Engineering Localized Surface Plasmon Interactions in Gold by

Feb 6, 2017 - It is becoming increasing clear that intrinsic high losses associated with plasmons can be utilized to create new device concepts to har...
0 downloads 11 Views 3MB Size
Letter pubs.acs.org/NanoLett

Engineering Localized Surface Plasmon Interactions in Gold by Silicon Nanowire for Enhanced Heating and Photocatalysis Daksh Agarwal,† Carlos O. Aspetti,† Matteo Cargnello,‡ MingLiang Ren,† Jinkyoung Yoo,∥ Christopher B. Murray,†,‡ and Ritesh Agarwal*,† †

Department of Materials Science and Engineering and ‡Department of Chemistry, University of Pennsylvania, Philadelphia, Pennsylvania 19104, United States ∥ Center for Integrated Nanotechnologies, Los Alamos National Laboratory, Los Alamos, New Mexico 87545, United States S Supporting Information *

ABSTRACT: The field of plasmonics has attracted considerable attention in recent years because of potential applications in various fields such as nanophotonics, photovoltaics, energy conversion, catalysis, and therapeutics. It is becoming increasing clear that intrinsic high losses associated with plasmons can be utilized to create new device concepts to harvest the generated heat. It is therefore important to design cavities, which can harvest optical excitations efficiently to generate heat. We report a highly engineered nanowire cavity, which utilizes a high dielectric silicon core with a thin plasmonic film (Au) to create an effective metallic cavity to strongly confine light, which when coupled with localized surface plasmons in the nanoparticles of the thin metal film produces exceptionally high temperatures upon laser irradiation. Raman spectroscopy of the silicon core enables precise measurements of the cavity temperature, which can reach values as high as 1000 K. The same Si−Au cavity with enhanced plasmonic activity when coupled with TiO2 nanorods increases the hydrogen production rate by ∼40% compared to similar Au−TiO2 system without Si core, in ethanol photoreforming reactions. These highly engineered thermoplasmonic devices, which integrate three different cavity concepts (high refractive index core, metallodielectric cavity, and localized surface plasmons) along with the ease of fabrication demonstrate a possible pathway for designing optimized plasmonic devices with applications in energy conversion and catalysis. KEYWORDS: Localized surface plasmons, thermoplasmonics, metallo-dielectric cavity, silicon, cavity heating, Raman spectroscopy, nanowire, photoreforming

M

of high refractive index core (Si) along with an effective metallic cavity confines light to an intense mode that leads to strong absorption. Furthermore, the evanescent field from this mode extends into the Au particles of the thin-film causing much stronger excitation of localized surface plasmons (LSPs) thereby heating the cavity to temperatures close to 1000 K. To test the enhancement in heating due to engineered Si− Au resonant cavity structures, devices were fabricated by coating a 10 nm Au layer on Si nanowires dispersed on glass substrates (see Supporting Information). The Si nanowires typically have a 2 nm thick native oxide layer on the surface.14 The 10 nm thick Au film breaks into small particles with an average interparticle separation of ∼5 nm (Figure 1a, right inset). The extent of cavity heating which is directly proportional to the degree of plasmonic excitation was assessed by calculating the temperature of the Si core by measuring the change in temperature dependent phonon energy of Raman

etal nanostructures possess interesting optical properties because of presence of collective oscillations of electrons called surface plasmon resonances (SPRs), which when integrated with semiconductors influence the properties of the system significantly.1−4 The hot electrons generated via plasmon excitation can either be used to aid electron transfer in chemical reactions or to generate local heat for various applications such as photothermal therapy for cancer treatment, nanosurgery, photothermal drug delivery, photothermal imaging, and nanochemistry.5−11 Au is the most commonly used plasmonic metal because of its inert nature and its plasmonic resonances lying in the visible to infrared range,12,13 which can be tuned by altering the shape and size of Au nanostructures. Most plasmonic devices fabricated to date rely on the naturally occurring resonances of the synthesized nanostructures to harvest electromagnetic radiation for heating and/or catalytic applications. However, if the SPR excitations can be significantly enhanced by an engineered external cavity, it can lead to intense heating or increased catalytic activity. Here, by using a semiconductor−metal nanowire cavity we enhance the plasmonic properties of Au nanoparticles leading to intense heat generation at cavity mode resonance. The combined effect © 2017 American Chemical Society

