Plasmonic response and SERS modulation in electrochemical applied potentials

We study the optical response of individual nm-wide plasmonic nanocavities using a nanoparticle-on-mirror design utilised as an electrode in an electrochemical cell. In this geometry Au nanoparticles are separated from a bulk Au film by an ultrathin molecular spacer, giving intense and stable Raman amplification of 100 molecules. Modulation of the plasmonic spectra and the SERS response is observed with applied voltage under a variety of electrolytes. Different scenarios are discussed to untangle the various mechanisms that can be involved in the electronic interaction between NPs and electrode surfaces.

We study the optical response of individual nm-wide plasmonic nanocavities using a nanoparticle-on-mirror design utilised as an electrode in an electrochemical cell. In this geometry Au nanoparticles are separated from a bulk Au film by an ultrathin molecular spacer, giving intense and stable Raman amplification of 100 molecules.

Modulation of the plasmonic spectra and the SERS response is observed with applied voltage under a variety of electrolytes. Different scenarios are discussed to untangle the various mechanisms that can be involved in the electronic interaction between
NPs and electrode surfaces.
Understanding the dynamics of charge transport across nanostructures is a key challenge in the creation of ultrathin functional devices. Electrochemistry plays a strong role for such charge transport in many devices, including the new generations of resistive random access memory involving metal/thin-insulator/nanomaterial constructs, as well as in photocatalytic systems, and in surface (bio)chemical sensors. Probes of the mechanisms of surface electrochemistry have mainly utilised frequency-dependent electrical measurements, leaving many questions unanswered such as the location of double layers and the spatial distribution of co/counter-ions in nanostructured materials. We choose here the simple example of a molecule-coated metal electrode, and show how optical spectroscopy using plasmonics and surface-enhanced Raman scattering (SERS) can provide more insight into these challenges. In the simple system comprised of a metal electrode coated with an organic self-assembled monolayer (SAM), the rate of electron transfer through the SAM has been shown to decay exponentially as the chain length of the monolayer is increased 1 . Adsorption of metal nanoparticles (NPs) onto such layers results in an overall charge transfer across the modified electrode 2 .
Surprisingly, several electrochemical investigations of nanoparticle-mediated electron transfer across the organic layers [2][3][4][5] show distance-independent charge transfer between two metals when Page 2 of 9 separated by gaps as wide as 6.5 nm. 3 Charge transfer across the gaps has proven to be much faster than electron transfer between metal and the dilute redox species in solution (estimates suggest up to 10 12 times faster through an organic layer than redox transfer at the metal surface 5 ). In this description, the metal/thin-insulator/metal stack effectively short-circuits when NPs adsorb 3 , making the NPs an extension of the electrode underneath the organic layer 5 . In all this work, electronic transport between NP and electrode surfaces has been investigated with impedance measurements.
Characterization of electrodes at the single NP level is thus challenging, with few experimental studies reported [6][7][8][9] . An improved understanding of this surface chemistry is however crucial for catalysis, as well as a host of photo-electrochemical applications.  10 We first explore several possible scenarios arising from the application of a potential to this system and discuss their implications on the optical and SERS spectra. The first scenario (S1) is the penetration of charged ions into the hydrophobic SAM ( Figure 2a). An immediate effect would be seen in the SERS with shifts and weakening in the Raman lines due to displacement of the Raman  depends strongly on ߜ so that only the most tightly-bound ions in the inner Helmholtz layer are considered ( Figure 2c). This model 12 yields Au conductivity ߪ ൌ ݊݁ߤ ൌ 1/ߩ, where the electron mobility ߤ ൌ ሺ8݁/ߨ݊ ሻ ሺ݇ ி ߜሻ ଷ with Fermi wavevector ݇ ி ൌ ሺ3ߨ ଶ ݊ሻ ଵ/ଷ set by the Au electron density ݊. As we will show, using realistic values for these parameters shows that negative voltages which increase the ion separation ߜ, lead to blue shifts and line broadening (Figure 3b), as observed in recent experiments [13][14][15] . Thus modulation of the double layer changes the surface conductivity of gold, thus modifying the plasmonic coupling.
To quantify the spectral and SERS changes expected from this modulation of surface conductivity, we employ an analytic circuit model for coupled plasmonic dimers separated by small gaps developed by Benz et al. 16 The normalised gap capacitance ߟ is given by ߟ ൌ ‫ܥ‬ ‫ܥ/‬ ௦ ൌ ݊ ଶ ln ሺ1 ܴΘ ଶ /݀ሻ, with ‫ܥ‬ the gap capacitance of the coupled system, ‫ܥ‬ ௦ the sphere capacitance due to the fringing field, ݊ the gap refractive index, ܴ the NP radius, ݀ the separation between nanoparticle and gold film and Θ~0.1π is an angle parametrising the laterally localised electric field.
The impedance of the dimer can be written as 16 where ߳ is the background medium permittivity, ߳ the vacuum permittivity, and ܴ and ‫ܮ‬ are the gap resistance and inductance respectively. Here we consider capacitive coupling (ܴ ൌ ‫ܮ‬ ൌ 0) so where ߱ ൌ ߱/߱ and the damping term is given by Γ ൌ ߳ ߱ ߩሺVሻ, with ߱ the plasma frequency and ߩሺVሻ the potential dependent Au resistivity. From ܼ we can extract the plasmon resonance peak wavelength and peak width for different ion separations from the Au, ߜ (Figure 3b).

