Letter pubs.acs.org/NanoLett
Looking Inside a Working SiLED Florian Maier-Flaig,*,† Christian Kübel,*,‡ Julia Rinck,§ Tobias Bocksrocker,† Torsten Scherer,‡ Robby Prang,‡ Annie K. Powell,§,‡ Geoffrey A. Ozin,*,∥ and U. Lemmer*,† †
Light Technology Institute (LTI) and DFG Center for Functional Nanostructures (CFN), Karlsruhe Institute of Technology (KIT), Kaiserstraße 12, 76131 Karlsruhe, Germany ‡ Institute of Nanotechnology (INT) and Karlsruhe Nano Micro Facilty (KNMF), Karlsruhe Institute of Technology (KIT), Hermann-von-Helmholtz Platz 1, 76344 Eggenstein-Leopoldshafen, Germany § Institute of Inorganic Chemistry, Karlsruhe Institute of Technology, Engesserstraße 15, 76131 Karlsruhe, Germany and DFG Center for Functional Nanostructures (CFN), Karlsruhe Institute of Technology, Kaiserstraße 12, 76131 Karlsruhe, Germany ∥ Department of Chemistry, University of Toronto, 80 St. George Street, Toronto, ON M5S 3H6, Canada, and DFG Center for Functional Nanostructures (CFN), Karlsruhe Institute of Technology, Kaiserstraße 12, 76131 Karlsruhe, Germany S Supporting Information *
ABSTRACT: In this study, we investigate for the first time morphological and compositional changes of silicon quantum dot (SiQD) light-emitting diodes (SiLEDs) upon device operation. By means of advanced transmission electron microscopy (TEM) analysis including energy filtered TEM (EFTEM) and energy dispersive X-ray (EDX) spectroscopy, we observe drastic morphological changes and degradation for SiLEDs operated under high applied voltage ultimately leading to device failure. However, SiLEDs built from size-separated SiQDs operating under normal conditions show no morphological and compositional changes and the biexponential loss in electroluminescence seems to be correlated to chemical and physical degradation of the SiQDs. By contrast, we found that, for SiLEDs fabricated from polydisperse SiQDs, device degradation is more pronounced with three main modes of failure contributing to the reduced overall lifetime compared to those prepared from size-separated SiQDs. With this newfound knowledge, it is possible to devise ways to increase the lifetimes of SiLEDs. KEYWORDS: Silicon nanocrystals, morphology, degradation, hybrid quantum dot organic light emitting diode, nanoparticle diffusion and migration, percolative transport
Q
limited applicability due to their toxicity to humans and the environment6,7 and legal constraints on use.8 In contrast, silicon quantum dots (SiQDs) seem to be ideally suited for optoelectronics and biomedical applications, based on the nontoxic character and earth abundance of silicon. Recent literature reports indicate that silicon nanoparticles are either nontoxic or the least toxic of all known nanomaterials.9−11 Even though a lot of work has been published on QDLEDs,2−4,12−15 including the emerging field of silicon quantum dot based light emitting diodes (SiLEDs),16−20 an in-depth study of the morphological and compositional properties of these devices before and after device operation is lacking in the literature. Such kinds of study are especially important since lifetimes of QD-LEDs are still in their infancy2,20 and the lack of degradation studies in the literature is still most apparent.21
uantum dot light emitting diodes (QD-LEDs) offer great scientific, technological, and economic potential. Compared to their purely organic counterparts, which have already reached industrial production and the consumer market, the organic emissive layer of these devices is replaced by solution-processable functionalized quantum dots (QDs). QDs and QD-LEDs feature a wide spectral tunability and high color purity.1−4 Furthermore, as the efficiency of these devices is not intrinsically limited by spin statistics as in the case of fluorescent organic LEDs, they have the potential to reach much higher efficiencies. This, together with recently reported all-inorganic QD-LEDs potentially without the need for encapsulation,5 shows the great potential of this novel generation of light sources in terms of efficiency and stability of the devices. QD-LEDs based on II−VI semiconductor compounds have been intensively investigated reporting a wide spectral tunability as well as high external quantum efficiency (EQE) values.1−4 However, the compounds used for II−VI QD-LEDs such as CdS, CdSe, and their Pb containing counterparts have © XXXX American Chemical Society
