Letter pubs.acs.org/NanoLett
Single Photon Counting from Individual Nanocrystals in the Infrared Raoul E. Correa,† Eric A. Dauler,‡ Gautham Nair,† Si H. Pan,‡ Danna Rosenberg,‡ Andrew J. Kerman,‡ Richard J. Molnar,‡ Xiaolong Hu,§ Francesco Marsili,§ Vikas Anant,§ Karl K. Berggren,§ and Moungi G. Bawendi*,† †
Department of Chemistry, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, United States Lincoln Laboratory, Massachusetts Institute of Technology, Lexington, Massachusetts 02420, United States § Research Laboratory of Electronics, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, United States ‡
S Supporting Information *
ABSTRACT: Experimental restrictions imposed on the collection and detection of shortwave-infrared photons (SWIR) have impeded single molecule work on a large class of materials whose optical activity lies in the SWIR. Here we report the successful observation of room-temperature single nanocrystal photoluminescence at SWIR wavelengths using a highly efficient multielement superconducting nanowire single photon detector. We confirm that the photoluminescence from single lead sulfide nanocrystals is strongly antibunched, demonstrating the feasibility of performing sophisticated photon correlation experiments on individual weak SWIR emitters, and, more broadly, paving the way for sensitive measurements of spectral observables on infrared quantum systems that are incompatible with current detection techniques. KEYWORDS: Infrared spectroscopy, lead sulfide, SNSPD, single photon counting, photon correlation
O
Key to improving the functionality and performance of SWIR nanocrystals in the applications described above is a comprehensive picture of their excitonic lifecycle. Light emission is a fitting probe of nanoscale exciton dynamics; as revealed in single CdSe NC fluorescence studies,9−12 much can be learned about the microscopic phenomena governing recombination of band-edge and higher-order excitons. Most notable was the discovery of fluorescence intermittency, or “blinking”, which came about as a result of single NC microscopy and prompted further attempts to understand and control its behavior.1,13,14 However, very little is known about the exciton lifecycle in single SWIR-active colloidal NCs because current single-photon detector technologies do not have the sensitivity to measure their weak emission with adequate signal-to-noise.15 For example, the authors of ref 16 attempted single PbS NC spectroscopy with InGaAs avalanche photodiodes but were unable to locate individual nanocrystals, concluding that their approach was insufficient given the prohibitively large background detector noise and low detection quantum efficiency. Another recent experiment utilized frequency upconversion to detect single 1300 nm-wavelength photons from self-assembled epitaxial quantum dots (SAQDs) by shifting them into the visible, but limited phase-matching bandwidth in the nonlinear medium restricts the applicability to
ur understanding of processes that govern the excitonic lifecycle in molecules and nanomaterials is intimately linked to the measurements we can make to probe them. For example, efforts toward the efficient collection and detection of single photons from individual quantum systems in the visible region of the spectrum now enable us to study heterogeneous and dynamical observables, helping reveal a vast array of new phenomena hidden in ensemble-averaged experiments.2,4 Unfortunately, despite the maturation and near-routine nature of single molecule spectroscopy (SMS) in the visible, measurements of this kind remain extraordinarily difficult at infrared wavelengths.15 Luminescence detection from single SWIR-active systems is a natural extension of SMS with a growing number of applications targeting this wavelength regime. Lead chalcogenide (PbX, X = S, Se) colloidal semiconductor nanocrystals (NCs) with band gaps tunable throughout the shortwaveinfrared photons (SWIR) have been successfully exploited in solution processed solar cells, photodetectors and light emitting devices, while other novel NC materials from the semimetal family (e.g., Cd3As2, HgTe) remain relatively unexplored optically but hold great potential as active media in midinfrared photodetectors.5,7 It has also been argued that in vivo fluorescence imaging at SWIR wavelengths (1−2 μm) can achieve deeper penetration with minimal scattering losses across a wide range of tissues.8 Pushing the sensitivity of SWIR photon collection and detection to the ultimate limit, single photons from a single molecule, is desirable in this respect. © XXXX American Chemical Society
Received: February 15, 2012 Revised: April 26, 2012
A
dx.doi.org/10.1021/nl300642k | Nano Lett. XXXX, XXX, XXX−XXX
Nano Letters
Letter
Figure 1. Experimental apparatus. (a) False-color scanning electron micrograph of the multielement SNSPD chip, showing each of the four independently biased NbN nanowires in a different color. The multielement SNSPD has a total system detection efficiency of 36% at 1550 nm with the signal split evenly between the four wires. (b) Home-built sample-scanned confocal microscope in the epi-fluorescence configuration (SM, single-mode; FL, focusing lens; BS, beamsplitter; LP, long pass (650 and 700 nm); TCSPC, time-correlated single photon counter). The excitation light (red) is spatially filtered through a 633 nm single-mode fiber before exciting single NCs, with the emission (gray) collected, spectrally filtered and transported to the detectors. (c) Ensemble emission spectra for PbS/CdS (blue) and InAs/CdSe (red) NCs used in our experiments.