Received: December 12, 2016 Revised: January 24, 2017 Published: February 6, 2017 1839

DOI: 10.1021/acs.nanolett.6b05147 Nano Lett. 2017, 17, 1839−1845

Letter

Nano Letters

Figure 1. Extraordinary heating in Au-coated Si nanowire. (a) Schematic of the device on which heating experiments were performed. Green arrow indicates direction of the incident light, and solid and dashed black arrows indicate direction of polarization of electric field of incident light in transverse magnetic (TM) and transverse electric polarization (TE), respectively. Left inset shows the schematic of the Au-coated Si nanowire cavity cross-section indicating the diameter of Si nanowire and thickness of the Au film. Right inset shows the scanning electron microscopy image of a representative Au-coated Si nanowire. Scale bar: 200 nm. (b) Raman spectra of a bare Si nanowire (100 nm diameter) at different incident powers using a 532 nm wavelength pump in TM polarization with a spot size of 1 μm. The legend indicates laser excitation power values. (c) Temperature versus laser power for bare and 10 nm Au-coated Si nanowire of diameter 100 nm under TM and TE polarization excitation calculated from the Raman spectra shown in (b) and other corresponding Raman spectra. (d) Numerical simulations of absorption spectra of bare and 10 nm Au-coated Si nanowire of diameter 100 nm under TM and TE polarization excitation.

same nanowire after coating 10 nm Au film shows a dramatic increase in temperature of the system and a peak cavity temperature of ∼1000 K (ΔT ∼ 700 K) was observed in the TM polarization at the same laser intensity of 5.7 × 105 W/cm2 (Figure 1c, solid golden curve). To understand this enhanced heating effect, absorption spectra of bare and Au-coated Si nanowire of diameter 100 nm (Figure 1d) was calculated via the finite difference time domain (FDTD) methodology (see Supporting Information). Calculations for bare Si nanowire (Figure 1d, solid black curve) indicated that the cavity mode was resonant with the 532 nm excitation wavelength with a peak absorption value of ∼8% in TM polarization. Si is an indirect bandgap material (bandgap, Eg ∼ 1.12 eV)20 and because the pump wavelength (532 and 659.4 nm) is above the bandgap, the absorption is high; however, during relaxation of carriers, the rate of nonradiative recombination is ∼106 times faster than that of radiative recombination.21 As a result most of the charge carriers generated during excitation decay nonradiatively emitting a large number of phonons generating heat. Calculations also indicate that after deposition of a 10 nm Au layer on the Si nanowire (Figure 1d, solid golden curve), the cavity absorption increased to 14% (TM polarization), leading to increased heat generation in the cavity because only a small fraction (∼10−10) of the absorbed light is emitted radiatively via Au d-band transitions.22 Polarization-dependent measurements of cavity heating revealed that both bare Si and the Si−Au nanowire were heated to a higher temperature in the TM polarization

active optical phonon of Si via Raman scattering.15 The energy of this phonon mode is ∼521 cm−1 at 295 K and decreases with an increase in temperature (phonon softening),15 which has been demonstrated to provide a robust method for temperature calculation of the system (see Supporting Information).16 Raman spectroscopy on the devices was performed at two different excitation wavelengths (532 and 659.4 nm, spot size ∼1 μm) with the incident electric field either parallel (transverse magnetic, TM polarization) or perpendicular (transverse electric, TE polarization) to the nanowire long axis (Figure 1a, see Supporting Information). Power-dependent Raman measurements on a bare Si nanowire of diameter 100 nm (Figure 1b) show that with an increase in laser power the Raman shift decreased in energy suggesting an increase in temperature of the nanowire. The change in the phonon energy (Raman shift) of the Si nanowire as a function of laser power was fitted with the model proposed by Balkanski et.al15 to obtain the temperature of the system as a function of laser power. Compared to previous reports in metal−semiconductor structures in which indirect estimates of temperatures attained during optical excitation using either the rate of chemical reactions or simulations8,17−19 have been made, use of Si in this work provides a method to directly measure the temperature of the system. The temperatures (Figure 1c) obtained from the Raman spectra of bare Si nanowire show that a peak temperature of ∼500 K (ΔT ∼ 200 K) was achieved in the TM polarization (solid black curve) at a laser intensity of 5.7 × 105 W/cm2. However, Raman measurements performed on the 1840