Received: March 15, 2013 Revised: May 29, 2013
A
dx.doi.org/10.1021/nl400975u | Nano Lett. XXXX, XXX, XXX−XXX
Nano Letters
Letter
A detailed understanding of the structural and compositional changes during device operation is necessary to appreciate the mode of degradation and thus enable significant improvements of QD-LED performance and device lifetime to be realized in practice. Device failure of purely organic LEDs is often related to migration or diffusion of elements forming nonconductive barriers, which lead to higher internal resistance and, as a consequence, destroy the device.22 Gao et al.23 report on current-induced oxidation of the aluminum cathode leading to device defects. Cumpston24,25 stated that localized degradation of the cathode leads to formation of dark spots and degradation of the device. In addition, device failure can also be caused by a damaged ITO anode as reported by Gautier.26 In a detailed study on composite multilayer devices made of a polymer and CdTe nanoparticles Gallardo et al.22 showed that aluminum diffusion could also explain device degradation. Despite these various reports, no information exists in the literature on how morphological and compositional changes in working QDLEDs affect the operation of the device. In this communication, we describe for the first time the results of an in-depth morphological and compositional study of SiLEDs fabricated using colloidally stable SiQDs. A comparison between as-fabricated and electrically driven SiLEDs as well as SiLEDs prepared using monodisperse and polydisperse SiQDs has been carried out to correlate morphological and compositional features, from nanoscopic to macroscopic length scales, with the degradation behavior and device lifetimes. The resulting knowledge on variations of SiLED morphology and composition under working conditions should be easily transferrable to other QD-based LEDs as well as OLEDs and even QD-solar cells. After a first brief description of the experimental details and the intrinsic properties of the investigated SiLEDs in their initial pristine state, we will focus on the following three main questions: (1) Does the morphology change upon device operation? Could migration or some kind of morphological changes inside the SiLED explain the fast initial decay of the electroluminescence (EL) observed in long-term measurements? (2) What are the consequences of high versus low operation voltage on the nanoscopic and microscopic structure of the SiLEDs? (3) Does the morphology change when polydisperse SiQD samples are used, and does this account for the reduced lifetimes of these devices compared to the longer lifetimes of devices fabricated using monodisperse SiQD samples? Experimental Section. The SiLEDs were built using allylbenzene capped SiQDs synthesized following the method detailed in ref 27. For device fabrication, we used freshly prepared size-separated particles and compared their device performance with a set of samples of not size-separated (polydisperse) nanoparticles (see also ref 20). The device architecture of the investigated SiLEDs is schematically shown in Figure 1. A thin layer of PEDOT:PSS (∼15 nm) was spincoated on top of a structured ITO-covered glass substrate followed by a ∼10 nm thick layer of poly-TPD. The silicon nanoparticles were processed by spin-coating from toluene and form a ∼35 nm thick emissive layer. The hole blocking layer TPBi (∼35 nm) and the LiF/Al (0.8 nm/200 nm) cathode were both deposited by thermal evaporation. All fabrication steps were done in a nitrogen purged glovebox under inert conditions. Each SiLED consists of four different active luminescent areas (LA) of 25 mm2, which can be individually
Figure 1. Schematic representation of the SiLED multilayer stack. The silicon quantum dots are sandwiched between the liquid processed PEDOT:PSS/poly-TPD anode and the thermally evaporated TPBi/ LiF/Al cathode. The investigated cross-sections were cut by FIB as shown by the lamella on the right-hand side.