Figure 2. Intermittent fluorescence from single PbS/CdS (blue) and InAs/CdSe (red) nanocrystals. (a) Excerpt of a photoluminescence intensity versus time trace for a single PbS/CdS NC; photons were collected on all four channels of the multielement SNSPD. (b) Excerpt of a PL intermittency trace for InAs/CdSe, collected from a single channel. Histograms of the PL intensity values are shown to the right of each trace, clearly revealing two-state behavior.
narrow line width systems.17 Direct SWIR photon detection using SNSPDs circumvents this problem and has been utilized in the past to study the optical properties of SAQDs, as well as single plasmon excitations in gold waveguides.18,19,29,32 In these reports, the low system detection efficiency (ca. 5−10%) of the
SNSPD was compensated for by the extreme photostability and fast radiative rate of the SAQD emitter, providing an adequate signal-to-noise ratio to measure intensity autocorrelations. Infrared colloidal nanocrystals on the other hand produce a lower emission photon flux compared to their epitaxial B
dx.doi.org/10.1021/nl300642k | Nano Lett. XXXX, XXX, XXX−XXX
Nano Letters
Letter
counterparts; the excitation pumping rate is usually much higher for SAQDs due to the larger absorption cross-section when excited nonresonantly, as is the single exciton radiative rate.18,32 Since nanocrystals photobleach after a finite amount of time, single SWIR wavelength NC experiments demand efficient collection and detection of the weak luminescence, something that we demonstrate in this report. In order to meet the challenges facing single NC interrogation, we combined confocal microscopy with a novel multielement superconducting nanowire single photon detector (SNSPD)3 to image individual core/shell PbS/CdS and InAs/ CdSe nanocrystals. The detector was vital to the experiment’s success, so we highlight its operation briefly. The SNSPD consists of four interleaved niobium nitride (NbN) nanowires grown on a sapphire substrate and cooled to ∼2.6 K with each nanowire individually biased near the critical current giving a total detection efficiency of 36% at 1550 nm (9% per channel) (Figure 1a). When a single SWIR photon is absorbed by one of the four nanowires, it causes the absorbing nanowire to transiently switch into its normal conducting state before selfresetting a short time later (∼9 ns). The transient superconductor-normal transition produces a measurable voltage pulse that is detected electrically and used to record the photon’s arrival time; with negligible cross-talk, all four channels functioned identically and independently from each other. The quantized nature of single photon absorption obviates the need for a physical beam splitter to split the photon stream to multiple detectors, a potential source of experimental artifacts when measuring second-order intensity correlations (g(2)(τ)). A dilute PbS/CdS nanocrystal sample with emission centered at 1100 nm (Figure 1c) was excited using 633 nm cw-light at a moderate flux of ∼1.1 kW/cm2 with the resulting photoluminescence (PL) focused into a single-mode fiber coupled to the SNSPD (Figure 1b) (see Methods for more detail). PL intensity versus time traces were subsequently generated by binning arrival events from all four channels into 50 ms intervals; they clearly reveal the presence of intermittency (Figure 2a), stochastically switching between a fluorescent ON state and a nonfluorescent OFF state. Similarly, blinking was observed from single InAs NCs overcoated with a thick CdSe shell (∼7 monolayers) with the ensemble emission spectra centered at 1300 nm (Figure 2b). We verified that the OFF state count rate in all experiments was equal to the dark count rate measured from a pristine microscope coverslip under equivalent experimental conditions, supporting our assignment of binary blinking. In order to prove the independent operation of each channel, we collected the fluctuating PL intensity of the InAs nanocrystal in Figure 2b using a single nanowire detector; the