Ultrafast Charge Transfer Pathways Through A Prototype Amino

Feb 2, 2016 - 19, 1000 Ljubljana, Slovenia ... For a more comprehensive list of citations to this article, users are encouraged to perform a search in...
1 downloads 0 Views 2MB Size
Letter pubs.acs.org/NanoLett

Ultrafast Charge Transfer Pathways Through A Prototype AminoCarboxylic Molecular Junction Gregor Kladnik,† Michele Puppin,‡,§ Marcello Coreno,∥ Monica de Simone,§ Luca Floreano,§ Alberto Verdini,§ Alberto Morgante,‡,§ Dean Cvetko,*,† and Albano Cossaro*,§ †

Faculty of Mathematics and Physics, University of Ljubljana, Jadranska ul. 19, 1000 Ljubljana, Slovenia Dipartimento di Fisica, Università di Trieste, via A. Valerio 2, I-34127 Trieste, Italy § CNR-IOM Laboratorio TASC, Basovizza SS-14, km 163.5, I-34012 Trieste, Italy ∥ CNR-ISM, UOS Trieste, Area Science Park Basovizza, 34149 Trieste, Italy ‡

S Supporting Information *

ABSTRACT: Charge transport properties of a vertically stacked organic heterojunction based on the amino-carboxylic (A-C) hydrogen bond coupling scheme are investigated by means of X-ray resonant photoemission and the core-hole clock method. We demonstrate that hydrogen bonding in molecular bilayers of benzoic acid/cysteamine (BA/CA) with an A-C coupling scheme opens a site selective pathway for ultrafast charge transport through the junction. Whereas charge transport from single BA layer directly coupled to the Au(111) is very fast and it is mediated by the phenyl group, the interposition of an anchoring layer of CA selectively hinders the delocalization of electrons from the BA phenyl group but opens a fast charge delocalization route through the BA orbitals close to the A-C bond. This evidences that hydrogen bonding established upon A-C recognition can be exploited to spatially/orbitally manipulate the charge transport properties of heteromolecular junctions. KEYWORDS: Organic heterojunction, resonant photoemission, charge transport, amino-carboxylic, core-hole-clock, hydrogen bond

H

technique has been employed also to molecular junctions based on the carboxylic autorecognition.12 Notably, both fluorescence and STM measurements reveal that the ET efficiency of a hydrogen-bonded junction is comparable with6,13 or even higher12 than that of a σ-covalently bonded junction. This aspect suggests that the hydrogen bonding can be exploited in the design of organic-based devices not only for tailoring the morphology of their active organic part but also for enhancing their ET properties. In this context, we present here the study of the ET properties of a hetero-organic interface based on the amino-carboxylic (A-C) recognition and hydrogen bond formation. In particular, we describe the effects of the A-C junction on the delocalization dynamics of electrons excited into the unoccupied molecular orbitals (MOs) spatially spread over the carboxylic end group of the molecule. Ultrafast delocalization is observed for the electrons excited close to the established hydrogen bond. On the contrary, the delocalization dynamics is unaffected for MOs localized on atoms not directly involved in the junction. The A-C coupling scheme can be therefore more generally identified as ET promotor at heteroorganic interfaces. At the same time, the importance of the electronic coupling between different molecular building units is evidenced. Moreover, the comparison of these results with those obtained with the same carboxylic molecule directly

ydrogen bonding represents a powerful mechanism for controlling the morphology of the organic architectures in the supramolecular self-assemblies at surfaces.1−3 By mimicking the mechanism of growth of organic crystals,4 molecules with the proper functional groups can be driven to assemble on surfaces following a predictable recognition scheme, which allows one to obtain complex geometries with the desired structural properties. The technology based on these systems is a promising route toward the design of efficient electronic devices. In this view, not only the morphology but most importantly the electron transport (ET) properties of the organic architectures become a key design issue. The characterization of the intermolecular hydrogen bonds in terms of their influence on the transport properties of the supramolecular assembly is therefore a topic of relevant importance in organic electronics. The hydrogen bond has been already shown to be directly involved in the ET processes of quite complex biomolecules like proteins or cytochromes by both theoretical5 and experimental6−8 studies. In particular, a quenching of the fluorescence decay intensities of selected photoinduced transitions occurs when a molecule is involved in a hydrogen-bonded donor−acceptor interface. An alternative characterization of the hydrogen-bond-mediated ET is based on scanning tunneling microscope (STM) measurements and in particular on the recognition tunnelling,9 which has been successfully exploited to describe the conductance properties of intermolecular junction based on the Watson−Crick9−11 bonding scheme. More recently, the same STM-based © XXXX American Chemical Society

