Directional Excitation of Surface Plasmon Polaritons via Molecular

Jun 11, 2019 - Using the molecular tunneling junctions, we provide the first direct ... summary of the directivity analysis; NEXAFS characterization (...
0 downloads 0 Views 3MB Size
Letter Cite This: Nano Lett. 2019, 19, 4634−4640

pubs.acs.org/NanoLett

Directional Excitation of Surface Plasmon Polaritons via Molecular Through-Bond Tunneling across Double-Barrier Tunnel Junctions Wei Du,†,⊥ Yingmei Han,†,⊥ Hongting Hu,† Hong-Son Chu,‡ Harshini V. Annadata,† Tao Wang,†,¶ Nikodem Tomczak,†,§ and Christian A. Nijhuis*,†,∥ †

Department of Chemistry, National University of Singapore, 3 Science Drive 3, 117543 Singapore Department of Electronics and Photonics, Institute of High Performance Computing, A*STAR (Agency for Science, Technology and Research), 1 Fusionopolis Way, #16-16 Connexis, 138632 Singapore § Institute of Materials Research and Engineering, A*STAR (Agency for Science, Technology and Research), 2 Fusionopolis Way, Innovis, 138634 Singapore ∥ Centre for Advanced 2D Materials and Graphene Research Centre, National University of Singapore, 6 Science Drive 2, 117546 Singapore

Downloaded via KEAN UNIV on August 2, 2019 at 15:11:53 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.



S Supporting Information *

ABSTRACT: Directional excitation of surface plasmon polaritons (SPPs) by electrical means is important for the integration of plasmonics with molecular electronics or steering signals toward other components. We report electrically driven SPP sources based on quantum mechanical tunneling across molecular double-barrier junctions, where the tunneling pathway is defined by the molecules’ chemical structure as well as by their tilt angle with respect to the surface normal. Self-assembled monolayers of S(CH2)nBPh (BPh = biphenyl, n = 1−7) on Au, where the alkyl chain and the BPh units define two distinct tunnel barriers in series, were used to demonstrate and control the geometrical effects. The tilt angle of the BPh unit with respect to the surface normal depends on the value of n, and is 45° when n is even and 23° when n is odd. The tilt angle of the alkyl chain is fixed at 30° and independent of n. For values of n = 1−3, SPPs are directionally launched via directional tunneling through the BPh units. For values of n > 3, tunneling along the alkyl chain dominates the SPP excitation. Molecular level control of directionally launching SPPs is achieved without requiring additional on-chip optical elements, such as antennas, or external elements, such as light sources. Using the molecular tunneling junctions, we provide the first direct experimental demonstration of molecular double-barrier tunneling junctions. KEYWORDS: Molecular junction, surface plasmon, tunneling, double barrier, tilt angle

P

relatively bulky light sources. Tunneling electrons can directly excite SPPs in a single step, eliminating the need for external light sources and elements to overcome the mismatch in momentum between light in free space and SPPs,5−14 with up to 1−2% excitation efficiencies.5,6 Using junctions based on scanning tunneling microscopy (STM), directional SPP excitation has been demonstrated by positioning the STM tips near the edge of a metallic cavity20,21 or nanoparticle.22 Directional light emission from an optical antenna electrically

lasmonics holds promise to bridge the gap between nanoelectronics and photonics for next-generation information technologies, such as stacked high-bandwidth memory or three-dimensional integrated circuits.1−4 To integrate plasmonics with nanoelectronics, it is important to develop electrical excitation sources that directly convert electrical signals into plasmonic signals.5−14 In this context, electronic−plasmonic transducers that directionally launch propagating surface plasmon polaritons (SPPs) by electrical means are highly desirable. So far, most of the strategies for directional launching of SPPs rely on optical excitation by employing carefully designed metasurfaces15,16 or other optical elements,17−19 but these approaches are not suitable for onchip applications as they are diffraction-limited and require © 2019 American Chemical Society

