). In Figure 2 b, the higherenergy shoulder seen in the temporally integrated emission spectrum of C341 in Figure 1 c is much more pronounced in
1665
Ó 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Articles
Figure 2. Comparison between the prompt and delayed luminescence of core-only and QDQW CdSe/ZnS nanocrystals. a) The prompt emission from 0-4 ns after optical excitation (black curve) compared with the delayed emission from 4–8 ns (red curve) for the core-only particle (MX: core multiexciton). b), c) Similar comparison of prompt emission from 0–25 ns (black curves) to delayed emission from 25–50 ns (red curves) for the C341 and C422 QDQW nanocrystals, respectively. As depicted in the inset schematics, higher-energy excited states typically thermalize non-radiatively (solid black arrows) to lower-lying states as opposed to relaxing radiatively (sinusoidal arrows) to the ground state. This competition between relaxation rates usually results in emission from higher-energy states only appearing in prompt luminescence and necessitates inhibition of non-radiative relaxation processes for it to be observed in the delayed luminescence.
the prompt gated spectrum (0–25 ns, black curve) and completely absent in the corresponding delayed emission (25– 50 ns, orange curve). The relative quantum yield of the higherenergy state with respect to the lowest-energy exciton can be roughly estimated by the ratio of the respective intensities in ChemPhysChem 2015, 16, 1663 – 1669
www.chemphyschem.org
the fast (i.e. prompt) emission spectrum. In the C341 structure, the relative quantum yield of the higher-energy state is at least an order of magnitude larger than that of the MX state in the core-only structure. We exclude the possibility that this higher-energy shoulder in the C341 PL spectrum arises from MX states in either the QD or QW alone because of the following: 1) the relative quantum yield of the state is significantly higher than that of the conventional MX state in the core-only structure; 2) its spectral position agrees well with the exciton energy of a 4 monolayer thick QW;[14] and 3) its energetic position relative to the prominent, single-exciton peak does not coincide with that of the higher-energy multiexciton state that is observed in the core-only particle (Figure 2 a). Rather, we attribute this emission to quasi-isolated exciton states in the QD and QW where fast emission from the QW occurs before thermalization to the lower-lying QD exciton state: in the C341 architecture, inter-well thermalization between the QW and QD is only slowed and not completely suppressed. While the spectroscopic evidence is strong, unambiguous confirmation of this assertion ultimately requires that the intensity of this band is measured as a function of pump fluence to confirm that it scales linearly at low excitation intensities as would be expected for an isolated single exciton state of a quantum well.[28] However, such power-dependent experiments are highly susceptible to material degradation, in contrast to the transient spectroscopy at constant power used here. In stark contrast, the prompt (0–25 ns) and delayed (25–50 ns) emission spectra for the C422 QDQW nanocrystals with increased separation between the potential wells (Figure 2 c) are nearly identical. The long-lived nature of both of these states clearly differentiates them from multiexciton states of a conventional semiconductor nanocrystal that consists only of a single quantum well. Here, the QD and QW in the C422 structure can both be occupied independently by a single exciton, and the high degree of carrier localization in the QD and QW not only slows but completely inhibits the intra-particle, inter-well relaxation between the two potential wells, leading to efficient radiative recombination from both exciton states.[10, 13] On longer timescales that exceed the lifetime of primary quantum-confined direct excitons in semiconductor nanocrystals, the origin of the band-edge emission remains to be unambiguously established and has been attributed to the detrapping of carriers from dark electronic states into the bright excitonic state of the nanocrystal,[32–36] as well as direct radiative relaxation from shallow trap or surface states.[37, 38] Further ambiguity is found in the interpretation of the evolution of the emission spectrum on these timescales, where the observed spectral shifts could arise from a number of effects, including energy-dependent relaxation rates due to a phonon bottleneck,[39, 40] inter-particle energy transfer processes,[41] the quantum-confined Stark effect arising from a build-up of surface charges in the nanocrystals[42–45] or subtle photo-induced changes in the chemical state of the surfaces of the nanocrystals.[34] Although the precise origin of this delayed emission is unclear, these spectral dynamics likely reflect a complex interplay of the nanocrystals with their surrounding environment as well as their own surface and defect states. The influence of
