NANO LETTERS
Imaging of Resonant Quenching of Surface Plasmons by Quantum Dots
2006 Vol. 6, No. 12 2833-2837
Manuel J. Romero,* Jao van de Lagemaat, Ivan Mora-Sero,† Garry Rumbles, and Mowafak M. Al-Jassim National Renewable Energy Laboratory, 1617 Cole BouleVard, Golden, Colorado 80401-3393 Received August 24, 2006; Revised Manuscript Received October 20, 2006
ABSTRACT In scanning tunneling microscopy (STM), confinement of surface plasmons to the optical cavity formed at the metallic tunneling gap stimulates the emission of light. We demonstrate that quantum dots (QDs) found in such a cavity give rise to discrete, observable transitions in the tunneling luminescence spectrum due to the resonant extinction of the plasmon. The observed resonances represent a fingerprint of the QD and occur at the optical band gap owing to the nearly simultaneous transfer of carriers from both sides of the tunneling gap to the QD. The resonant quenching of surface plasmons enables a new imaging technique, dubbed plasmon resonance imaging, with a spatial resolution potentially similar to that of STM and the energy resolution of optical spectroscopies. This detection and imaging strategy is not restricted to QDs, being of great interest to an entire spectrum of nanostructures, from molecular assemblies and biomolecules to carbon nanotubes.
Recent developments in the fabrication and manipulation of nanoscale systems have made possible a broad range of emerging applications in (opto)electronics, biotechnology, energy conversion, and a multitude of other research fields. Easily synthesizable, quantum-confined semiconductors with well-controlled properties have made possible a new class of excitonic solar cells1 and electroluminescent diodes of record efficiency.2 The recent discovery of multiple-exciton generation in certain classes of nanoscale semiconductors3-7 could be exploited to overcome the Shockley-Queisser limit for solar-energy-conversion efficiency.8 For such a vision to be realized, a better understanding and control of electrons and excitons confined to quantum dimensions, the energy flow between nanoscale entities, and collective behavior emerging on larger scales are needed. A good example of confinement-induced phenomena is found in the field of plasmonics.9 Plasmonics interrogates the properties of surface plasmons (SPs)scollective oscillations of free electrons in conductorssand their use in active photonic elements such as optical guides, filters, and amplifiers. Confinement of SPs to the surface of a noble metal nanoparticle has a clear effect on its optical resonance, which, next to particle size, also depends on the density of free electrons.10 The strong coupling of plasmons with incoming photons is responsible for the electric-field enhancement on the surface of noble metal nanoparticles. This approach to light concentration is used in surface-enhanced Raman spectroscopy, where a 3 order-of-magnitude increase in * Corresponding author. E-mail:
[email protected]. † Current address: Departament de Cie ` ncies Experimentals, Universitat Jaume I, Av. de Vicent Sos Baynat, 12071 Castello´ de la Plana, Spain. 10.1021/nl061997s CCC: $33.50 Published on Web 11/14/2006
© 2006 American Chemical Society
detection efficiency is commonplace.11,12 This enhancement effect could also be used to increase the efficiency of certain types of solar cells and light-emitting diodes.13,14 In scanning tunneling microscopy (STM), SPs can be excited by the inelastic scattering of hot tunneling electrons. The frequency distribution of the plasmon is determined by the optical modes sustained by the cavity formed at the tunneling gap. SPs annihilate nonradiatively, transferring their energy to secondary processes such as phonon excitation, or radiatively, exciting luminescencescommonly referred to as scanning tunneling luminescence (STL). The latter is significantly enhanced by the confinement of plasmons in the optical cavity, and consequently, the STM is a perfect environment to investigate these phenomena. This combination of ultrahigh resolution possible with STM and the energetic resolution of luminescence spectroscopy was used to show that different parts of a single porphyrin molecule emit at different wavelengths,15 and to induce luminescence from individual C60 molecules on gold surfaces which then provides a light source of only 4 Å in size.16 It is therefore conceivable that the radiative decay of plasmons can be used as a powerful nanoscale light source for optical spectroscopy and imaging. The electromagnetic field associated with the plasmon provides strong coupling with the objectssuch as a molecule, quantum dot (QD), or similar entitysplaced in the cavity. In analogy to Raman scattering, the molecular absorption is possibly enhanced significantly by the electromagnetic field. In this paper, we demonstrate that (CdSe)ZnS QDs of core-shell structure can be optically detected and mapped
Figure 1. Schematic of the STM developed for plasmonic resonance imaging.