Received: April 22, 2019 Revised: June 11, 2019 Published: June 11, 2019 4634

DOI: 10.1021/acs.nanolett.9b01665 Nano Lett. 2019, 19, 4634−4640

Letter

Nano Letters driven by a tunnel junction has also been reported.14 In recent years, “molecular electronic plasmonics”23 emerged where the molecules inside tunnel junctions play an active role and determine both the electronic and plasmonic properties of the system. Molecular tunneling junctions have been used, for example, to tune the tunneling charge transfer plasmon mode24−29 and to excite9,13 or detect SPPs.30,31 In principle, the shape of the tunneling barrier of molecular junctions can be readily changed. For example, molecules with a conductive conjugated and relatively insulating aliphatic part should result in molecular double-barrier junctions,32−40 but it has not been experimentally verified whether the charge carriers probe two distinct barriers or an average barrier because in two-terminal junctions only the total currents across the junctions can be measured and interpretation of data therefore relies on theoretical models with associated simplifications and assumptions. Here, we demonstrate directional SPP launching by changing the shape of a molecular double-barrier junction by controlling the tilt angle of an active unit via odd−even effects. Our results also experimentally confirm, for the first time, that the molecular double-barrier picture holds. Figure 1 shows the

molecular junctions schematically. The large-area molecular junctions consist of self-assembled monolayers (SAMs) of S(CH2)nBPh with n = 1−7, where BPh stands for a biphenyl unit; these SAMs have been well-characterized in the literature.41−43 In these SAMs, n determines the tilt angle of the BPh units, αBPh (in degrees), which is larger when n is even (45°) than when n is odd (23°).41 In other words, these SAMs display odd−even effects in the value of αBPh as a function of n. This type of odd−even effect is driven by the fact that the gold−sulfur−carbon bond angle is fixed at 104°; this has been well-documented.44 We confirmed this odd−even effect in our SAMs with angle-dependent near-edge X-ray absorption fine structure (NEXAFS) spectroscopy at the C K-edge (see Supporting Information). Recently, odd−even effects in the tunneling rates of molecular junctions have been demonstrated where J could be modulated by a factor of 3−5.45−47 Figure 1 also indicates the tilt angle αalk (in degrees) of the alkyl chain with respect to the surface normal, which is 30° for aliphatic SAMs on Au.48 In principle, molecular junctions are interesting as the shape of the tunneling barrier can be precisely tuned by simply changing the chemical structure of the molecule.49,50 Often, the mechanism of charge transport across molecular junctions is approximated with the general tunneling equation J = J0 e−dmolβmol with βmol =

jij 2me Ψ zyz jj zz k ℏ {

(1) 2

where J is the current density (in A/cm ), J0 is the preexponential factor (in A/cm2) which can be seen as the effective contact resistance R0 (=1/J0), dmol is the length of the molecule, βmol is the tunneling decay coefficient along the molecule, Ψ is the tunneling barrier height, me is the effective electron mass, and ℏ is the reduced Planck’s constant. Equation 1, however, fails to capture the shape of the tunneling barrier. It has been suggested that molecular junctions can be seen as “series tunneling junctions” where each molecular component forms a distinct part of the tunnel junction.32−40 According to this view of molecular junctions, here, the alkyl chain and BPh units pose two distinct tunneling barriers in series where the tunneling barrier widths are defined by the lengths of the alkyl chain (which depends on n), dalk, and the BPh unit, dBPh, and tunneling barrier heights are defined by the alkyl chain, φalk, and BPh unit, φBPh (Figure 1c). Thus, in principle, the junctions are double-barrier junctions with the corresponding tunneling decay coefficients along the alkyl chain, βalk, and BPh units, βBPh, given by eq 2 J = J0 e−(dBPhβBPh + dalkβalk )

(2)

In general, the aromatic part poses a lower φ than the aliphatic molecules (as indicated in Figure 1c); consequently, aromatic molecules have typical values of β in the range of 0.2−0.4 Å−1,51−54 whereas aliphatic molecules have β of 0.8 Å−1.55 In addition, others have suggested that intermolecular tunneling (where charges tunnel from one molecule to another) may be important for molecules with large tilt angles.56−58 In case the double-barrier picture holds, the tunneling rates are determined by the BPh units for small values of n, and the red arrows in Figure 1a indicate that the tunneling direction should be determined by the tilt angle of the BPh units, which, in turn, should influence the directional plasmon launching as indicated by the blue arrows. For large values of n, the most resistive part of the junction is defined by the alkyl chain and