addressed. A detailed description the device fabrication is given in ref 20. The morphological and compositional studies were carried out by transmission electron microscopy (TEM) using an image corrected Titan 80−300 (FEI Company) operated at an acceleration voltage of 300 kV and equipped with an S-UTW EDX detector (EDAX) and a Tridiem 863 energy filter (Gatan). Energy filtered and element specific images (3window technique) were obtained using a US1000 slow-scan charge-coupled device (CCD) camera mounted at the back of the Tridiem. Initial imaging in BF-TEM and HAADF-STEM mode was performed under strict low-dose conditions with a total dose of ∼100 e/nm2 for the first image. However, as no morphology changes were visible over an extended image series, the final images and elemental maps were acquired close to standard imaging conditions without noticeable structural changes compared to the initial low-dose images. TEM cross-section samples of the SiLEDs were prepared by in situ lift-out in a focused ion beam (FIB) system (FEI Strata 400 S). The initial cutting was performed using 30 kV Ga+ ions with final polishing at 5 and 2 kV. SEM imaging at 1 kV was limited to a minimum to exclude beam damage of the SiLED during preparation. The photoluminescence (PL) properties of the SiQD film shown in Figure 4 were studied using an actively Q-switched Nd:YVO4 laser (AOT-YVO-20QSP, AOT Ltd.) emitting short pulses of ca. 0.5 ns length at a wavelength of 355 nm. PL spectra were acquired using a cooled intensified CCD camera (PiMax512, Princeton Instruments) coupled to a spectrometer (SpectraPro-300i, Aceton Research Corporation). Results and Discussion. Six different SiLED samples were investigated in this study. Besides a reference sample of a pristine LA, which has not been operated (SiLED A0), two SiLEDs (SiLED A1−1/A1−2, see Figure 4) driven at 8 mA/ cm2 for 14 min and 4 h, respectively, were studied. For comparison, we investigated an additional sample A2, which was driven at high voltage of about 15 V (current ∼150 mA/ cm2) for around 90 min. In order to comment on the reduced lifetime of polydisperse (not size-separated) silicon nanoparticle emitters (see discussion in ref 20), we investigated polydisperse SiLEDs in the pristine state (SiLED B0) and driven at 6 mA/cm2 for 14 min (SiLED B1). The SiLEDs with polydisperse nanoparticles were fabricated using the same experimental conditions as for the other SiLEDs studied here. B
dx.doi.org/10.1021/nl400975u | Nano Lett. XXXX, XXX, XXX−XXX
Nano Letters
Letter
Figure 2. Zero-loss filtered BF-TEM image of a cross-section of the pristine SiLED stack of sample A0 with low-loss EFTEM images acquired at different energies with a 5 eV energy slit width and elemental maps for silicon, carbon, oxygen, and nitrogen (note: the oxygen map does not show the oxygen content of the ITO layer due to the In−M edge). The scale bars are 50 nm for all images.
Figure 3. Cross-section of SiLED A1−2 operated for several hours. The HAADF-STEM image (left) shows the well-defined layer structure as well as individual SiQDs in the SiQD layer. In the EDX maps (right) of the area indicated in the STEM image, the individual layers can be clearly distinguished based on their elemental composition. The scale bars are 20 nm for all images.
