2.1.

Enhancement Factor of Single-Molecule SERS

The required enhancement factor for single-molecule SERS is a basic and critical value that can help researchers intuitively estimate how difficult it is to realize SERS detection with single-molecule sensitivity. Unfortunately, this value is still controversial. When the single-molecule SERS was first observed around 20 years ago, it was claimed that an enhancement factor of $\sim {10}^{14}$ was essential for single-molecule detection.^27,28 However, the EM-induced enhancement factor by the SERS substrates was only $\sim {10}^{10}$,²⁶ which can hardly be used to account for the high enhancement factor for single-molecule detection. A number of efforts have been made to explain this large enhancement gap by considering new enhancement mechanisms, such as chemical enhancement and resonance Raman enhancement.⁴² Later on, it was argued that an enhancement factor of $\sim {10}^8$ is sufficient for realizing single-molecule detection.⁴³ It is worthwhile to note that the sensitivity of SERS measurement does not only rely on SERS substrates but also relies on the detected molecules and the optical setup used for the measurement. On the one hand, various molecules may exhibit various Raman polarizabilities under the same SERS measurement condition. Especially, when the photon energy of excitation light accidentally satisfies the electronic transition of a specific molecule, resonance Raman scattering occurs, which is capable of providing an enhancement of Raman polarizability up to $\sim {10}^6$.⁴⁴ The difference in the Raman polarizability results in the difference of the lowest detectable concentrations for different molecules. On the other hand, even for the same type of molecules, the sensitivities or detection limits of SERS measurement may also be different using different optical setups.⁴⁵ The parameters of optical components in the optical setups, such as numerical apertures of objective lens, sensitivities of spectrometers, wavelengths of incident light, and polarization states of incident light, also significantly influence the sensitivity of SERS measurement.⁴⁶ The numerical apertures determine the collection efficiency of Raman scattering for the SERS measurement. A high numerical aperture enables a high collection efficiency, which is beneficial for improving the sensitivity for SERS measurement.⁴⁷ The sensitivity of the spectrometer used for the SERS measurement determines the lowest Raman intensity coupled to the spectrometer that can be detected. A high sensitivity of the spectrometer enables a low detectable Raman intensity, which is also desired for increasing the sensitivity of SERS measurement. Moreover, the excitation wavelength and polarization state of the incident light determine the excitation of SPR on the SERS substrate. When the excitation wavelength is selected at the resonance wavelength of SPR while the polarization state is selected for the highest electromagnetic enhancement of the SERS substrate, the enhanced Raman signal is the strongest.⁴⁸ This indicates that the selection of an optimum excitation wavelength and an optimum polarization state of the incident light can also be utilized for increasing the sensitivity of the SERS measurement. In summary, the required single-molecule SERS enhancement factor varies for different molecules at different experimental conditions.

How to experimentally estimate the enhancement factor of single-molecule SERS is another important point that needs to be taken care of (see Refs. 2627.28.–29). The enhancement factor of single-molecule SERS is defined identically to that of conventional SERS, which can be obtained by comparing Raman intensities of the same type of molecules with and without enhancement of SERS substrates. For the measurement without enhancement of SERS substrates, which is the ground truth for estimation of the enhancement factor, the concentration of the measured sample is relatively high. Molecular solutions are preferred as the samples for the ground truth measurement as their molecular concentrations are precisely controllable. For the measurement with enhancement of SERS substrates, the concentration of the measured sample is extremely low. Especially for single-molecule detection, the concentration of the measured sample needs to be decreased to the level at which only one molecule on average exists in a measured volume. Such low concentration is probable to result in an inhomogeneous distribution of molecules adsorbed on the SERS substrates. Therefore, the practical concentration of molecules in the measured volume on the SERS substrate requires prudential calibration. Some experimental evidences are conventionally needed to verify the practical concentration is at the single-molecule level.

2.2.

Experimental Evidences of Single-Molecule SERS

SERS intensity fluctuation is one of the representative experimental evidences of single-molecule SERS. The fluctuation is partially due to the instability of surfaces of SERS substrates. The migration of surface atoms of SERS substrates (e.g., amorphous Au substrates) can influence the measured single-molecule signal.⁴⁰ Moreover, when the concentration of molecules is at the single-molecule level, only one molecule on average exists in the measured volume. The average here indicates the average in both time and space domains. As the current single-molecule SERS is mainly achieved by the SPR-based EM, hotspots with electromagnetic field enhancement high enough for single-molecule detection are sparsely distributed in the measured volume. Thus, there is a large probability that the single molecule that dynamically moves in the measured volume does not locate at the hotspots at any time. This means that if we continuously monitor the SERS intensity of the single-molecule sample in time and space domains, the SERS intensity is expected to strongly fluctuate due to the dynamic motion of the single molecule in and out of the hotspots.^49–53

Bianalyte approach is capable of offering more experimental evidence of single-molecule SERS. In this approach, a mixture of two analytes with distinguishable Raman spectra is used as the sample for the SERS measurement.^54,55 When the sample concentration is high, more than one molecule locates at the hotspots within the measured volume. The measured SERS signal is expected to be a mixture of the SERS spectra of both analytes. If the sample concentration is gradually decreased, and finally to the single-molecule level, only one molecule on average exists within the measured volume. The measured SERS signal is expected to come from only one of the two analytes. If we continuously monitor the SERS spectrum of the bianalyte sample in time and space domains at such concentration, the measured SERS spectrum is the spectrum of either one analyte or the other, rather than the mixture of two spectra. This phenomenon observed using the bianalyte approach can be used as another experimental evidence of single-molecule SERS.

2.3.

Data Analysis of Single-Molecule SERS

As we have discussed in the last part, the SERS signals will fluctuate in time and space domains when the concentration of detected molecules is decreased down to the single-molecule level. It is essential to employ statistical methods to analyze the experimental data of single-molecule SERS. In order to make full use of statistical soundness, a sufficient number of SERS spectra need to be measured for statistical analysis. Principal component analysis (PCA) as a classical analytical method is widely used for characterization of single-molecule SERS.^56–58 For example, the PCA is powerful for decomposition of SERS spectra of a mixture in the bianalyte approach. The decomposition with statistical soundness is usable for characterization of SERS substrates, such as precise estimation of resonance wavelengths of hotspots and evaluation of enhancement efficiencies of SERS substrates. In addition, digitization of SERS intensities has been used for quantification of molecular concentration in single-molecule SERS. An intensity threshold is set to digitize single-molecule SERS signals with intensity fluctuation.^45,59 Based on the digitization, a calibration curve indicating the correlation between the signal intensity and sample concentration is constructed, which can be used for quantification of the sample at an extremely low concentration. Moreover, cluster analysis is also usable for classification of measured SERS spectra of a mixture in the bianalyte approach based on the correlation between the measured SERS spectra and the ground-truth spectra of two target molecules. The cluster analysis method partitions the measured SERS spectra into two groups of similar spectra, which correspond to two target molecules.⁵⁶ Recently, the emergence of advanced chemometric techniques, such as least-square regression and artificial neural networks, provides the possibility for quantitative analysis of detected molecules by SERS measurement as it provides much deeper analysis of variables implied in the measured SERS spectra. Deep learning as a popular data analysis method is advantageous for classification of complex mixed analytes with a large number of features in the measured SERS spectra. Rapid classification of honey varieties has been recently demonstrated using deep learning to analyze the measured SERS spectra.⁶⁰ More data analysis methods with strong statistical soundness are desirable in the future to improve the reproducibility and quantification of single-molecule SERS.