trace also shows discrete switching between two states. Additional single nanocrystal traces are plotted in the Supporting Information, all of which show analogous dynamics to those presented in Figure 2. We note that the lower signalto-noise ratio observed in the InAs/CdSe experiments most likely arises due to reduced coupling to the SNSPDs as the emission wavelength approaches the collection fiber’s singlemode wavelength. While two state blinking is a signature of single NC localization,1 this observation was further confirmed by measuring the second-order intensity autocorrelation or g(2)(τ) of the luminescence photon stream
g(2)(τ ) =
⟨I1(t )I2(t + τ )⟩ ⟨I1(t )⟩⟨I2(t + τ )⟩
(1)
whose value at τ = 0 reports on the statistical nature of the single photon source.20 In our experiment, the photon arrival times recorded on each nanowire were correlated with the other three channels, achieving a 50% increase in coincidence counts compared to a two-channel setup. We emphasize that even in the correlation measurements, each emission photon has a 36% chance of detection (rather than 9% if only one channel was active); this is because the photon’s spatial mode overlaps with all four nanowire channels on the interleaved chip, increasing the probability of detection to that of all four elements combined. One can readily make the analogy between our method and a setup that splits the emission photon stream to four independent detectors, each with a detection efficiency of 36%. Figure 3a depicts an unnormalized g(2) measurement for a single PbS/CdS nanocrystal; the 0-time dip was found to be g(2) 0 = 0.415 when normalized by the coincidence count value at longer times, strong evidence that the SWIR emission is nonclassical and is composed primarily of single photons. The full extent of antibunching is reduced due to the accumulation of uncorrelated background-background and signal-background coincidence counts during the integration period, obscuring the intrinsic NC g(2) 0 . For four detection channels, the uncorrelated
Figure 3. Unnormalized second-order intensity autocorrelations for PbS/CdS nanocrystals. (a) Strong antibunching from a single blinking SWIR nanocrystal, characteristic of nonclassical luminescence. After accounting for the uncorrelated background, the normalized g0(2) ≈ 0.19. The solid red line is a double-sided single exponential fit, which after accounting for the excitation pumping rate γ gives an extracted excited state lifetime of ca. 115 ns (see text). (b) The same measurement on a cluster of nanocrystals displayed no dip at zero time delay as expected from a group of independent emitters. C
dx.doi.org/10.1021/nl300642k | Nano Lett. XXXX, XXX, XXX−XXX
Nano Letters
Letter
density distributions P(ton/off) are estimated by weighting each value H(ton/off,i) by a factor of 0.5(ton/off,i+1−ton/off,i−1),23 a method that extends the temporal dynamic range of the histogrammed data sets, especially when the frequency of longtime events is small. Figure 4 presents P(ton/off) for PbS/CdS NCs plotted on doubly logarithmic axes, showing that the off-time events
coincidence counts can be calculated using C(τ) = ∑i =4 1∑j >4 i(NiNj + NiSj + NjSi)wT, where Ni (Si) is the average dark (signal) count rate on channel i, w is the bin time for g(2)(τ), and T is the integration time (note that S + N = IPL, the average count rate measured on each channel). For this experiment, Ni = 280 counts/s, Si = 266 counts/s, w = 40 ns, and T = 1200 s, giving C(τ) = 65. The intrinsic 0-time dip is therefore g(2) 0 = 0.19, a much stronger indicator of single emitter localization. It has been shown in CdSe nanocrystals that multiexciton radiative emission can corrupt the single photon stream even in the limit of low excitation fluence due to a small but finite biexciton quantum yield.12,28 However, one would expect the biexciton quantum yield to be low based on ensemble kBX,rad and kBX,non‑rad measurements for PbS (ca. 107 and 1010 s−1, respectively).30,31 While the probability of a