Photoluminescence of CdSe/ZnS core/shell quantum dots enhanced by energy transfer from a phosphorescent donor

We demonstrate exciton energy transfer from a thin film of phosphorescent dye fac tris(2-phenylpyridine) iridium (Ir(ppy)3) to a monolayer of colloidal CdSe/ZnS core/shell quantum dots (QDs). The energy transfer is manifested in time-resolved photoluminescence (PL) measurements as elongation of the QD PL time constant from 40 to 400 ns, and a concomitant 55% increase of time-integrated QD PL intensity. The observed PL dynamics are shown to be dominated by exciton diffusion within the Ir(ppy)3 film to the QD layer. 2006 Elsevier B.V. All rights reserved. Over the past several years the optical and electronic properties of colloidaly synthesized nanocrystals [1], or quantum dots (QDs), of CdSe have been extensively studied with the aim of using QD films in solid state opto-electronic devices. Efficient exciton generation in CdSe QDs suggests use of nanocrystal composite films in photovoltaic cells [2], while high luminescence quantum yields and tunability of QD emission wavelengths over the entire visible spectrum suggests QD film use in light emitting devices (LEDs) [3]. These developments are a consequence of advances in colloidal QD synthesis that allow for increased control over the shape, size, and emission wavelength of nanocrystals [4], and the development of methods for forming QD thin films of controlled structure and composition [5]. Utilizing these advances, in the present Letter we fabricate hybrid organic thin film/QD structures which demonstrate triplet exciton energy transfer (ET) from a thin film of phosphorescent molecules to a monolayer of CdSe/ZnS core/shell QDs. Triplet exciton harvesting and transfer to an efficient lumo0009-2614/$ see front matter 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.cplett.2006.04.009 * Corresponding author. E-mail address: [email protected] (P.O. Anikeeva). phore has been previously used in advancing organic light emitting device (OLED) technology [6], and has the potential to similarly benefit the emerging field of quantum-dotLEDs (QD-LEDs) [7]. OLEDs based on phosphorescent materials exhibit high quantum efficiencies [8]. For example, organic phosphors containing d Ir complexes, such as the fac tris(2-phenylpyridine) iridium (Ir(ppy)3) that is used in this Letter (see inset of Fig. 1) [9,10], show record efficient electro-phosphorescence at room temperature with external quantum efficiencies as high as 19% [11]. In these compounds spin– orbit coupling leads to the mixing of the spin-singlet and spin-triplet excited states [10,12], which enables the radiative relaxation of the triplet-state and leads to a fast phosphorescent decay (<1 ls) and high phosphorescence efficiency that benefits OLED performance. In QDs, the presence of transition metal atoms, such as Cd, leads to electron–hole exchange interaction and spin– orbit coupling [13] that mix the electron and hole spin states. In CdSe QDs electron spin-mixing results in a non-emissive lowest energy exciton, so called ‘dark exciton’ [14], which is between 0.13 meV (as in CdSe bulk) and 12.5 meV (for the smallest QD with few nm diameter) Fig. 2. Time-integrated PL spectra of samples I, II and III. All measurements were obtained at the same excitation source power of k = 395 nm light. The PL spectrum of sample III can be constructed from a linear superposition of the PL spectra of samples I and II. Fig. 1. Overlap between CdSe/ZnS QD absorption (solid line) and Ir(ppy)3 emission (dash-dot-dot) spectra suggests energy transfer from Ir(ppy)3 to QDs. (Note. QD absorption spectrum was obtained by the direct measurement in a thin film, consequently it exhibits a red tail due to the scattering of the organic ligands in a solid film.) Inset: Schematic drawing of a ZnS overcoated CdSe QD and Ir(ppy)3 structural formula. P.O. Anikeeva et al. / Chemical Physics Letters 424 (2006) 120–125 121 below the emissive excitonic state [15]. Thermal mixing of the dark and emissive exciton results in a room temperature QD radiative lifetime in the range from 3 to 30 ns (depending on the QD core/shell structure, size distribution, and