Received: December 22, 2015 Revised: January 29, 2016

A

DOI: 10.1021/acs.nanolett.5b05231 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters

Figure 1. (A) Comparison of experimental VB spectra (light-blue curve) with the DFT calculated energies of filled orbitals for a free BA molecule (vertical bars and Gaussian broadening = dark blue curve). The uppermost HOMOs obtained by DFT calculations are shown as isosurface plots and tagged to corresponding peaks in the spectra. (B) Spectral decomposition according to the calculated partial density of states (PDOS) projected on the phenyl and carboxylic groups of the molecule are shown with respective resonant valence band spectra (blue and red curve).

involved in a junction can be investigated.28 The measured conductance in a molecular junction is determined by both the electronic structure of the junction and the energetic alignment of the (un-) occupied MOs involved in the transport process. On the contrary, CHC is limited to the analysis of the electronic delocalization over chosen empty MOs, which in turn may be localized on a particular molecular building unit. The CT dynamics measured with CHC may therefore not strictly mirror the single molecule conductive properties. Nevertheless, the CHC technique is chemically specific and allows for disentangling the CT characteristics of the different building units of a molecular junction. This aspect is well demonstrated in the present work, where we show how the A-C interaction differently affects the CT properties of the two different units constituting the carboxylic molecule. In particular, we determine the delocalization dynamics of core electrons excited in the MOs of the carboxylic and of the phenyl groups of the BA molecule involved in the junction. In order to give a proper interpretation of the valence band (VB), near edge X-ray absorption fine structure (NEXAFS) and RPES of the BA/CA/Au(111) complex, we first characterized the electronic properties of the BA molecule. Figures 1 and 2 report the VB and the NEXAFS spectra of the isolated molecule in gas phase and its comparison with the DOS DFT calculation. The calculated curves reproduce the experimental findings well; the different features of the experimental curves can be assigned to respective occupied and unoccupied MOs (highest occupied MOs (HOMOs) and lowest unoccupied MOs (LUMOs)) as reported in the figure. The details about the calculation methods are reported in the Supporting Information, as well as the complete assignment of the spectral features. Here the nature of the MOs closer to the vacuum level is discussed, which is important for the presentation of the RPES maps. The C K-edge NEXAFS spectrum, shown in Figure 2, presents two separated spectral features related to the transitions from C 1s core levels on the aromatic ring and on the carboxylic group to the π*1−3 orbitals at ∼285 and ∼288 eV, respectively. These assignments are in agreement also with the previously reported spectral identification.29 The experimental evidence of the symmetries of the assigned MOs comes from the RPES spectra at the carbon K-edge. Here the photon

coupled to the gold substrate, reveals that the insertion of the intermediate layer with A-C interaction can be exploited to selectively tailor the ET transparency/opacity of the organometallic interface for spatially different parts of the molecule. The system is entirely grown under ultrahigh vacuum conditions. It consists of a self-assembled monolayer (SAM) of a small carboxylic molecule (benzoic acid, BA) grown on top of the Au(111) surface, previously functionalized by an aminoterminated alkanethiol (cysteamine, CA). A detailed description of the chemical and morphological description of the system has been given elsewhere14 and is summarized in the Supporting Information. As shown by the X-ray photoemission spectroscopy (XPS), the growth of the vertically stacked heterojunction is driven by the A-C recognition process, which establishes a hydrogen bonding scheme between the functional groups of the two molecular species. The ET properties of the A-C coupling have been recently investigated by break junction experiments, which revealed that the efficiency of the interface was considerably lower than in the case of a covalently bonded junction.15 The system was in that case grown in solution, with the A-C interface formed upon the electrostatic attraction between the amino and the carboxylic groups in their ionic state. Moreover, the geometry of the A-C interaction was driven by the break junction measurement requirement. The nature of the A-C interaction was therefore different with respect to our system. Here we investigate the ET efficiency at the hydrogen-bonded A-C interface by means of X-ray resonant photoemission spectroscopy (RPES) and the core-hole clock (CHC) method.16−20 CHC method has been successfully employed to demonstrate covalent bonding promoting charge transport (CT) on a few hundred attosecond time scale,21,22 donor−acceptor bonding allowing subfemtosecond electron dynamics,23 π−Au coupling enabling electron transfer within a few femtoseconds,24 but also π−π coupling between aromatic molecules occurring on 1−60 fs, depending on the interring separation.25 CHC has been also successfully employed to study subfemtosecond dynamics of electron delocalization in wet DNA-based systems26 and to probe the electron delocalization in liquid water and ice.27 In fact, CHC represents a powerful tool for the characterization of the CT dynamics in molecular junctions, complementary to the break junction experiments. In the latter, the CT across the entire molecule B