1666
Ó 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Articles wave function localization and excited-state thermalization processes on these long-lived excitations is unknown and could provide further hints regarding the underlying physics of these states and their dynamics. Therefore, the QDQW nanocrystal architectures present an opportunity to explore how the different degrees of localization of the QD and QW excitons as well as the presence of a higher-energy metastable intraparticle state alter these relaxation phenomena. Figure 3 shows the spectral dynamics of the band-edge luminescence from 0 ns to 1300 ns (in 25 ns bins) after laser excitation for the core-only nanocrystals and the C341 QDQW system that exhibits partially inhibited thermalization from the QW to the QD. At delay times longer than ~ 10 ns, the majority of excitons and all of the multiexcitons that formed directly from photoexcitation have relaxed. The remaining emission arises from the long-lived excitations as discussed above. This long-lived emission, which can also be isolated accurately using time-resolved magnetic resonance techniques,[32, 46] is the
Figure 3. Spectral dynamics of the band-edge emission from long-lived excitations in the QD core of the core-only and C341 nanocrystals. a),b) Emission spectra as a function of delay time after optical excitation, plotted on a logarithmic false-color scale. The emission spectrum at each time-slice was fitted with a Gaussian to extract peak position and intensity values. The emission peak position is overlaid on top of the false-color images (white data points) and displayed at higher energy resolution in (c) and (d). e), f) Peak intensity, normalized to the initial value, for the core-only and C341 nanocrystals, respectively.
ChemPhysChem 2015, 16, 1663 – 1669
www.chemphyschem.org
focal point of these datasets. The top panels (Figure 3 a and Figure 3 b) report the time-resolved emission spectra of each system on a logarithmic false-color scale. The emission spectra for both systems exhibit two distinct features: the band-edge emission peak as well as a broadband lower energy feature that is attributed to direct radiative recombination of deeplevel traps in the nanocrystals.[47, 48] The spectral position of the band-edge emission peaks are overlaid on top of the temporal traces of the emission spectra and displayed in higher resolution in Figure 3 c and Figure 3 d. Evidently, the band-edge emission exhibits remarkably similar behavior in both systems: emission intensity decays with similar dynamics (Figure 3 e and Figure 3 f), and the emission peak shows a subtle shift to lower energies with increased delay. In terms of long-scale emission dynamics, the presence of the QW does not alter the spectral dynamics of the QD core from that of the core-only particle when thermalization is permitted from the higher-energy QW state to the QD. As shown in Figure 4, the emission dynamics of the longlived excitations in the core-only system are distinctly different to those of the QDQW nanocrystals when thermalization between the lowest-energy exciton states of the QD and QW is inhibited. Figure 4 a again displays the band-edge emission spectra as a function of delay time after laser excitation for the uncoupled QDQW nanocrystal system (C422). Two distinct emission peaks can be discerned in the trace of the luminescence spectrum over the entire delay range: the high-energy peak that is attributed to band-edge recombination in the CdSe QW and the lower-energy peak that is attributed to band-edge recombination in the CdSe QD. The spectral positions of these two peaks over the course of the delay time are overlaid on top of the emission spectra and are reported at a higher resolution in Figure 4 b. Unlike the monotonic decrease in emission energy exhibited by the core-only and coupled QDQW nanocrystal architectures in Figure 3, the QD emission here shows a much faster initial drop in emission energy (from 0–250 ns) followed by an increase (250–750 ns), and then seemingly stabilizes with increased delay (750–1300 ns). Surprisingly, the shell emission behaves in a distinctly different manner and continually increases in energy by ~ 20 meV from its initial value at 0 ns. Despite the dramatically different spectral dynamics, the QD and QW both exhibit similar decay dynamics of the emission intensity (Figure 4 c) as seen in the coupled QDQW and core-only nanocrystals. Strikingly, the higherenergy emission from the QW decays at a slower rate than the lower-energy emission of the QD, suggesting a complete absence of energy transfer processes (such as FRET) from the QW to the QD. Again, we note that the dynamics reported here occur on timescales that exceed the lifetimes of multiexciton states in individual QDs or QWs. Additional observations emerge from the spectral dynamics that are presented in Figures 3 and 4. The progression of the band-edge emission of the QD core in the QDQW nanocrystals mimics that of the core-only nanocrystals when a thermalization pathway exists between the QD and QW as in the case of the C341 nanocrystals in Figure 3. However, the scenario is very different when thermalization is inhibited between the