3.1.

Electromagnetic Enhancement

Electromagnetic enhancement, as the most widely used strategy for SERS, plays a fundamental role in single-molecule SERS. At the early stage of SERS development, metal substrates with rough surfaces were used, which were capable of offering decent Raman enhancement for detected molecules.^62–67 However, such metal substrates without delicate morphological designs can hardly satisfy the need for single-molecule detection due to their insufficient electromagnetic enhancement. Later on, extremely high electromagnetic enhancement has been demonstrated, which enables the measurement of SERS signals of nonresonant molecules with Raman differential cross sections typically in the range of ${10}^{-29}$ to ${10}^{-30}{\mathrm{cm}}^2/\mathrm{sr}$. Blackie et al.⁶⁸ demonstrated single-molecule SERS of two nonresonant molecules, 1,2-di-(4-pyridyl)-ethylene (BPE) and adenine with enhancement factors of $5\times {10}^9$ and ${10}^{11}$, respectively. The experiment not only verifies the capability of SERS for detecting nonresonant molecule at the single-molecule level but also provides quantitative enhancement factors required for single-molecule SERS detection of nonresonant molecules. However, single-molecule SERS with high reproducibility still remains a challenge. In this section, we will introduce several recently developed designs with highly controllable morphological structures that enable simultaneously high electromagnetic enhancement and reproducibility for single-molecule SERS.

3.1.1.

Plasmonic nanogaps with precise size control

Plasmonic nanogaps, as a representative type of simple and effective structures for SERS applications, are frequently utilized for single-molecule SERS as they are capable of providing extremely high electromagnetic enhancement. However, the enhancement factors enabled by plasmonic nanogaps are highly sensitive to the sizes of nanogaps. Nanometer-scale error in the size can lead to an enhancement factor difference up to several orders of magnitude. Plasmonic nanogaps with precise size control are essential for single-molecule SERS as they are capable of offering controllable ultrahigh electromagnetic enhancement and excellent reproducibility. Several specific plasmonic nanogaps with controllable gap sizes will be introduced in this section.

In 2003, Wang et al.⁴⁷ for the first time demonstrated an optical antenna to observe the highly directional emission of Raman scattering from single molecules. The optical antenna consists of a silver ring and a silver dimer with a precise size control as shown in Fig. 1(a). By virtue of the strong electromagnetic enhancement of the plasmonic nanogap in the silver dimer, the directional emission of Raman signals from single molecules can be observed in the far-field. The single-molecule SERS spectra of R6G-d0/R6G-d4 enabled by the optical antenna are shown in Fig. 1(b). A distinct difference is observed between the two spectra, which is useful for distinguishing isotopologues. More importantly, with the precise size control of the plasmonic nanogap, an unprecedented near-unity fraction of the optical antennas has single-molecule sensitivity, indicating their excellent reproducibility for single-molecule SERS. In 2010, Lim et al. proposed and experimentally demonstrated a gold-silver core-shell nanodumbbell (GSND) with an engineerable nanogap for highly reproducible single-molecule SERS.⁶⁹ The GSND consists of two gold-silver core-shell nanoparticles linked by a single DNA molecule as shown in Fig. 1(c). It is fabricated with a high-yield synthetic method using a single-target-DNA hybridization, which is achieved by interconnecting two DNA sequences. One of the two DNA sequences, which is linked to a gold nanoparticle, is used to precapture a target Raman-active molecule, and then hybridized with the other DNA sequence, which is linked to another gold nanoparticle. The hybridization process enables linking the two gold nanoparticles as well as locating the target molecule within the nanogap of the two gold nanoparticles. Silver shells with precisely controllable thickness are subsequently grown on the surfaces of the two gold nanoparticles to engineer the gap size. The single-molecule SERS signals measured by the fabricated GSND are shown in Fig. 1(d). Due to the precise size-controllability of the fabrication method, SERS detection with single-molecule sensitivity and high reproducibility was successfully achieved.⁶⁹ Thacker et al.⁷⁰ proposed and experimentally realized a similar plasmonic nanogap with a controllable gap size based on a DNA origami technique. The proposed fabrication technique is advantageous for accurate positioning of nanoparticles. Gold nanoparticle dimers with reliable sub-5-nm nanogaps were successfully fabricated using the DNA origami technique, which is shown in Fig. 1(e). A high Raman enhancement factor of $\sim {10}^7$ with high reproducibility was realized as shown in Fig. 1(f). The results clearly exhibit the promising potential of this building technique for highly reproducible single-molecule SERS.⁷⁰

Fig. 1

Plasmonic nanogaps with precise size control for single-molecule SERS. (a) Scanning electron microscope image of an optical antenna. Inner and outer radii of silver rings are 300 and 380 nm, respectively. Left-hand rod is 80 nm long and 70 nm wide. Right-hand rod is 68 nm long and 62 nm wide. (b) Representative Raman spectra of R6G-d0 and R6G-d4 single-molecule level events. The R6G-d4 spectrum is shifted in the vertical direction by 2000 counts for display purposes.⁴⁷ (c) Nanometer-scale silver-shell growth-based gap-engineering in the formation of the SERS-active GSND. (d) Raman spectra taken from Cy3-modified oligonucleotides (redline) and Cy3-free oligonucleotides (blackline) in NaCl-aggregated silver colloids.⁶⁹ (e) Schematic of the NP dimers assembled on the DNA origami platform. The NPs are coated with an ssDNA brush to prevent aggregation as well as facilitate attachment to the origami platform. (f) Measured Raman spectra of the ssDNA coating of 19 bases of thymine (sequence 1) on the NP dimers.⁷⁰ (g) SECARS enhancement map. (h) Three representative SECARS spectra showing a pure $p$-MA event (top), a pure adenine event (middle), and a mixed event (bottom).⁴⁸ Figures reprinted with permission: (a) and (b) Ref. 47, © 2013 by the American Chemical Society; (c) and (d) Ref. 69, © 2009 by the Nature Publishing Group; (e) and (f) Ref. 70, © 2014 by the Nature Publishing Group; and (g) and (h) Ref. 48, © 2014 by the Nature Publishing Group.

