We investigate the plasmonic properties of a self-assembled 2D array of Ag nanospheres (average particle diameter/inter-particle separation distance of 9/3.7 nm). The structures of the individual particles and their assemblies are characterized using high-resolution transmission electron microscopy (HR-TEM). The plasmonic response of the nanoparticle network is probed using two-photon photoemission electron microscopy (TP-PEEM). HR-TEM and TP-PEEM statistics reveal the structure and plasmonic response of the network to be homogeneous on average. This translates into a relatively uniform surface-enhanced Raman scattering (SERS) response from biphenyl,4-4-dithiol (BPDT) molecules adsorbed onto different sites of the network. Reproducible, bright, and low-background SERS spectra are recorded and assigned on the basis of density functional theory calculations in which BPDT is chemisorbed onto the vertex of a finite tetrahedral Ag cluster consisting of 20 Ag atoms. A notable agreement between experiment and theory allows us to rigorously account for the observable vibrational states of BPDT in the ∼200–2200 cm−1 region of the spectrum. Finite difference time domain simulations further reveal that physical enhancement factors on the order of 106 are attainable at the nanogaps formed between the silver nanospheres in the 2D array. Combined with modest chemical enhancement factors, this study paves the way for reproducible single molecule signals from an easily self-assembled SERS substrate.

Nanometric confinement of light waves using assemblies of plasmonic nanoparticles is the underlying concept behind on-going efforts aimed at ultrasensitive chemical detection. In systems comprising such assemblies, the plasmonic Eigenmodes of the individual building blocks hybridize to form new collective modes in the network.1 These collective modes exhibit unique properties when compared to individual particles, such as the extreme localization and enhancement of incident and scattered radiation fields.2–4 This is bolstered by demonstrations of surface-enhanced Raman scattering5–8 (SERS) from molecules coaxed into plasmonic nanojunctions.9 In this work, we probe the enhanced Raman response of biphenyl-4,4-dithiol (BPDT) molecules adsorbed onto a self-assembled hexagonal close-packed distribution of small silver nanospheres. The nanoparticles are nearly spherical (average diameter of 9 nm), and feature average inter-particle separations of 3.7 nm in the network.

Various ordered assemblies of small nanoparticles have been reported, fashioned using pulsed laser irradiation,10 biologically11 and chemically12 templated assembly, as well as ion radiation techniques.13 The latter study, in which Sweatlock et al. generated linear arrays of small Ag nanoparticles (10 nm in diameter) with variable inter-particle spacing (0–4 nm), is of particular relevance to our work because of the dimensions considered.13 For linear arrays of small particles separated by 2–4 nm, it was concluded that (i) the plasmon resonance of the assembly shifts by less than 0.4 eV with respect to its analogue for the isolated particle, (ii) the maximum field enhancement (E/E0)4 attainable for on-resonance excitation is 6.4 × 103, and (iii) the response is confined to the inter-particle region and maximally enhanced at the midpoints between the metal particles.13 Although the field enhancement in the weakly coupled Ag assemblies (2–4 nm particle separation) is modest when compared to what was classically predicted for strongly coupled nanoparticles (0–1 nm particle separation), more recent quantum analyses of plasmonic nanojunctions suggest that the further separated assemblies are more suited for our intended application.4,14,15 In the case of nearly touching plasmonic nanostructures (sub-nm separation), conductive overlap is established, and the onset of electron tunneling drastically reduces the plasmonic field enhancements predicted by classical simulations. Raman scattering at the quantum limit of plasmons is a subject of on-going investigation in our group.15,16 In this regime, the molecular response (line spectra) competes with a broad and nearly featureless response associated with tunneling plasmons.15–17 In contrast, the particle networks described herein operate exclusively in the classical regime of plasmons, and take advantage of plasmonic mode confinement to enhance the optical response of the adsorbed molecules. An added advantage of our described approach is the ease with which the hexagonal close-packed distribution of small Ag nanospheres is readily self-assembled from a colloidal solution of isolated Ag nanoparticles.