secondary adjacent nanocrystal emitting into the collection volume is very low (the blinking trace exhibits clear two-state instead of three-state blinking), we note that our calculation of C(τ) accounts for this by using the OFF-state intensity rather than the detector dark count rate for Ni. We also collected fluorescence from a cluster of nanocrystals that did not blink but exhibited continuous intensity fluctuations, and calculated the resulting g(2)(τ). No 0-time antibunching was seen (Figure 3b), confirming that the observed single NC antibunching was not an artifact of the experiment. At short times the normalized autocorrelation takes on the form g(2)(τ) = 1 − exp[−ktotτ], where ktot = γ + kL, γ = Iexσ/ℏω is the excitation pumping rate, kL is the single exciton radiative rate, σ is the absorption cross-section at 633 nm, and Iex is the excitation intensity.20 In this experiment we estimate γ ≤ (1100 W/cm2 × 1 × 10−15 cm2)/(3.14 × 10−19 W) = 3.5 × 106 s−1. The absorption cross-section at the excitation wavelength was obtained by scaling σband‑edge by the optical density at 633 nm, using band-edge values from ref 33. Given the uncertainties in accurately determining γ (i.e., most likely underestimating the excitation spot size), we can at most provide an upper bound to the excited state lifetime. The extracted time constant from a double-sided single exponential fit to g(2) gives k−1 tot = 82 ns, resulting in an excited state lifetime ≤ 115 ns, shorter than the ensemble photoluminescence k−1 L lifetime in solution (ca. 500−1000 ns). A detailed study of single PbX NC lifetimes over a large selection of nanocrystals would be desirable before making assertions about the distribution; nonetheless, we can confidently assert that the fluorescence from single SWIR-emitting PbS/CdS NCs was isolated, collected and detected at room temperature, with the photon statistics revealing sub-Poissonian behavior. Using the observations of two state blinking and antibunching as criteria for single emitter localization, we can further analyze the fluorescence intensity versus time traces of individual NCs to gain a better understanding of the SWIR blinking process. Previous studies have shown that nearly all nanocrystal systems to date exhibit power-law-distributed ON and OFF periods (ton/off) of emission, indicative of a highly distributed mechanism for quenching and reviving the luminescence. The functional form of the probability density P(ton/off) is governed by the nature of carrier relaxation within the NC, and informs studies of the stochastic processes responsible for switching single NCs on and off.21,22 To calculate the ON- and OFF-time probability density distributions, an ON/OFF intensity threshold is set at roughly 3σ above the average background signal for each trace and ON/ OFF event histograms H(ton/off) are generated. The probability
Figure 4. On and off-time probability densities for PbS/CdS NCs, plotted on double logarithmic axes. (a) The on-time probability density (open circles), fit to eq 2. (b) The off-times (open squares) are similarly power-law distributed but do not display an appreciable exponential cutoff at longer waiting times. The rarity of extremely long off-times (toff > 30 s) skews the data away from the power-law fit (eq 3), but do not justify the use of an exponential cutoff (see main text).
decrease almost linearly while the on times display a cutoff for longer times. This is in line with the findings of CdSe, CdTe, InP, and a host of other NCs, a remarkable fact given the different electronic, crystal and chemical structure of PbS/CdS NCs. The distributions were fitted using a nonlinear leastsquares method to the following functions ⎛ t ⎞ −αon P(ton) ∝ ton exp⎜ − on ⎟ ⎝ τsat ⎠
(2)
−αoff P(toff ) ∝ toff
(3)