DOI: 10.1021/acs.nanolett.6b05147 Nano Lett. 2017, 17, 1839−1845

Letter

Nano Letters

Figure 2. Effect of electric-field confinement on heating in bare and Au-coated Si nanowire of diameter 100 nm. (a,b) Spatial distribution of electric field intensity inside the cavity under TM excitation for a bare and a 10 nm Au-coated Si nanowire respectively at 532 nm. (c) Electric field distribution inside a bare Si nanowire under TM excitation at 659.4 nm. (d) Temperature versus laser power for Au-coated and bare Si nanowire under TM polarization at 659.4 nm excitation with a spot size of ∼1 μm.

compared to TE polarization (Figure 1c): a temperature of 500 K (TM) in comparison to 360 K (TE) in bare Si nanowire and 1000 K (TM) compared to 860 K (TE) in Si−Au nanowire. Absorption values from FDTD calculations support the experimentally observed polarization dependence at 532 nm (Figure 1d) with higher absorption in the TM polarization in both bare Si (7.5% TM and 5% TE) and Si−Au (14% TM and 9% TE) nanowires. In order to better understand the mechanism of cavity absorption and heating, spatial distribution of electric field intensity was calculated from the FDTD calculations in bare and Au-coated Si nanowire of diameter 100 nm. Calculations for the bare Si nanowire show that at cavity mode resonance (532 nm excitation in TM polarization), the electric field mode is strongly confined inside the Si nanowire (Figure 2a) with a low mode volume (∼3.5 × 10−4 μm3 inside the cavity versus ∼0.15 μm3 in free space, see Supporting Information) and a 4fold enhancement in field intensity because of high refractive index of Si. The intense electric field in Si leads to high cavity absorption (8%, see Supporting Information) and consequently heating. Calculations for the Au-coated Si nanowire show that the mode volume in this cavity is similar to the bare Si nanowire but the field inside the Si core extends into the Au film (Figure 2b) leading to a significant increase in cavity absorption (14%) that causes stronger heating. To further investigate the effect of mode confinement, evanescent fields in Au film and cavity absorption on heating, measurements were performed on the same 100 nm diameter Si nanowire (bare and after coating Au) at 659.4 nm excitation wavelength. This wavelength was chosen because according to FDTD calculations, electric field is poorly confined by the 100

nm Si nanowire at this wavelength. Calculations show that the peak electric field intensity inside the cavity is approximately three times lower at 659.4 nm (Figure 2c) than at 532 nm (Figure 2a) in a bare Si nanowire in TM polarization. Moreover both Au23 and Si24 are less lossy at 659.4 nm compared to at 532 nm (ε″ (Au) = 2.2 (532 nm) and 1 (659.4 nm), ε″ (Si) = 0.39 (532 nm) and 0.12 (659.4 nm)). These factors lead to lower cavity absorption in bare (0.6%) and in Si−Au nanowires (1.4%) (Figure 1d). Experiments performed on the bare Si nanowire at 659.4 nm revealed that the temperature of bare nanowire was ∼320 K (Figure 2d) at laser power intensity of ∼6 × 105W/cm2, which is in agreement with the low absorption value obtained from calculations. Because there is only a marginal increase in absorption value after coating Au (1.4% versus 0.6%), only a small increase in temperature of the Si−Au nanowire is expected at comparable laser excitation intensities. Surprisingly, a substantial enhancement in heating was still observed and a temperature of ∼550 K was attained in the Si−Au cavity at ∼6 × 105W/cm2 laser intensity. As mentioned earlier, the thin Au layer breaks into small particles of size range 20−30 nm (Figure 1a, right inset) that support LSP modes in the visible wavelength range25 and could play a significant role in cavity heating. To ascertain the contribution of LSPs toward heating, two control experiments were performed. In the first experiment, to prevent LSP excitation at the Si−Au interface, a cavity was fabricated by coating a 150 nm thick Au film on Si nanowire (diameter 100 nm). Because the field penetration depth in Au at 532 nm is ∼50 nm, laser irradiation was done through the glass substrate (Figure 3a, inset). As a result, the electric field would be too weak to excite any LSPs at the air−Au interface 1841