Figure 1. Schematic illustrations of the Au−S(CH2)nBPh//EGaIn junctions where “−” indicates the metal−thiolate bond and “//” the noncovalent interface of the SAMs with the EGaIn electrode. (a) Orientation of the BPh unit on Au with an odd and even number of carbon atoms for n = 1−3. The red arrows indicate the direction of the tunneling electrons defined by the tilt angle of the BPh unit (αBPh), and the blue arrows indicate the directional SPP launching. (b) For junctions with n > 3, the tunneling direction is defined by the tilt angle of the alkyl unit (αalk). (c) Energy level diagram of the double-barrier junction where the two barrier widths, dalk and dBPh, and heights, φalk and φBPh, are defined by the alkyl chain and BPh units. 4635

DOI: 10.1021/acs.nanolett.9b01665 Nano Lett. 2019, 19, 4634−4640

Letter

Nano Letters

junctions with neven (−1.34 ± 0.48 A/cm2). These results show that the origin of the odd−even effect is due to the difference in the contact resistance. The obtained values of β are lower than that of alkanethiolate SAMs55 but higher than that of aromatic SAMs.51−54 We believe that the mismatch in diameter between BPh units and alkyl chains weakens the van der Waals interactions and, given the short length of the alkyl chains, results in liquid-like packing and a lower observed β value. We studied the plasmonic properties of the junctions by leakage radiation microscopy using a previously reported method9 (see Supporting Information section S5). Although most SPPs decay thermally, some SPPs decay radiatively due to defects in the electrode materials. We studied the light emission from these junctions in the far-field through the bottom electrode (30 nm thick gold electrode) using an oil immersion objective with numerical aperture (NA) of 1.49. Figure 3a shows the optical micrograph of a junction with a

the tunneling direction, and associated SPP excitation is determined by αalk, as indicated in Figure 1b. Therefore, this system provides us the opportunity to experimentally determine whether the double-barrier junction picture holds. The synthesis of the HS(CH2)nBPh precursors (section S1), fabrication of the bottom electrodes (section S2), SAM formation (section S3), and junction fabrication (section S4) are described in detail in the Supporting Information. Briefly, we formed the SAMs on ultraflat template-stripped Au with a thickness of 200 nm, which were then contacted using the wellknown EGaIn technique (section S4).9,59 This technique uses a moldable, non-Newtonian liquid metal of a eutectic gallium− indium alloy covered with a self-limiting, highly conductive native oxide, which gives the alloy non-Newtonian properties. The EGaIn technique yields SAM-based junctions in >90% yields in working devices with high reproducibility.59,60 We measured the J(V) characteristics of the Au−S(CH2)nBPh// EGaIn junctions with n = 1−7 by recording statistically large numbers of J(V) curves (typically 400 J(V) curves obtained from 20 to 25 different junctions). For each measured voltage, we plotted all measured values of log10|J| in histograms (see Figure S22 for histograms of J at +0.5 V) against which we fitted Gaussians to determine the Gaussian mean of log10|J|, ⟨log10|J|⟩G, which was used to reconstruct the ⟨log10|J|⟩G versus V curves shown in Figure 2a.

Figure 3. (a) Optical micrograph of the junction with a S(CH2)2BPh SAM imaged through the semitransparent Au electrode and (b) corresponding EMCCD light emission image recorded at −2.0 V with a 100× oil immersion objective (NA = 1.49) and 10 s integration time. (c) UV/vis spectra recorded from the entire junction area as a function of the applied bias. (d) High energy photon cutoff as a function of the applied bias; the dashed line indicates the quantum cutoff.