Faceting and nanometric asperities on metal nanoparticles complicate the structure of the enhanced local electric fields around plasmonic nanoparticles and their assemblies.18,19 This has direct implications on optical measurements which take advantage of plasmonic field confinement and enhancement.18–20 In this regard, combined high-resolution microscopic and spectroscopic investigations are invaluable. Herein, we combine transmission and two-photon photoemission electron microscopy (TEM and TP-PEEM)—which characterize the structure and plasmonic response of the Ag self-assemblies—with SERS spectroscopy. Although the measurements are not directly correlated at present, the combination of the three techniques is unprecedented. That said, we provide enough statistical evidence to ensure the uniformity of structure, plasmonic response, and as a result, the optical response measured from BPDT molecules adsorbed at various sites of the substrate. The recorded Raman spectra are assigned on the basis of density functional theory calculations, and the local electromagnetic fields in the SERS experiments are simulated using finite difference time domain simulations.

Acetone (ACS, Amresco), silver nitrate (99%, Sigma-Aldrich), oleylamine (tech., 70%), ethanol (anhydrous, 95%, Aldrich), and toluene (anhydrous, 99.8%, Aldrich) were used as purchased. All reactions were performed under argon, using the standard air free Schlenk technique, unless otherwise noted. The centrifuge used for precipitation operated at 7200 rpm. Ag nanocrystals were synthesized in a one-pot procedure which employed oleylamine ligands for nanoparticle stabilization. In a typical synthesis, 0.011 g of AgNO3 was placed in a one-neck flask and dissolved in 3 ml of oleylamine through sonication. This part of the procedure is designed to facilitate the formation of Ag-oleate complexes that serve as precursors for Ag nucleation. In the second step, the temperature of the reaction system is raised to 100–120 °C and maintained at that level for 30 min. After cooling down to room temperature, the reaction mixture was transferred to 2 centrifuge tubes, topped with ethanol, and centrifuged for 3 min. The clear supernatant was poured off. The Ag nanoparticles that had precipitated were dissolved in 6 ml toluene. These particles were then precipitated a second time with the addition of an excess amount of ethanol and subsequent centrifugation. The final precipitate was suspended in 4 ml of toluene.

TEM images were recorded using a Cs-corrected environmental transmission electron microscope (FEI, Titan 80-300) operated at 300 kV in high vacuum mode. TEM samples were prepared by drop casting a 1 μl volume of the nanoparticle suspension on a single-layer graphene sheet on lacey carbon, supported by a copper grid (Ted Pella, Inc.).

Our PEEM setup has been previously described elsewhere.18,19 In this work, the illumination source is based on a 90 MHz Titanium-Sapphire oscillator (Griffin-10, KM Labs) delivering sub-20 fs laser pulses centered at 800 nm. The fundamental output of the oscillator is frequency doubled through a thin beta barium borate crystal, steered towards the microscope, attenuated to 5 mW, and focused onto the sample surface using a 20 cm focal length lens. The polarization of the incident radiation field is controlled using a half wave plate. PEEM samples were prepared by drop casting a 1 μl volume of the nanoparticle suspension on a 15 nm thick Ag film prepared on a freshly cleaved mica substrate by arc-discharge physical vapor deposition. The sputtering target was purchased from Ted Pella Inc. (99.99% purity), and the film thickness was monitored in situ using a quartz crystal microbalance. The typical surface roughness measured from freshly prepared Ag films indicates a root mean square height distribution of 3–5 nm.

SERS measurements were conducted using an inverted optical microscopy setup (Axiovert 200, Zeiss). The incident 514 nm continuous wave laser light (Innova 300, Coherent) is attenuated using a variable neutral density filter wheel (10–100 μW/μm2), reflected off a dichroic beamsplitter, and focused onto the sample surface using an oil-immersion objective (1.3 NA, 100×). The polarization of the incident laser is controlled using a half wave plate, which allows rotating the light polarization in the sample plane. Unless otherwise stated, this is the condition used throughout. Alternatively, we deliver light polarized along the surface normal, in the light propagation direction. This is achieved using a radial polarization converter (Altchena). The backscattered radiation is collected through the same objective, transmitted though the dichroic beamsplitter, and filtered through a long pass filter. The resulting light is detected by a liquid nitrogen cooled charge coupled device (CCD) coupled to a spectrometer (Holespec f/1.8i, Kaiser Optical System). The effective instrument resolution in the micro-Raman experiments is 8 cm−1. SERS active substrates were prepared by spin casting a 0.1 mM solution (volume = 2 μl) of BPDT onto the above-described TEM and PEEM substrates.