giving exponents αon = 1.46 ± 0.05, αoff = 1.51 ± 0.05 and a cutoff time τsat = 1.3 ± 0.1 s. The low-probability data in the OFF-time plot appear to indicate a cutoff at long times (toff > 30 s), but the majority of the data set excludes a power-law fit with an exponential tail (analogous to eq 2). The values of the power-law exponents indicate that an average OFF time does not exist for PbS/CdS nanocrystals,24 detrimental to the operating performance of light-emitting devices fabricated from this material. D
dx.doi.org/10.1021/nl300642k | Nano Lett. XXXX, XXX, XXX−XXX
Nano Letters
Letter
Notes
The exponent values are also similar to those found in visible emitting NCs (1 < αon/off < 2), furthering the theory that the processes governing stochastic blinking in colloidal nanocrystals are universal, seemingly insensitive to microscopic material properties and perhaps better characterized by physical properties like shape, surface passivation and the local chemical environment. A number of theories have invoked mechanisms that do not rely on intrinsic chemical properties such as the material’s band gap, electron/hole effective masses, or the dielectric constant.21,22 Recently, statistical analysis of luminescence spectral diffusion was conducted using a bath of environmental two-level systems in the same way ref 22 analyzed single NC blinking statistics with both approaches finding anomalous power-law distributions of the experimental observables.25 Our experiments help lend credence to these proposals and will hopefully guide future theoretical efforts to explain NC intermittency. In summary, we have experimentally realized a platform to count single infrared photons from individual quantum systems under ambient conditions, directly probing the excitonic lifecycle in single SWIR nanocrystals without plasmonic or low-temperature perturbations. The method is general and opens new avenues for discovery; for example, quantifying the extent of antibunching (g(2) 0 ) from photostable carbon nanotubes that emit at SWIR wavelengths would benefit those who wish to realize an efficient room-temperature single photon source. For nanocrystals, this quantity is directly proportional to the biexciton luminescence quantum yield, an observable that is difficult to extract from ensemble measurements but is of significant importance for lasing applications. Finally, one could envision using intensity correlations to extract spectral information from single emitters,27 circumventing detectivity limitations in traditional InGaAs arrays when conducting spectroscopy on weakly emitting systems. Methods. Core/shell PbS/CdS nanocrystals were chemically synthesized by making PbS core particles following literature guidelines.6 After precipitating the cores in isopropanol and redissolving in hexanes, a large excess of cadmium oleate was added and the resulting mixture was heated at 100 °C for 24 h. InAs/CdSe NCs were synthesized following Peng and co-worker’s preparation methods.26 All nanocrystals were precipitated once and redissolved in hexanes, before forming an ∼10−10 M solution with 9:1 hexane/octane. The resulting solution was dropcast onto microscope coverslip slides (No. 1), and loaded face-down in a home-built sample-scanned confocal microscope (oil immersion objective: Nikon 100×, NA = 1.25). Single-mode 633 nm light was focused through a f = 15 cm lens and reflected into the objective with a 1 in. thick BK7 glass beamsplitter (90:10 transmission/reflection). PL emission was collected and focused onto the core of a single-mode fiber (single-mode for wavelengths >1260 nm), which transports the photon stream to the multielement superconducting nanowire detector. Photon arrival times are recorded on all four channels using a Picoquant Hydraharp 400.
■
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS This work was supported by the Center for Excitonics, a Department of Energy EFRC (Award No. DE-SC0001088). At Lincoln Laboratory, the work is sponsored by the United States Air Force under Air Force Contract No. FA8721-05-C-0002. Opinions, interpretations, recommendations and conclusions are those of the authors and are not necessarily endorsed by the United States Government. The authors would like to thank Mr. J. Daley for technical assistance. This work made use of MIT’s shared scanning-electron-beam-lithography facility in the Research Laboratory of Electronics (SEBL at RLE).