S(CH2)2BPh SAM with the EGaIn stabilized in a through-hole in a transparent rubber.59 Although the EGaIn forms noninvasive contacts to the SAM, the effective contact area is small relative to the geometrical contact area defined by the diameter of the through-hole. The corresponding light emission image (Figure 3b, recorded at −2.0 V) shows that the photons originate from discrete emission spots. These observations are in agreement with our previous findings where we concluded that the effective electrical contact area of EGaIn is about 106 lower than the geometrical contact area.61 Figure 3c,d shows the UV/vis spectra and the corresponding cutoff photon energy (at the high-frequency end) as a function of the applied bias. The spectra show a broad peak which blue shifts with increasing bias following the quantum cutoff law hν ≤ eVbias (where e is the electron charge and ν is the frequency of the photon), in agreement with previous reports.5−14

Figure 2. (a) J(V) characteristics of the Au−S(CH2)nBPh//EGaIn junctions. (b) Plot of ⟨log10|J|⟩G versus n at 0.5 V. The error bar in a represents the 95% confidence intervals, and the error bar in b represents the natural logarithm standard deviation. The red solid lines are fits to eq 1.

Figure 2b shows the plot of ⟨log10|J|⟩G at 0.5 V as a function of n, which clearly shows an odd−even effect (black dashed line). The red solid lines are fits to eq 1, from which we determined the value of βodd of 0.61 ± 0.09 n−1 and βeven of 0.79 ± 0.26 n−1. Here, the error of βeven is much larger than that of βodd, which is due to fewer data points for even carbon number than odd carbon number. The value of ⟨log10|J0|⟩G for junctions with nodd (−2.55 ± 0.18 A/cm2) is lower than that of 4636

DOI: 10.1021/acs.nanolett.9b01665 Nano Lett. 2019, 19, 4634−4640

Letter

Nano Letters

Figure 4. (a) Experimentally defocused images of junctions with n = 1−7 recorded at −2.1 V bias. (b) Simulated defocused patterns based on the images shown in a with the dipole orientation indicated as [θ;Φ]. (c) Definitions of orientation of the dipole emitter used in the simulations based on two angles: out-of-plane angle θ and in-plane angle Φ. The z-axis corresponds to the optical axis of the microscope. (d) Plot of θ as a function of n. The error bars represent the 95% confidence intervals.

The values of both θ and Φ were derived from the experimentally defocused images using simulated defocused images with the values of θ and Φ indicated for each individual emission spot (Figure 4b and Supporting Information section S6). Figure 4d shows a statistical analysis of θ based on 109− 127 defocused spots (from 6 to 17 junctions) for each value of n (see Supporting Information, section S7, Table S2, and Figure S24). In our analysis, we only used non-overlapping spots with good signal-to-noise ratio (see examples in Figures 4a,b). For the junctions with n = 1−3, we observed a clear odd−even effect in θ of 22 ± 1° and 25 ± 1° for n = 1 and 3, but 44 ± 3° for n = 2. These values are in very good agreement with literature values for αBPh of 23 ± 7° for n = odd and 45 ± 10° for n = even measured by NEXAFS reported in literature41 and our own NEXAFS data shown in the Supporting Information. The odd−even effect fades quickly and is weak

To measure whether the SPPs are excited in certain directions, we recorded defocused images of the light emission from the junctions by moving the objective away from the image plane by ∼1 μm. Figure 4a shows representative experimentally defocused images recorded from junctions with n = 1−7 at a bias of −2.1 V. The defocused emission spots have a bright spot in the middle, and their symmetry depends on the value of n. Figure 4c defines the orientation of the emitter by two angles: the out-of-plane angle θ and the inplane angle Φ. The value of θ is directly related to the molecular tilt angles αalk and αBPh (determined through NEXAFS; see Table S4) and, as we show below, the relative lengths of the molecular components, that is, the values of dalk and dBPh. The value of Φ cannot be controlled in our experiments and will be different for different domains of the SAMs due to the polycrystalline nature of the Au electrode. 4637

DOI: 10.1021/acs.nanolett.9b01665 Nano Lett. 2019, 19, 4634−4640

Letter

Nano Letters

Figure 5. (a) Cross section intensity profiles of the defocused patterns obtained for junctions with n = 2 (left) and n = 3 (right); the dashed line indicates Φ. Insets show the corresponding defocused images from which the profiles were taken. (b) Directivity D as a function of n. The error bars represent the 95% confidence intervals.