1667
Ó 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Articles only nanocrystals could arise due to several effects. One tempting explanation is that the distinct degrees of exciton localization in the QD and QW of the C422 nanocrystals probe the spatial dependence of the spectral dynamics of the longlived (presumably trapped) excitations, which are otherwise masked by efficient thermalization in the other systems. For example, it is plausible that the shell is more sensitive to surface-related mechanisms, effectively shielding the interior of the nanocrystal and leaving the core susceptible to effects that are internal to the nanocrystal. On the other hand, a Coulombic interaction between the charge carriers of the QD exciton state and the metastable QW exciton state could amplify the spectral dynamics in analogy to enhanced spectral diffusion in multichromophoric conjugated polymers, where the range of random spectral dynamics is increased when multiple chromophores couple to each other.[49] With further studies that perhaps implement an external bias[35] or magnetic resonance techniques,[32, 48] the well-defined exciton localization in these systems may provide a novel route to further elucidate the origins of these intriguing spectral dynamics.
3. Conclusions
Figure 4. Spectral dynamics of the band-edge emission from long-lived excitations in the QD and QW in the C422 nanocrystals. a) Gated emission spectra plotted on a logarithmic false-color scale with spectral peak positions (extracted by fitting the summation of two Gaussian peaks to each timeslice) of the QD (black data points) and QW (red data points) overlaid. The peak positions are reported at a higher resolution in (b). c) Decay of emission intensity in the QD and QW channels normalized to the initial intensity.
two wells in the C422 nanocrystals (Figure 4). First, the emission from the long-lived excited states in the QD and QW evolves in clearly distinct manners from that of the core-only nanocrystals. Second, the energetic ranges of the QD and QW dynamics (~ 20 meV in Figure 4 c) are over 2 Õ larger than those of the QD in the core-only and C341 nanocrystals (~ 10 meV in Figure 3 b). This marked contrast indicates that the spectral dynamics of the long-lived excited states of the QD core are significantly altered when a higher-energy metastable state exists in the QW. Moreover, the emission spectra of the QD and QW in C422 (Figure 4 c) do not evolve in a concerted manner. These pronounced differences in the spectral dynamics of the QD and QW states of C422 with respect to both each other as well as to the QD of the C341 and coreChemPhysChem 2015, 16, 1663 – 1669
www.chemphyschem.org
We have compared the spectral dynamics of long-lived excitations in QDQW nanocrystal architectures with and without intraparticle, interwell thermalization to those of a single-well, core-only system. Higher-energy relaxation is only observable in the long-lived emission when the wave function overlap between the excitons of the QD and QW on the individual nanoparticles is minimized, effectively eliminating the thermalization pathway between the two states. Interestingly, when this thermalization pathway persists in the nanocrystal system, the spectral dynamics—the gradual drifting of the emission spectrum—mimic those of the core-only particle. In contrast, when thermalization between the QW and QD is suppressed, the spectral dynamics of QW and QD emission are distinctly different. This effective blocking of excited-state thermalization, analogous to a breach of Kasha’s rule in molecular materials which has been studied widely,[50] is in agreement with prior reports that reversible shuttling of excitation energy between direct and indirect excitonic states by external electric fields can raise the excitonic lifetime by up to five orders of magnitude.[35, 36] These new results therefore demonstrate how engineering of exciton localization in QDQW nanocrystal architectures can be exploited to reveal subtle details about intraparticle interstate interactions, which are often overlooked in conventional steady-state spectroscopies. An intriguing challenge remains in identifying the precise chemical nature of the trap states responsible for delayed population of the direct exciton. Time-resolved optically-detected magnetic resonance techniques have recently been shown to offer, in principle, chemical fingerprinting of trap states[32, 48] based on delayed luminescence. The nanocrystals reported here, characterized by inhibited trap-state relaxation, are likely to be particularly suited to investigations involving this technique.