DOI: 10.1021/acs.nanolett.6b05147 Nano Lett. 2017, 17, 1839−1845

Letter

Nano Letters

Figure 3. Role of localized surface plasmons in heating of an Au coated Si nanowire of diameter 100 nm. (a) Temperature versus laser power of a bare and 150 nm Au-coated Si nanowire under TM excitation (532 nm excitation, spot size, 1 μm). Inset shows the cavity cross-section schematic indicating that laser excitation is done through glass. (b) Numerical simulations of absorption spectra of bare and Au-coated Si nanowire cavity in (a) under TM excitation at 532 nm. (c) Numerical simulations of the absorption spectra of a bare and 10 nm Ti-coated Si nanowire of diameter 100 nm under TM and TE excitation. (d) Experimentally observed temperature versus laser power of bare and 10 nm Ti coated Si nanowire of diameter 100 nm under TM and TE excitation at 532 nm wavelength excitation with a spot size of 1 μm.

reached a maximum of ∼650 K at a pump power of ∼5.9 × 105W/cm2 (Figure 3d), which is much lower than in the 10 nm Au coated cavity (1000 K, at ∼5.7 × 105 W/cm2). These observations can be explained via significant contribution of LSPs to heat generation in the Si−Au cavity. Without any LSP activity in Au particles, temperature of the Si−Au cavity should be lower than 650 K (ΔT < 350 K). But contributions from LSPs excited from a strong evanescent field extending into the Au particles increase the cavity heat generation. Moreover because the Au particles in the Au film are separated by an average distance of ∼5 nm (Figure 1a, right inset), the particles can also strongly interact with each other leading to further enhancement of the LSP excitations.12 These factors lead to a drastic increase in cavity heating causing the cavity temperature to increase to ∼1000K (ΔT ∼ 700 K, Figure 1c). Furthermore, as shown by FDTD calculations, when a cavity mode exists at the excitation wavelength the evanescent field intensity in the Au film is enhanced, which should lead to stronger LSP excitations and consequently higher heating as was experimentally observed in the Si nanowire of 100 nm diameter at 532 nm excitation (Figure 1c) where temperature of the cavity increased from 500 to 1000 K (ΔT = 500 K) after coating Au. On the other hand, decreasing the evanescent field strength in the Au film by detuning the cavity mode of the Au-coated Si nanowire (at 654.9 nm) should reduce the strength of LSPs excitation (and cavity absorption) and should lead to lower heat generation as was experimentally observed in the same nanowire with a 659.4 nm excitation (temperature increased from 320 to 550 K, ΔT = 230 K, after coating with Au, Figure 2d). Thus, the FDTD calculations, which treat the Au layer to be a continuous film as opposed to individual particles, explain