for n = 4−7, where the value of θ is close to 30°. These observations can be explained as follows. (i) The odd−even effect in θ follows the odd−even effect in αBPh for junctions with n = 1−3; this observation indicates that the BPh unit is the dominant tunnel barrier in this regime. (ii) The value of θ is weakly dependent on n for junctions with n ≥ 4 and close to value of αalk of 30° as reported in literature48 (and in agreement with our NEXAFS data, Figure S26); this observation indicates that the alkyl chain is the dominant tunneling barrier in this regime. (iii) The transition in the value of θ at n = 4 indicates a transition in the dominant tunneling barrier which can be directly related to the relative tunneling efficiencies along the two molecular components (Figure 1c and eq 2). These observations indicate that through-molecular bond tunneling is the dominant mechanism of charge transport across the junctions and show that the momentum of the charge carrier is preserved in the momentum of the plasmons. Based on these results, we conclude that tunneling through molecular bond dominates over intermolecular tunneling even for SAMs with large tilt angles. The above results confirm that the tunnel junctions can be described as double-barrier junctions. To be more quantitative, we estimate the value of n at which the transition of the dominant tunneling barrier occurs as follows. According to eq 2, the transition at which the alkyl chain or the BPh is the dominant barrier is given by e−dalkβalk = e−dBPhβBPh. Using previously reported tunneling decay coefficients of βalk = 0.8 Å−1, βBPh = 0.3 Å−1, and dBPh = 8 Å, the value of dalk can be estimated at which this transition occurs. The estimated transition length is dalk = 3 Å, which corresponds to n = 3 and is in good agreement with the experimental data. Therefore, for n ≤ 3, the dominant tunnel barrier is defined by the BPh unit and SPP excitation follows an odd−even effect in αBPh, whereas for n ≥ 4, the dominant tunnel barrier is the alkyl chain and the direction of SPP excitation is determined by αalk (30°). The directivity of the SPP excitation can also be obtained from the defocused images by analyzing the cross section intensity profiles of the defocused spots. Figure 5a shows representative profiles along the direction of the in-plane angle Φ for junctions with n = 2 and 3. Based on the cross section intensity profile, we define the directivity of plasmon excitation D as

D=

IL − IR IL + IR

that IL > IR). Based on the simulated defocused patterns (Figure S23), the value of D increases with θ in the range of 0−65°. We note that, for a large value of θ, D decreases with increasing θ, and thus this method cannot be used to determine directional plasmon excitation. In this study, however, the highest tilt angle is 45° for αBPh with n = 2. Figure 5b shows the experimentally obtained values of D for each value of n (see Supporting Information section S8, Table S3, and Figure S25) from which we can see that n = 2 with θ of 44° gives the highest D, n = 1 and 3 with θ of 22−25° gives the lowest D, and n = 4−7 with θ close to 30° gives intermediate values of D. In summary, we studied the SPP excitation in a doublebarrier molecular tunnel junctions based on S(CH2)nBPh SAMs which gave the following three new insights that are interesting to both the plasmonics and molecular electronics communities. (i) The direction of SPP excitation can be controlled by changing the tunneling direction which, in turn, depends on the tilt angle of the molecular components with respect to the surface normal. (ii) Through-molecular bond tunneling dominates over intermolecular tunneling even for highly tilted SAMs with tilt angles of 45°. (iii) Junctions with SAMs of S(CH2)nBPh can be treated as double-barrier junctions where the aliphatic and aromatic parts pose distinct tunneling barriers. We believe that these results are interesting toward integration of plasmonics with nanoelectronics because SPPs can be excited unidirectionally by quantum mechanical tunneling junctions without the need for complex optical elements.



ASSOCIATED CONTENT

* Supporting Information S

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.nanolett.9b01665.



(3)

Synthesis of the molecules; preparation of the AuTS surface; preparation of the SAMs; formation and electrical characterization of the junctions; optical characterization of the junctions; simulation of the defocused images; summary of the θ angle analysis; summary of the directivity analysis; NEXAFS characterization (PDF)

AUTHOR INFORMATION

Corresponding Author

where IL and IR are the maximum intensities of the left and right lobes of the defocused spot (here, we always used Φ such