using SPs. QDs give rise to well-defined transitions in the tunneling luminescence spectrum due to the resonant extinction of the radiative decay of plasmons, the position of which can be correlated with their respective fluorescence. The physics of the resonances is also discussed. Atomically flat gold substrates were prepared by annealing vapor-deposited gold in a hydrogen flame. These substrates were covered with a self-assembled monolayer of 1,6hexanedithiol by immersion in a 5 mM solution in methanol for several hours. Submonolayer self-assembled QD structures for STM observation were prepared by immersion of the flame-annealed and as-deposited gold substrates in a diluted (about 10 mg/mL) toluene solution of (CdSe)ZnS QDs of core-shell structure (Evident Technologies). The coverage is controlled by the immersion time. The samples are labeled by the fluorescence emission (nominal) maximum of their respective solutions in nanometers. Figure 1 shows the PRI experimental setup, which is similar to that described by Egusa et al. and others.17-19 We employed a customized STM based on a multiaxis nanopositioning platform integrated inside a scanning electron microscope. In this configuration, the ultrasharp tip (10-nm diameter, nominal) is accessible to observation in the electron microscope and positioned in the focal point of the parabolic mirror for maximum collection of the luminescence. A CCD camera (Roper Scientific Silicon EEV 1340 × 400) is used to detect the STL emission and can be set to acquire either the emission pattern or the emission spectrum. Interference filters can be used to distinguish different processes involved in the emission of photons or to increase the signal-to-noise ratio. With this setup, a sensitivity of 106 cps/nA in the detection of the light emission by the plasmon in bare gold substrates was achieved. The current setup can also distinguish between light generated at the apex of the tip or more diffusely around the tip by studying the image projected on the CCD camera when no spectrometer is inserted between the mirror and the CCD camera. Contrary to scanning tunneling spectroscopy (STS), where the feedback is temporarily disabled and the tunneling current is acquired while scanning the bias applied to the tip, in the scanning tunneling luminescence spectroscopy (STLS) measurements reported here, the photon intensity is recorded while scanning the bias applied to the tip and simultaneously maintaining a constant tunneling current. This approach will make the tip height vary during the experiment as the feedback loop adjusts for 2834
Figure 2. Tunneling luminescence spectroscopy for a series of self-assembled colloidal (CdSe)ZnS quantum dots of core-shell structure on flame-annealed gold substrates, labeled by the nominal fluorescence emission of their respective solutions (in nanometers). The plasmon emission spectrum for the bare gold is shown for comparison (top). Resonances in the extinction of the plasmoninduced emission are indicated by vertical dash lines. The PL spectrum for each type of dot on the gold substrates is also displayed.
variations in current that occur while the bias is varied. This was necessary to avoid the tip crashing during the long data acquisition necessary due to the low light intensities. In future experiments, we will attempt speeding up data acquisition and limiting drift by cooling to cryogenic temperatures, so that the measurements can be done without active feedback. Figure 2 shows the STL spectrum of bare gold, along with spectra corresponding to the series of (CdSe)ZnS dots. In this case, we measure the photon intensity while scanning the bias applied to the tip while maintaining a constant tunneling current. On bare gold, the spectrum (Figure 2, on top) reveals the different SP modes of increasing energy associated with the cavity (n ) 1, 2, and 3). n ) 1 corresponds to the first detectable mode starting at 1.15 eV, the detection limit of the CCD in our setup. The gradual decrease in the photon intensity between the plateaus of a single mode (because of the scale it is not very clear for n ) 1 but it is similar to that of n ) 2) can be explained by a gradual decrease of the tunneling gap owing to the increase of the applied bias while maintaining a constant tunneling current. Nano Lett., Vol. 6, No. 12, 2006
the potential of the tip and therefore Ef,s - Ef,tip ) -q∆V) needs to fulfill the condition -q(1 - η)∆V ) Ese + Σse
Figure 3. Energy level schemes of the substrate-quantum dottip ensemble. (a) Situation in resonance, the substrate communicates with the LUMO level, and the tip communicates with the HOMO level of the quantum dot. Recombination occurs in the quantum dot, either by light emission at energies associated with the normal PL maximum of the quantum dots or by nonradiative recombination. (b) Situation out of resonance, tunneling occurs directly to the substrate and recombination occurs by generation of SPs, which subsequently causes light emission.