thereby decreasing the heat generation in this cavity compared to a cavity with only 10 nm Au film. Raman measurements performed with pump intensity up to ∼6 × 105W/cm2 revealed (Figure 3a) that whereas when the bare Si nanowire was heated to a temperature of 500 K, the temperature of the cavity after coating with 150 nm of Au reached only 380 K, even though the cavity absorption increased 4-fold after coating with Au (27% vs 7%, Figure 3b). This decrease in temperature is because of two reasons: prevention of LSP excitation in the cavity with a 150 nm Au layer and availability of a larger heat sink (thicker Au film) to dissipate heat. In the second experiment, to prevent the formation of LSPs in the thin metallic film while maintaining the same cavity structure and thickness of metal, a 10 nm thick titanium (Ti) layer (instead of Au) was coated on a Si nanowire of diameter 100 nm. While Ti does not support any plasmon modes at 532 nm,24 it is more lossy than Au23 at 532 nm (ε″ (Ti) = 9.3 versus ε″ (Au) = 2.2). This should lead to higher absorption in the Ti cavity compared to the Au cavity, as is also shown by FDTD calculations (Figure 3c), leading to higher heating if LSPs were not contributing to heat generation in the Au cavity. Because the thermal conductivity of Ti (21.9 W/m/K)26 is much lower than that of Au (317 W/m/K)26 and both have similar specific heat capacities (C(Ti) = 2.35 J/cm3/K, C(Au) = 2.49 J/cm3/K),26 temperature in Ti-coated nanowire is expected to be higher than in the Au-coated Si nanowire. In other words, if LSPs do not contribute to heating in the Au cavity, then under similar experimental conditions in a Ti cavity, temperature of 1000 K should be attained at intensities much lower than 5.7 × 105W/cm2 at 532 nm excitation. But experiments revealed that temperature in the Ti-coated cavity 1842

DOI: 10.1021/acs.nanolett.6b05147 Nano Lett. 2017, 17, 1839−1845

Letter

Nano Letters

Figure 4. Study of cavity mode resonance of bare and Au-coated Si nanowire of 240 nm diameter and enhanced photocatalysis using Au coated Si nanowire cavities. (a) Simulated absorption spectra of a bare and 10 nm Au-coated Si nanowire of diameter 240 nm under TM and TE excitation indicating the presence of modes at 659.4 nm pump wavelength. (b) Experimentally measured temperature versus laser power of bare and 10 nm Au-coated Si nanowire of diameter 240 nm under TM and TE excitation at 659.4 nm wavelength with a spot size of ∼1 μm. (c) Schematic of the photocatalysis setup. Inset shows the schematic of the Au-coated Si nanowire devices with TiO2 nanorods fabricated on a glass substrate. Au−TiO2 devices do not contain Si nanowires and Si−Au cavity does not have the TiO2 film on top of Au. (d) H2 evolution volume during catalytic photoreforming of ethanol using Au-coated Si nanowire cavities dispersed on a glass substrate with TiO2 nanorod catalysts while being irradiated with a 300 W Hg/Xe lamp, kept at a distance of 12.5 cm from the center of the photoreactor.

predicted from FDTD calculations. However, the experimentally measured temperatures in this cavity (Si nanowire diameter 240 nm) are much higher than the temperatures attained in the cavity with Si nanowire of diameter 100 nm, which is not resonant at 659.4 nm (Figure 2d) and can be attributed to the higher absorption in the cavity and stronger LSP excitation in the Au particles of the Au film highlighting the tunability of cavity heat generation. Thus, the Si−Au engineered cavities simultaneously take advantage of a high refractive index dielectric Si, and the enhanced cavity mode resulting from the Au shell and LSPs of the Au particles in the shell. Moreover the cavity mode resonances are tunable by changing the Si nanowire diameter and are unaffected by changes in shape and size of Au particles in the film. The strong plasmonic activity in the Au layer is also attractive for enhancing the rate of various chemical reactions.11 In this work, we take advantage of the highly engineered plasmonic devices to drive photothermochemical transformations for conversion of ethanol to generate H2 using the combined effect of light and heat. Photocatalytic harvesting of H2 from renewable sources such as alcohols is important for its sustainable production because it is a crucial industrial building block and a promising clean fuel.27 This is attained by reducing protons to H2 and oxidizing carbon-containing compounds to CO2 via photogenerated electrons and holes, respectively, in a semiconductor catalyst, a process called photoreforming. It has been shown that use of plasmonic nanostructures such as Au