Coherent anti-Stokes Raman scattering (CARS) as a specific four-wave mixing process involves the coherent interaction of two pump waves, one Stokes wave, and one anti-Stokes wave through the third-order polarizability of the vibronic modes of a molecule.⁴⁸ The CARS signal is conventionally obtained by measuring the anti-Stokes wave generated with the input of two pump waves and one Stokes wave in the four-wave mixing process. By virtue of the molecular coherence, CARS signal is usually orders of magnitude stronger than spontaneous Raman signal. However, compared with SERS, conventional surface-enhanced CARS (SECARS) suffers from relatively low enhancement factors due to time-scale effects. Zhang et al.⁴⁸ proposed a plasmonic nanogap with Fano resonance for single-molecule SECARS with an enhancement factor up to ${10}^{11}$. By combination of high electromagnetic enhancement of the nanogap and high sensitivity of nonlinear Raman spectroscopy, single-molecule sensitivity was experimentally realized. The plasmonic nanogap was formed by a plasmonic quadrumer, as shown in Fig. 1(g), which was fabricated by a standard complementary metal oxide semiconductor process. Electron-beam lithography was used to define its pattern and precisely control its size. With such a design, the plasmonic nanogap enables simultaneous electromagnetic enhancement of pump light, Stokes scattering, and anti-Stokes scattering. A total enhancement factor of $\sim {10}^{11}$ over spontaneous Raman signal was experimentally realized for single-molecule detection. The bianalyte approach was also used to experimentally verify the capability of the proposed technique for single-molecule detection, as shown in Fig. 1(h). At the low concentration, either the SECARS spectrum of $p$-MA or adenine was measured for a mixed sample, exhibiting the signature of single-molecule detection.⁴⁸

3.1.2.

Plasmonic sharp tips

Plasmonic sharp tips are another type of configuration that enables ultrahigh Raman enhancement for single-molecule SERS. This is mainly due to their sharp geometric configuration that is capable of realizing high concentration and localization of electromagnetic field. Besides chemically synthesized nanoparticles with sharp tips (e.g., sea-urchin-like nanoparticles),⁷¹ another representative technique based on the plasmonic sharp tips is tip-enhanced Raman spectroscopy (TERS), which combines a plasmonic sharp tip with a controllable mechanical system with a shifting precision of subnanometer. TERS, as an extension technique of SERS, not only provides a single-molecule sensitivity but also enables Raman measurement in space domain with a subnanometer resolution.

In 2011, Liu et al.⁷² used fishing-mode TERS to study the molecular structure of single-molecule junctions in different conduction states. The fishing-mode TERS, as shown in Fig. 2(a), is capable of simultaneously measuring conductance and TERS signals of a single-molecule junction locating at the apex of a sharp gold tip. A bias voltage is applied between the tip and a gold substrate to tune the conduction states of the single-molecule junction. The measured TERS signals of a 4,4’-bipyridine (4bipy) molecular junction around Raman peak of $1609{\mathrm{cm}}^{-1}$ at different voltages are shown in Fig. 2(b). By virtue of the high sensitivity of the plasmonic sharp tip, the splitting of the Raman peak of the single-molecule junction was observed at a high bias voltage, which is inaccessible by other methods. Zhang et al.⁷³ utilized TERS for chemical mapping of a single molecule as shown in Fig. 2(c). By a combination of high electromagnetic enhancement of a plasmonic sharp tip and precise tuning capability of a scanning tunneling microscopy (STM), TERS mapping of a single molecule with a subnanometer spatial resolution for different vibrational modes was realized, which is shown in Fig. 2(d).⁷³ The above results verify that plasmonic sharp tips are an ideal and promising geometric configuration for single-molecule SERS.

Fig. 2

Plasmonic sharp tips for single-molecule SERS. (a) An illustration of the FM-TERS system for simultaneous conductance and TERS measurement of single-molecule junctions. A gold (111) surface is preadsorbed with molecules. The distance between the gold tip and the gold (111) surface is then controlled by an STM equipped with a conductance-measuring circuit. The red contour indicates the distribution of electromagnetic field strength between the tip and the substrate. Wavy arrows indicate Raman scattering by the molecules. (b) The Raman band of 4bipy ($1609{\mathrm{cm}}^{-1}$) is seen to change with the bias voltage. Inset: current versus bias curve for 4bipy obtained with a mechanical break-junction setup.⁷² (c) Schematic of tunneling-controlled TERS in a confocal-type side-illumination configuration, in which $V_{\mathrm{b}}$ is the sample bias and $I_{\mathrm{t}}$ is the tunneling current. (d) The top panels show experimental TERS mapping of a single molecule for different Raman peaks ($23\times 23$, $\sim 0.16\mathrm{nm}$ per pixel), processed from all individual TERS spectra acquired at each pixel (120 mV, 1 nA, 0.3 s; image size: $3.6\times 3.6{\mathrm{nm}}^2$). The middle panels show the theoretical simulation of the TERS mapping. The left bottom panel shows a height profile of a line trace in the inset STM topograph (1 V, 20 pA). The right bottom panel shows a TERS intensity profile of the same line trace for the inset Raman map associated with the $817{\mathrm{cm}}^{-1}$ Raman peak.⁷³ (e) The hollow conical taper and the taper with spiral corrugations along the conical surface (top panel). The simulated $E$-field intensity distributions in the $x\text{-}z$ plane for the two structures with $x$-polarized incident light at the wavelength of 800 nm (bottom panel). (f) The side-view SEM image of fabricated spiral taper (left panel) and the false-color image of focus pattern of gold-coated spiral taper under linearly polarized excitation with $x$-direction polarization (right panel).⁷⁴ Figures reprinted with permission: (a) and (b) Ref. 72, © 2011 by the Nature Publishing Group; (c) and (d) Ref. 73, © 2013 by the Nature Publishing Group; and (e) and (f) Ref. 74, © 2014 by Wiley-VCH Verlag.

However, experimental fabrication of these plasmonic sharp tips with extremely sharp apexes composed of several metal atoms to satisfy the requirement of ultrahigh electromagnetic enhancement for single-molecule SERS remains a challenge. As we have discussed in Sec. 2.1, the SERS sensitivity for a specific molecule does not only depend on the enhancement factors of SERS substrates but also on other experimental conditions. The conventional plasmonic sharp tips in TERS suffer from low transfer efficiencies of input light from light sources to hotspots of the sharp tips, which results in extremely high requirement of electric-field enhancement for single-molecule detection. To solve the above problem, Li and coworkers^74,75 proposed and experimentally demonstrated a plasmonic spiral tip for efficient transfer and focusing of input light. The high-transfer efficiency enabled by the spiral tip provides an additional pathway to increase the detection sensitivity toward single-molecule SERS. The theoretical design of the plasmonic spiral tip is shown in Fig. 2(e). By making use of the spiral design, the mirror symmetry of the plasmonic tip is broken, avoiding the destructive interference of the coupled surface plasmon polaritons at the apex of the tip. Based on the theoretical design, they first fabricated the template of the spiral tip using three-dimensional (3-D) direct laser writing, and then coated a uniform gold film with a thickness of 100 nm on the surface of the template via the magnetron sputtering technique. The SEM image of the fabricated plasmonic spiral tip is shown in the left panel of Fig. 2(f). The fabricated plasmonic spiral tip exhibits excellent focusing performance. A bright hot spot independent of the polarization of the input light is observed at the apex of the spiral tip as shown in the right panel of Fig. 2(f), indicating a high optical flux. Compared with the conventional plasmonic sharp tips with extreme pursuit of high electric-field enhancement, the plasmonic spiral tip opens up a new avenue toward single-molecule SERS.