The STLS spectrum for the series of (CdSe)ZnS QDs is shown in the bottom panes of Figure 2. The spectra are labeled by the fluorescence emission (nominal) of their respective solutions in nanometers: QD488, QD556, QD557, and QD614. A band-pass filter with high transmittance between 300 and 775 nm is inserted to increase the signalto-noise ratio of the QD resonances. In absence of any perturbation and using this filter, a monotonous increase with the bias is expected, as indicated by the dashed line on the substrate spectrum. This is the universal spectrum on which one can find superimposed the resonances characteristic of the object (in this case, QDs) present in the cavity. The PL spectrum is also displayed to establish the mutual correspondence. The tunneling luminescence spectrum of individual dots shows very distinct transitions owing to the resonant quenching of radiative emission by plasmons. These transitions closely follow the fundamental optical transition of the dots. At a characteristic resonant energys2.63 eV for QD488, 2.41 eV for QD536, 2.27 eV for QD557, and 2.04 eV for QD614splasmon-induced emission becomes nearly completely extinguished. Additional transitions can be seen on the STLS (notably at 2.72 eV for QD536, and at 2.56 eV for QD614), but they are not as strong or well-defined as the fundamental resonances. The extinction of emission of photons by plasmons in the presence of QDs can speculatively be explained by a simple model in which, at the resonant condition, the Fermi levels of substrate and tip are aligned with the states of the QD and electrons and holes are transferred into the dot (Figure 3). This contrasts with the situation out of resonance, where electrons tunnel through the QD and transfer their energy to SPs followed by light emission. The conditions for simultaneous transfer can be calculated by using accepted models for tunneling spectroscopy.20 If it is assumed that first an electron is transferred into the dot from the substrate, the applied bias ∆V (defined as potential of the substrate minus Nano Lett., Vol. 6, No. 12, 2006
(1)
where η is the fraction of the potential that drops over the QD-tip gap (η ) (Vdot - Vtip)/∆V), Ei is a single-particle energy level associated with a certain state i of the dot, and Σi are the polarization energies associated with a carriers in that same state. It is assumed that the applied bias drops only between tip and dot, and substrate and dot, implying that 1 - η ) (Vsubstrate - Vdot)/∆V and therefore that there is only a negligible electrical field inside the dot.20 Subsequently, a hole is transferred from the tip into the dot at qη∆V ) Esh - Σsh + Je-h
(2)
where Je-h is the electron-hole Coulomb attraction energy, which occurs in this equation since there is already a carrier of opposite sign on the QD. Solving eqs 1 and 2 for ∆V yields for the applied bias at the first condition of resonance q∆V ) -∆Ese-sh - Σse - Σsh + Je-h ) -Eopt gap
(3)
where Eopt gap is the optical band gap of the quantum dot. Note that we neglect smaller energies such as exchange interactions in this first-order calculation. This does not impact the conclusion. The sequence of carriers transferred into the dot (i.e., whether first a hole goes or an electron, or whether these events occur simultaneously) also does not matter for the applied bias. The above result can be understood intuitively by the fact that energetically it does not matter whether an exciton on the QD is created by optical excitation or by injection from tip and substrate. In accordance with eq 3, and because in these QDs the PL emission occurs at energies slightly smaller than the optical band gap (the Stokes shift is of the order of 30 meV as determined from PL excitation spectroscopy of QD solutions (not shown)), Figure 2 shows that the plasmon emission is turned off when the applied bias equals the optical band gap. It has to be remarked here that injection from the substrate into the dot occurs at the LUMO level and from the tip into the HOMO level of the dot, yielding a dot in the excited state (Figure 3a). The dot can subsequently relax by radiative emission and nonradiative processes. From the reasoning above, it could be expected that at applied biases above the condition calculated for the simultaneous transfer into HOMO and LUMO, other resonances should occur. Such higher-bias resonances are observed in some cases (Figure 3, lower panel); however, one would expect many more resonances and more closely spaced ones. We cannot currently explain this discrepancy. Some further alternative explanations for the observed resonances are given in the discussion surrounding Figure 5. From the specificity of the observed resonances, a strategy to plasmon resonance imaging is demonstrated based on 2835
Figure 4. Constant current STM image (a) and corresponding plasmon resonance image of a multidot layer consisting of QD536, QD557, and QD614 (b). The color in (b) represents the resonant energy for each dot: blue for QD536, green for QD557, red for QD614. A line scan over this image is shown in (c), where a resolution better than 10 nm is demonstrated. (d) Drift-corrected superposition of (a) and (b). (a) V ) 0.5 V and It ) 100 pA. (b) It ) 5 nA. 125 ms/pixel. The bias is modulated between 1.9 and 2.7 V on each pixel to search for the resonant energy.