However, no size-separation of the SiQDs was carried out; only some large (not colloidally stable) particles were removed by a simple washing step with ethanol/methanol. A cross-section of the pristine reference device of SiLED A0 is shown in Figure 2. Before operation, the SiLED features a very well-defined layer structure with extremely smooth interfaces of the organic layers and the SiQD layer. Besides the strong material contrast of the QD-layer in BF-TEM, the organic materials can also be distinguished by low-loss EFTEM imaging and elemental mapping. The layer sequence corresponds well to the stack shown schematically in Figure 1 with nitrogen visible in the TPBi layer and silicon (with some oxygen) present in the SiQD layer. In addition, we observe a thin silicon rich layer between PEDOT and poly-TPD, which can be attributed to a monolayer of SiQDs (diameter ≈ 2.5 nm). This SiQD monolayer also contains significant amounts of oxygen as visible in the oxygen EFTEM and the STEM-EDX maps (Figures 3 and S1, Supporting Information). This layer is
present in all samples and seems to originate from intermixing of poly-TPD and SiQDs during processing of the SiLED; in a SiQD-free reference system, which was otherwise processed identically, this layer is not present (data not shown). Furthermore, the oxygen map shows a thin oxidized layer at the interface between the TPBi and the aluminum cathode, which is presumably due to slight oxidation during preparation. Figure 3 shows a HAADF-STEM image of SiLED A1−2 driven at 8 mA/cm2 for 4 h. Even though we operated this SiLED for several hours, the layer structure is still very welldefined, and no individual nanoparticles outside the SiQD layer are observed, whereas individual SiQDs with a diameter of around 2.5 nm are visible in the SiQD layer. The individual materials and layers can again be clearly attributed to their constituting elements as shown in the STEM-EDX images of Figure 3: Carbon is present throughout the organic multilayer system, sulfur is dominant in the PEDOT:PSS layer and partially also diffused into the poly-TPD layer (similar to the C
dx.doi.org/10.1021/nl400975u | Nano Lett. XXXX, XXX, XXX−XXX
Nano Letters
Letter
Figure 4. (a) EL-decay of SiLED A1 driven at 8 mA/cm2. The inset shows the PL-decay of a SiQD film under pulsed laser excitation in vacuum (p = 10−6 mbar; λexc = 355 nm; pulse duration: 0.5 ns). The numbers correspond to the nomenclature of the SiLEDs described in the text. (b) EL-spectra at different points in time of the decay. The inset shows the J−V characteristics of the device. The spectra are cut at wavelengths above 800 nm due to experimental limitations.
Figure 5. Cross-section of SiLED A2 operated under high current. The HAADF-STEM image shows SiQDs in the TPBi layer. A schematic representation of the diffusion process is shown on the left-hand side.
migration of nanoparticles toward the cathode during operation as well as intermixing effects during preparation of the SiLED as some of these high density areas were also found in the pristine SiLED A0. However, the statistical information on the distribution of these high-density areas is too limited to clearly separate intermixing and diffusion contributing to their formation. Overall, we conclude that under normal voltage/current conditions no significant migration of large particles or longrange disorder of the organic layers occurs even after operation for several hours. Only some small density fluctuations inside the TPBi layer are present at this stage and might act as seed for subsequent device failure. However, considering that the electroluminescence intensity of SiLED A1−2 was already reduced to ∼20% of the initial intensity, we have to assume that this fast initial decay of the electroluminescence (Figure 4) cannot be attributed to morphological changes but is mainly attributed to chemical processes (creation of defects and dark particles). This conclusion is also supported comparing the typical decay characteristics (Figure 4) of the EL of SiLED A1 with the photoluminescence decay of a thin SiQD film (inset Figure 4). In both cases, we observe a fast initial decay on a time scale of only a few minutes, followed by a second slower