*Email: [email protected]. 4638

DOI: 10.1021/acs.nanolett.9b01665 Nano Lett. 2019, 19, 4634−4640

Letter

Nano Letters ORCID

(21) Dong, Z.; Chu, H. S.; Zhu, D.; Du, W.; Akimov, Y. A.; Goh, W. P.; Wang, T.; Goh, K. E. J.; Troadec, C.; Nijhuis, C. A.; Yang, J. K. W. ACS Photonics 2015, 2, 385−391. (22) Le Moal, E.; Marguet, S.; Rogez, B.; Mukherjee, S.; Dos Santos, P.; Boer-Duchemin, E.; Comtet, G.; Dujardin, G. Nano Lett. 2013, 13, 4198−4205. (23) Wang, T.; Nijhuis, C. A. Appl. Mater. Today 2016, 3, 73−86. (24) Zhu, W.; Esteban, R.; Borisov, A. G.; Baumberg, J. J.; Nordlander, P.; Lezec, H. J.; Aizpurua, J.; Crozier, K. B. Nat. Commun. 2016, 7, 11495. (25) Tan, S. F.; Wu, L.; Yang, J. K. W.; Bai, P.; Bosman, M.; Nijhuis, C. A. Science 2014, 343, 1496−1499. (26) Cha, H.; Yoon, J. H.; Yoon, S. ACS Nano 2014, 8, 8554−8563. (27) Jung, H.; Cha, H.; Lee, D.; Yoon, S. ACS Nano 2015, 9, 12292−12300. (28) Ciracì, C.; Hill, R. T.; Mock, J. J.; Urzhumov, Y.; FernándezDomínguez, A. I.; Maier, S. A.; Pendry, J. B.; Chilkoti, A.; Smith, D. R. Science 2012, 337, 1072−1074. (29) Hajisalem, G.; Nezami, M. S.; Gordon, R. Nano Lett. 2014, 14, 6651−6654. (30) Conklin, D.; Nanayakkara, S.; Park, T. H.; Lagadec, M. F.; Stecher, J. T.; Chen, X.; Therien, M. J.; Bonnell, D. A. ACS Nano 2013, 7, 4479−4486. (31) Noy, G.; Ophir, A.; Selzer, Y. Angew. Chem., Int. Ed. 2010, 49, 5734−5736. (32) Liao, K. C.; Hsu, L. Y.; Bowers, C. M.; Rabitz, H.; Whitesides, G. M. J. Am. Chem. Soc. 2015, 137, 5948−5954. (33) Hsu, L.; Wu, N.; Rabitz, H. Phys. Chem. Chem. Phys. 2016, 18, 32087−32095. (34) Holmlin, R. E.; Haag, R.; Chabinyc, M. L.; Ismagilov, R. F.; Cohen, A. E.; Terfort, A.; Rampi, M. A.; Whitesides, G. M. J. Am. Chem. Soc. 2001, 123, 5075−5085. (35) Querebillo, C. J.; Terfort, A.; Allara, D. L.; Zharnikov, M. J. Phys. Chem. C 2013, 117, 25556−25561. (36) Holmlin, R. E.; Ismagilov, R. F.; Haag, R.; Mujica, V.; Ratner, M. A.; Rampi, M. A.; Whitesides, G. M. Angew. Chem., Int. Ed. 2001, 40, 2316−2320. (37) Magoga, M.; Joachim, C. Phys. Rev. B: Condens. Matter Mater. Phys. 1999, 59, 16011−16021. (38) Joachim, C. Nat. Nanotechnol. 2012, 7, 620−621. (39) Wang, G.; Kim, T. W.; Lee, H.; Lee, T. Phys. Rev. B: Condens. Matter Mater. Phys. 2007, 76, 205320. (40) Zhang, Y.; Soni, S.; Krijger, T. L.; Gordiichuk, P.; Qiu, X.; Ye, G.; Jonkman, H. T.; Herrmann, A.; Zojer, K.; Zojer, E.; Chiechi, R. C. J. Am. Chem. Soc. 2018, 140, 15048−15055. (41) Rong, H. T.; Frey, S.; Yang, Y. J.; Zharnikov, M.; Buck, M.; Wühn, M.; Wöll, C.; Helmchen, G. Langmuir 2001, 17, 1582−1593. (42) Lee, S.; Puck, A.; Graupe, M.; Colorado, R., Jr; Shon, Y. S.; Lee, T. R.; Perry, S. S. Langmuir 2001, 17, 7364−7370. (43) Cyganik, P.; Buck, M.; Strunskus, T.; Shaporenko, A.; WiltonEly, J. D. E. T.; Zharnikov, M.; Wöll, C. J. Am. Chem. Soc. 2006, 128, 13868−13878. (44) Tao, F.; Bernasek, S. L. Chem. Rev. 2007, 107, 1408−1453. (45) Jiang, L.; Sangeeth, C. S. S.; Nijhuis, C. A. J. Am. Chem. Soc. 2015, 137, 10659−10667. (46) Toledano, T.; Sazan, H.; Mukhopadhyay, S.; Alon, H.; Lerman, K.; Bendikov, T.; Major, D. T.; Sukenik, C. N.; Vilan, A.; Cahen, D. Langmuir 2014, 30, 13596−13605. (47) Baghbanzadeh, M.; Simeone, F. C.; Bowers, C. M.; Liao, K. C.; Thuo, M.; Baghbanzadeh, M.; Miller, M. S.; Carmichael, T. B.; Whitesides, G. M. J. Am. Chem. Soc. 2014, 136, 16919−16925. (48) Love, J. C.; Estroff, L. A.; Kriebel, J. K.; Nuzzo, R. G.; Whitesides, G. M. Chem. Rev. 2005, 105, 1103−1169. (49) Vilan, A.; Aswal, D.; Cahen, D. Chem. Rev. 2017, 117, 4248− 4286. (50) Jeong, H.; Kim, D.; Xiang, D.; Lee, T. ACS Nano 2017, 11, 6511−6548.