DFT calculations were performed using the methodologies implemented in Gaussian 09.21 We employ the B3LYP22 exchange-correlation functional in conjunction with the def2-TZVP basis set23 with matching pseudo potentials for Ag. Unconstrained geometry optimization was performed to locate the global minimum of BPDT chemisorbed onto the vertex of a tetrahedral Ag cluster consisting of 20 Ag atoms. Static Raman spectra were computed at the optimized Ag20-BPDT geometry, and the derived frequencies were linearly scaled by an empirical factor of 0.976 to align the computed spectra with their experimental analogues.

FDTD simulations were performed using a commercially available software (Lumerical Inc.) running on a local computer cluster. The computational models used replicate the SERS samples and experimental geometry by accounting for sample permittivity, laser wavelength, polarization, and angle of incidence. The representative calculations shown in Sec. III incorporate a silver nanoparticle quadrimer (dimensions corresponding to limiting cases inferred from the high resolution TEM images) atop a 15 nm thick silver layer coated on a glass substrate, all parsed in a three dimensional simulation volume using a Gaussian source at a 0° angle of incidence. The calculations yield the spatially resolved relative intensities of the electric field components as a function of time. Standard Fourier transforms result in the corresponding spatial and frequency resolved relative field magnitudes.

The synthesis of isolated Ag nanocrystals via the reduction of metal salts by oleylamine ligands has been demonstrated in previous reports.24–28 Therein, the reaction of AgNO3 with oleylamine proceeds in a single phase solvent (e.g., toluene) at 60–80 °C, leading to a slow reduction of silver ions into neutral clusters. Alternatively, the nanoparticle growth kinetics can be controlled by employing oleylamine, which acts both as reducing agent and solvent for the reaction. Close monitoring of the growth kinetics during the synthesis revealed27,29 that oleylamine molecules formed complex aggregates with the silver salt instantly upon mixing of the two reagents. At an elevated temperature (>80 °C), the complexes decomposed into very small particles, which eventually recombined into larger/thermally stable nanoparticles featuring average diameters in the 5–10 nm range. Using this method, the nanoparticle growth rate can be tuned either by adjusting the concentration of oleylamine in the toluene solution, or by varying the reaction temperature. The absorption profile of the synthesized Ag nanoparticles exhibited a sharp plasmon resonance centered at 415 nm, indicative of a narrow size distribution of nanoparticle diameters in chloroform. TEM images corroborate the UV-Vis observables, revealing a relatively narrow size distribution of particles and an average nanoparticle diameter of 9 ± 2 nm, see Figure 1. The recorded TEM images also reveal that drop casting the nanoparticle solution forms patches of uniformly self-assembled Ag nanoparticle networks. A fast Fourier transform of the low magnification TEM image of a prototypical particle assembly gives a long-range lattice distance of 11 nm. Dividing the measured lattice distance by |$\frac{{\sqrt 3 }}{2}$|32 and subtracting the average particle size yields an inter-particle separation distance of 3.7 nm in the network, best described as a hexagonal closed packed distribution of silver nanospheres. The high resolution TEM image shown in the right Panel of Figure 1 additionally reveals that the synthesized nanocrystals are nearly spherical. This feature is important for our intended application, as large variations in the plasmonic response of particles and their assemblies were previously associated with deviations from spherical shapes.18,19

FIG. 1.

Left Panel: TEM image of a prototypical self-assembled Ag nanoparticle network on a single layer of graphene. Statistics of the particle size distribution (top left inset) show an average particle diameter of ∼9 nm and a relatively narrow size distribution. Notice the well-defined spots in the cropped fast Fourier transform (FFT, bottom right inset) of the image. The average inter-particle distance (calculated using the lattice distance derived from the FFT and the average particles size) is ∼3.7 nm. Right Panel: High-resolution TEM image revealing a nearly spherical morphology of the synthesized nanocrystals.

FIG. 1.

Left Panel: TEM image of a prototypical self-assembled Ag nanoparticle network on a single layer of graphene. Statistics of the particle size distribution (top left inset) show an average particle diameter of ∼9 nm and a relatively narrow size distribution. Notice the well-defined spots in the cropped fast Fourier transform (FFT, bottom right inset) of the image. The average inter-particle distance (calculated using the lattice distance derived from the FFT and the average particles size) is ∼3.7 nm. Right Panel: High-resolution TEM image revealing a nearly spherical morphology of the synthesized nanocrystals.