pristine SiLED A0 shown in Figure S1, Supporting Information), and nitrogen is detected in the TPBi layer (the STEM-EDX signal is very weak, but nitrogen is clearly visible in the EFTEM maps in Figure S2, Supporting Information). The thin silicon and oxygen-rich nanoparticle monolayer between PEDOT and poly-TPD mentioned before is also clearly apparent in the STEM-EDX maps of SiLED A1−2 as is silicon, oxygen, and a very small amount of fluorine in the SiQD layer itself. Finally, the nominally 0.8 nm LiF deposited with the Al cathode results in a clear fluorine signal visible at the interface between TPBi and Al, which coincides with the oxygen signal already mentioned for the pristine SiLED A0. A detailed inspection of the HAADF-STEM and STEM-EDX images shows that besides the well conserved layer structure, some high and low density areas are present in the TPBi layer (see Figure S3, Supporting Information). The low-density areas are probably due to small voids present in the TPBi layer, whereas the high density areas can be linked to an increased silicon and oxygen signal in the STEM-EDX maps and a reduced carbon and nitrogen signal in EFTEM. This suggests that the high density areas are due to very small (oxidized) silicon nanoparticles, which are not resolved individually in the STEM images. These structures are probably due to electroD
dx.doi.org/10.1021/nl400975u | Nano Lett. XXXX, XXX, XXX−XXX
Nano Letters
Letter
Figure 6. SiLED A2 operated at high current densities. (a) HAADF-STEM overview image of hot spot defect. The Al-cathode is partially/entirely peeled off (left/right part of image). (b) Scanning electron microscopy overview image of a large defect area with several hot spots (position of the TEM cross-section for detailed analysis marked in white; scale bar, 5 μm). (c) Cross-section of area indicated in topmost image. Very close to defect, the organic layers are still intact (scale bar, 50 nm). (d) Organic and Al layers are partially peeled off. Glass and ITO (bright layer) have melted together (scale bar, 100 nm).
Figure 7. Cross-section of SiLED B0 prepared from polydisperse SiQDs. Examples for preferential electrical current paths along larger particles are indicated as white lines (right image).
PEDOT:PSS layers seem not to be affected by strong interdiffusion and even the SiQD monolayer between them is well preserved. Given the fact that the SiQD layer features the lowest electrical conductivity of the materials used,28 most of the voltage drops over this specific layer resulting in the high electromigration of the SiQDs, presumably with counter diffusion of the small molecule TPBi. This leads to the observed drastic morphological changes of these two layers as a first step toward the catastrophic failure of the SiLEDs. In addition to the migration of SiQDs visible everywhere, the device driven at high current also suffers from macroscopic defects as shown in Figure 6. Theses dramatic defects are due to current paths that already form at early stages of operation. Along these paths, electrically induced heat cannot be dissipated quickly enough, accumulates, and locally destroys the LED. As a consequence, prominent local hot-spots appear where the aluminum cathode gets partially exfoliated, and even melting of the ITO and underlying glass substrate is observed (Figure 6a,b,d). Nevertheless, even close to this fatal device failure, the organic layers are well-defined with only the mixing of TPBi and SiQDs present (Figure 6c). In addition, compared
decay with time constants of several hours suggesting that the decrease in EL-intensity is not due to the organic parts of the LED but rather related to the SiQDs. Obviously, the EL decays much more than the PL. We attribute this behavior to the harsh conditions for electrical excitation under which chemical changes are induced by the high electrical field and high carrier densities. This might lead to more pronounced degradation as compared to purely optical excitation. In both, EL and PL experiments, we assume that a specific amount of poorly functionalized or already defective nanocrystals gets switched off after only a short period of time. This bleaching is most probably due to defects at the nanocrystals’ surface leading to dark particles. The remaining brightly emitting SiQDs explain the leveling off of the signal in both cases. Under operation in the high current/voltage regime things change significantly. As shown in Figure 5, the initially welldefined layer structure is strongly disordered: SiQDs as well as TPBi suffer from strong migration and layer intermixing. As detailed in Figure 5, the individual nanoparticles migrate toward the cathode and completely mix with the upper-lying hole blocking layer. Interestingly, the poly-TPD and the E
dx.doi.org/10.1021/nl400975u | Nano Lett. XXXX, XXX, XXX−XXX
Nano Letters
Letter
Figure 8. Cross-section of SiLED B0 prepared from polydisperse SiQDs. HAADF-STEM overview image (left) showing a large number of high density areas in the TPBi layer. STEM-EDX maps of the indicated area show a high silicon and oxygen concentration for the high-density area. The scale bars are 10 nm.