acquiring a tunneling luminescence spectrum for each pixel on the image. This is performed by synchronizing the readout of the CCD (photon intensity) with the scanning of the STM tip while the bias is modulated. To illustrate the potential of this optical microscopy in mapping and tagging molecular species, a self-assembled multidot submonolayer was prepared by soaking a non-flame-annealed gold substrate in a 1:1:1 weight by mixture solution of QD536, QD557, and QD614 for 1 h. This substrate is chosen over the flamed version because the submicrometer features of the metallic surface further enhance SP confinement, increasing the photon intensity and speeding up spectrum acquisition. Figure 4a shows a constant current STM image for the multidot structure. Figure 4b is the corresponding plasmon resonance image acquired in a second scan, which is produced by searching for the resonances of the multidot structure on each individual STLS spectrum. The bias is modulated between 1.9 and 2.7 V and the setpoint for the tunneling current is It ) 5 nA, with an overall acquisition time of 125 ms for each spectrum. In this image, the color represents the resonant energysblue, green, and red for the characteristic resonance of QD536, QD557, and QD614, respectively (see Figure 2)s and the brightness the strength of the resonancesthe brightest pixel in the image corresponds to the total extinction of the plasmon at the resonant energy. Thirty three events of resonance are recorded for QD536, 79 for QD557, and 147 2836
for QD614. Another finding worth noticing is that the resonance is much stronger for QD536 and QD557 than for QD614soverall, blue and green are much brighter than red in the plasmon resonance image (see Figure 4b and line scan in Figure 4c). Three explanations are possible for the observation of many more events due to the larger dots than for the smaller dots: I, bigger dots more readily adsorb to the surface; II, bigger dots are more likely to give rise to the quenching resonance; III, energy transfer occurs between dots in a cascade effect toward the lowest-band gap or biggest dots. It was also found that the blue and green resonances (i.e., QD536 and QD557) are much stronger than those corresponding to QD614 in the plasmon resonance image (see Figure 4b and line scan in Figure 4c). The reason for this remains unclear, but we speculate that perhaps the plasmonics of the cavity are tuned such that only specific resonances are amplified. To provide a more complete answer to this question, experiments with different ratios of the particle sizes should be undertaken. Finally, Figure 4d is a superposition of parts a and b of Figure 4 which shows that, statistically, resonant transitions are preferentially observed near the metallic surface grain boundaries. Also, as indicated by arrows, a few of the individual metallic grains are inactive. This observation further suggests that the local plasmonics of the substrate is an important factor in determining whether resonance occurs. The extinction (including reabsorption) of light emission by plasmons at the resonant energy may also be assisted by an electromagnetic field enhancement effect due to the SP in the optical cavity. Figure 5a shows a schematic illustration of the plasmon modes relevant to our discussion. In addition to the localized SP (LSP) in the optical cavity under the tip, there is a propagating SP (PSP) emerging from the cavity that diffuses along the gold surface. The LSP and PSP modes are coupled to each other by the optical cavity but can be distinguished from each other in the far-field emission pattern. A QD placed in this cavity is subject to the oscillating electric field of the LSP, as illustrated in Figure 5b. Ultimately, the massive enhancement of the electromagnetic field in the tunneling gap could stimulate the extinction of photon emission by plasmons in resonance with the optical transition of the dot. This effect might be aided by the fact that the spectra in Figures 2 and 3 were taken with current feedback intact, which implies that during the measurements the tip changes height and therefore dynamically alters the plasmon modes in the cavity. An essential condition for the observation of an ideal tunneling luminescence spectrum at the metallic gap is a stable flow of plasmons. Charge storage in the dot or its local environment disrupts the plasmon dynamics of the cavity as well as its release and the SP emission becomes unstable. A better understanding of the coupling between the plasmon and the charge storage in the dot or traps at the QD’s surface can be exploited in investigating the dynamics of charging-discharging, responsible for the blinking in colloidal QDs.21-23 Therefore, it is crucial to expedite the acquisition of STLS to avoid the charge storage likely to Nano Lett., Vol. 6, No. 12, 2006
Acknowledgment. This work was supported by the U.S. Department of Energy under Contract No. DE-AC3699GO10337. References
Figure 5. Plasmonics involved in the STL measurements. (a) Localized and propagating modes of the surface plasmon (LSP and PSP, respectively) are coupled in the optical cavity formed by the tunneling gap. They can be distinguished on the emission pattern in the far field. PSP modes are seen on the image of the focus of the mirror projected in the center of the CCD (see Figure 1). LSP modes, on the other hand, contribute to the annular feature on the emission pattern. (b) A quantum dot under the tip is subject to the oscillating electric field of the plasmon sustained by the cavity.
occur at long time scales or to slow down such dynamic charging by cooling down the system to cryogenic temperatures. In summary, owing to the combination of the superb resolution inherent to STM with the detection of single molecular species by the interaction with surface plasmons, plasmon resonance imaging shows great potential in nanoscale science. Because the resonance measured by STLS is a fingerprint of the object embedded in the optical cavity, this imaging strategy will enable the investigation of nanoscale systems of great complexitysfrom cascade superstructures of quantum dots, to hybrids of dots and carbon nanotubes, to biomolecules and molecular assemblies.