Tao Wang: 0000-0003-2250-7569 Christian A. Nijhuis: 0000-0003-3435-4600 Present Address ¶

Institute of Functional Nano & Soft Materials (FUNSOM), Jiangsu Key Laboratory for Carbon-Based Functional Materials & Devices, Soochow University, Suzhou 215123, Jiangsu, China.

Author Contributions ⊥

W.D. and Y.H. contributed equally to this work.

Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS We acknowledge the National Research Foundation (NRF) for supporting this research under Prime Minister’s Office, Singapore, under its medium sized centre programme, and the Competitive Research Programme (CRP) program (NRFCRP17-2017-08). We acknowledge Dr. Dongchen Qi and Dr. Anton Tadich for technical assistance during NEXAFS measurements performed at the soft X-ray beamline of the Australian Synchrotron, which is a part of ANSTO.



REFERENCES

(1) Ozbay, E. Science 2006, 311, 189−193. (2) Brongersma, M. L.; Shalaev, V. M. Science 2010, 328, 440−441. (3) Arakawa, Y.; Nakamura, T.; Urino, Y.; Fujita, T. IEEE Commun. Mag. 2013, 51, 72−77. (4) Sun, Y.; Jiang, L.; Zhong, L.; Jiang, Y.; Chen, X. Nano Res. 2015, 8, 406−417. (5) Qian, H.; Hsu, S.; Gurunatha, K.; Riley, C. T.; Zhao, J.; Lu, D.; Tao, A. R.; Liu, Z. Nat. Photonics 2018, 12, 485−488. (6) Du, W.; Wang, T.; Chu, H. S.; Nijhuis, C. A. Nat. Photonics 2017, 11, 623−627. (7) Kern, J.; Kullock, R.; Prangsma, J.; Emmerling, M.; Kamp, M.; Hecht, B. Nat. Photonics 2015, 9, 582−586. (8) Parzefall, M.; Bharadwaj, P.; Jain, A.; Taniguchi, T.; Watanabe, K.; Novotny, L. Nat. Nanotechnol. 2015, 10, 1058−1063. (9) Du, W.; Wang, T.; Chu, H.-S.; Wu, L.; Liu, R.; Sun, S.; Phua, W. K.; Wang, L.; Tomczak, N.; Nijhuis, C. A. Nat. Photonics 2016, 10, 274−280. (10) Lambe, J.; McCarthy, S. L. Phys. Rev. Lett. 1976, 37, 923−925. (11) Berndt, R.; Gimzewski, J. K.; Johansson, P. Phys. Rev. Lett. 1991, 67, 3796−3799. (12) Bharadwaj, P.; Bouhelier, A.; Novotny, L. Phys. Rev. Lett. 2011, 106, 226802. (13) Ivashenko, O.; Bergren, A. J.; McCreery, R. L. J. Am. Chem. Soc. 2016, 138, 722−725. (14) Gurunarayanan, S. P.; Verellen, N.; Zharinov, V. S.; James Shirley, F.; Moshchalkov, V. V.; Heyns, M.; Van de Vondel, J.; Radu, I. P.; Van Dorpe, P. Nano Lett. 2017, 17, 7433−7439. (15) Lin, J.; Mueller, J. P. B.; Wang, Q.; Yuan, G.; Antoniou, N.; Yuan, X.; Capasso, F. Science 2013, 340, 331−334. (16) Huang, L.; Chen, X.; Bai, B.; Tan, Q.; Jin, G.; Zentgraf, T.; Zhang, S. Light: Sci. Appl. 2013, 2, No. e70. (17) López-Tejeira, F.; Rodrigo, S. G.; Martín-Moreno, L.; GarcíaVidal, F. J.; Devaux, E.; Ebbesen, T. W.; Krenn, J. R.; Radko, I. P.; Bozhevolnyi, S. I.; González, M. U.; Weeber, J. C.; Dereux, A. Nat. Phys. 2007, 3, 324−328. (18) Yang, J.; Xiao, X.; Hu, C.; Zhang, W.; Zhou, S.; Zhang, J. Nano Lett. 2014, 14, 704−709. (19) Liu, Y.; Palomba, S.; Park, Y.; Zentgraf, T.; Yin, X.; Zhang, X. Nano Lett. 2012, 12, 4853−4858. (20) Wang, T.; Boer-Duchemin, E.; Comtet, G.; Le Moal, E.; Dujardin, G.; Drezet, A.; Huant, S. Nanotechnology 2014, 25, 125202. 4639