the importance of cavity mode at excitation wavelength to excite stronger LSPs in the Au particles of the film for enhanced heat generation. The broadband and polarization independent nature of these LSP excitations, because of the random shape, size, and orientation of the Au particles in the film, makes the cavity heating less anisotropic despite the geometrical anisotropy of the Au coated Si nanowire devices. Finally, to show the wavelength tunability of Si nanowire cavity device, experiments were conducted on a Si nanowire of 240 nm diameter with and without a 10 nm Au film. This diameter was chosen because while both Au and Si are less lossy at 659.4 nm than at 532 nm, a cavity mode exists in this nanowire for 659.4 nm excitation (Figure 4a) which should cause high absorption in bare and Si−Au nanowires as well as strong LSP excitation in the Au particles in the Si−Au nanowires. Experiments with a 659.4 nm pump in TM polarization indicated that the bare Si nanowire was heated to a temperature of ∼400 K and after coating Au the cavity attained a temperature of 900 K (Figure 4b) at pump intensities of ∼5.3 × 105 and ∼6.7 × 105 W/cm2, respectively. The temperatures attained in the TE polarization for bare (380 K) and Au-coated (760 K) Si nanowire at the same pump intensities are lower than in the corresponding cavities in TM polarization. These results are in accordance with lower absorption values in TE polarization (2% and 10% in bare and Au coated nanowire respectively) than in TM polarization (3% and 13% in bare and Au-coated nanowire, respectively) 1843

DOI: 10.1021/acs.nanolett.6b05147 Nano Lett. 2017, 17, 1839−1845

Nano Letters



leads to an improvement in the rate of photocatalytic reactions.28,29 To show enhanced photoreforming rates by cavity enhanced plasmonic structures, we fabricated devices for photoreforming (schematic of the photocatalysis set up in Figure 4c) of ethanol under continuous UV/vis illumination from a 300 W Hg/Xe lamp (C2H5OH + 3H2O →2CO2 + 6H2). These devices were fabricated by drop-casting a solution of titania nanorods in hexanes/octane (see Supporting Information) to form a layer of colloidal TiO2 nanorods (length ∼40 nm and diameter ∼5 nm) on top of the Au-coated Si nanowires dispersed on a glass substrate with an area coverage density of ∼1% (Figure 4c, inset).30 By tuning the concentration and volume of the drop-casting solution, a 5−7 nm thick layer of TiO2 nanorods was deposited on the substrate, on top of the Si−Au cavities. The thin, porous titania film increases reactant transport to the active sites that are located at the Au−TiO2 interface.31 Analysis of photocatalytic reactions under various conditions revealed that films containing the Si/Au/TiO2 cavity catalyst showed superior H2 production rates compared to films that only had Au/TiO2 catalyst (Figure 4d), showing the importance of cavityenhanced plasmonics in driving the reaction rate. Rates of photocatalytic hydrogen production, calculated by fitting the data of H2 production versus time (Figure 4d) for the first 5 h to a straight line yielded an intrinsic rate of H2 production of 90 μmol g−1 min−1 versus 66 μmol g−1 min−1 for Si/Au/TiO2 and Au/TiO2 samples, respectively, demonstrating that a significant activity increase of ∼40% was obtained by introducing Aucoated Si nanowire cavities with only ∼1% area coverage. Furthermore, while the Au/TiO2 catalyst progressively deactivated under reaction conditions, as evidenced by a change in the slope of H2 production versus time curve after ∼5 h, the Si nanowire/Au/TiO2 sample produced H2 at a constant rate for as long as three consecutive days after taking the evaporation of the solution into account. Experiments with only Au/Si cavity (no TiO2) did not show any H2 production activity even after several hours under illumination (Figure 4c) confirming the role of titania as a catalyst in the photoreforming reaction. Thus, our results show that cavity-enhanced plasmonics provide a new, easy and stable way to significantly enhance the activity of photocatalyst for H2 production from ethanol compared to other H2-evolving systems.32 The rate of photocatalysis can be further improved by improving the density of devices and optimization of their geometry and cavity resonances for optimized reaction rates. In conclusion, we have demonstrated that by engineering the cavity modes of Si nanowire−Au shell cavity, it is possible to obtain highly intense optical modes inside the cavity, which can be tuned with Si nanowire diameter. This leads to increased cavity absorption as well as stronger excitation of LSPs in the Au particles of the thin-film, which causes significant cavity heating with temperatures reaching ∼1000 K. These cavities also increase the H2 generation rate in photoreforming reaction of ethanol by ∼40% with only ∼1% areal coverage. Our work shows that highly engineered nanoscale plasmonic cavity structures are promising for energy conversion applications that can be further improved via shape and optical engineering of nanostructures.33,34