Close modal

Two-photon photoemission electron microscopy images (30 × 30 μm2) of a self-assembled Ag nanoparticle film on silver are shown in Figure 2. The two images were recorded using p- (Panel (a)) and s-polarized (Panel (c)) 400-nm laser illumination. The histograms shown next to the two images graph the normalized counts (number of pixels) as a function of TP-PEEM intensity to assess the uniformity of the plasmonic response from the substrate.18,19,30 Namely, in TP-PEEM, an intense laser pulse acts twice on the plasmonic substrate; the first interaction creates a standing polarization state in Ag, and the second probes the prepared state coherently through photoemission.30,31 As such, the photoelectron yield is proportional to the square of the laser intensity (fourth power of the electric field, E4).

FIG. 2.

Two-photon photoemission electron microscopy (TP-PEEM) images (30 × 30 μm2) of self-assembled Ag nanoparticle films supported on a 15 nm thick Ag layer evaporated on a freshly cleaved mica substrate. Shown are two images recorded following irradiation with p- (Panel (a)) and s-polarized (Panel (c)) laser illumination, and a corresponding statistical analysis of the images (for p- and s-polarization in Panels (b) and (d), respectively). The histograms graph the normalized counts (number of pixels) as a function of TP-PEEM intensity to gauge the distribution of intensities in the two images shown. The histograms shown in the insets are similarly obtained from TP-PEEM images of the blank Ag substrate (analogous PEEM images not shown). Note that the relative TP-PEEM signal strengths are preserved in Panel (b), Panel (d), and their insets.

FIG. 2.

Two-photon photoemission electron microscopy (TP-PEEM) images (30 × 30 μm2) of self-assembled Ag nanoparticle films supported on a 15 nm thick Ag layer evaporated on a freshly cleaved mica substrate. Shown are two images recorded following irradiation with p- (Panel (a)) and s-polarized (Panel (c)) laser illumination, and a corresponding statistical analysis of the images (for p- and s-polarization in Panels (b) and (d), respectively). The histograms graph the normalized counts (number of pixels) as a function of TP-PEEM intensity to gauge the distribution of intensities in the two images shown. The histograms shown in the insets are similarly obtained from TP-PEEM images of the blank Ag substrate (analogous PEEM images not shown). Note that the relative TP-PEEM signal strengths are preserved in Panel (b), Panel (d), and their insets.

Close modal

Several considerations arise in interpreting the recorded TP-PEEM images and intensity statistics. First, the pixel size in the TP-PEEM images shown is 60 nm2. As such, each pixel in the image averages over the response of some 30 Ag particles and the photoemission hotspots between them, vide infra. In this regard, the averaged TP-PEEM intensity in a single pixel is orders of magnitudes lower than what would be expected from a single hotspot. Second, the intensity statistics obtained from the blank Ag substrate (inset of Panel (d) in Figure 2) carry uncertainty due to the weak coupling between the s-polarized light and the blank Ag substrate. We refrain from over-analyzing the recorded images, and note that intensity statistics from the image recorded following s-polarized laser irradiation (Panel (d) in Figure 2) reveal that the plasmonic response of the substrate is fairly homogeneous at a resolution of 60 nm2. All the recorded TP-PEEM intensities are well within an order of magnitude from the average photoemission, and 90% of the recorded intensities are within a factor of 4 from the average. Note that the distribution of PEEM intensities is even narrower for p-polarized laser excitation (Panels (a) and (b) of Figure 2).

The SERS response arises from the interaction of molecular polarizability tensors with local electromagnetic fields. We begin our discussion of the SERS results by describing the enhanced local electromagnetic fields in the silver nanoparticle network. In this regard, FDTD simulations are informative. Figure 3 shows the simulated electric field distribution for silver nanosphere quadrimers atop a 15 nm thick silver slab following irradiation with 514 nm light polarized in the x-direction. These calculations reveal that the response is confined to and optimally enhanced at the junction formed between the individual nanospheres. The attainable physical enhancement factors (E/E0)4 at the hotspots formed between two particles in the network range between 1–2 × 106 (3 nm separation, Figures 3(a) and 3(b)) and 6–8 × 105 (4 nm separation, Figures 3(c) and 3(d)). Combined with rather modest chemical enhancement factors amenable to silver-bound BPDT molecules,20 single molecule detection sensitivity is potentially feasible from the 2D silver nanoparticle network.

FIG. 3.