3 and 4), here the small particles were not removed by sizeseparation and are still part of the solution used for processing. Hence, already during SiLED processing and probably also at low applied electric fields, these particles diffuse into the TPBi layer and create the observed local inhomogenieties. These material fluctuations might contribute to a percolative charge transport in the device and finally lead to shorter device operation lifetimes In general, compared to SiLEDs based on size-separated particles, processing of polydisperse SiQD films of high quality is rather difficult. Macroscopic defects (even after filtering of the solution) and large thickness variations are present in all polydisperse SiLEDs (not shown here). These inhomogeneities, in addition to the large nanoparticles included in the emissive layer as well as the density fluctuation of the TPBi, are presumably responsible for the reduced device lifetimes. In summary, three major differences between mono- and polydisperse SiLEDs have been observed that potentially explain the shorter device lifetimes and the accelerated degradation of polydisperse SiLEDs: (1) short-circuits caused by large particles inside the SiQD layer (Figure 7), (2) mixing/ diffusion of very small SiQDs into the TPBi layer (Figure 8), and (3) layer inhomogeneities during processing of the polydisperse film. All three factors cause or at least facilitate percolative transport through the entire LED stack, which in turn generates hot spots leading to significantly reduced lifetimes of the polydisperse devices. Conclusions. In summary, the results of a detailed morphological and compositional study of silicon based light emitting diodes (SiLEDs) have been presented for the first time. Before operation, the pristine SiLEDs feature a very welldefined layer structure with extremely smooth interfaces of both the organic layers and the silicon quantum dot (SiQD) layer. This well-defined structure is preserved even after an extended period of time under normal operation conditions. Together with the similarity of the fast EL decay of the SiLEDs and the PL decay of a SiQD film, this suggests that the fast initial EL-decay of the devices is due to chemical processes at the atomic level in the SiQD layer rather than due to morphological changes. In contrast, under high applied electric
to SiLED A1 the overall EL behavior does not change at high applied voltage. Only a small spectral shift of the EL-signal, which we attribute to the remaining polydispersity of the SiQDsample,20 can be observed (Figure S4, Supporting Information). As described by Maier-Flaig et al. earlier,20 overall device operation lifetimes significantly increase when size-separated SiQDs are used. In contrast, SiLEDs based on polydisperse SiQDs show a significant number of defects, which initially manifest themselves as bright spots in the otherwise homogeneous LA. These spots serve as seeds for subsequent failure leading to short device lifetimes. To understand how polydisperse SiQDs affect the SiLED performance, we investigated the morphology of SiLEDs prepared from polydisperse SiQDs in the as-prepared state as well as after brief operation. Figure.7 shows the cross-section of SiLED B0. Even though no large particles stand out of the SiQD layer, the polydispersivity of the sample is obvious. Besides homogeneous-sized particles, as in the case of SiLED A0 and A1, plenty of large particles are visible throughout the emissive layer. These particles might form percolation paths for the current flowing through the device. As discussed in context with Figure 6, these paths potentially serve as seeds for short-circuits, which lead to local device failure and hence reduced device lifetimes.20 For illustration, three potential current paths are indicated by white lines in Figure 7. However, except for one case, we never observed any large SiQDs outside the main SiQD layer in any of our cross-sections. This suggests that the larger SiQDs are not causing percolation paths in any other layer than the emissive SiQD layer. Furthermore, a large number of high density areas is present in the TPBi layer in all investigated polydisperse SiLEDs. STEM-EDX analysis (Figure 8) shows similar features as seen before for the high density areas in the size-separated samples. However, for the nonsize-separated SiLEDs, the silicon and oxygen content of these high-density areas is very obvious, and we attribute the high density areas to significant amounts of very small nanoparticles, which intermix with the TPBi layer. Compared to the size-separated samples shown before (Figures F
dx.doi.org/10.1021/nl400975u | Nano Lett. XXXX, XXX, XXX−XXX
Nano Letters
Letter
fields, we observe drastic morphological changes, including large macroscopic defects. Strong electromigration of nanoparticles and significant intermixing of the SiQDs and the hole blocking layer is obvious and leads to fast device degradation. A detailed study of SiLEDs fabricated using polydisperse SiQDs revealed three major aspects that explain the shorter operation lifetimes compared to their size-separated (monodisperse) counterparts: short circuits due to larger particles inside the SiQD layer, interlayer diffusion of very small particles, and inhomogenieties of the overall SiQD layer. We therefore conclude that in QD-LEDs size-separated nanoparticles are to be recommended in order to increase the lifetime of the devices. Furthermore, as interlayer diffusion of very small SiQDs is one of the limiting steps, cross-linking or the use of polymeric hole blocking layers in addition to even higher precision size-separation might further increase device lifetimes. Finally, new classes of electrically active surface functional ligands could significantly boost electrical conductivity of the SiQD layer. Lower applied fields would be needed for operation and hence enhanced device stability could be achieved.