DOI: 10.1021/acs.nanolett.5b05231 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters

indicated in Figure 1A. The complete set of calculated HOMOs is reported in the Supporting Information. We next determine the dynamics of core-excited electrons when molecules are coupled through the A-C recognition scheme. Within the CHC,16,24 based on the comparison between the intensities of the valence band resonances measured in our coupled system and in the isolated one, we determine the dynamics of the core-excited electron on different excitation sites influenced by the amino-carboxylic coupling. Details of the CHC method implemented here are given in the Supporting Information. Briefly, in an isolated system the core-excited electron cannot delocalize and hence either participates or spectates the core-hole decay via emission of an Auger electron (autoionization). The former gives rise to resonant peaks in the VB photoemission (participator intensity) degenerate with direct photoemission from the filled orbitals (HOMO−n; n = 0, 1, 2, 3...), whereas the latter gives rise to Auger electron emission with two holes in the VB (spectator intensity), and a Coulomb energy shift due to the screening by the spectator electron. Both these processes, participator and spectator decay, are however quenched in the electronically coupled system, where the delocalization dynamics of the excited electron competes with the dynamics of the core-hole decay, and the resulting spectrum consists of a normal Auger emission. By comparing the participator intensity in the coupled (Ip) versus isolated system (Ip0), we can obtain the electron delocalization dynamics as τ = τchIp(Ip0 − Ip)−1, where τch = 6 fs is the core-hole lifetime of the carbon 1s inner shell.30 Here we compare the RPES intensities of the BA in the gas phase (BA-gas) with the intensities of the benzoic acid anchored in the BA/CA/Au architecture. In particular, we make use of RPES maps shown in Figure 3 that were obtained as a series of valence band spectra taken with photon energy tuned across the C K-edge in steps of 0.1 eV and represented in two-dimensional-like intensity map, I(EK,hν) with the electron kinetic energy and the photonenergy scales on the horizontal and vertical axis, respectively. The nonresonant part of the photoemission has been measured at a photon energy 5 eV below the π* resonance and subtracted from all spectra in the I(EK,hν) map. The corresponding NEXAFS spectrum is also reported aside, showing the typical BA resonances due to excitations on the phenyl and carboxyl groups.14 Single photoemission spectra measured at photon energies corresponding to C 1s → πphenyl and C 1s → πcarboxyl resonances are shown separately in the lower panel in comparison to the BA-gas phase. The participator intensity in the coupled and isolated systems is evaluated from the intensities of the HOMO−n; n = 0−2 resonances. We note that participator resonances on the phenyl excitation (at hν ∼ 285 eV) show almost no quenching with respect to the BA-gas phase, indicating that delocalization dynamics over empty orbitals of the phenyl ring exceeds the upper limit of the CHC method (τphenyl > 10·τch = 60 fs). On the contrary the participator resonances on the carboxyl group (hν ∼ 288 eV) of the BA/CA/Au are quenched yielding an estimate of the delocalization time τcarboxyl ≤ 20 fs, which we attribute to the specific intermolecular coupling of the aminocarboxylic molecular anchor. Interestingly, we also measured the RPES map for the BA/Au system, namely for the BA monolayer adsorbed directly on the Au(111) surface. In this case, the BA molecules adsorb flat14 (see Supporting Information, Figure S2) and this geometry promotes a very efficient electronic coupling between the molecules and the Au.