3.1.3.

Plasmonic complex nanostructures

Plasmonic complex nanostructures based on new physics and novel nanofabrication techniques have recently emerged as a new strategy for single-molecule SERS. Compared with plasmonic simple nanostructures, such as the nanogaps and sharp tips introduced above, the plasmonic complex nanostructures enable versatile functions and extraordinary optical properties to improve the experimental conditions for single-molecule SERS. Improving the experimental conditions of SERS measurement, such as increasing transfer efficiency of input light to hotspots and collection efficiency of Raman signals, can be another promising alternative for enhancing the sensitivity of SERS measurement. Several classical plasmonic complex nanostructures that enable improved experimental conditions for SERS measurement will be introduced in this section.

In 2013, Lin et al. proposed and experimentally demonstrated a complex SERS substrate with the function of electrostatic precipitation for effective localized collection and identification of molecules.⁷⁶ The schematic diagram of the SERS substrate comprising of plasmonic complex nanostructures is shown in Fig. 3(a). By applying an external biased voltage to the substrate, the electrostatic precipitation can be introduced into the substrate to increase collection efficiency of target molecules. A charged photoresist layer with circular openings as shown in the inset SEM image of Fig. 3(a) was fabricated for localizing charged molecules to further increase the collection efficiency of the target molecules. Compared with the identical nanostructured SERS substrate without the function of electrostatic precipitation, the sensitivity of the proposed complex SERS substrate is improved by two orders of magnitude as shown in Fig. 3(b). The proposed complex SERS substrate with the function of electrostatic precipitation indicates that increasing the collection efficiency of molecules on the substrate is a promising method to enhance the sensitivity of SERS measurement. Such a strategy also increases adsorptivity of molecules on the SERS substrate, which may also be applied to single-molecule SERS to decrease the SERS intensity fluctuation and increase the reproducibility of SERS measurement.⁷⁶

Fig. 3

Plasmonic complex nanostructures for single-molecule SERS. (a) Concept of a complex SERS substrate with the function of electrostatic precipitation for effective localized collection and identification of molecules. The inset is the scanning electron microscope image of the SERS substrate. (b) Raman intensity on the biased substrates is increased by a factor of 615.⁷⁶ (c) Schematic diagram of the setup for nanoslit SERS. The nanoslit chip is sealed in a flow cell, which separates the electrolyte solution into two compartments. The top chamber can accommodate a water-immersion objective lens. A 785-nm laser with 8 mW is focused on the gold nanoslit. Axon patch 200B amplifier is used to apply the transmembrane voltages and monitor the ionic currents between two Ag/AgCl electrodes. The inset shows a top-view SEM image of the nanoslit structure, consisting of an inverted prism nanoslit cavity with Bragg-mirror gratings. The scale bar is $1\mu \mathrm{m}$. (d) The contour maps of SERS of a typical asynchronous blinking of the mixed isotopic adenines (top panel) and a typical fluctuation of the single adenine (bottom) recorded in 5 s, respectively, in single-molecule sensing.⁷⁷ (e) 3-D finite-difference time-domain simulations for a Ag nanoparticle on a warped Au substrate. A prominent field enhancement is demonstrated due to the effective gradient of permittivity. The right panel shows the SEM image of the substrate where the scale bar is 500 nm. (f) Typical spectra for bianalyte analysis.⁷⁸ Figures reprinted with permission: (a) and (b) Ref. 76, © 2013 by the Nature Publishing Group; (c) and (d) Ref. 77, © 2018 by the Nature Publishing Group; and (e) and (f) Ref. 78, © 2018 by the Nature Publishing Group.

Chen et al. utilized a plasmonic complex nanostructure consisting of a plasmonic nanoslit, a cavity, and two Bragg-mirror gratings to achieve single-molecule SERS.⁷⁷ The schematic diagram of the entire device is shown in Fig. 3(c). The cavity is able to efficiently couple the incident light into surface plasmon polaritons and guide them into the nanoslit, which provides an efficient transfer pathway for the incident light from the light source to the hotspot. The two Bragg-mirror gratings are used to reflect surface plasmon polaritons back to the nanoslit, further increasing the transfer efficiency. By making use of such a plasmonic complex nanostructure, extremely high electromagnetic enhancement is achieved in the nanoslit, which enables a high sensitivity up to the single-molecule level. In addition, a bias voltage is applied to the SERS substrate to attract negatively charged DNA molecules to pass through the nanoslit for SERS measurement. The bi-analyte approach was also used to experimentally verify the capability of the complex nanostructure for single-molecule detection, which is shown in Fig. 3(d). The proposed plasmonic complex nanostructure indicates that realization of high transfer efficiency of incident light from the far-field to the near-field can be a promising strategy for single-molecule SERS.⁷⁷

Also Mao et al.⁷⁸ utilized a plasmonic complex nanostructure designed by transformation optics to achieve broadband enhancement of hotspots in both magnitude and volume for single-molecule SERS. The schematic diagram of the proposed plasmonic complex nanostructure, that is, a silver nanoparticle on a warped gold substrate, is shown in the inset of Fig. 3(e). The warped substrate is designed using transformation optics, which achieves an immense enhancement to form a larger and brighter hotspot as shown in Fig. 3(e). Compared with a silver nanoparticle on a flat gold substrate, an additional threefold enhancement for electric field is realized, enabling a much higher sensitivity. By employing such a plasmonic complex nanostructure, a Raman enhancement factor up to $\sim {10}^9$ was experimentally realized, which is applicable for single-molecule detection. A bianalyte experiment was also conducted to verify the capability of the proposed plasmonic complex nanostructure for single-molecule detection, which is shown in Fig. 3(f). The proposed nanostructure indicates that developing SERS substrates based on new physics is another alternative for single-molecule SERS.⁷⁸

3.2.

Chemical Enhancement

Chemical enhancement as another commonly applied strategy for SERS is also applicable for single-molecule SERS. Compared with electromagnetic enhancement, chemical enhancement achieves a relatively low enhancement factor up to ${10}^7$.²⁵ Although single-molecule SERS enabled by chemical enhancement alone has not been demonstrated, chemical enhancement can definitely be employed as complementary to the electromagnetic enhancement for single-molecule SERS, or to relax the rigorous requirement on electromagnetic enhancement to fuel broad applications. Furthermore, chemical enhancement enables exploration of a variety of materials, such as semiconducting metal oxides and 2-D materials, as SERS substrates, thus providing more options for single-molecule SERS under various physical and chemical environments.