1668
Ó 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Articles Acknowledgements We thank Dr. Kipp van Schooten and Alex Thiessen for helpful discussions and technical assistance with the preparation of the figures. The authors are indebted to the Volkswagen Foundation for financial support through grant I/86 242. JML is a David & Lucile Packard Foundation Fellow. Keywords: nanostructures · quantum dots · quantum wells · semiconductors · time-resolved spectroscopy [1] D. V. Talapin, J.-S. Lee, M. V. Kovalenko, E. V. Shevchenko, Chem. Rev. 2010, 110, 389. [2] C. de Mello Doneg, Chem. Soc. Rev. 2011, 40, 1512 – 1546. [3] H. Zhu, T. Lian, Energy Environ. Sci. 2012, 5, 9406 – 9418. [4] E. A. Dias, S. L. Sewall, P. Kambhampati, J. Phys. Chem. C 2007, 111, 708 – 713. [5] A. J. Nozik, Annu. Rev. Phys. Chem. 2001, 52, 193 – 231. [6] U. Soni, A. Pal, S. Singh, M. Mittal, S. Yadev, R. Elangovan, S. Sapra, ACS Nano 2014, 8, 113 – 123. [7] E. Yoskovitz, G. Menagen, A. Sitt, E. Lachman, U. Banin, Nano Lett. 2010, 10, 3068 – 3072. [8] P. Tyagi, P. Kambhampati, J. Phys. Chem. C 2012, 116, 8154 – 8160. [9] M. S¸ahin, S. Nizamoglu, O. Yerli, H. V. Demir, J. Appl. Phys. 2012, 111, 023713. [10] S. Nizamoglu, H. V. Demir, Opt. Express 2008, 16, 3515 – 3526. [11] J. Berezovsky, O. Gywat, F. Meier, D. Battaglia, X. Peng, D. D. Awschalom, Nat. Phys. 2006, 2, 831 – 834. [12] S. Sapra, S. Mayilo, T. A. Klar, A. L. Rogach, J. Feldmann, Adv. Mater. 2007, 19, 569 – 572. [13] E. A. Dias, J. I. Saari, P. Tyagi, P. Kambhampati, J. Phys. Chem. C 2012, 116, 5407 – 5413. [14] D. Battaglia, B. Blackman, X. Peng, J. Am. Chem. Soc. 2005, 127, 10889 – 10897. [15] S. Chakrabortty, G. Xing, Y. Xu, S. W. Ngiam, N. Mishra, T. C. Sum, Y. Chan, Small 2011, 7, 2847 – 2852. [16] Z. Deutsch, O. Schwartz, R. Tenne, R. Popovitz-Biro, D. Oron, Nano Lett. 2012, 12, 2948 – 2952. [17] R. Tenne, A. Teitelboim, P. Rukenstein, M. Dyshel, T. Mokari, D. Oron, ACS Nano 2013, 7, 5084 – 5090. [18] A. Eychmìller, A. Mews, H. Weller, Chem. Phys. Lett. 1993, 208, 59 – 62. [19] A. Mews, A. Eychmìller, M. Giersig, D. Schooss, H. Weller, J. Phys. Chem. 1994, 98, 934 – 941. [20] D. Schooss, A. Mews, A. Eychmìller, H. Weller, Phys. Rev. B 1994, 49, 17072 – 17078. [21] A. Eychmìller, T. Voßmeyer, A. Mews, H. Weller, J. of Lumin. 1994, 58, 223 – 226. [22] A. Mews, A. Eychmìller, Ber. Bunsenges. Phys. Chem. 1998, 102, 1343 – 1357. [23] D. Dorfs, A. Eychmìller, Z. Phys. Chem. (Muenchen Ger.) 2006, 220, 1539 – 52.