Figure 2. Carbon K-edge NEXAFS of BA. Top panel: experimental data in the gas phase (filled circles) compared to DFT calculated spectrum (light blue curve). Inset: sketch of the BA molecule with carbon site labels (C1−C7) and calculated LUMO (isosurface plot). Bottom panel: spectral decomposition of DFT calculated spectrum according to core-hole excitation site (C1−C7).

energy is chosen to correspond to the resonant excitation from C 1s core level to the unoccupied orbitals on the phenyl and carboxyl groups at ∼285 and ∼288 eV, respectively. In addition to direct photoemission, the valence band spectra present also electron emission intensity due to resonant excitation to LUMO(+n); n = 0, 1, 2, 3, and so forth, followed by core-hole de-excitation with autoionization from the valence orbitals HOMO(−m); m = 0, 1, 2, 3, and so forth. In both cases, the final state is left with a single hole in the VB but the latter is a resonant process, requiring a high degree of spatial overlap of the three involved wave functions, the core, the intermediate LUMO(+n), and the HOMO(−m) of the final hole. Hence, VB resonances in the spectra occur when there is strong spatial correlation of the (HOMO−m, LUMO+n) pair over carbon sites, which helps us to identify the orbitals and to properly assign the features in the VB and NEXAFS spectra. Figure 1B shows the comparison of the calculated partial density of occupied states (PDOS) over the phenyl and carboxyl molecular groups, together with the respective resonant valence band spectra taken at the core → phenyl (at ∼285 eV) and core → carboxyl π* (∼288 eV) excitation. HOMO and HOMO−1, which exclusively occupy carbon atoms of the phenyl mostly resonate with the π* excitation to empty orbitals on the phenyl ring, whereas lower HOMO−m; m = 2, 3 peaks, which reside on the carboxyl group resonate mostly with the excitation to empty orbitals over carboxyl carbons. Spatial representations (isosurface plot) of a few highest occupied orbitals of BA are C

DOI: 10.1021/acs.nanolett.5b05231 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters

Figure 3. Carbon K-edge RPES map for BA/CA/Au. Top panel: Resonant photoemission intensity I(EK, hν). The nonresonant intensity has been subtracted. On the right-hand axis side the corresponding NEXAFS intensity is reported. Bottom panel: single VB spectra taken at resonance excitations from the C atoms of the phenyl (black) and carboxyl groups (red) and at photon energies indicated by the arrows. The respective spectra of the BA in gas phase are also reported (gray and red fill-to-zero lines). Participator quenching is indicated by the curved arrows and the obtained CT times are indicated.

In fact, the CHC analysis for this system reveals a very fast CT from the phenyl-localized orbitals of BA to the Au continuum of states. We note that for BA/Au(111) no fast CT occurs from the carboxylic MOs. Moreover, we also know from our CHC analysis that the CA monolayer is well coupled to Au(111) allowing for ultrafast transfer over the CA empty orbitals (see Supporting Information for the CHC analysis of the BA/Au and CA/Au). The CA intermediate layer in BA/CA/Au therefore not only couples efficiently BA and CA through the amino-carboxylic coupling scheme14 but also mediates ultrafast delocalization to the Au continuum and is hence almost transparent for the BA → CA → Au electron transport over carboxylic MOs, whereas it efficiently inhibits ultrafast CT over the phenyl-localized MOs. In conclusion, we have shown that A-C recognition at a hetero-organic interface can be exploited to promote ultrafast CT over empty molecular orbitals involved in the coupling scheme. We also find that whereas for metal−organic interfaces of BA/Au the CT is ultrafast from BA orbitals on the phenyl groups, the insertion of an intermediate anchoring layer of CA closes fast CT channels from the BA empty orbitals on the phenyl, while opening those on the carboxyl group. Therefore, through the A-C coupling scheme, the intermediate anchoring layer is selectively transparent/opaque for electron transport over SAM’s carboxyl/phenyl orbitals to the Au substrate. Our findings suggest that the A-C functionalization of a heteroorganic interface can conveniently represent a route to tailor