Chemical enhancement enhances Raman signal intensity by providing additional pathways of charge-transfer resonance.⁸⁰ A typical energy diagram of a substrate-molecule hybrid system consisting of a metal SERS substrate and an adsorbed molecule is shown in Fig. 4(a). The Raman intensity, $I_{\mathrm{R}}$, is proportional to the product of the intensity of incident light, $I_{\mathrm{L}}$, and the square of the Raman polarizability, ${\alpha }_{\sigma \rho }$, of the molecule:

Fig. 4

Chemical enhancement for single-molecule SERS. (a) A typical energy diagram of a substrate–molecule hybrid system consisting of a metal SERS substrate and an adsorbed molecule.⁸⁰ (b) Scanning electron microscopy image for ${\mathrm{W}}_{18}{\mathrm{O}}_{49}$ sample. Inset in (b): high-resolution transmission electron microscopy image on one single nanowire illustrating clear lattice fringe of 0.38 nm, which suggested that the nanowire growth was along the [010] direction. (c) Improved SERS properties of ${\mathrm{W}}_{18}{\mathrm{O}}_{49}$ samples after $\mathrm{Ar}/{\mathrm{H}}_2$ annealing treatment. Raman signals of R6G molecule on pristine ${\mathrm{W}}_{18}{\mathrm{O}}_{49}$ and the samples after annealing treatment (in $\mathrm{Ar}/{\mathrm{H}}_2$ for 1 h). The tested concentration of R6G was $1\times {10}^{-6}\mathrm{M}$. (d) A comparison of Raman EF for the two respective vibration modes P1 and P3. Data reported in this histogram resulted from Raman spectra acquired over 30 different regions per sample and provide an indication of the EF for each Raman mode. The ${\mathrm{H}}_2$-treated sample shows the greatest enhancement at band P1, the EF for which was evaluated to be $3.4\times {10}^5$.⁸ (e) Raman spectra collected for oxygen-incorporated ${\mathrm{MoS}}_2$ sample annealed for 40 min at four different concentrations, ${10}^{-4}$, ${10}^{-5}$, ${10}^{-6}$, and ${10}^{-7}\mathrm{M}$, suggesting the detection limit was as low as ${10}^{-7}\mathrm{M}$. (f) SERS spectra on the oxygen-incorporated ${\mathrm{MoS}}_2$ samples with different annealing temperatures. (g) and (h) SERS spectra of R6G on a series of other semiconductor materials ${\mathrm{WS}}_2$ and ${\mathrm{MoSe}}_2$ respectively.²³ Figures reprinted with permission: (a) and (b) Ref. 80, © 1986 by the American Institute of Physics; (c) and (d) Ref. 8, © 2015 by the Nature Publishing Group; (e)–(h) Ref. 23, © 2015 by the Nature Publishing Group.

Through chemical enhancement, a series of nonmetal substrates with high Raman enhancement capabilities were developed by material engineering. Cong et al. demonstrated a Raman enhancement factor of $3.4\times {10}^5$ with urchin-like ${\mathrm{W}}_{18}{\mathrm{O}}_{49}$ substrate [Fig. 4(b)], in which the oxygen vacancies contributed significantly to the chemical enhancement of Raman intensity.⁸ Specifically, as shown in Figs. 4(c) and 4(d), the additional oxygen vacancies in ${\mathrm{W}}_{18}{\mathrm{O}}_{49}$ created with $\mathrm{Ar}/{\mathrm{H}}_2$ annealing treatment enriched the surface states of the material, which provided magnified affinity for the adsorbent–adsorbate interaction and enhanced the Raman signal through charge-transfer resonance. The highest enhancement factor of $3.4\times {10}^5$ was experimentally realized by utilizing oxygen vacancies. On the contrary, by incorporating oxygen into ${\mathrm{MoS}}_2$ substrate, Cong et al. also achieved high chemical enhancement with a SERS detection limit of rhodamine 6G molecules down to the concentration of ${10}^{-7}\mathrm{M}$ ($1\mathrm{M}=1\mathrm{m}\mathrm{o}\mathrm{l}/\mathrm{L}$) [Figs. 4(e) and 4(f)].²³ Some other substrates, including ${\mathrm{WS}}_2$ and ${\mathrm{MoSe}}_2$ [Figs. 4(g) and 4(h)], were also successfully equipped with chemical enhancement for SERS by the oxygen incorporation approach. The chemical enhancement significantly broadens the potential material platform for SERS, which is expected to promote single-molecule SERS by combination with electromagnetic enhancement.

3.3.

Resonance Enhancement

Resonance enhancement is a classic and well-studied strategy for enhancing the weak Raman scattering signals. Apart from its wide employment for measuring the spontaneous Raman spectra of various molecules in solid state or liquid solutions, attempts have been made to utilize resonance enhancement in SERS measurement. To date, the Raman enhancement factors contributed by resonance enhancement have been found to be between ${10}^4$ and ${10}^6$, which are unable to facilitate single-molecule SERS alone but still provide considerable enhancement effect. More importantly, achieving resonance enhancement is readily feasible regardless of the substrate since it directly results from the interaction between the excitation photon and the probed molecule. Therefore, similar to chemical enhancement, resonance enhancement can be used in parallel with other strategies to further amplify the Raman signals.

The mechanism of resonance enhancement is relatively simple and straightforward compared with the strategies above, as the substrate is not involved. Typically, resonance enhancement occurs when the energy of the incident photon is close to the energy required for electronic transition between states of $|\mathit{K}⟩$ and $|\mathrm{I}⟩$, in the probed molecule as shown in Fig. 4(a). Under the resonance condition, the term $A$ in Eq. (5) is drastically enhanced through the resonance denominator, ${\omega }_{KI}^2-{\omega }^2$, when $\omega$ is close to ${\omega }_{\mathit{KI}}$. Similar to the chemical enhancement, the molecular Raman polarizability, ${\alpha }_{\sigma \rho }$, is also enhanced, leading to the enhancement of the Raman intensity $I_{\mathrm{R}}$.

Resonance enhancement has long been employed for SERS by virtue of its great convenience. Hildebrandt et al. studied the surface-enhanced resonance Raman scattering (SERRS) of rhodamine 6G adsorbed on colloidal silver.⁸¹ By virtue of the high detection sensitivity enabled by resonance enhancement, they were able to distinguish the two different kinds of adsorption sites on the surface of the silver particles with different enhancement mechanisms. Hildebrandt et al. also investigated the SERRS of cytochrome c (cyt) at room and low temperatures.⁸² From the SERRS spectra of ${\mathrm{cyt}}^{3+}$ adsorbed on Ag glycerol sol at different temperatures, they demonstrated the adsorption-induced partial transition of ${\mathrm{cyt}}^{3+}$ molecules from the low-spin (LS) state to the high-spin (HS) state. As one of the important Raman enhancement strategies, resonance enhancement offers alternative or complementary chances for realizing single-molecule SERS.

3.4.