■
(6) Derfus, A. M.; Chan, W. C. W.; Bhatia, S. N. Nano Lett. 2003, 4, 11−18. (7) Hoshino, A.; Hanada, S.; Yamamoto. Arch. Toxicol. 2011, 1−14. (8) Anc, M. J.; Pickett, N. L.; Gresty, N. C.; Harris, J. A.; Mishra, K. C. ECS J. Solid State Sci. Technol. 2013, 2, R3071−R3082. (9) Wang, Q.; Bao, Y.; Zhang, X.; Coxon, P. R.; Jayasooriya, U. A.; Chao, Y. Adv. Healthcare Mater. 2012, 1, 189−198. (10) Bimbo, L. M.; Sarparanta, M.; Santos, H. A.; Airaksinen, A. J.; Mäkilä, E.; Laaksonen, T.; Peltonen, L.; Lehto, V.-P.; Hirvonen, J.; Salonen, J. ACS Nano 2010, 4, 3023−3032. (11) Alsharif, N. H.; Berger, C. E. M.; Varanasi, S. S.; Chao, Y.; Horrocks, B. R.; Datta, H. K. Small 2009, 5, 221−228. (12) Colvin, V. L.; Schlamp, M. C.; Alivisatos, A. P. Nature 1994, 370, 354−357. (13) Tessler, N.; Medvedev, V.; Kazes, M.; Kan, S.; Banin, U. Science 2002, 295, 1506−1508. (14) Caruge, J. M.; Halpert, J. E.; Wood, V.; Bulovic, V.; Bawendi, M. G. Nat. Photonics 2008, 2, 247−250. (15) Cho, K.-S.; Lee, E. K.; Joo, W.-J.; Jang, E.; Kim, T.-H.; Lee, S. J.; Kwon, S.-J.; Han, J. Y.; Kim, B.-K.; Choi, B. L.; Kim, J. M. Nat. Photonics 2009, 3, 341−345. (16) Cheng, K.-Y.; Anthony, R.; Kortshagen, U. R.; Holmes, R. J. Nano Lett. 2010, 10, 1154−1157. (17) Puzzo, D. P.; Henderson, E. J.; Helander, M. G.; Wang, Z.; Ozin, G. A.; Lu, Z. Nano Lett. 2011, 11, 1585−1590. (18) Cheng, K.-Y.; Anthony, R.; Kortshagen, U. R.; Holmes, R. J. Nano Lett. 2011, 11, 1952−1956. (19) Anthony, R.; Cheng, K.-Y.; Holman, Z. C.; Holmes, R. J.; Kortshagen, U. R. Nano Lett. 2012, 12, 2822−2825. (20) Maier-Flaig, F.; Rinck, J.; Stephan, M.; Bocksrocker, T.; Bruns, M.; Kübel, C.; Powell, A. K.; Ozin, G. A; Lemmer, U. Nano Lett. 2013, 13, 475−480. (21) Lim, J.; Bae, W. K.; Kwak, J.; Lee, S.; Lee, C.; Char, K. Opt. Mater. Express 2012, 2, 594−628. (22) Gallardo, D.; Bertoni, C.; Dunn, S.; Gaponik, N.; Eychmüller, A. Adv. Mater. 2007, 19, 3364−3367. (23) Gao, M.; Richter, B.; Kirstein, S.; Möwald, H. J. Phys. Chem. B 1998, 102, 4096−4103. (24) Cumpston, B. H.; Jensen, K. F. Appl. Phys. Lett. 1996, 69, 3941− 3943. (25) Cumpston, B. H.; Parker, I. D.; Jensen, K. F. J. Appl. Phys. 1997, 81, 3716−3720. (26) Gautier, E.; Lorin, A.; Nunzi, J.