Letter

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.nanolett.6b05147. Methods for nanowire and cavity fabrication, optical measurements, temperature and cavity mode numerical calculations, and photocatalysis (PDF)



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. ORCID

Ritesh Agarwal: 0000-0002-1289-4334 Present Address

(M.C.) Department of Chemical Engineering, Stanford University, Stanford, CA 94305. Author Contributions

D.A., C.O.A., and R.A. conceived the concept. D.A. and R.A. designed the optical experiments. D.A. and M.C. designed the photocatalysis experiments. J.Y. synthesized silicon nanowires. D.A. fabricated the nanoscale cavities and performed optical experiments and numerical calculations. M.C. performed photocatalysis experiments. D.A., C.O.A., M.C., M.L.R., C.B.M., and R.A. analyzed the results. D.A., M.C., and R.A. wrote the manuscript. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS We thank the University Research Foundation at Penn for their support. This work was performed in part at CINT, a United States Department of Energy, Office of Basic Energy Sciences User Facility at Los Alamos National Laboratory (Contract DEAC52-06NA25396) and Sandia National Laboratories (Contract DE-AC04-94AL85000).



REFERENCES

(1) Maier, S. A. Plasmonics: Fundamentals and Application; Springer: New York, 2007. (2) Liu, W.; Lee, B.; Naylor, C. H.; Ee, H.-S.; Park, J.; Johnson, A. T. C.; Agarwal, R. Nano Lett. 2016, 16, 1262−1269. (3) Lee, B.; Park, J.; Han, G. H.; Ee, H.-S.; Naylor, C. H.; Liu, W.; Johnson, A. T. C.; Agarwal, R. Nano Lett. 2015, 15, 3646−3653. (4) Cho, C.-H.; Aspetti, C. O.; Park, J.; Agarwal, R. Nat. Photonics 2013, 7, 285−289. (5) Cho, C.-H.; Aspetti, C. O.; Turk, M. E.; Kikkawa, J. M.; Nam, S.; Agarwal, R. Nat. Mater. 2011, 10, 669−675. (6) Atwater, H. A.; Polman, A. Plasmonics for improved photovoltaic devices. Nat. Mater. 2010, 9, 205−213. (7) Mubeen, S.; Lee, J.; Singh, N.; Kramer, S.; Stucky, G. D.; Moskovts, M. Nat. Nanotechnol. 2013, 8, 247−251. (8) Adleman, J. R.; Boyd, D. A.; Goodwin, D. G.; Psaltis, D. Nano Lett. 2009, 9, 4417−4423. (9) Jain, P. K.; El-Sayed, I. H.; El-Sayed, M. A. Nano Today 2007, 2, 18−29. (10) Christopher, P.; Xin, H.; Linic, S. Nat. Chem. 2011, 3, 467−472. (11) Baffou, G.; Quidant, R. Laser Photon. Rev. 2013, 7, 171−187. (12) Wang, H.; Levin, C. S.; Halas, N. J. J. Am. Chem. Soc. 2005, 127, 14992−14993. (13) Cao, L.; Barsic, D. N.; Guichard, A. R.; Brongersma, M. L. Nano Lett. 2007, 7, 3523−3527. 1844