Other Potential Strategies

Besides the above mature strategies, new strategies need to be continuously proposed and developed to promote the advance of single-molecule SERS. Potential new strategies may come from new understanding and interpretation of single-molecule SERS, as the mechanism of single-molecule SERS is still not completely understood. For example, as we have introduced in Sec. 3.1.2, the state-of-the-art TERS enables not only single-molecule sensitivity but also subnanometer spatial resolution.^83–89 However, numerical simulation based on the conventional SERS theory indicates that the hotspot size generated by the plasmonic sharp tip used in the experiment is still in the order of 10 nm, as shown in Fig. 5(a), which is much larger than the experimentally realized spatial resolution.⁹⁰ How a 10-nm hotspot enables subnanometer spatial resolution has aroused great interest and discussion.

Fig. 5

Potential strategy for single-molecule SERS enabled by Rayleigh scattering-assisted Raman enhancement theory. Schematic configuration of TERS system used for Raman mapping of molecule (a) with and (b) without molecule selfinteraction. (c) The excitation process in classical Raman physics, where the incident light is scattered by the Ag tip–substrate nanogap to form highly localized plasmonic gap mode with greatly enhanced electric field intensity. $E_0$ and $E_s$ (yellow arrows) are the incident and scattering electric fields, summing up to the local field $E$. (d) The radiation process in classical Raman physics, where the Raman signal emitted by the molecule is scattered by the Ag–substrate gap to have greatly enhanced far-field intensity. $G_0·\mu$ and $G_s·\mu$ (blue arrows) are the direct Raman radiation of molecule (described by the dipole moment $\mu$) and greatly enhanced scattering Raman radiation by the tip–substrate nanogap, where $G_0$ and $G_s$ are free-space and scattering dyadic Green’s function. (e) The excitation process in the extended Raman physics considering the molecule selfinteraction with the Ag tip–substrate gap through multiple elastic scattering (white arrows, abbreviated by $E.S$.) for the incident light. $E_{\mathit{m},\mathit{s}}$ is the modified excitation field due to the $E.S$. mechanism and $p$ is the induced dipole moment. (f) The radiation process in the extended Raman physics when considering the molecule selfinteraction. The multiple $E.S$. mechanism (white arrows) strongly modifies the molecule dipole moment to a value ${\mu }_{N,R}$.⁹⁰ (a)–(f) Figures reprinted with permission from  Ref. 90, © 1986 by the American Chemical Society.

To interpret this puzzling question, Li and coworkers theoretically investigated it and found that Rayleigh scattering of the molecule that is previously ignored in conventional SERS theory plays an unexpectedly critical role, which may be one of the main contributors to the subnanometer spatial resolution.⁹⁰ With the assist of multiple Rayleigh scattering, a super hotspot with a shrunken size as shown in Fig. 5(b) is generated for achieving a high spatial resolution. Based on this discovery, they proposed an extended Raman theory to amend the conventional Raman theory. In conventional Raman theory, the influence of the molecule on the hotspot excited by the incident light is totally ignored, as shown in Figs. 5(c) and 5(d). In the extended Raman theory, as shown in Figs. 5(e) and 5(f), they took the influence of the molecule on its surrounding electromagnetic background into consideration. When the molecule size is comparable with the nanogap size and the distance to the plasmonic sharp tip, its Rayleigh scattering within the nanogap will respond and modify the localized electric field of both the excitation light and the Stokes (anti-Stokes) radiation light. This response and modification is called the electromagnetic selfinteraction process of the molecule, which is characterized through Raman polarizability of the molecule and the induced dipole moment. The effective localized electric field of the incident light that includes the influence of the Rayleigh scattering of the molecule will be determined selfconsistently as

4.1.

Process Monitoring of Chemical Catalysis

The central issue of catalytic chemistry is to understand how single molecules interact with each other, which provides us with the possibility to precisely design and control various chemical reactions that are critical to our lives.^91–95 Elucidating the molecular mechanisms of chemical reactions requires the observation of single-molecule behavior at subnanometer scale, which is beyond the detection limit of most analytical tools. Fortunately, by virtue of its extremely high sensitivity, single-molecule SERS enables the real-time tracking of molecules undergoing chemical reactions. In addition, the surface plasmons in subnanometer scale cavities generate hot charge carriers that can catalyze chemical reactions or induce redox processes of molecules located at the hotspots.²⁹

Using single-molecule SERS, Zhang et al. studied the concentration-dependent intermolecular and intramolecular reactions catalyzed by surface plasmons on gold dimers.²⁹ By analyzing the single-molecule SERS spectra of $p$-nitrothiophenol ($p\mathrm{NTP}$) molecules, they found that $p\mathrm{NTP}$ molecules tended to dimerize into 4,4-dimercaptoazobenzene (DMAB) at higher concentrations [Figs. 6(a) and 6(b)] while $p\mathrm{NTP}$ exclusively reacted to thiophenol (TP) at or close to the single-molecule level [Fig. 6(c)] where intermolecular reactions could not occur. Nijs et al. employed single-molecule SERS to study the hot-electron-induced redox processes in single aromatic molecules that are located at the nanocavity with a nanoparticle-on-mirror configuration, as shown in Fig. 6(d).³⁰ Specifically, they observed the change in SERS spectra of a single molecule during a time period of 0.9 s, as shown in Fig. 6(e), showing the occurrence of reduction and oxidation processes at the single-molecule level. In addition to reactive molecules, single-molecule SERS can also be applied to study the catalytic activity of nanoparticles. Recently, Zhang et al. demonstrated the direct SERS tracking of the reduction of a 4-nitrothiophenol (4-NTP) catalyzed by a single 13-nm small gold nanoparticle (S-GNP) in aqueous solution.³¹ Based on a configuration composed of a large gold nanoparticle (L-GNP) and an S-GNP [Fig. 6(f)], the time-resolved single-molecule SERS spectra [Fig. 6(g)] were recorded to demonstrate the borohydride addition during the reaction process. It can be seen from the above examples that single-molecule SERS has become a powerful tool for process monitoring of chemical catalysis at the single-molecule level.

Fig. 6

Process monitoring of chemical catalysis by single-molecule SERS. (a) SERS spectra of ${10}^{-7}\mathrm{M}$ pNTP excited at 3, 15, and 30 mW laser power, respectively. (b) SERS spectra of ${10}^{-8}\mathrm{M}$ pNTP excited at 0.015, 0.15, and 1.5 mW laser power, respectively. (c) From top to bottom: Time-dependent SERS spectra of reacting pNTP ($c={10}^{-9}\mathrm{M}$) at 3 mW laser at 2, 6, and 30 min; normal SERS spectra of TP and $p\mathrm{NTP}$, respectively.²⁹ (d) Selfassembled 80-nm nanoparticle-on-mirror geometry used to elicit field enhancements required for single-molecule spectroscopy. (e) Short (0.9 s) segment from NPoM time scan showing the reduction and oxidation processes.³⁰ (f) Schematic diagram showing the configuration of an L-GNP/S-GNP dimer linked with 1,6-hexanedithiol. (g) (top) Schematic illustration showing the S-GNP catalyzed reduction of 4-NTP in the presence of borohydride. (bottom) Color-coded intensity map of time-dependent single-NP SERS spectra after borohydride addition, with a range of Raman shifts between 900 and $1800{\mathrm{cm}}^{-1}$ for a 1 s integration time, taken every 5 s at 638 nm.³¹ Figures reprinted with permission: (a)–(c) Ref. 29, © 2015 by the Royal Society of Chemistry; (d) and (e) Ref. 30, © 2017 by the Nature Publishing Group; and (f) and (g) Ref. 31, © 2019 by the Royal Society of Chemistry.