FDTD simulations illustrating electric field localization and enhancement at 3 ((a) and (b)) and 4 nm ((c) and (d)) gaps formed between the nanoparticles which comprise the 2D assembly. Two different quadrimer orientations (limiting cases taken from HR-TEM images) with respect to the incident electromagnetic field are shown for each separation distance considered. The incident 514 nm light wave is polarized in the x-direction. The scale bars increase from black (Re(E) = 0) to dark red (high values of Re(E)).

FIG. 3.

FDTD simulations illustrating electric field localization and enhancement at 3 ((a) and (b)) and 4 nm ((c) and (d)) gaps formed between the nanoparticles which comprise the 2D assembly. Two different quadrimer orientations (limiting cases taken from HR-TEM images) with respect to the incident electromagnetic field are shown for each separation distance considered. The incident 514 nm light wave is polarized in the x-direction. The scale bars increase from black (Re(E) = 0) to dark red (high values of Re(E)).

Close modal

As a general statement, we expect the geometry and homogeneity in structure (Figure 1) as well as the measured plasmonic response (Figure 2) of the network to translate into a uniform and bright SERS response, given an even distribution of molecular reporters. The SERS spectra of BPDT molecules adsorbed at 10 different sites of the substrate—separated by several microns—are indicative of the latter, see Figure 4. The integrity of the molecular reporters and their optical response under the experimental conditions used in these measurements is first ensured by monitoring the temporal evolution of Raman scattering activity at 4 frequency shifts corresponding to the aromatic C = C stretching vibration at 1580 cm−1, the C–C stretching vibration at 1280 cm−1, the in-plane C–H bending motion at 1202 cm−1, and the C–S(H) stretching vibration at 1094 cm−1, see Table I. Each resulting spectrum shown (Panel (a) in Figure 4) was obtained by time averaging 100 spectra, individually integrated for 0.1 s. The small standard deviation from the global average of 1000 spectra measured (inset of Panel (a) in Figure 4) asserts the uniformity of the response recorded from the various sites probed. These results are consistent with the intensity statistics derived from the TP-PEEM images, which suggest that the plasmonic response of the substrate is uniform over a length scale of tens of microns.

FIG. 4.

Raman scattering from BPDT molecules adsorbed onto graphene-supported Ag nanoparticle networks (see Figure 1). Shown in Panel (a) are 10 different spectra recorded from different regions of the substrate, separated by several microns. Each spectrum is obtained by time-averaging 100–150 spectra individually integrated for 0.1 s. The insets of Panel (a) show the global average obtained by summing over some 1000 spectra and the calculated standard deviation in two spectral regions. Temporal variations in intensities at 4 different frequency shifts corresponding to 4 vibrational resonances of BPDT are plotted in Panel (b). They reveal that the molecule and its signal are stable over a 30 s time period under the experimental conditions used in these measurements (514 nm, 100 μW/μm2).

FIG. 4.

Raman scattering from BPDT molecules adsorbed onto graphene-supported Ag nanoparticle networks (see Figure 1). Shown in Panel (a) are 10 different spectra recorded from different regions of the substrate, separated by several microns. Each spectrum is obtained by time-averaging 100–150 spectra individually integrated for 0.1 s. The insets of Panel (a) show the global average obtained by summing over some 1000 spectra and the calculated standard deviation in two spectral regions. Temporal variations in intensities at 4 different frequency shifts corresponding to 4 vibrational resonances of BPDT are plotted in Panel (b). They reveal that the molecule and its signal are stable over a 30 s time period under the experimental conditions used in these measurements (514 nm, 100 μW/μm2).

Close modal
Table I.

Assignments of the observed and calculated vibrational modes of BPDT. The relative scattering activities (calculated) and intensities (measured) are normalized to the intensity of the predominant C = C stretching vibration at 1581/1582 cm−1 (calculated/measured).