DOI: 10.1021/acs.nanolett.6b05147 Nano Lett. 2017, 17, 1839−1845

Letter

Nano Letters (14) Aspetti, C. O.; Cho, C. H.; Agarwal, R.; Agarwal, R. Nano Lett. 2014, 14, 5413−5422. (15) Balkanski, M.; Wallis, R.; Haro, E. Phys. Rev. B: Condens. Matter Mater. Phys. 1983, 28, 1928−1934. (16) Gupta, R.; Xiong, Q.; Adu, C. K.; Kim, U. J.; Eklund, P. C. Nano Lett. 2003, 3, 627−631. (17) Fasciani, C.; Alejo, C. J. B.; Grenier, M.; Netto-Ferreira, J. C.; Scaiano, J. C. Org. Lett. 2011, 13, 204−207. (18) Haas, K. M.; Lear, B. J. Nanoscale 2013, 5, 5247−51. (19) Fang, Z.; Zhen, Y.-R.; Neumann, O.; Polman, A.; Abajo, F. J. G.; Nordlander, P.; Halas, N. J. Nano Lett. 2013, 13, 1736. (20) Sze, S. M.; Kwok, K. N. Physics of semiconductor devices, 3rd ed.; J. Wiley & Sons Inc.: New York, 2006. (21) Liang, D.; Bowers, J. E. Nat. Photonics 2010, 4, 511−517. (22) Lin, A.; Son, D. H.; Ahn, I. H.; Song, G. H.; Han, W. T. Opt. Express 2007, 15, 6374−6379. (23) Johnson, P. B.; Christry, R. W. Phys. Rev. B 1972, 6, 4370−4379. (24) Palik, E. D. Handbook of Optical Constants of Solids, 1st ed.; Academic Press: Burlington, MA, 1998. (25) Bardhan, R.; Mukherjee, S.; Mirin, N. A.; Levit, S. D.; Nordlander, P.; Halas, N. J. J. Phys. Chem. C 2010, 114, 7378−7383. (26) Lide, D. R. CRC Handbook of Chemistry and Physics, 84th ed.; CRC Press, Taylor & Francis Group: Boca Raton, FL, 2003. (27) Barreca, D.; Carraro, G.; Gombac, V.; Gasparotto, A.; Maccato, C.; Fornasiero, P.; Tondello, E. Adv. Funct. Mater. 2011, 21, 2611− 2623. (28) Zhang, Z.; Wang, W.; Gao, E.; Sun, S.; Zhang, L. J. Phys. Chem. C 2012, 116, 25898−25903. (29) Christopher, P.; Xin, H.; Marimuthu, A.; Linic, S. Nat. Mater. 2012, 11, 1−7. (30) Cargnello, M.; Montini, T.; Smolin, S. Y.; Priebe, J. B.; Jaén, J. J. D.; Doan-Nguyen, V. V. T.; McKay, I. S.; Schwalbe, J. A.; Pohl, M.; Gordon, T. R.; Lu, Y.; Baxter, J. B.; Brückner, A.; Fornasiero, P.; Murray, C. B. Proc. Natl. Acad. Sci. U. S. A. 2016, 113, 3966−3971. (31) Duonghong, D.; Borgarello, E.; Graetzel, M. J. Am. Chem. Soc. 1981, 103, 4685−4690. (32) Ma, Y.; Wang, X.; Jia, Y.; Chen, X.; Han, H.; Li, C. Chem. Rev. 2014, 114, 9987−10043. (33) Day, R. W.; Mankin, M. N.; Gao, R.; No, Y.-S.; Kim, S.-K.; Bell, D. C.; Park, H.-G.; Lieber, C. M. Nat. Nanotechnol. 2015, 10, 345− 352. (34) Luo, Z.; Jiang, Y.; Myers, B. D.; Isheim, D.; Wu, J.; Zimmerman, J. F.; Wang, Z.; Li, Q.; Wang, Y.; Chen, X.; Dravid, V. P.; Seidman, D. N.; Tian, B. Science 2015, 348, 1451−1455.

1845

DOI: 10.1021/acs.nanolett.6b05147 Nano Lett. 2017, 17, 1839−1845