the fidelity of the organic capping layer), and luminescence efficiencies exceeding 80% in solution [16,17]. Since QDs are efficient lumophores and Ir(ppy)3 is an efficient triplet exciton harvester, in this Letter we consider energy transfer (ET) from Ir(ppy)3 to CdSe/ZnS core/shell QDs in order to enhance the luminescence intensity of QD lumophores. Overlap of Ir(ppy)3 luminescence and QD absorption spectra suggests the possibility of efficient ET (Fig. 1). Our Letter follows earlier experiments that investigated Förster ET to CdSe/ZnS QDs from both fluorescent organic hosts as well as from inorganic substrate layers [18,19]. QDs have proven themselves as efficient exciton donors in energy transfer experiments with various organic dyes and bioorganic molecules [20,21]; however, there remains debate in the literature over the demonstration of ET from an organic donor to CdSe/ZnS core/shell QD (see e.g. [22]). In contrast, the present Letter definitively confirms that CdSe/ZnS core/shell QDs can efficiently accept excitons from an organic donor by demonstrating ET of triplet excitons from a phosphorescent dye to QD lumophores. We fabricated three thin film structures: sample I is a 40 nm thick film of 10% Ir(ppy)3 doped into 4,4 0-N,N 0-dicarbazole-byphenyl (CBP) thermally evaporated onto a glass substrate. Sample II is a monolayer of CdSe/ZnS QDs (7 nm QD diameter) printed [23] onto a glass substrate. Finally, sample III is a hybrid structure consisting of a monolayer of CdSe/ZnS QDs printed onto a 40 nm thick film of 10% Ir(ppy)3 in CBP on glass. Comparing the PL signatures of the three samples we observe a 21 ± 4% decrease of Ir(ppy)3 time-integrated PL intensity in sample III as compared to sample I and a concomitant 55 ± 5% increase in CdSe/ZnS QD film PL intensity in sample III as compared to sample II (see Fig. 2). The change in PL is calculated by numerically decomposing the sample III spectrum into CdSe/ZnS QD and Ir(ppy)3 components. The PL change suggests ET from the Ir(ppy)3 film to the QD monolayer. We note that simple reabsorption of Ir(ppy)3 luminescence by the QD film does not account for the observed PL change since the 7 nm thick QD monolayer has very weak absorption (<1.5%) over the Ir(ppy)3 PL spectrum. Assuming a QD PL efficiency on the order of 0.1 (typical of QD films), we find that reabsorption of Ir(ppy)3 photons by the QD layer can lead to small QD PL flux increase of at most 0.0015 times the Ir(ppy)3 photon flux, or a roughly three orders of magnitude smaller than the observed QD intensity. (To provide an upper limit on the reabsorption effect, we assume all of the Ir(ppy)3 flux is directed through the QD film.) Consequently, reabsorption does not significantly contribute to the observed increase in QD PL in sample III. (Note that because the film thicknesses are much less than the wavelengths of the emitted light, and the refractive index contrasts between the layers are small, optical cavity effects are not expected to be significant.) Data from time-resolved PL measurements are shown in Fig. 3. The PL of CdSe/ZnS QDs in sample II (data set E) exhibits two time constants with a shorter time constant of s 1 1⁄4 10 ns and a longer time constant of s QD 2 1⁄4 40 ns. The Ir(ppy)3 PL decay also exhibits bi-exponential behavior, with a dominant time constant of sIrðppyÞ3 1⁄4 610 ns (as obtained from data set A). In sample III, however, the QD PL decay (data set D) is substantially elongated, leading to a longer time constant of s 2;sIII 1⁄4 500 ns, which is Fig. 3. Time resolved PL measurements for samples I, II and III, performed over (a) a 5000 ns time window and (b) a 500 ns time window (first 200 ns shown). The black lines and dots represent the experimental measurements, and the thick grey lines represent numerical fits using the proposed diffusion model. To obtain data sets A and B, the sample PL was integrated over the wavelength range of k = 511 nm to k = 568 nm, to yield in each case the time dependence of the Ir(ppy)3 PL. Similarly, to obtain data sets C and E, a wavelength range of k = 600 nm to k = 656 nm was used. Data set C therefore reflects the intensity of combined Ir(ppy)3/QD PL near the QD PL peak. Data set E reflects the intensity of solely the QD PL. To obtain data set D, the intensity due to the Ir(ppy)3 PL was subtracted from C to yield just the QD PL intensity in sample III. Note that the grey fit lines assume a single exponential time decay for the Ir(ppy)3, and so are only expected to fit the Ir(ppy)3 at early times (where the single exponential decay dominates). 