and manipulate the ET properties of the system. In addition, we have shown how the resonant photoemission CHC method can successfully be applied to investigate the ET properties of molecular interfaces based on the hydrogen bond, alternatively to STM-based recognition methods, which in general can affect the morphology of the system. Methods. The solid-state measurements were performed at the ALOISA beamline31 of the Elettra Synchrotron. Gas phase spectra have been acquired at the gas phase photoemission beamline,32 Elettra, Trieste-Italy. RPES at the carbon K-edge was conducted by taking a series of XPS scans with incident photon energies between 280 and 310 eV in steps of 0.1−0.2 eV. For each photon energy, XPS spectrum covering ∼60 eV kinetic energy range was measured to construct a photoemission map as a function of photon energy and electron kinetic energy. In the gas-phase, selected scans were performed in correspondence to the photon energies of interest, as determined by NEXAFS spectra. The nonresonant spectrum was measured in the pre-edge region at a photon energy of 283 eV, I(EK, hν = 283 eV), and was subtracted from each XPS spectrum in the RPES map, I(EK,hν). In the CHC analysis, we compared the resonant photoemission spectra at hν = 285 and 288 eV for the electronically coupled BA in the BA/CA/Au complex and for the electronically isolated molecules in the BA gas phase. All spectra have been normalized to the integrated Auger intensity. The CT time has been evaluated as described in the manuscript. Participator intensities have been evaluated D

DOI: 10.1021/acs.nanolett.5b05231 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters

(15) Yi, Z.; Trellenkamp, S.; Offenhäusser, A.; Mayer, D. Chem. Commun. (Cambridge, U. K.) 2010, 46, 8014−8016. (16) Brühwiler, P. A.; Karis, O.; Mårtensson, N. Rev. Mod. Phys. 2002, 74, 703−740. (17) Cao, L.; Gao, X.-Y.; Wee, A. T. S.; Qi, D.-C. Adv. Mater. 2014, 26, 7880−7888. (18) Vilmercati, P.; Cvetko, D.; Cossaro, A.; Morgante, A. Surf. Sci. 2009, 603, 1542−1556. (19) Menzel, D. Chem. Soc. Rev. 2008, 37, 2212−2223. (20) Wang, L.; Chen, W.; Wee, A. T. S. Surf. Sci. Rep. 2008, 63, 465− 486. (21) Schnadt, J.; Brühwiler, P. A.; Patthey, L.; O’Shea, J. N.; Södergren, S.; Odelius, M.; Ahuja, R.; Karis, O.; Bässler, M.; Persson, P.; Siegbahn, H.; Lunell, S.; Mårtensson, N. Nature 2002, 418, 620− 623. (22) Föhlisch, A.; Feulner, P.; Hennies, F.; Fink, A.; Menzel, D.; Sanchez-Portal, D.; Echenique, P. M.; Wurth, W. Nature 2005, 436, 373−376. (23) Schiros, T.; Kladnik, G.; Prezzi, D.; Ferretti, A.; Olivieri, G.; Cossaro, A.; Floreano, L.; Verdini, A.; Schenck, C.; Cox, M.; Gorodetsky, A. A.; Plunkett, K.; Delongchamp, D.; Nuckolls, C.; Morgante, A.; Cvetko, D.; Kymissis, I. Adv. Energy Mater. 2013, 3, 894−902. (24) Kladnik, G.; Cvetko, D.; Batra, A.; Dell’Angela, M.; Cossaro, A.; Kamenetska, M.; Venkataraman, L.; Morgante, A. J. Phys. Chem. C 2013, 117, 16477−16482. (25) Batra, A.; Kladnik, G.; Vázquez, H.; Meisner, J. S.; Floreano, L.; Nuckolls, C.; Cvetko, D.; Morgante, A.; Venkataraman, L. Nat. Commun. 2012, 3, 1086. (26) Ikeura-Sekiguchi, H.; Sekiguchi, T. Phys. Rev. Lett. 2007, 99, 228102. (27) Nordlund, D.; Ogasawara, H.; Bluhm, H.; Takahashi, O.; Odelius, M.; Nagasono, M.; Pettersson, L. G. M.; Nilsson, A. Phys. Rev. Lett. 2007, 99, 217406. (28) Xiang, D.; Jeong, H.; Lee, T.; Mayer, D. Adv. Mater. 2013, 25, 4845−4867. (29) Bâldea, I.; Schimmelpfennig, B.; Plaschke, M.; Rothe, J.; Schirmer, J.; Trofimov, a. B.; Fanghänel, T. J. Electron Spectrosc. Relat. Phenom. 2007, 154, 109−118. (30) Coville, M.; Thomas, T. D. Phys. Rev. A: At., Mol., Opt. Phys. 1991, 43, 6053−6056. (31) Floreano, L.; Cossaro, A.; Gotter, R.; Verdini, A.; Bavdek, G.; Evangelista, F.; Ruocco, A.; Morgante, A.; Cvetko, D. J. Phys. Chem. C 2008, 112, 10794−10802. (32) Prince, K. C.; Blyth, R. R.; Delaunay, R.; Zitnik, M.; Krempasky, J.; Slezak, J.; Camilloni, R.; Avaldi, L.; Coreno, M.; Stefani, G.; Furlani, C.; De Simone, M.; Stranges, S. J. Synchrotron Radiat. 1998, 5, 565− 568.