Ag20-BPDT (harm × 0.976 cm−1)Scattering Act. (Norm.)Experimental (cm−1)Intensity (Norm.)Description
411 0.02 411 0.06 Out-of-plane C–H wag 
529 0.01 543 0.04 In-plane skeletal deformation 
695 699 0.04 In-plane ring deformation 
713 712 0.04 Out-of-plane ring deformation 
768 0.01 772 0.04 In-plane ring deformation 
817 0.01 823 0.05 Out-of-plane C–H wag 
1008 0.01 1014 0.01 In-plane ring deformation 
1078 0.15 1082 0.30 C–S(Ag) stretch 
1089 0.05 1094 0.24 C–S(H) stretch 
1188 0.12 1202 0.18 In-plane C–H bend 
1276 0.29 1280 0.35 C–C stretch 
1499 0.02 1499 0.06 In-plane HC = CH rock 
1581 1.00 1582 C = C stretch 
Ag20-BPDT (harm × 0.976 cm−1)Scattering Act. (Norm.)Experimental (cm−1)Intensity (Norm.)Description
411 0.02 411 0.06 Out-of-plane C–H wag 
529 0.01 543 0.04 In-plane skeletal deformation 
695 699 0.04 In-plane ring deformation 
713 712 0.04 Out-of-plane ring deformation 
768 0.01 772 0.04 In-plane ring deformation 
817 0.01 823 0.05 Out-of-plane C–H wag 
1008 0.01 1014 0.01 In-plane ring deformation 
1078 0.15 1082 0.30 C–S(Ag) stretch 
1089 0.05 1094 0.24 C–S(H) stretch 
1188 0.12 1202 0.18 In-plane C–H bend 
1276 0.29 1280 0.35 C–C stretch 
1499 0.02 1499 0.06 In-plane HC = CH rock 
1581 1.00 1582 C = C stretch 

We then performed similar measurements which probe the SERS response of BPDT molecules coated on (i) a 15 nm thick blank silver substrate evaporated on a 0.1 μm thick microscope slide, and (ii) an Ag particle network self-assembled on (i). The global SERS averages of (i) and (ii) (1000 spectra from 10 different spots on each sample) are plotted in Figure 5, and compared to the global SERS average response from the graphene-supported network. The difference in the average SERS signal strength from the two Ag network featuring substrates can be rationalized as follows. First, the underlying silver substrate acts as a neutral density filter (OD = 0.35) both in the incident and scattered directions of our inverted optical Raman microscopy setup. This renders the optical response brighter for the optically transparent graphene featuring substrate. Second, the response from the nanoparticle network on the silver substrate is further enhanced when compared to its analogue on a transparent dielectric, as gauged from FDTD simulations. The two effects nearly cancel out in practice, amounting to a minor difference in the measured SERS response ((E/E0)4); that is well within the heterogeneity of the plasmonic response of the nanoparticle assemblies in the TP-PEEM picture, see Figure 2. The difference in the Raman activity of BPDT molecules adsorbed on (i) the Ag film and (ii) the nanosphere assembly on the same film can also be reconciled with the TP-PEEM results. For s- and p-polarized laser irradiation, the ratios of spatially averaged TP-PEEM intensities from the assembly over the blank Ag substrate are 27 and 4 ((E/E0)4), respectively. This is consistent with average SERS ratios of 12 and 7 ((E/E0)4) at 1580 cm−1, measured for light polarized along the sample plane (in the xy plane corresponding to s-polarized light in TP-PEEM) and along the light propagation direction (in the z-direction corresponding to p-polarized light in TP-PEEM), respectively. The corresponding s/p ratio of spatially averaged TP-PEEM intensities from the assembly is 1.8, consistent with a Pz/Pxy SERS ratio of 1.6 at 1580 cm−1. The observed variations are again well within the spread in the plasmonic response of the substrate, see Figure 2. We acknowledge that geometrical factors which affect the electron collection efficiency certainly come into play in interpreting the TP-PEEM images. That said, the agreement between the PEEM and SERS suggests that such factors are mostly averaged out in the statistical approach adopted to interpret the TP-PEEM results herein.

FIG. 5.

The SERS spectra of BPDT adsorbed onto different substrates, all measured using identical conditions for direct comparison. The reader is referred to the text for more details.

FIG. 5.

The SERS spectra of BPDT adsorbed onto different substrates, all measured using identical conditions for direct comparison. The reader is referred to the text for more details.

Close modal

The signal-to-noise achieved in the SERS measurements which probe the optical response of BPDT molecules adsorbed on the graphene-supported Ag nanosphere assembly allows us to directly compare our recorded spectra with the B3LYP/def2-TZVP spectral simulations based on the Ag20-BPDT model, see Figure 6 and Table I. We note that the recorded spectra in Figure 6 (and throughout this report) have not been background subtracted; we only subtract a straight line to rid the spectra of the dark counts of the detector. The low-background optical response is attributed to the uniformity of structure and plasmonic response from the particle array. Overall, the agreement between experiment and theory is notable, and allows us to confidently assign some of the low-frequency SERS active vibrations of BPDT for the first time. In future reports, we will exploit this particular feature of our substrate to assess the theoretical models used to simulate SERS spectra.