DOI: 10.1021/acs.nanolett.9b01665 Nano Lett. 2019, 19, 4634−4640

Letter

Nano Letters (51) Bowers, C. M.; Rappoport, D.; Baghbanzadeh, M.; Simeone, F. C.; Liao, K. C.; Semenov, S. N.; Zaba, T.; Cyganik, P.; Aspuru-Guzik, A.; Whitesides, G. M. J. Phys. Chem. C 2016, 120, 11331−11337. (52) Peng, G.; Strange, M.; Thygesen, K. S.; Mavrikakis, M. J. Phys. Chem. C 2009, 113, 20967−20973. (53) Penner, P.; Zhang, X.; Marschewski, E.; Behler, F.; Angelova, P.; Beyer, A.; Christoffers, J.; Gölzhäuser, A. J. Phys. Chem. C 2014, 118, 21687−21694. (54) Xie, Z.; Bâldea, I.; Smith, C. E.; Wu, Y.; Frisbie, C. D. ACS Nano 2015, 9, 8022−8036. (55) Akkerman, H. B.; De Boer, B. J. Phys.: Condens. Matter 2008, 20, 013001. (56) Song, H.; Lee, H.; Lee, T. J. Am. Chem. Soc. 2007, 129, 3806− 3807. (57) Qi, Y.; Ratera, I.; Park, J. Y.; Ashby, P. D.; Quek, S. Y.; Neaton, J. B.; Salmeron, M. Langmuir 2008, 24, 2219−2223. (58) Frederiksen, T.; Munuera, C.; Ocal, C.; Brandbyge, M.; Paulsson, M.; Sanchez-Portal, D.; Arnau, A. ACS Nano 2009, 3, 2073−2080. (59) Wan, A.; Jiang, L.; Sangeeth, C. S. S.; Nijhuis, C. A. Adv. Funct. Mater. 2014, 24, 4442−4456. (60) Reus, W. F.; Thuo, M. M.; Shapiro, N. D.; Nijhuis, C. A.; Whitesides, G. M. ACS Nano 2012, 6, 4806−4822. (61) Chen, X.; Roemer, M.; Yuan, L.; Du, W.; Thompson, D.; Del Barco, E.; Nijhuis, C. A. Nat. Nanotechnol. 2017, 12, 797−803.

4640

DOI: 10.1021/acs.nanolett.9b01665 Nano Lett. 2019, 19, 4634−4640