Nano Lett., Vol. 6, No. 12, 2006
(1) Gur, I.; Fromer, N. A.; Geier, M. L.; Alivisatos, A. P. Science 2005, 310, 462. (2) Coe, S.; Woo, W. K.; Bawendi, M.; Bulovic, V. Nature 2002, 420 (6917), 800-803. (3) Schaller, R. D.; Agranovich, V. M.; Klimov, V. I. Nat. Phys. 2005, 1 (3), 189-194. (4) Ellingson, R. J.; Beard, M. C.; Johnson, J. C.; Yu, P. R.; Mic´ic´, O. I.; Nozik, A. J.; Shabaev, A.; Efros, A. L. Nano Lett. 2005, 5 (5), 865-871. (5) Qi, D. F.; Fischbein, M.; Drndic´, M.; Selmic, S. Appl. Phys. Lett. 2005, 86 (9), 093103. (6) Schaller, R. D.; Klimov, V. I. Phys. ReV. Lett. 2004, 92 (18), 186601. (7) Murphy, J. E.; Beard, M. C.; Norman, A. G.; Ahrenkiel, S. P.; Johnson, J. C.; Yu, P.; Mic´ic´, O. I.; Ellingson, R. J.; Nozik, A. J. J. Am. Chem. Soc. 2006, 128 (10), 3241-3247. (8) Nozik, A. J. Physica E 2002, 14 (1-2.), 115-120. (9) Barnes, W. L.; Dereux, A.; Ebbesen, T. W. Nature 2003, 424 (6950), 824-830. (10) Xia, Y. N.; Halas, N. J. MRS Bull. 2005, 30 (5), 338-344. (11) Michaels, A. M.; Nirmal, M.; Brus, L. E. J. Am. Chem. Soc. 1999, 121 (43), 9932-9939. (12) Jiang, J.; Bosnick, K.; Maillard, M.; Brus, L. J. Phys. Chem. B 2003, 107 (37), 9964-9972. (13) Westphalen, M.; Kreibig, U.; Rostalski, J.; Luth, H.; Meissner, D. Sol. Energy Mater. Sol. Cells 2000, 61 (1), 97-105. (14) Choulis, S. A.; Mathai, M. K.; Choong, V.-E. Appl. Phys. Lett. 2006, 88 213503. (15) Qiu, X. H.; Nazin, G. V.; Ho, W. Science 2003, 299 (5606), 542546. (16) Berndt, R.; Gaisch, R.; Gimzewski, J. K.; Reihl, B.; Schlittler, R. R.; Schneider, W. D.; Tschudy, M. Science 1993, 262 (5138), 14251427. (17) Egusa, S.; Liau, Y. H.; Scherer, N. F. Appl. Phys. Lett. 2004, 84 (8), 1257-1259. (18) Nilius, N.; Benia, H. M.; Salzemann, C.; Rupprechter, G.; Freund, H. J.; Brioude, A.; Pileni, M. P. Chem. Phys. Lett. 2005, 413 (1-3), 10-15. (19) Sakurai, M.; Thirstrup, C.; Aono, M. Appl. Phys. A 2005, 80 (6), 1153-1160. (20) Jdira, L.; Liljeroth, P.; Stoffels, E.; Vanmaekelbergh, D. l.; Speller, S. Phys. ReV. B 2006, 73, 115305. (21) Neuhauser, R. G.; Shimizu, K. T.; Woo, W. K.; Empedocles, S. A.; Bawendi, M. G. Phys. ReV. Lett. 2000, 85 (15), 3301-3304. (22) Shimizu, K. T.; Neuhauser, R. G.; Leatherdale, C. A.; Empedocles, S. A.; Woo, W. K.; Bawendi, M. G. Phys. ReV. B 2001, 6320 (20). (23) Kuno, M.; Fromm, D. P.; Gallagher, A.; Nesbitt, D. J.; Mic´ic´, O. I.; Nozik, A. J. Nano Lett. 2001, 1 (10), 557-564.
NL061997S
2837