FIG. 6.

The global average (see Figure 3 definition) and selected time averaged response from BPDT adsorbed onto graphene-supported Ag nanoparticle networks (see Figure 1) along with the B3LYP/def2-TZVP-PP spectrum based on the Ag20-BPDT model. A remarkable agreement between experiment and theory, which captures the fine details in the zoomed in version of the plot in the lower panel, is noted. The calculated and measured frequencies and intensities are summarized in Table I.

FIG. 6.

The global average (see Figure 3 definition) and selected time averaged response from BPDT adsorbed onto graphene-supported Ag nanoparticle networks (see Figure 1) along with the B3LYP/def2-TZVP-PP spectrum based on the Ag20-BPDT model. A remarkable agreement between experiment and theory, which captures the fine details in the zoomed in version of the plot in the lower panel, is noted. The calculated and measured frequencies and intensities are summarized in Table I.

Close modal

Through a combination of TEM, TP-PEEM, and SERS, we characterize the structural and plasmonic properties of self-assembled Ag nanoparticles. We find that the uniformity of structure and plasmonic response translates into a reproducible, uniform, and low-background SERS response from BPDT molecules adsorbed at different sites of the substrate. The reported assembly protocol potentially allows us to tune and optimize the attainable optical enhancement factors at the junctions formed between the particles via ligand-guided self-assembly. Herein, we show that enhancement factors of 106 are attainable, rendering the described SERS substrate potentially capable of broadcasting the feeble optical response of a few molecules. Although the sample requirements for TEM (electron transparency), TP-PEEM (conductivity), and SERS (optical transparency) are somewhat restrictive, on-going efforts in our group aim at correlating the three measurements to gain further insights into the operating physics behind the optical experiments described. Variations of the substrates are also envisioned, e.g., an Ag nanoparticle network on a dithiol-coated Ag film which takes advantage of the SERS hotspots formed between the particles, and the hotspots formed between the particles the underlying plasmonic surface. This will be the subject of future works.

P.Z.E. acknowledges support from the Laboratory Directed Research and Development Program through a Linus Pauling Fellowship at Pacific Northwest National Laboratory (PNNL), an allocation of computing time from the National Science Foundation (TG-CHE130003), and the use of the Extreme Science and Engineering Discovery Environment. W.P.H. acknowledges support from the U.S. Department of Energy (DOE), Office of Science, Office of Basic Energy Sciences, Division of Chemical Sciences, Division of Chemical Sciences, Geosciences and Biosciences. The TEM work was supported through the Chemical Imaging Initiative, under the Laboratory Directed Research and Development Program at PNNL. Part of this work was performed using EMSL, a national scientific user facility sponsored by the DOE's Office of Biological and Environmental Research and located at PNNL. PNNL is a multiprogram national laboratory operated for DOE by Battelle.