■
ASSOCIATED CONTENT
S Supporting Information *
Additional information and figures. This material is available free of charge via the Internet at http://pubs.acs.org.
■
REFERENCES
(1) Nirmal, M.; Dabbousi, B. O.; Bawendi, M. G.; Macklin, J. J.; Trautman, J. K.; Harris, T. D.; Brus, L. E. Nature 1996, 383, 802. (2) Dahan, M.; Sabine, L.; Luccardini, C.; Rostaing, P.; Riveau, B.; Triller, A. Science 2003, 302, 442. (3) Dauler, E. A.; Kerman, A. J.; Robinson, B. S.; Yang, J. K. W.; Voronov, B.; Gol’tsman, G.; Hamilton, S. A.; Berggren, K. K. J. Mod. Opt. 2009, 56, 364. (4) Moerner, W. E. J. Phys. Chem. B 2002, 106, 910. (5) Sargent, E. H. Nat. Photonics 2009, 3, 325. (6) Zhao, N.; Osedach, T. P.; Chang, L. Y.; Geyer, S. G.; Wanger, D.; Binda, M. T.; Arango, A. C.; Bawendi, M. G.; Bulovic, V. ACS Nano 2010, 4, 3743. (7) Keuleyan, S.; Lhuillier, E.; Brajuskovic, V.; Guyot-Sionnest, P. Nat. Photonics 2011, 5, 489. (8) Lim, Y. T.; Kim, S.; Nakayama, A.; Stott, N. E.; Bawendi, M. G.; Frangioni, J. V. Mol. Imaging 2003, 2, 50. (9) Kuno, M.; Fromm, D.; Hafmann, H.; Gallagher, A.; Nesbitt, D. J. Chem. Phys. 2000, 112, 3117. (10) Schlegel, G.; Bohnenberger, J.; Potapova, I.; Mews, A. Phys. Rev. Lett. 2002, 88, 137401. (11) Biadala, L.; Louyer, Y.; Tamarat, P.; Lounis, B. Phys. Rev. Lett. 2010, 105, 157402. (12) Park, Y.-S.; Malko, A. V.; Vela, J.; Chen, Y.; Ghosh, Y.; GarcíaSantamaría, F.; Hollingsworth, J. A.; Klimov, V. I.; Htoon, H. Phys. Rev. Lett. 2011, 106, 187401. (13) Wang, X.; Ren, X.; Kahen, K.; Hahn, M. A.; Rajeswaran, M.; Maccagnano-Zacher, S.; Silcox, J.; Cragg, G. E.; Efros, A. L.; Krauss, T. D. Nature 2009, 459, 686. (14) Mahler, B.; Spinicelli, P.; Buil, S.; Quelin, X.; Hermier, J.-P.; Dubertret, B. Nat. Mater. 2008, 7, 659. (15) Hadfield, R. H. Nat. Photonics 2009, 3, 696. (16) Rakher, M. T.; Bose, R.; Wong, C. W.; Srinivasan, K. Opt. Express 2011, 19, 1786. (17) Rakher, M. T.; Ma, L.; Slattery, O.; Tang, X.; Srinivasan, K. Nat. Photonics 2010, 4, 786. (18) Zinoni, C.; Alloing, B.; Monat, C.; Zwiller, V.; Li, L. H.; Fiore, A.; Lunghi, L.; Gerardino, A.; de Riedmatten, H.; Zbinden, H.; Gisin, N. Appl. Phys. Lett. 2006, 88, 131102. (19) Zinoni, C.; Alloing, B.; Li, L. H.; Marsili, F.; Fiore, A.; Lunghi, L.; Gerardino, A.; Vakhtomin, Y. B.; Smirnov, K. V.; Gol’tsman, G. Appl. Phys. Lett. 2007, 91, 031106. (20) Loudon, R. The quantum theory of light; Oxford University Press: Oxford, 2000. (21) Verberk, R.; van Oijen, A. M.; Orrit, M. Phys. Rev. B 2002, 66, 233202. (22) Frantsuzov, P.; Volkan-Kacso, S.; Janko, B. Phys. Rev. Lett. 2009, 103, 207402. (23) Hodak, J. H.; Downey, C. D.; Fiore, J. L.; Pardi, A.; Nesbitt, D. J. Proc. Natl. Acad. Sci. U.S.A. 2005, 102, 10505. (24) Margolin, G.; Barkai, E. Phys. Rev. Lett. 2005, 94, 080601.
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected].
E
dx.doi.org/10.1021/nl300642k | Nano Lett. XXXX, XXX, XXX−XXX
Nano Letters
Letter
(25) Plakhotnik, T.; Fernee, M. J.; Littleton, B.; Rubinsztein-Dunlop, H. Phys. Rev. Lett. 2010, 105, 167402. (26) Xie, R.; Peng, X. Angew. Chem., Int. Ed. 2008, 47, 7677. (27) Marshall, L. F.; Cui, J.; Brokmann, X.; Bawendi, M. G. Phys. Rev. Lett. 2010, 105, 053005. (28) Nair, G.; Zhao, J.; Bawendi, M. G. Nano Lett. 2010, 11, 1136. (29) Heeres, R. W.; Dorenbos, S. N.; Koene, B.; Solomon, G. S.; Kouwenhoven, L. P.; Zwiller, V. Nano Lett. 2010, 10, 661. (30) Nair, G.; Geyer, S. G.; Chang, L. Y.; Bawendi, M. G. Phys. Rev. B 2008, 78, 125325. (31) McGuire, J. A.; Joo, J.; Pietryga, J. M.; Schaller, R. D.; Klimov, V. I. Acc. Chem. Res. 2008, 41, 1810. (32) Hadfield, R. H.; Stevens, M. J.; Mirin, R. P.; Nam, S. W. J. Appl. Phys. 2007, 101, 103104. (33) Cademartiri, L.; Montanari, E.; Calestani, G.; Migliori, A.; Guagliardi, A.; Ozin, G. A. J. Am. Chem. Soc. 2006, 128, 10337.
F
dx.doi.org/10.1021/nl300642k | Nano Lett. XXXX, XXX, XXX−XXX