from the integrated intensities of the upper valence band peaks (HOMO at 6 eV for the ∼285 eV spectrum, and HOMO−2 for the ∼288 eV one). The Auger intensity from the CA/Au layer beneath has been evaluated by comparing NEXAFS of the BA, CA/Au, and BA/CA/Au spectra (see Figure S8 in Supporting Information). More details about the spectra acquisition conditions, the preparation of the sample, and the implementation of the CHC method are given in the Supporting Information.



ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.nanolett.5b05231. The details about the DFT calculations, the experimental methods, the CHC analysis as well as the measurements on the CA/Au and BA/Au systems. (PDF)



AUTHOR INFORMATION

Corresponding Authors

*E-mail: [email protected]. *E-mail: [email protected]. Present Address

(M.P.) Fritz-Haber Institut der MPG, Berlin, Germany. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This work was supported by ANCHOR project of the FIRB 2010 call of MIUR (ref RBFR10FQBL). D.C. and G.K. acknowledge support from the Slovenian Research Agency (Program P1-0112 at Jožef Stefan Institute, Ljubljana and Project Z1-6726). M.C. and M.d.S. acknowledge Italian MIUR (PRIN 2010ERFKXL_006).



REFERENCES

(1) Bartels, L. Nat. Chem. 2010, 2, 87−95. (2) De Feyter, S.; De Schryver, F. C. Chem. Soc. Rev. 2003, 32, 139− 150. (3) Barth, J. V. Annu. Rev. Phys. Chem. 2007, 58, 375−407. (4) Desiraju, G. R. Angew. Chem., Int. Ed. Engl. 1995, 34, 2311−2327. (5) Beratan, D. N.; Betts, J. N.; Onuchic, J. N. Science 1991, 252, 1285−1288. (6) Turro, C.; Chang, C. K.; Leroi, G. E.; Cukier, R. I.; Nocera, D. G. J. Am. Chem. Soc. 1992, 114, 4013−4015. (7) de Rege, P. J.; Williams, S. A.; Therien, M. J. Science 1995, 269, 1409−1413. (8) Therien, M. J.; Selman, M.; Gray, H. B.; Chang, I. J.; Winkler, J. R. J. Am. Chem. Soc. 1990, 112, 2420−2422. (9) Ohshiro, T.; Umezawa, Y. Proc. Natl. Acad. Sci. U. S. A. 2006, 103, 10−14. (10) Huang, S.; Chang, S.; He, J.; Zhang, P.; Liang, F.; Tuchband, M.; Li, S.; Lindsay, S. J. Phys. Chem. C 2010, 114, 20443−20448. (11) Chang, S.; He, J.; Kibel, A.; Lee, M.; Sankey, O.; Zhang, P.; Lindsay, S. Nat. Nanotechnol. 2009, 4, 297−301. (12) Nishino, T.; Hayashi, N.; Bui, P. T. J. Am. Chem. Soc. 2013, 135, 4592−4595. (13) Sessler, J. L.; Sathiosatham, M.; Brown, C. T.; Rhodes, T. A.; Wiederrecht, G. J. Am. Chem. Soc. 2001, 123, 3655−3660. (14) Cossaro, A.; Puppin, M.; Cvetko, D.; Kladnik, G.; Verdini, A.; Coreno, M.; De Simone, M.; Floreano, L.; Morgante, A. J. Phys. Chem. Lett. 2011, 2, 3124−3129. E

DOI: 10.1021/acs.nanolett.5b05231 Nano Lett. XXXX, XXX, XXX−XXX