-M.; Schalchli, A.; Benattar, J.-J.; Vital, D. Appl. Phys. Lett. 1996, 69, 1071−1073. (27) Mastronardi, M. L.; Maier-Flaig, F.; Faulkner, D.; Henderson, E. J.; Kübel, C.; Lemmer, U.; Ozin, G. A. Nano Lett. 2011, 12, 337−342. (28) Mastronardi, M. L.; Henderson, E. J.; Puzzo, D. P.; Chang, Y.; Wang, Z. B.; Helander, M. G.; Jeong, J.; Kherani, N. P.; Lu, Z.; Ozin, G. A. Small 2012, 8, 3647−3654.
ASSOCIATED CONTENT
S Supporting Information *
Additional TEM, EFTEM, and STEM-EDX results complementing Figures 2 and 3. This material is available free of charge via the Internet at http://pubs.acs.org.
■
AUTHOR INFORMATION
Corresponding Author
*E-mail: florian.maier-fl
[email protected] (F.M.-F.); christian.
[email protected] (C.K.);
[email protected] (G.A.O.);
[email protected] (U.L.). Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS G.A.O. is Government of Canada Research Chair in materials chemistry and nanochemistry. He thanks the Natural sciences and Engineering Council of Canada and the Ministry of Science, Research, and the Arts of Baden−Württemberg for strong and sustained support of this work at UoT and KIT. F.M.-F. acknowledges generous support by the Karlsruhe School of Optics & Photonics (KSOP). Part of this work was sponsored by the Karlsruhe Nano Micro Facility (KNMF), a large scale infrastructure facility operated at Karlsruhe Institute of Technology (KIT). We thank Moritz Stephan at KIT for the SiLED preparation.
■
REFERENCES
(1) Sun, Q.; Wang, Y. A.; Li, L. S.; Wang, D.; Zhu, T.; Xu, J.; Yang, C.; Li, Y. Nat. Photonics 2007, 1, 717−722. (2) Kwak, J.; Bae, W. K.; Lee, D.; Park, I.; Lim, J.; Park, M.; Cho, H.; Woo, H.; Yoon, D. Y.; Char, K.; Lee, S.; Lee, C. Nano Lett. 2012, 12, 2362−2366. (3) Bourvon, H.; Le Calvez, S.; Kanaan, H.; Meunier-Della-Gatta, S.; Philippot, C.; Reiss, P. Adv. Mater. 2012, 24, 4414−4418. (4) Sun, L.; Choi, J. J.; Stachnik, D.; Bartnik, A. C.; Hyun, B.-R.; Malliaras, G. G.; Hanrath, T.; Wise, F. W. Nat. Nanotechnol. 2012, 7, 369−373. (5) Wood, V.; Panzer, M.; Caruge, J.-M.; Halpert, J. E.; Bawendi, M. G.; Bulovic, V. Nano Lett. 2009, 10, 24−29. G
dx.doi.org/10.1021/nl400975u | Nano Lett. XXXX, XXX, XXX−XXX