ChemPhysChem 2015, 16, 1663 – 1669
www.chemphyschem.org
[24] C. Mauser, E. Da Como, J. Baldauf, A. L. Rogach, J. Huang, D. V. Talapin, J. Feldmann, Phys. Rev. B 2010, 82, 081306. [25] C. Galland, S. Brovelli, W. K. Bae, L. A. Padilha, F. Meinardi, V. I. Klimov, Nano Lett. 2013, 13, 321 – 328. [26] E. A. Dias, A. F. Grimes, D. S. English, P. Kambhampati, J. Phys. Chem. C 2008, 112, 14229 – 14232. [27] R. Tan, D. A. Blom, S. Ma, A. B. Greytak, Chem. Mater. 2013, 25, 3724 – 3736. [28] V. I. Klimov, Annu. Rev. Phys. Chem. 2007, 58, 635 – 673. [29] P. Kambhampati, Acc. Chem. Res. 2011, 44, 1 – 13. [30] P. Kambhampati, J. Phys. Chem. C 2011, 115, 22089 – 22109. [31] P. Kambhampati, J. Phys. Chem. Lett. 2012, 3, 1182 – 1190. [32] K. J. van Schooten, J. Huang, W. J. Baker, D. V. Talapin, C. Boehme, J. M. Lupton, Nano Lett. 2013, 13, 65 – 71. [33] M. Jones, S. S. Lo, G. D. Scholes, J. Phys. Chem. C 2009, 113, 18632 – 18642. [34] M. Jones, J. Nedeljkovic, R. J. Ellingson, A. J. Nozik, G. Rumbles, J. Phys. Chem. B 2003, 107, 11346 – 11352. [35] S. Liu, N. J. Borys, J. Huang, D. V. Talapin, J. M. Lupton, Phys. Rev. B 2012, 86, 045303. [36] R. M. Kraus, P. G. Lagoudakis, A. L. Rogach, D. V. Talapin, H. Weller, J. M. Lupton, J. Feldmann, Phys. Rev. Lett. 2007, 98, 017401. [37] A. Eychmìller, A. Hsselbarth, L. Katsikas, H. Weller, Ber. Bunsenges. Phys. Chem. 1991, 95, 79 – 84. [38] G. Morello, M. Anni, P. D. Cozzoli, L. Manna, R. Cingolani, M. De Giorgi, J. Phys. Chem. C 2007, 111, 10541 – 10545. [39] D. C. Hannah, N. J. Dunn, S. Ithurria, D. V. Talapin, L. X. Chen, M. Pelton, G. C. Schatz, R. D. Schaller, Phys. Rev. Lett. 2011, 107, 177403. [40] G. Rainý, I. Moreels, A. Hassinen, T. Stçferle, Z. Hens, R. F. Mahrt, Nano Lett. 2012, 12, 5224 – 5229. [41] A. Javier, D. Magana, T. Jennings, G. F. Strouse, Appl. Phys. Lett. 2003, 83, 1423 – 1425. [42] S. A. Empedocles, M. G. Bawendi, Science 1997, 278, 2114 – 2117. [43] R. G. Neuhauser, K. T. Shimizu, W. K. Woo, S. A. Empedocles, M. G. Bawendi, Phys. Rev. Lett. 2000, 85, 3301 – 3304. [44] J. Mìller, J. M. Lupton, A. L. Rogach, J. Feldmann, D. V. Talapin, H. Weller, Phys. Rev. Lett. 2004, 93, 167402. [45] J. Mìller, J. M. Lupton, A. L. Rogach, J. Feldmann, D. V. Talapin, H. Weller, Phys. Rev. B 2005, 72, 205339. [46] K. J. van Schooten, C. Boehme, J. M. Lupton, ChemPhysChem 2014, 15, 1737 – 1746. [47] S. Chandramohan, B. D. Ryu, H. K. Kim, C.-H. Hong, E.-K. Suh, Opt. Lett. 2011, 36, 802 – 804. [48] K. J. van Schooten, J. Huang, D. V. Talapin, C. Boehme, J. M. Lupton, Phys. Rev. B 2013, 87, 125412. [49] F. Schindler, J. M. Lupton, J. Feldmann, U. Scherf, Proc. Natl. Acad. Sci. USA 2004, 101, 14695 – 14700. [50] T. Itoh, Chem. Rev. 2012, 112, 4541 – 4568.
Received: February 23, 2015 Revised: January 30, 2015 Published online on March 24, 2015
1669
Ó 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