4.2.

Imaging of Vibrational Modes

Visualizing single molecules provides the most straightforward answers to a lot of chemical and biological questions with unprecedented detail. Atomic force microscopy (AFM) is commonly used to image the morphology of single molecules but fails to recognize chemical information of single molecules. In contrast, single-molecule SERS probes the vibrational modes of single molecules, which contain rich information of their chemical bonds and functional groups. With its chemical specificity, single-molecule SERS opens up a new window for high-content vibrational imaging of single molecules.

In 2015, by detecting vibrational fingerprints with an STM-controlled TERS, as shown in Fig. 7(a), Jiang et al. successfully distinguished two adjacent molecules with similar morphologies, zinc-5,10,15,20-tetraphenyl-porphyrin (ZnTPP) and meso-tetrakis (3,5-di-tertiarybutyl-phenyl)-porphyrin (H₂TBPP), which are indistinguishable with AFM.³² The measured Raman spectra of the two molecules to be distinguished, as shown in Fig. 7(b), exhibit obvious distinction. To further enhance the signal-to-noise ratio of the measured single-molecule Raman spectra for subnanometer-resolved vibrational imaging, Jiang et al. utilized multivariate analysis based on STM images and TERS images [Figs. 7(c)–7(g)] to achieve identification of single ZnTPP and ${\mathrm{H}}_2\mathrm{TBPP}$ molecules.³³ In addition, instead of directly detecting TERS signals of the probed molecules for imaging, electrostatic fields of single molecules can also be mapped with an intermediate single-molecule probe. With TERS-relayed molecular force microscopy (TERS-mfm), Lee et al. successfully imaged the distribution of electrostatic fields of single molecules.³⁴ As shown in Fig. 7(h), TERS was utilized to measure the vibrational modes of a tip-attached single CO molecule, which was perturbed by the electrostatic fields of the probed molecule. The reconstructed distributions of electrostatic fields of cobalt(II)-tetraphenylporphyrin (CoTPP) and zinc(II)-etioporphyrin (ZnEtio) molecules on Au(111) substrate are shown in Figs. 7(i) and 7(j). By virtue of its high flexibility and submolecular level spatial resolution, single-molecule SERS is a promising technique for various high-content chemical imaging.

Fig. 7

Imaging of vibrational modes by single-molecule SERS. (a) Schematic of STM-controlled TERS in a confocal-type side-illumination configuration on Ag (111) using Ag tips. (b) TERS spectra acquired above ZnTPP or ${\mathrm{H}}_2\mathrm{TBPP}$ molecular islands (0.1 V, 1 nA, 30 s). For comparison, powder Raman spectra and Raman spectra calculated via DFT are also shown. The brown spectrum was taken on bare Ag (111) (0.1 V, 1 nA, 30 s) and the black spectrum was measured on top of a molecular island but with the tip retracted 5 nm from the surface (0.1 V, 30 s).³² (c), (d) TERS images reconstructed based on single-peak analysis for the Raman peaks at $\sim 701{\mathrm{cm}}^{-1}$ (c, integrated over 687 to $736{\mathrm{cm}}^{-1}$) and $\sim 918{\mathrm{cm}}^{-1}$ (d, integrated over 890 to $959{\mathrm{cm}}^{-1}$). (e) STM image of two adjacent porphyrin molecular domains ($-1\mathrm{V}$, 5 pA). (f) STM image simultaneously acquired during TERS imaging of the area denoted by the dashed square in (e) (–0.1 V, 1 nA, $7\text{\hspace{0.17em}nm}\times 7\text{\hspace{0.17em}nm}$, $32\times 32$pixels, 1 s per pixel). The boundary between the molecular domains is highlighted by a yellow line. (g) TERS spectra, averaged over the blue and green squares ($3\times 3$pixels) shown in (f), extracted from the datacube for ZnTPP and ${\mathrm{H}}_2\mathrm{TBPP}$ molecules, respectively.³³ (h) Diagram of the system to detect the distribution of the electrostatic fields of single molecules. (i) Electrostatic field surface obtained by color-coding the Stark on the STM CH topography. The common image size is $23⏧\times 23⏧$. The set point is 0.1 nA, 1.2 V. (j) Electrostatic field mapped on the isosurface of local density of states. The common image size is $27⏧\times 27⏧$. The set point is 0.1 nA, 1.2 V.³⁴ Figures reprinted with permission: (a) and (b) Ref. 32, © 2015 by the Nature Publishing Group; (c)–(g) Ref. 33, © 2017 by the Nature Publishing Group; (h)–(j) Ref. 34, © 2018 by the American Association for the Advancement of Science.

In addition, imaging vibrational modes of molecules can retrospectively help interpret the nature of SERS substrates. To elucidate the time-dependent signal fluctuation in single-molecule SERS, Lindquist et al. studied the hotspot dynamics on single nanoparticles by high-speed single-molecule SERS imaging.⁴⁰ Surprisingly, the surface of a fully functionalized ${\mathrm{SiO}}_2@\mathrm{Ag}$ nanoparticle remains dark more than 98% during the measurement, suggesting that the hotspots are formed transiently due to the random reconstruction of the metal surface. By further analyzing the influence of several factors, such as temperature and excitation wavelength, the authors concluded that the signal fluctuation mainly results from the transient formation of hotspots driven by the random reconstruction of the metal surface. The work provides a promising method to investigate the fast dynamic properties of single-molecule SERS hotspots.

4.3.

Observation of Charge Transfer in Nanoelectronics

Nanoelectronic devices are of essential importance to advanced electronic products, such as central processing units, electronic memory storages, and optoelectronic devices. Molecular nanoelectronics that utilize single molecules as active components are capable of realizing information systems with extremely high performance and low power consumption. Studying molecular charge transfer provides insights into the design and fabrication of molecular nanoelectronic devices. Similar to the difficulties in investigating single molecules, conventional methods are usually unable to probe such subtle and dynamic phenomena due to their limited sensitivities. Single-molecule SERS, which is highly sensitive to the variations in molecular charge transfer processes, has great potential to fulfill the requirements of characterizing molecular nanoelectronic devices.