View Article Online
We see that for increasing ߜ, the peak wavelength blueshifts and gives a sharper plasmonic resonant peak, meaning that this damping contribution is reduced. As we show below, this model fairly reproduces the trends we observe experimentally, as well as those recently obtained on nanoparticles in solution 13 . It predicts that the more negatively charged is the substrate, the more blue-shifted and less damped the plasmonic resonance should appear, because the solvated surface ions scatter less the electrons accelarated by the light at the Au surface. The damping Γ (Figure 3b) incorporates only the Drude damping contribution and underestimates the overall experimental damping, which broadens the plasmon resonance observed. However if applying a potential only varies the ion separation and thus the Drude damping, all other contributions to the overall damping should not be modified. Therefore we believe this gives a reasonable estimate of the tuning in peak width, as indeed also observed (see below). We note that this model suggests that in order to observe 10nm plasmonic shifts from this model, unfeasibly large ion separations from the surface are needed >10nm while Debye lengths for the double layer are typically < 1 nm.
A corollary of this mechanism which reduces Drude damping as solvated ions retract from the surface, is the increase of the trapped optical field strength, thus also increasing the SERS signal. This SERS strength would scale through a figure of merit proportional to Q ସ , set by the Q factor of the resonance (Figure 3c). 17,18 An enhancement of the SERS signal should then be observed for negative applied potentials and correspondingly a reduction for positive potentials 3 . We note that because we do not include the additional contributions to the damping discussed above, this SERS dependence is overestimated. Our model would also suggest that the enhancement of SERS should be seen for insulating as well as conducting molecules.
Finally, we consider a scenario (S4) exploring the effect of the reversible reduction of H + in the aqueous solution to form H 2 gas which is surface bound around the NP. This can be modelled as a thin shell of gas around the NP with a refractive index ݊ ≃ 1. Using a finite-element calculation, we estimate that this would yield a shift of the coupled plasmonic of ~ 15nm (Figure 3a) with only minor changes to the SERS intensity 19 .
To explore these effects experimentally, we exploit the high sensitivity to field-induced changes occurring in the nano-gap to investigate resonant light scattering under changing electric potential 13 . Significant increases in the coupled peak intensity, together with peak sharpening and spectral blue shifts, are observed when a negative voltage (Au substrate negatively charged) is applied (Figure 4bd). The opposite behaviour is observed (decreased amplitude, broadening, and redshifts) for positive potential. No significant differences are observed between different electrolytes (TBA, MgSO 4 , NaNO 3 and Na 2 SO 4 ) 13 which disproves the ion penetration scenario (S1). Nor are significant differences in plasmon shifts seen between self-assembled monolayers of different conductivity 13 .
Similar behaviour is seen for many different NPoM showing this is a very general behaviour. The dynamics observed for double-layer charging shows a sharp initial current spike, saturating after ~2 sec (Figure 4a). This is in contrast to the scattering spectral changes (Figure 4b surface at these potentials, as expected. We experimentally observe plasmon shifts of the order of 5 to 10 nm, which is not compatible with the modulation of refractive index in bulk salt solutions (S2), since it would have to change by an unfeasible Δ݊=0.1 to explain the spectral shifts observed. This observed shift is however more compatible with the refractive index change induced by a thin layer of H 2 gas around the NP (S4). In this scenario however, it is not directly obvious why a 5-fold SERS enhancement is observed (Figure   4e-f). As we show elsewhere 13 , we believe that the previously-established idea that the AuNP and the Au surface are at the same potential is in fact incorrect, and field-induced modulation of the molecular SERS is responsible. The spectral shifts on the other hand seem to arise from electrochemical surface reactions in water at the Au NP, which reversibly form and oxidise H 2 , thus modulating the plasmonic response. Since the SAM layers used are hydrophobic, we do not believe that any such gas evolution can take place in the nanoscale gap, but instead is located on the upper uncoated surfaces of the NP. Because an initial positive voltage already red-shifts the plasmon, this would suggest that already reduced H 2 gas is already present as a surface layer. It is however also possible that the modulation of Au conductivity from the local solvated ions (S3) also plays a role.
In conclusion, we study the optical response of Au NPs in a NPoM geometry, separated from bulk Au

View Article Online
Dark-Field spectroscopy. Optical dark-field images are recorded on a custom Olympus GX51 inverted microscope. Samples are illuminated with a focused white light source (halogen lamp). The scattered light is collected through a 50× dark-field objective (LMPLFLN-BD, NA 0.8) and analysed with a fiber-coupled (50 µm optical fiber) Ocean Optics QE65000 cooled spectrometer. We use a standard diffuser as a reference to normalize white light scattering.

SERS analysis.
SERS experiments are performed on the same modified Olympus GX51 inverted microscope used for dark-field spectroscopy. A monochromatic 633 nm HeNe laser beam is focused on the sample using a 50× objective (NA 0.8). Raman scattering is collected through the center of the objective and analysed with a Shamrock SR-303i spectrometer (600 l/mm 650nm blazed grating) coupled with an EMCCD camera cooled to −85°C. Rayleigh scattering is filtered out with a long pass 633nm filter. The system is calibrated using a silicon substrate as a reference. Spectral acquisitions are taken using an integration time of 1 s and the laser power on the sample is 30μW.