122 P.O. Anikeeva et al. / Chemical Physics Letters 424 (2006) 120–125 identical to the dominant time constant of the Ir(ppy)3 PL from the same sample (data set B), strongly suggesting that this ‘delayed’ QD PL is due to energy transfer of Ir(ppy)3 excitons to the QD film. An investigation of the first 200 ns of QD PL (Fig. 3b) reveals a slight increase in the initial PL intensity and a small increase of the short time constant, yielding s 1;sIII 1⁄4 12 ns. Note that in Fig. 3, the data are obtained by integrating the PL spectra over the wavelength ranges specified in the figure caption. Furthermore, the QD PL decay for sample III (data set D) is obtained by subtracting the Ir(ppy)3 PL spectrum from the total signal, as illustrated in Fig. 4, and then integrating the difference signal over the specified wavelength range. We again note that the time-dependent contribution of reabsorption to the Fig. 4. (a) PL spectra of sample III are shown at times t = 0 ns (dash-dot-dot) integrated PL spectrum (dotted line) is shown for comparison (not to scale). ( from the PL of sample III is shown at t = 0 ns (dash-dot-dot), t = 500 ns (das QD PL is at most 0.0015 times the Ir(ppy)3 flux intensity (obtained from sample I). Therefore, for all times reabsorption contribution is at most 3% of the observed QD PL signal. Since reabsorption does not contribute significantly to the QD PL intensity in sample III, the observed QD PL enhancement and elongation of QD PL lifetime can be attributed to exciton energy transfer from Ir(ppy)3 molecules to the QD monolayer. For quantitative analysis of the data, we note that the observed PL time dependence (Figs. 3 and 4) is shaped by four physical processes that govern exciton dynamics in the Ir(ppy)3:CBP film: Ir(ppy)3 exciton radiative decay, non-radiative decay, ET to the QD film, and diffusion. The two decay mechanisms combine to determine the observed radiative lifetime of 610 ns and a , t = 500 ns (dashed line), t = 1500 ns (solid line) after excitation. Ir(ppy)3 b) QD PL in sample III obtained by subtracting scaled Ir(ppy)3 spectrum hed), t = 1500 ns (solid) after excitation. P.O. Anikeeva et al. / Chemical Physics Letters 424 (2006) 120–125 123 PL quantum efficiency of 15% (as calculated from optimized electroluminescence efficiencies of 12% doped Ir(ppy):CBP OLEDs [24]). The ET mechanism leads to the observed quenching of the Ir(ppy)3 PL and the associated enhancement of the QD PL, and capturing the Ir(ppy)3 excitons that are closest to the QD film. Finally, the diffusion mechanism induces a net flow of Ir(ppy)3 excitons towards the QD film, due to depletion by ET of Ir(ppy)3 excitons near the QD interface. To numerically model the combined processes we model the ET mechanism as an instant energy transfer of any Ir(ppy)3 exciton that is within distance, LET, from the QD film. In this model the dynamics of the ET process is controlled entirely by Ir(ppy)3 exciton diffusion, which determines the rate at which excitons are supplied to the region within LET of the QD film. A reasonable value for LET is determined by considering the possible ET mechanisms. In the case of ET by correlated electron exchange, a mechanism also known as Dexter transfer [25], the exciton capture region consists of a single Ir(ppy)3 layer adjacent to QDs, since the transfer rate falls off exponentially with distance, reducing to negligible the ET contributions from all the more distant molecular layers. Since the Ir(ppy)3 molecule is on the order of 1 nm in extent, then we estimate that LET = 1 nm for Dexter energy transfer. In the case of resonant ET, a mechanism also known as Förster transfer, the capture region can be larger. For Förster transfer, the rate of ET between a donor (D) and an acceptor (A), is given by [26]