Li et al. studied the voltage-driven tuning of vibrational mode energies of single molecules.³⁵ By applying different bias voltages to the ${\mathrm{C}}_{60}$ molecules in gold junctions [Fig. 8(a)], they observed shifts in molecular vibrational modes, which presumably resulted from the addition of electronic charge to the molecule, instead of simple Stark effect, through the analysis of single-molecule SERS spectra [Figs. 8(b) and 8(c)]. In contrast, Han et al. used single-molecule SERS to directly investigate the mechanisms of charge transfer.³⁶ For the first time, they achieved the assessment of the contribution of SPR to charge transfer by tuning the surface plasmon absorption of Au nanorods with a series of excitation wavelengths [Figs. 8(d) and 8(e)]. They also proposed the physical mechanisms of plasmon-assisted charge transfer in such semiconductor/organic molecule/Au assemblies [Fig. 8(f)]. Therefore, single-molecule SERS is highly expected to lead to fundamental breakthroughs in the field of nanoelectronics as it allows for the direct interrogation of charge transfer processes.

Fig. 8

Observation of charge transfer in nanoelectronics by single-molecule SERS. (a) Raman analysis of ${\mathrm{C}}_{60}$ in an electromigrated junction. (Main plot) Example of a SERS spectrum of ${\mathrm{C}}_{60}$ in an electromigrated junction. Surrounding diagrams illustrate examples of the complicated displacements associated with Raman-active modes, calculated for an isolated, symmetric, gas-phase molecule. Each such mode is fivefold degenerate in the absence of symmetry breaking. (Inset) Scanning electron microscopic image of an electromigrated junction. (b) Raman response of device 3 as a function of bias ($x$ axis) and Raman shift ($y$ axis). The sudden change in the intensity at around 0.1 V is the result of blinking. (c) Vibrational energy shift as a function of bias for three particular modes: 1258 and $1592{\mathrm{cm}}^{-1}$. The discretized Raman shift data result from pixilation of the detector.³⁵ (d), (e) SERS spectra of 4-MBA in the different ${\mathrm{TiO}}_2$–MBA–Au NR assemblies at excitation wavelengths of (d) 633 nm and (e) 785 nm. The numbers on the left side in each figure indicate the SP absorption peaks of the ${\mathrm{TiO}}_2$–MBA–Au NR assemblies. (f) Main physical mechanisms involved in plasmon-assisted charge transfer.³⁶ Figures reprinted with permission: (a)–(c) Ref. 35, © 2014 by the National Academy of Sciences; (d)–(f) Ref. 36, © 2018 by the Royal Society of Chemistry.

4.4.

Combination with Other Techniques

Besides the stand-alone application of single-molecule SERS, single-molecule SERS has also been combined with some other techniques to create new tools for studying various phenomena at the single-molecule level. On the one hand, the combined techniques may either compensate for the intrinsic drawbacks of single-molecule SERS or provide tunable circumstances for single-molecule SERS measurement, thus enabling systematic investigation of single-molecule events. On the other hand, the single-molecule SERS can equip the combined techniques with ultrahigh sensitivity and high-content information of molecular vibration. For example, single-molecule SERS can be combined with electrochemical or mechanical techniques to monitor single-molecule chemical reactions. In this section, we will provide several examples of combination of single-molecule SERS with other techniques and discuss their potential applications.

In 2010, by integrating an electrochemical setup with single-molecule SERS [Fig. 9(a)], Cortes et al. tracked the redox processes of single molecules in a time-dependent manner and compared the SERS spectra of single- and many-molecule events.⁹⁶ Specifically, using the bianalyte method, they collected the SERS signals of rhodamine 6G (RH6G) and nile blue (NB) in an open-frame electrochemical cell. Using SERS intensity as a tag, they demonstrated that the single-molecule events differ from one molecule to another at the same position [Fig. 9(b)]. Also, as shown in Fig. 9(c), the behaviors of one molecule vary between two successive cycles. In addition, Zheng et al. combined mechanically controllable break junction (MCBJ) with single-molecule SERS to study the conductance discrepancy of benzene-1,4-dithiol (BDT).⁹⁷ As shown in Fig. 9(d), the single-molecule SERS and electric conductance of BDT molecules are measured in an MCBJ setup. By comparing the SERS spectra of BDT under different conditions [Fig. 9(e)], they clarified that the dimerization of BDT was responsible for the low conductance status before the junction break [Figs. 9(f) and 9(g)]. Besides, single-molecule SERS itself can also be improved by combination with other techniques. For example, Patra et al. demonstrated the reversible assembly of plasmonic nanoparticles for single-molecule SERS, which overcame the limitations of irreversible chemical aggregation.⁹⁸ Notably, as shown in Fig. 9(h), they applied a single evanescent-wave excitation beam to achieve the nanoparticle assembly and single-molecule SERS simultaneously. They also created interacting nanoparticle assemblies using two excitation beams [Fig. 9(i)]. To summarize, the combination of single-molecule SERS with other techniques significantly expands the application of conventional single-molecule SERS, which is expected to contribute to diverse applications, such as study of protein–protein interaction and design of molecular motors.

Fig. 9

Combination of single-molecule SERS with other techniques. (a) Ag colloids premixed with a suitable combination of bianalyte SERS partners (RH6G and NB at 2- to 5-nM concentration) are deposited on a working electrode (Ag) in an open-frame electrochemical cell (with a wide-area Pt counter electrode and a Ag/AgCl reference electrode). (b) Many molecule and SM-SERS intensity cases for NB. The varied durations of the signals in the “on” (oxidized) state reveal the different redox potentials for that particular molecule. (c) Variations in redox properties for two consecutive cycles, as determined from the SERS intensity of the $590{\mathrm{cm}}^{-1}$ peak of NB along two voltammetric cycles: single molecule; many molecules.⁹⁶ (d) Schematic of the MCBJ-SERS setup and the molecular structures of BDT and dimeric-BDT. (e) SERS spectra collected when the molecular junction was mechanically controlled at the regimes of high conductance (red), low conductance (green), and breakage (gray), respectively. An ordinary Raman spectrum of BDT powder (brown) is displayed for comparison. Laser excitation: 785 nm. (f) The distributions of the conductance deviations from the master curve within the high conductance (red) and low conductance (green) regimes. Bin size: 0.02. (g) The hypothesized evolution of the microscopic configuration as the conductance evolves from high conductance to low conductance.⁹⁷ (h) Schematic illustration of the simultaneous plasmonic assembly of nanoparticles and SERS. Black arrows indicate the direction of 532 nm laser (green beam) used for SPP excitation of the metal film deposited over glass coverslip. Gray spheres are the metal nanoparticles dispersed in the water medium and red dots indicate the SERS-active molecules. Inset on the right shows the TEM image of Ag core–Au shell nanoparticle used in this experiment. Scale bar is 20 nm. (i) Schematic representation of dual assembly of Ag nanoparticles on Au film coupled to a prism: green lines are the two laser beams of 532 nm introduced at two opposite ends of the prism. Black arrows indicate the direction of the laser beams.⁹⁸ Figures reprinted with permission: (a)–(c) Ref. 96, © 2010 by the American Chemical Society; (d)–(g) Ref. 97, © 2018 by the Royal Society of Chemistry; and (h) and (i) Ref. 98, © 2014 by the Nature Publishing Group.