1.
N. J.
Halas
,
S.
Lal
,
W.-S.
Chang
,
S.
Link
, and
P.
Nordlander
,
Chem. Rev.
111
,
3913
(
2011
).
2.
E.
Hao
and
G. C.
Schatz
,
J. Chem. Phys.
120
,
357
(
2004
).
3.
F. J.
Garcıa de Abajo
,
J. Phys. Chem. C
112
,
17983
(
2008
).
4.
D. C.
Marinica
,
A. K.
Kazansky
,
P.
Nordlander
,
J.
Aizpurua
, and
A. G.
Borisov
,
Nano Lett.
12
,
1333
(
2012
).
5.
M.
Fleischmann
,
P. J.
Hendra
, and
A. J.
McQuillan
,
Chem. Phys. Lett.
26
,
163
(
1974
).
6.
D. L.
Jeanmaire
and
R. P.
Van Duyne
,
J. Electroanal. Chem.
84
,
1
(
1977
).
7.
M.
Moskovits
,
J. Chem. Phys.
69
,
4159
(
1978
).
9.
E.
Le Ru
and
P. G.
Etchegoin
,
Annu. Rev. Phys. Chem.
63
,
65
(
2012
).
10.
M.
Kaempfe
,
H.
Graener
,
A.
Kiesow
, and
A.
Heilmann
,
Appl. Phys. Lett.
79
,
1876
(
2001
).
11.
R. A.
McMillan
,
C. D.
Paavola
,
J.
Howard
,
S. L.
Chan
,
N. J.
Zaluzec
, and
J. D.
Trent
,
Nat. Mater.
1
,
247
(
2002
).
12.
S. M.
Adams
,
J. D.
Caldwell
,
F.
Bezares
,
J.
Culbertson
, and
R.
Ragan
,
Small
8
,
2239
(
2012
).
13.
L. A.
Sweatlock
,
S. A.
Maier
,
H. A.
Atwater
,
J. J.
Penninkhof
, and
A.
Polman
,
Phys. Rev. B
71
,
235408
(
2005
).
14.
R.
Esteban
,
A. G.
Borisov
,
P.
Nordlander
, and
J.
Aizpurua
,
Nat. Commun.
3
,
1
(
2012
).
15.
P. Z.
El-Khoury
,
D.
Hu
,
V. A.
Apkarian
, and
W. P.
Hess
,
Nano Lett.
13
,
1858
(
2013
).
16.
P. Z.
El-Khoury
and
W. P.
Hess
,
Nano Lett.
14
,
4114
(
2014
).
17.
M.
Banik
,
P. Z.
El-Khoury
,
A.
Nag
,
A.
Rodriguez-Perez
,
N.
Guarrottxena
,
G. C.
Bazan
, and
V. A.
Apkarian
,
ACS Nano
6
,
10343
(
2012
).
18.
S. J.
Peppernick
,
A. G.
Joly
,
K. M.
Beck
, and
W. P.
Hess
,
J. Chem. Phys.
134
,
034507
(
2011
).
19.
S. J.
Peppernick
,
A. G.
Joly
,
K. M.
Beck
, and
W. P.
Hess
,
J. Chem. Phys.
138
,
154701
(
2013
).
20.
M.
Banik
,
A.
Nag
,
P. Z.
El-Khoury
,
A.
Rodriguez-Perez
,
N.
Guarrotxena
,
G. C.
Bazan
, and
V. A.
Apkarian
,
J. Phys. Chem. C
116
,
10415
(
2012
).
21.
M. J.
Frisch
,
G. W.
Trucks
,
H. B.
Schlegel
 et al., GAUSSIAN 09, Revision D.01, Gaussian, Inc., Wallingford, CT,
2009
.
22.
A. D.
Becke
,
J. Chem. Phys.
98
,
5648
(
1993
).
23.
A.
Schafer
,
C.
Huber
,
R.
Ahlrichs
,
J. Chem. Phys.
100
,
5829
(
1994
).
24.
H.
Hiramatsu
and
F. E.
Osterloh
,
Chem. Mater.
16
,
2509
(
2004
).
25.
M.
Chen
,
M. G.
Feng
,
X.
Wang
,
T. C.
Li
,
J. Y.
Zhang
, and
D. L.
Qian
,
Langmuir
23
,
5296
(
2007
).
26.
X.
Lu
,
M. S.
Yavuz
,
H. Y.
Tuan
,
B. A.
Korgel
, and
Y.
Xia
,
J. Am. Chem. Soc.
130
,
8900
(
2008
).
27.
X. M.
Lu
,
H. Y.
Tnan
,
B. A.
Korgel
, and
Y. N.
Xia
,
Chem. – Eur. J.
14
,
1584
(
2008
).
28.
N.
Pazos-Perez
,
D.
Baranov
,
S.
Irsen
,
M.
Hilgendorff
,
L. A.
Liz-Marzan
, and
M.
Giersig
,
Langmuir
24
,
9855
(
2008
).
29.
X.
Liu
,
M.
Atwater
,
J.
Wang
,
Q.
Dai
,
J.
Zou
,
J. P.
Brennan
, and
Q.
Huo
,
J. Nanosci. Nanotechnol.
7
,
3126
(
2007
).
30.
C.
Wiemann
,
D.
Bayer
,
M.
Rohmer
,
M.
Aeschlimann
, and
M.
Bauer
,
Surf. Sci.
601
,
4714
(
2007
).
31.
A.
Kubo
,
K.
Onda
,
H.
Petek
,
Z. J.
Sun
,
Y. S.
Jung
, and
H. K.
Kim
,
Nano Lett.
5
,
1123
(
2005
).